RURAL SETTLEMENT AND ECONOMIC ACTIVITY: OLIVE OIL AND ...
369
RURAL SETTLEMENT AND ECONOMIC ACTIVITY: OLIVE OIL AND AMPHORAE PRODUCTION ON THE TARHUNA PLATEAU DURING THE ROMAN PERIOD Thesis submitted for the degree of Doctor of Philosophy at the University of Leicester by Mftah A. M. Ahmed (DoA Tripoli, Libya) School of Archaeology and Ancient History University of Leicester March 2010
Transcript of RURAL SETTLEMENT AND ECONOMIC ACTIVITY: OLIVE OIL AND ...
RURAL SETTLEMENT AND ECONOMIC ACTIVITY: OLIVE OIL
AND AMPHORAE PRODUCTION ON THE TARHUNA PLATEAU
DURING THE ROMAN PERIOD
Thesis submitted for the degree of
Doctor of Philosophy
at the University of Leicester
by
Mftah A. M. Ahmed (DoA Tripoli, Libya)
School of Archaeology and Ancient History
University of Leicester
March 2010
ii
Contents
List of figures and tables......................................................................................... v
Abstract .................................................................................................................. x
Acknowledgment ................................................................................................... xi
Chapter 1: The Gebel Tarhuna .......................................................................... 13
1.1 Introduction ..................................................................................................... 13
1.2 Geographical and climate conditions of the study area .................................. 15
1.3 The agricultural importance of the Tarhuna plateau ...................................... 18
1.4 The Tarhuna plateau landscape archaeology ................................................. 26
1.5 Background of the study .................................................................................. 28
Chapter 2: The Tarhuna Archaeological Survey .............................................. 32
2.1 Introduction ..................................................................................................... 32
2.2 Aims of the survey ........................................................................................... 33
2.3 Methodology .................................................................................................... 35
2.3.1 Intensive survey method ............................................................................... 39
2.3.2 Extensive survey method ............................................................................. 44
2.3.3 The application of GIS .................................................................................. 45
2.4 Wadis Turgut and Doga as a key study ........................................................... 46
2.5 Distribution of sites in the landscape .............................................................. 55
2.5.1 Site elevation ................................................................................................. 56
2.5.2 Distribution patterns of site types ................................................................. 59
2.6 Site typology .................................................................................................... 61
2.6.1 Villages (small towns) .................................................................................. 61
2.6.2 Oilery farms .................................................................................................. 65
2.6.3 Large farms .................................................................................................. 67
2.6.4 Small farms ................................................................................................... 68
2.6.5 Fortified farmhouses ..................................................................................... 70
2.6.6. Dams, Cisterns and Wells ............................................................................ 78
Chapter 3: Ancient rural settlement on the Tarhuna plateau ......................... 86
3.1. Introduction .................................................................................................... 86
3.2 Settlement type and organisation ..................................................................... 91
3.2.1 Site size ......................................................................................................... 91
3.2.2 Site type description ...................................................................................... 101
iii
3.2.3 Term of villa ................................................................................................. 102
3.2.4 The agricultural villages ............................................................................... 106
3.2.5 Oileries and large farms ................................................................................ 116
3.2.6 Associated mausolea and other tombs .......................................................... 126
3.2.7 Layout of the presses .................................................................................... 128
3.2.8 Small farms ................................................................................................... 134
3.3. Settlement construction and organisation ..................................................... 138
3.3.1. Farming sites ................................................................................................ 142
3.3.2. Rural baths ................................................................................................... 149
3.4. Settlement density and diversity ..................................................................... 154
3.5. Evaluation of settlement pattern over time ..................................................... 164
Chapter 4: Olive oil pressing facilities and pressing process ........................... 174
4.1 Introduction ..................................................................................................... 174
4.2 Type of the press on the Tarhuna plateau ........................................................ 178
4.3 Production process .......................................................................................... 180
4.3.1. Mills ............................................................................................................. 180
4.3.2. Arbores ......................................................................................................... 189
4.3.3. The press floors (arae) ................................................................................. 199
4.3.4 Counterweight blocks ................................................................................... 213
4.3.5 Tanks and vats .............................................................................................. 219
4.4 Standradisation ................................................................................................ 222
4.5. Capacity production of the Gebel Tarhuna olive presses ............................... 224
4.5.1 Calculation of capacity production ............................................................... 226
4.5.2 The production capacity of traditional lever presses in the Mesallata
region during the late Ottoman period ................................................................... 235
Chapter 5: Amphora production sites on the Tarhuna plateau .................... 248
5.1. Introduction .................................................................................................... 248
5.2. Distribution pattern of amphora kiln sites on the Tarhuna plateau ............... 250
5.3. Types of amphorae produced by the Tarhuna plateau kilns ........................... 261
5.3.1 Tripolitanian amphora I ................................................................................ 262
5.3.2 Tripolitanian amphora II ............................................................................... 264
5.3.3 Tripolitanian amphora III .............................................................................. 266
5.4. Construction of amphora kilns on the Tarhuna plateau ................................. 257
iv
5.4.1 The excavation of Arbaia pottery kiln (TUT48) ........................................... 269
5.5. The Tarhuna plateau amphora stamps ........................................................... 273
Chapter 6: Conclusion ......................................................................................... 288
6.1 Overview ……………………………………………………………………………... 288
6.2 Economic aspects of archaeological sites ....................................................... 286
Bibliography ................................................................................................... ...... 308
Appendix ............................................................................................................... 321
v
List of figures and tables Figure 1.1: Location of the Tarhuna region 16
Figure 1.2: Location of the Tarhuna plateau in Tripolitania. 17
Figure 1.3: The modern arable areas on the Gebel Tarhuna 21Table 1.1: Mean monthly and mean annual precipitation (mm) from the Tarhuna town meteorological station (1925 – 1978 18
Figure 2.1: An example of using the Google Earth to identify sites (DOG69). 32Figure 2.2: Location of the Tarhuna plateau and the surveyed area. 38Figure 2.3: The three areas intensively surveyed by the TAS 2007 40Figure 2.4: The first intensively surveyed area, the ‘Hajaj’ area (the Wadi Doga). 41Figure 2.5: The gasr Beni Mousa and its surrounding archaeological features. 42Figure 2.6: The Gasr Doga mausoleum 49Figure 2.7: The surveyed area of the Wadis Turgut and Doga 51Figure 2.8: A new boundary inscription fits within the Di Vita Evrard projected by line. 53Figure 2.9: Site location in relation to the topography. 57Figure 2.10: Site location and elevation the north‐eastern sector of the Gebel Tarhuna. 59
Figure 2.11: The Gasr Ed‐Dauun village 62Figure 2.12: Location of the Medina Doga 63Figure 2.13: Plan of Medina Doga 64Figure 2.14: Distribution of oilery sites in Wadis Turgut and Doga 66Figure 2.15: Distribution of large farms in Wadis Turgut and Doga. 68Figure 2.16: Distribution of the recorded small farms on the eastern Tarhuna plateau. 69
Figure 2.17: The middle sector of the Wadi Turgut 70Figure 2.18: Some of the fortified farms identified on Google Earth imagery of the Gebel Tarhuna 71Figure 2.19: Distribution of defended farms recorded from the high resolution imagery of the Google Earth. 72Figure 2.20: The Gasr Shāeir 73Figure 2.21: Location of 11 open farms replaced by fortified ones in the late Roman period. 74Figure 2.22: An example from the middle of Turgut shows how some of the hill‐top gsur built close to the wadi bed 75Figure 2.23: Chart showing the location of Type 1 (open/fortified sites 77Figure 2.24: Location of the Type 2 fortified farmhouses 77Figure 2.25: The distance between hill‐top gsur and other farming site 78Figure 2.26: The oilery farm of Loud Meghara (TUT43) 81Figure 2.27: Dams in the Wadi Turgut. 82Figure 2.28: A dam in the Wadi Turgut, (TUT24). 83Figure 2.29: DOG 111, large farm associated with cisterns, dam, kiln and Tank 84Table 2.1: Site location in relation to the elevation level 58Figure 3.1: Measuring site (TUT54) by the Google Earth 95Figure 3.2: The chart shows bands of 100 sites of known size which are located in the Wadis Turgut and Doga. 95Figure 3.3: Distribution of oileries and large farms in the Wadis Turgut and Doga. 96Figure 3.4: Senam al‐Halafi 1. 99Figure 3.5: Complex site of Senam Halafi 1. 100Figure 3.6: Senam Halafi 1; line of columns at eastern side. 100Figure 3.7: The agricultural villages and small towns in eastern section of the Tarhuna plateau. 108Figure 3.8: Ain Astail agricultural village. 109
vi
Figure 3.9: The Gasr Dehmesh village 111 Figure 3.10: Gasr Dehmesh occupies a top of small hill nearby a large farm building to the west 112Figure 3.11: A majen (cistern) in the Gasr Dehmesh village 112Figure 3.13: The early 1st c. AD fineware collected from a mausoleum in the vicinity of the Gasr Dehmesh village. 113Figure 3.14: Location of Halafi village. 114Figure 3.15: Halafi village 115Figure 3.16: Senam Aref (DOG 60). 116Figure 3.17: Farms and presses in Wadi Turgur and Doga. 118Figure 3. 18: The recorded oilery sites in the Gebel Tarhuna 119Figure 3.19, a) Distribution of oilery‐villas in the Wadis Turgut and Doga. 121Figure 3.19,b) The oilery‐villas in the Wadis Turgut and Doga and the luxury villas in the area between Lepcis Magna and Oea 125Figure 3.20: A Google Earth image shows the distribution of oilery‐villas in the wadis Turgut and Doga and measuring the distance between site no. 54 (Senam Semana) and the Mediterranean Sea 125Figure 3.21: Locations of the eight mausolea that have been recorded in the Gebel Tarhuna 126Figure 3.22: Corner Corinthian capitals from es‐Sonama mausoleum 127Figure 3.23: b) Plan of Senam Semana (TUT54). 130Figure 3.24: Plan of Henscir el‐Begar, Tunisia 131Figure 3.25: Orthostats and columns with trapezium capitals at Senam Semana (TUT54). 131Figure 3.26: Plan of the oilery‐villa of Henscir es‐Senam (TUT38). 132Figure 3.27: Sidi Eysawi, a large farm‐villa with pottery kiln. 133Figure 3.28: A base of column at Sidi Eysaw 133Figure 3.29: Comparative plan of some small farms (the upper plans from Oates 1953‐4). 136Figure 3.30: Location of an oilery‐villa with the nearby two small farms (the Google Earth background). 137
Figure 3.31: a) Use of opus quadratum and an arched gate at Sidi Madi (TUT52). 143 Figure 3.31: b) Fine ashlar masonry at Sidi Madi (TUT52). 143 Figure 3.32: Use of the opus quadratum at Senam Aref (DOG60 144 Figure 3.33: Some architectural elements left in a quarry close to TUT45 144Figure 3.34: Symbols (probably New‐Punic letters) mark limestone blocks at the large farm‐villa at Sidi Eysawi (TUT53). 145Figure 3.35: a) Senam Aref (DOG60). 146Figure 3.35: b) Sidi al‐Akhder (DOG66). 146Figure 3.36: TUT3. 147Figure 3.37: Distribution of rural baths in the Gebel Tarhuna. 149Figure 3.38: A bath‐house at Bir Twafga. 150Figure 3.39: A general view of Ain Guman villa/bath showing the opus africanum structure of its eastern wall. 150Figure 3.40: Mosaic (a) and tile, bronze pipe (b) at the Ain Guman villa/bath. 151 Figure 3.41: Location of villa/bath and the dam at Ain Guman. 152Figure 3.42: Distribution of the rural sites in the Wadis Turgut and Doga 157Figure 3.43: Density of rural sites in the middle of the Wadi Turgut 158Figure 3.44: The complex site of Henscir Assalha (TUT15) in the Wadi Turgut. 159Figure 3.45: Pie charts showing the settlement diversity and their percentages in the Wadis Turgut and Doga. 160Figure 3.46: The density of sites in the intensively surveyed area of the upper Wadi Guman 162Figure 3.47: A sample of the hoard of Numidia coins found in the upper Wadi Guman. 163
vii
Figure 3.48: The measurement of territory of Lepcis 165Figure 3.49: The number of rural sites have recorded/rerecorded by the TAS and their dates. 168Table 3.1: Oilery sites in the Wadis Turgut / Doga and their sizes. 97Table 3.2: Large farm sites in the Wadis Turgut / Doga and their sizes. 98Table 3.3: The amphora kilns identified by the TAS in the Tarhuna plateau. 98Table 3.4: The Silin Survey, rural villas and farms by periods 104Table 3. 5: Numbers of farms and numbers of presses in Wadi Turgut and Wadi Doga. 118Table 3. 6: The recorded oilery‐villas in the Wadis Turgut and Doga. 122Table 3.7: Some of the estimated figures of the rural settlements before and after the TAS in the Tarhuna plateau 163Table 3.8: A synoptic table of the rural archaeological sites (excluding dams) they have been recorded/rerecorded by the TAS 167Figure 4.1: Schematic drawing of the Tripolitanian olive press 178Figure 4.2: trapetum and mola olearia mill types 180Figure 4.3: Distribution of trapetum and mola olearia mill types in North Africa 181Figure 4.4: The chart shows the number of different press elements have been recorded by the TAS within a number of farming sites in the Tarhuna plateau. 182Figure 4.5: Mills of the three types of mola olearia have been recorded in the Kasserine survey. 183Figure 4.6: Types of mill recorded by the TAS in the Gebel Tarhuna. 184Figure 4.7: Two types of crushing stones found in the Wadi Turgut. 187Figure 4.8: A reconstruction elevation showing the possible milling process of the Gebel Tarhuna mills 188Figure 4.9: An olive press locates within the ancient oilery farm (TUT43) in the Wadi Turgut 189Figure 4.10: A selection of press orthostats from the Tarhuna plateau. 191Figure 4.11: Height of the selected press orthostats from the Tarhuna plateau arranged in descanting height order. 193Figure 4.12: a) Images of the highest press‐uprights of sites DUN128 and TUT9; 195Figure 4.12: b) Unusual press orthostats of site TUT3 (Type T6). 195Figure 4.13: An illustration of the Tarhuna plateau press orthostats types 197Figure 4.14: Schematic view of the main press elements 198Figure 4.15: A line of seven in situ press beds at Sidi Aboageala (TUT12). 199Table 4.8: Some of the Gebel Tarhuna press beds recorded by the TAS and selected for measurement. 200Figure 4.16: A circular channel of press slab with internal meanders. The circle indicates hypothetical diameter of stacked baskets. 203Figure 4.17: Examples of press beds with eroded meanders recorded in Tunisia. 203Figure 4.18: Some of press beds with traces of stacked baskets recorded by the TAS in the Wadi Turgut. 204Figure 4.19: A reconstruction of exilibus regulis (by Drachmann 1932: 150). 205Figure 4.20: Two press beds of circular channel with angle cuts. 206Figure 4.21: An illustration of the press bed (TUT29) shows the fitting of wooden slabs on top of it. 208Figure 4.22: A Google Earth image demonstrates the location of Lebda Cement Factory rural villa 208Figure 4.23: A press bed with angle cuts discovered in the Lebda Cement Factory 209Figure 4.24: Types of Tarhuna press floors 212Figure 4.25: An over view of the distribution of the main pressing elements in the site TUT14 213Figure 4.26: Types of counterweight recorded by the TAS in the Gebel Tarhuna. 214Figure 4.27: Reconstruction of a Tripolitanian lever and weights press with perforated piers and “Semana” type counterweight 214Figure 4.28: A “Semana” type of press counterweight found at Hendek Kale in Turkey 214Figure 4.29: An example of the imbedded in situ counterweight from the oilery TUT43 215Figure 4.30: A press counterweight recorded at oilery farm (DOG82). 216
viii
Figure 4.31: Types of press counterweights (Brun 1987) 217Figure 4.32: A press vat cut from a piece of limestone and installed in front of one of press beds at DUN128. 219Figure 4.33: A press vat discovered in 2007 at the Lebda Cement Factory 220Figure 4.34: Location of sites with the highest orthostat holes south of Gasr Ed‐Dauun 229Figure 4.35: Chart showing maximum, minimum and average of the circular channel diameters listed in the Table 4.11. 230Figure 4.36: The mill and the millstone at Alarabi family olive oil press. 235Figure 4.37: The document M.T.T 1888 238Figure 4.38: The document 64/M/Ch 1843. 240Figure 4.39: The document D/M/T/T 1863. 242Figure 4.40: The document (D/M/T/T 1863 and D/M/T/T 1874). 245Table 4.1: The Gebel Tarhuna olive mills. 185Table 4.2: The large olive mills listed by Brun 1986. 186Table 4.3: Dimensions of some millstones 187Table 4.4: Estimation of processing capacity for the recorded mills by the TAS 188Table 4.5: The dimensions of selected the Tarhuna plateau press orthostats. 192Table 4.6: Average, maximum and minimum measurements of the Tarhuna press orthostats recorded by the TAS. 194Table 4.7: Dimensions of some press orthostats recorded in Thelepte region in Tunisia (after Hermassi 2004). 196Table 4.9: The 44 classified press floors (recorded by the TAS 211Table 4.10: The visible surface size of selected counterweight blocks recorded by the TAS 216Table 4.11: Some of the Kasserine counterweights 217Table 4.12: Dimensions of some press vats recorded by the TAS. 219Table 4.13: Mattingly’s measurements of maximum and minimum operating heights of selected Tripolitanian presses 226Table 4.14: Maximum and minimum operating heights of selected Gebel Tarhuna presses as recorded by the TAS. 227Table 4.15: Diameter of 18 circular channels of the press beds recorded by the TAS. 230Table 4.16: Hypothetical oil yields from the Wadis Turgut and Doga presses of small, medium and large capacity. 232Figure 5.1: A map of pottery kilns in Tripolitania 249Figure 5.2: A map shows the distribution of pottery production sites on the Tarhuna plateau and their relationship to the Roman eastern Gebel 253Figure 5.3: The 16 amphora stamps collected by the writer in the last two decades from pottery production sites in the Gebel Tarhuna. 255Figure 5.4: An amphora stamp recorded in 1994 at GUM90. 257Figure 5.4: a) A stamp on Tripolitana I amphora found by the TAS at Henscir Assalha (TUT15). 257Figure 5.4:b) The same stamp on a Tripolitania I amphora found at the Laurons II wreck in Marseille 258Figure 5.6: The Tripolitanian amphorae I, II and III. 259Figure 5.7: Examples of The Tripolitanian amphora I produced in kilns recorded by the TAS 261Figure 5.8: The Tripolitanian amphorae I, II and III 262Figure 5.9: Examples of Tripolitanian amphora II recorded by the TAS 263Figure 5.10:a) Examples of Tripolitanian III of the Gebel Tarhuna recorded by the TAS; 264Figure 5.10:b) Some examples of Tripolitanian III of the fourth century AD (from Bonifay 2004). 264Figure 5.11: Sketch of Ain Scersciara and Hai al‐Andulas pottery kilns 266Figure 5.12: Sketch view of kiln no.1 in Arbaia (TUT48). 266Figure 5.13: A Google Earth image shows location of Arbaia pottery kilns within the surrounding 267
ix
landscape
Figure 5.14: A line of three below kilns cut by a modern track. 268Figure 5.15: Ain Scersciara pottery kiln which probably was able to fire hundred of amphorae during one load. 269Figure 5.16: Sample from the complete uncrushed olive stones found inside the stoke‐hole of kiln no.1 (TUT48). 269
Figure 5.17: Two Tripolitanian amphora stamps found in Bu Njem. 274Figure 5.18: Shows the Tripolitanian amphorae stamps recorded at Monte Testaccio (Rome) and published in CEIPAC database 283Figure 5.19: The main African amphorae types recorded at the Monte Testaccio. 285Table 5.1: The amphora kilns identified by the TAS in the Tarhuna plateau 254Table 5.2: Diameter of some amphorae kilns in the Gebel Tarhuna recorded by the TAS. 271Table 5.3: List of Tripolitania III amphora stamps and suggested identifications with individuals or families known from Lepcitanian epigraphy 276Table 5.4: List of amphora stamps identified at kiln‐sites by the TAS in the Gebel Tarhuna 277Table 5.5: The main Tripolitanian stamps from Monte Testaccio 281
x
Abstract
This thesis examines the rural settlement, landscape and two rural economic
activities: olive oil and amphorae production on the Tarhuna plateau of Tripolitania.
This was gendered from the Late Neo-Punic through the Roman period. Tripolitania is
considered one of the main olive oil production regions during the Roman imperial era.
Previous studies have tended to stress that presses of the Gebel Tarhuna were totally
used for olive oil production, but the new evidence identified by the Tarhuna
Archaeological Survey (TAS) has addressed that the wine was also produced to some
extent in this area during the Roman period. The study has shown that there was a close
relationship between olive oil and wine production and amphorae production by
identifying new 14 amphora kiln sites with a quite large number of amphora stamps.
These stamps reveal that these amphora workshops mostly located within estates belong
to the urban elite.
The dissertation is divided into six chapters. The first half of the thesis is dealing
with the geographic and literary background, the TAS and the ancient rural settlement
on the Tarhuna plateau. After this, chapters are devoted to examine pressing facilities
and the press element typology. This examination led me to estimate the capacity
production of about 200 presses recorded in the Wadis Turgut and Doga with their close
relationship to amphora production sites. Finally, some evaluation points are made with
attention paid to the importance of future work as a key factor for improvement the
knowledge about rural economic and settlement in this hinterland region of Tripolitania.
xi
Acknowledgements
Firstly, I would like to thank ALLAH for giving me the patience and health to
perform this task. Secondly, this study would never been completed without the help of
many people and organisations, to whom I owe a great deal of thanks. Inevitably, any
errors must remain my own. To my PhD supervisor, Professor David Mattingly, I owe
an immense amount of gratitude for his years of guiding, teaching and nurturing, and I
cannot begin to express my deep feeling of respect and esteem. He has always been the
epitome of helpfulness and encouragement. I also owe my thanks to Dr. Jermy Taylor
for his invaluable support.
I would like to express my gratitude to the staff of the School of Archaeology and
Ancient History (University of Leicester) for all valuable help I have gained during the
years of my study in the school: Mr. Ian Reeds, Anita Radini, Matt Hobson and Martin
Sterry. My thanks go also to the members of my survey group: Hafed Abdouli, Anita
Radini, Fares Mousa, Meyria Rodregez Gonzalis and Sabastian.
The thesis would not have been done without the assistance of some individuals and
foundations. I would like to express my deepest appreciation to those have encouraged
and corrected me along the way: Dr. Hafed Walda, Abdasslam Wheashi, J. Dore, Yosef
Khalefa and Ali Bobreaq. My sincere gratitude to the Libyan Department of
Antiquities, Department of Lepcis Magna and Tarhuna: Haj Jumaa Anag, Dr. Saleh
Agab, Musbah Asmia, Emhemmed Masaoud, Belgasm Makhyon, Juma A. Omar,
Hemali Saleh and Bashier Nayam, they all have been invaluable colleagues, offering
crucial assistance over many long days of fieldwork.
xii
Finally, this work dedicated to my family, especially my wife, who has been
steadfast in her support and enthusiasm and my twin brother Abdalati, my nephews
Jebril and Hamza, for their great support and assistance.
13
Chapter 1: The Gebel Tarhuna 1.1 Introduction
The Tarhuna plateau region of Tripolitania offers a unique opportunity to study
Roman rural settlement patterns and economic activity, on a landscape that combined
aspects of olive oil and amphora production. The context and origins of many rural sites
were not previously recorded, and my work confirms the suggestion that the Gebel
Tarhuna was one of the best olive oil productive areas in Tripolitania during the Roman
period (Goodchild 1976; Mattingly 1985; 1995; Oates 1953).
The thesis describes the results of the Tarhuna Archaeological Survey (TAS). In the
last forty years, there have been several important archaeological surveys in suburban
and rural landscapes in the Mediterranean World. However, the hinterland of
Tripolitanian coastal cities has rarely been the subject of a regional survey. The TAS
aims to gain a greater understanding of economic activity and settlement patterns in the
Tarhuna plateau during Roman period. Using an integrated approach, based on
archaeological materials, geographical data, landscape and surface survey, this research
investigates the economic aspects of archaeological sites, in particular those sites of
farms associated with presses and sites of amphora production. There is an urgent
necessity for carrying out further recording and investigating of these sites. Specifically,
a number of these sites have recently started to be disturbed by several operations of
spoilage, looting, vandalism, rubbish dumping, encroachment of modern agricultural
reclamation projects and urban spread, which in time will damage or erase the features
of these sites.
Prior to beginning this thesis I have had many years experience in excavations and
surveys. My background in excavations includes a participation in two excavations
14
conducted in urban and suburban of Lepcis Magna. The first has carried out in the Wadi
Ressaf (a suburban area very close to Lepcis Magna) during 1996-8 by the University
of Roma Tre (Munzi 1998; Munzi and Pentiricci 1997). The second excavation was
conducted by the Society of Libyan Studies in the area adjacent to Lepcis’ theatre
during 1996-8 (Walda 1997). I have benefited from my participation in my supervisor’s
project, the Desert Migration Project 2007-10 (Mattingly et al. 2008; 2009; 2007). I
have also directed rescue excavations as part of my work in the Tarhuna archaeological
department between years 1995-2003. Before embarking upon this research I also had
some experience in survey and analysis of artefacts by joining a number of survey
projects of areas surrounding Lepcis Magna and the Libyan Desert. The intensive focus
used for analysing finds collected by the above projects and careful excavation
techniques employed were important influences.
The Roman presses of the Tarhuna plateau first came to scholarly attention through
the work of Cowper (1897), who mistook them for prehistoric megaliths. An increase
in our knowledge of the archaeology of the Tarhuna plateau has occurred since
Cowper’s day (Cowper 1897) and permitted new conclusions to be drawn from the
analysis of the material concerning settlement pattern and economic activity during the
Roman period in this part of Tripolitania. In a period of approximately 12 weeks
(February – March and October – November 2007), I identified 111 sites, from 61 of
which I recorded pressing evidence (c. 210 presses). The fieldwork was mainly
concentrated on the region of Wadis Turgut and Doga (Fig. 2.2 and 2.7). Considering
the limited time I was able to spend in the field, and the small part of the Tarhuna
plateau that I surveyed (115 km2 out of 3500 km2), it is quite remarkable that I was able
to identify so many sites and to collect samples from them.
15
The study investigates the regional distribution, acquisition, and production patterns
of rural settlement of the Roman period on the Tarhuna plateau. Specific focus was
placed on the Wadis Turgut and Doga at the north eastern sector of the Gebel Tarhuna
(Chapter 2). That chapter also describes the methodology that is employed in the TAS
and the definition that is given to the identified sites. Chapter 3 deals with the Tarhuna
plateau’s settlement types and organisation, settlement construction, and settlement
density and diversity. The pressing facilities of olive oil (and wine) are examined in
Chapter 4, with particular attention paid to the quantitative analysis of pressing
materials and the estimated annual capacity production. The importance of the new
recorded amphora production sites in the Tarhuna plateau by the TAS and the
Tripolitanian amphora types and stamps are emphasised in Chapter 5. Chapter 6
concludes the thesis by summarising the most significant results, evaluating the project
and addressing the future need for more archaeological research regarding this region.
1.2 Geographical and climate conditions of the study area:
The Tarhuna plateau, or Gebel, lies in North-West Libya (Fig. 1.1), on the eastern
part of the Gebel Nafusa, called the Gebel Tarhuna-Msellata. It is bordered by the
Gefara plain on the north, the pre-desert area on the south, the Msellata plateau on the
east, and the Gebel Gharian on the west (Fig. 1.2). Although the northern boundary is
formed by a narrow strip of Gefara plain, topographically the region is dominated by
mountain plateau that ranges in its elevation from c.135 m at its north-eastern border to
610 m asl. in the Aurban village close to the Gebel Gharian.
The present-day aspect of the relief of the Tarhuna plateau territory reflects the
course of geomorphological evolution in the period of tectonic activity within the
Neogene-Quaternary period. The neotectonic uplift of the northern periphery of the
16
African continent inherited from the previous geological periods had led to the final
isolation of littoral lowlands and elevated plains-plateaux, which represent the largest
geomorphological elements of Tripolitania.
Figure 1.1: Location of the Tarhuna region (drawn by A. Wheshi).
Scholars have argued that the continental regime established in Tripolitania is earlier
than the one occurred in Cyrenaica (Desio 1971; Hecht et al. 1963). A series of faults
arose in the process of uplifting on the steeper northern limbs. The faults themselves
take the form of vast steps along the areas of the deepest and largest faults. In the relief
they represent the littoral plain and the steps of the Gebel Nafusa plateau (Desio 1971;
El-Hennawy M. 1975).
The Tarhuna plateau is crossed by several valleys (wadis) running to north or to
south. The most famous one is called the Wadi Targelat- Caam and considered the
largest wadi in the southern part of the plateau (Brehony 1960; Cowper 1897). Alluvial
deposits of the Tarhuna region form part of alluvial deposits of mixed genesis that
display a considerable distribution in the Gebel Nafusa east of the Gharian meridian on
th
re
co
ho
fe
m
19
Figure 1.2: L
he Gebel T
epresent a st
oarse fragm
Most of t
owever, the
eaturing tw
meteorologic
920s to 197
Gef
Location of th
Tarhuna-Mse
tratum of lo
ments of calc
the Gebel
e annual di
wice or mu
cal station (
78 ranges
fara plain
Oea
e Tarhuna pla
ellata to L
oamy sands,
careous or si
Tarhuna li
istribution
uch precip
(Table 1.1),
from 55 m
0
The Tarhun
The
ateau in Tripo
epcis Magn
, loams and
ilicon matte
ies between
of precipita
pitation. Ac
the amoun
mm (1925)
94.5
na plateau
e Mediterran
litania.
na on the
clays with
er (Shishov
n the 300
ation is lik
ccording to
nt of precipi
to 506 (19
18 27
The pre
ean Sea
littoral pla
interlayers
1980).
– 200 mm
kely non-un
o the data
itation for t
974), 28%
7 36Kil
e‐desert
Lepcis Mag
ain. These
of slightly
m rainfall i
niform, som
a of the T
the period f
of the year
High: 650m
Low :00
lometer
µgna
17
deposits
rounded
sohyets;
me years
Tarhuna
from the
rs being
rs
µ
18
relatively hot and dry with less than 200 mm of precipitation and 12% of the years
being slightly wet with more than 400 mm of precipitation (Jones 1971).
Jan Feb Mar
Apr May June July Aug Sep Oct Nov Dec Annual average
55.3 43.5 34.7 19.3 5.8 2.7 0.2 0.6 12.7 27.4 23.7 47.0 272.9 Table 1.1: Mean monthly and mean annual precipitation (mm) from the Tarhuna town meteorolo‐
gical station (1925 – 1978).
The territory under survey is situated in the zone of Mediterranean type climate, in
the belt of subtropical alternate atmospheric circulation. In summer the climate is
determined by the stable high pressure zone situated over the Mediterranean Sea, by a
peak pressure with descending tropical air currents. Descending, the air masses get
warm and dry; the air temperature reaches its maximum in this period, with the rainfall
at its minimum, making up from May to August about 2% of the yearly total. In the
autumn-winter-spring period the climate conditions are determined by the cyclonic
activity of the ascending air masses of the temperate zone. A number of cyclones,
moving over North and Central Europe, shift to the south together with tropical
maximum, get additionally saturated over the Mediterranean Sea and become stronger.
The mean air temperature in winter is two or three times lower than in summer. The
amount of precipitation from October to March is 85-90 % of annual precipitation, its
maximum being distinctly evident in winter because of increased cyclonic activity in
the Mediterranean zone (Shishov 1980).
1.3 The agricultural importance of the Tarhuna plateau:
The region’s economic resources, in particular its agricultural constituents, very
likely played a substantial role in making this region one of the most productive
agricultural hinterland in Tripolitania for olive trees cultivation and rural settlement. In
the same way, the density of archaeological remains of the olive oil production sites is
19
seen as a reflection of the involvement of the elite, especially from Lepcis Magna, in
exploiting this countryside as a prime source of their revenue contributing to the power
and wealth of the region (Manacorda 1977; Mattingly 1988a; Reynolds 1995). The high
level of civilization and monumental building, the huge public buildings and the scale
of use of luxury materials in the main coastal centres (Lepcis Magna, Oea and Sabratha)
reflect their wealth and vast financial resources. According to Romanelli this wealth
was developed mainly on the basis of the successful Roman agricultural expansion in
Tripolitania and had its climax before the third century AD (Romanelli 1929).
Similarly, Mattingly states that the extension of well-developed agricultural lands so far
into the Gebel (Gebel Tarhuna-Msellata) had a great impact on the wealth of many of
the Lepcitanian aristocracy from the Augustan period (Mattingly 1988a: 22-23; 1995:
141).
The people of the Tarhuna plateau encountered by Barth and Cowper in their travels
in the nineteenth century, were practising pastoral transhumance integrated with
multiple cropping on the wadi beds (Barth 1857; Cowper 1897), and this pattern
continued in places well into the twentieth century. The archaeological evidence reveals
that the region had a greater agricultural potential during the classical period. The large
amount of ancient farming remains visible in the landscape encouraged the Italian
colonial authority in the last century to re-establish many modern farms and to build a
number of agricultural village centres (Hornby 1945). The present-day cultivation
potential of the region is mainly concentrated in the northern sector of the plateau. The
southern wadis of the plateau are still used as pasture lands, with cropping only during
the years of higher precipitation. Figure 1.2 illustrates the northern areas of the Tarhuna
20
plateau occupied by the main arable lands. Olives remain the most important followed
by almond, grape and fig (Fig. 1.3).
A series of questions can be asked about the archaeology of the Tarhuna region.
What processes lay behind the development of the ancient agriculture on the Tarhuna
Gebel, given that it has the character of a semi-arid plateau separating the coastal plain
from the arid pre-desert zone? How did the Tarhuna plateau play an important role in
the economy of the main coastal centres, especially, Lepcis Magna, during Antiquity?
Can it be demonstrated that the Tarhuna plateau succeeded in providing a large scale of
olive oil production? To what extent do the archaeological data of the Tarhuna plateau
make the region an appropriate area for investigating larger research questions about the
Roman economy? How best can we mobilize field survey results in the context of
broader debates about rural settlement trends and Roman economy? To what extent
does the evidence can be illustrate economic growth, intensification of production,
urban-rural relationships and networks, the wider agricultural economy (including the
labour force) and landscape? To what extent is it legitimate to consider an increasing
level of specialisation as a proxy of economic growth, and to what extent is it legitimate
to consider it as such in the Roman world?
In considering this problem, a prime aim of this research is to find out the factors
responsible for permanent sedentary agricultural settlement in such a zone during the
classical period. A range of archaeological and historical evidence will be used to build
up a picture of ancient economic activity on the Tarhuna plateau. The evidence for two
main economic activities, which are olive oil production and amphorae manufacture,
will be analysed with reference to this framework, in order to assess the organisation
21
µ
07
143.5
Km
Olive and alm
ond
Olive and alm
ond
Olive and alm
ond
Olive and alm
ond
Grapes and
Figgs
Grap
es and F
iggs
Olive, alm
ond an
d fig
Wa
di
Tu
rg
ut
Wadi DogaM
eji
Dauu
nK
hadra
Tarh
una
Sidi e
s-Said
So
uk aj-joum
ah
13.533
408
13.533
408
13.754
246
13.754
246
13.975
084
13.975
084
32.576336
32.576336
07
14
3.5
Km
µ
Figure 1.3: The modern arable areas on the G
ebel Tarhuna. (Mapping of environm
ental resources project for agricultural use, National program
for development
of vegetation cover, Tripoli 2007).
22
and level of these economic activities in relation to settlement patterns. For this reason,
the archaeological remains will be investigated to assess and place the Tarhuna plateau
in the regional economic context and story, and to understand more clearly how the
elite of the coastal cities were involved in investment in agricultural or rural estates.
The argument by Mattingly is that the exploitation of the Gebel lands (Gebel
Tarhuna-Msellata) coincided with the development of the coastal plain during the
Roman period (Mattingly 1987:49). Initially, this agricultural development took place,
probably from the pre-Roman period (late Punic and early Numidian periods, 2nd-1st
centuries BC), in areas where the rainfall was especially favourable. The minimum
rainfall required for the cultivation of wheat and olives (200-400 mm) allows their safe
cultivation in only limited areas of the Gebel by dry farming (Barker 1984). According
to Mattingly, the Gebel Tarhuna-Msellata is more suitable for growing olive trees
because average rainfall of only of 300 mm per year is usually received in the region
(Mattingly 1985:31).
The production of olive oil has long been recognised as the most likely mainstay of
the agricultural economy of Tripolitania during the Roman period, though as we shall
see wine may have been more significant here than previously recognised. Mattingly
(1995) has emphasised that this situation was not unexpected or unnatural because the
olive tree is considered a hardy tree and very often is able to adapt to marginal
environments. The most common view is that the dry farming of olives was particularly
suited for the hillsides of the Tarhuna plateau; indeed the traces of olive farms and the
existence of hundreds of presses indicate the widespread nature of olive cultivation in
the region. Goodchild, for example, mentioned that the main area for the growing of
olive trees, in Tripolitania, was the whole of the eastern Gebel from Tarhuna and
23
Msellata hill region to the sea and from the Gebel escarpment to the Wadi Targhlat
(Goodchild 1952a:76). A very similar argument is presented by Mattingly, who sees
that the Gebel Tarhuna (3500 sq. km) includes the best documented surviving traces of
Roman olive farms in Tripolitania, even though the landscape and land-use had
changed in the post-Roman period (Mattingly 1988a:25). Oates (1954:91) suggested
that this change started from 4th century AD, due to the marauding raids of pre-desert
tribes such as the Austuriani who caused a decline of the prosperous agricultural society
of the first three centuries AD.
Some writers have claimed that the eastern Gebel lands were first opened up to
intensive olive growing early in the first century AD (Grahame 1998). However, this is
based on the fact that the previous surveys could not produce any proof that the
discovered farms in the region existed before the first century AD (Mattingly 1995:
140; Oates 1953: 110). The Roman-era agricultural boom did not start from nothing and
elements of the pre-Roman rural landscape are starting to be recognised (Munzi et al.
2004). The proposed hypothesis is that the region witnessed a degree of settled life and
agricultural practice from at least the second century BC. Black glazed ware of
“Campana A” production and Numidian coins found recently at some sites on the
Tarhuna plateau are evidence of the early exploitation system. This new evidence
supports Mattingly’s conclusion (Mattingly 1988a) that if Caesar’s fine of three million
pounds of olive oil (Bell. Afr. 97.3) was imposed on Lepcis Magna, this level of
production could only have been reached if the Gebel Tarhuna was already being
intensively farmed by the mid first century BC. It seems to be reasonable to suggest that
the sedentary agricultural was not only associated with the Tripolitanian coastal centres
but also extended to this internal plateau during the pre-Roman period (Munzi et al.
24
2004: 21). A very different picture is presented by Mark Grahame who has argued that
the city fined by Caesar was Leptis Minor and was not Lepcis Magna, and that the lands
of the Gebel were dominated by fully pastoralist peoples until the first century AD
(Grahame 1998: 107). An attempt will be made by this research to shed further light on
the crucial issue of the nature and scale of pre-Roman development of the Gebel
Tarhuna.
A basic problem with studying the traditional economy is that in fact it is hard to
find any detailed knowledge from the historical sources. There are few direct references
to farming and rural settlement behind the coastal cities in Tripolitania, or to what
extent the agricultural production in this hinterland extended in scale beyond
subsistence to surplus production in the classical period. A rare exception is presented
by Apuleius in his Apology which describes estates belonging to his wife (Aemilia
Pudentilla) in the hinterland of Oea. The problem can be more fully addressed by the
examination of the ample archaeological evidence for rural settlement including villas,
farmsteads, and presses (Cowper 1897; Goodchild 1951; 1952a; Mattingly 1985; 1987;
1988a/b/c; 1989; 1995; Oates 1953; 1954).
Because the literary sources were mainly produced by and for a leisured elite, they
say little about economics (Morris et al. 2007). Archaeological evidence examined by
field experts reveals that this perspective is misinformed. Since the beginning of the
second half of the twentieth century the archaeological data has expanded the
knowledge about ancient economic behaviour, and Greco-Roman economic historians
have become aware of new questions and have been using new methods to answer them
(Finley 1973; Garnsey and Saller 1987; Jones 1974; Mattingly and Salmon 2001;
Rathbone 1989; Rostovetzeff 1953; 1957).
25
The last three decades witnessed a great historical debate concerning the Greco-
Roman economy. Many historians have tried to find answers regarding the Roman
economy And in particular relating to the debate about whether it can best be described
in terms of concepts of ‘primitivism’ or ‘modernism’, and whether the limitations of
the Roman economy can be seen as a reflection of the influence of the consumer city.
The consumer city meant a lack in capital investment, limited technological
innovations, a low level of growth or surplus production or industrial specialization, and
little long-distance trade in non-luxury items. Proponents of this view believe that the
Roman economy only grew to a limited extent, if at all (Finley 1985; Garnsey and
Saller 1987). Other scholars have identified a different model for the Greco-Roman
economy to that which was developed in intricate and painstaking detail by Finley and
his school. This view is illustrated by the following statement by Hopkins that “the size
of surplus produced in the Mediterranean basin during the last millennium BC and the
first two centuries AD gradually increased ... The growth in the surplus produced and
extracted was largely the result of two factors, political change and the spread of
technical and social innovations” (Hopkins 1983: xiv). Archaeological data relating the
Greco-Roman agricultural production and trade, especially of olive oil and wine, has
revealed quite persuasive evidence of a real growth to set against the ‘irrational’ and
limited economic growth view (Amouretti 1986; Forbes and Foxhall 1978; Foxhall and
Forbes 1982; Hitchner 2002; Mattingly 1988b). If growth can be shown to have
occurred, underlying questions can be addressed. Accordingly, this study will highlight
evidence that there was economic growth, specialisation of olive oil production and
surplus production for export on the Tarhuna plateau during the Roman period,
particularly from the first to third centuries AD. In this case study, the possibility to use
data on the production and export of olive oil and amphorae opens a new potential for
26
debate about the social and infrastructural accelerators that underlay this growth in the
Roman period. In the concluding chapter of the thesis, I shall return to review the
impact of my data on the wider debates about growth, scale, specialisation and
standardisation in the Roman economy and to the inter-relationship between town and
country.
1.4 The Tarhuna plateau landscape archaeology A prime attraction of studying Tarhuna’s ancient landscape is that its prominent,
well-preserved and extensive rural sites still stand out in the landscape, especially sites
of oil presses. It is conventional to talk of these presses as ‘oil presses’, though as this
thesis will show some were evidently for wine production. Although these sites appear
to have been abandoned for most of the last 1500 years, a significant number of them
(more than 200 sites) remain standing-up, and some of them still have presses more
than 3 m high. I cannot emphasise too much that this semi-arid area south of the
cultivated coastal plain presents a remarkable image with its substantial well-built
settlements with numerous farm buildings and presses. The process of rural
exploitation, growth and decline occurred across more than five centuries leaving
distinct traces in the archaeological record (Cowper 1897; Oates 1953; 1954; Goodchild
1951; Mattingly 1985; 1987; 1995), though there is undoubtedly enormous
archaeological potential for renewed survey work here. For instance, Cowper (1897)
and Oates (1953) recorded more than 260 presses even though their works did not cover
the whole of the Tarhuna region. As we shall see, this thesis adds a significant number
of new rural sites which have not been previously recorded.
There are many archaeological remains on the Tarhuna region (Chapters 2 and 3).
However, these remains are extremely dispersed over most parts of the plateau. Another
27
interesting point is that the Gebel Tarhuna was a boundary land between two important
ancient coastal centres, Lepcis Magna and Oea (Di Vita-Evrard 1979; Mattingly 1987).
It is possible that this ancient reality may be reflected in settlement patterning (Chapter
2). Like many other cities in the Roman world, the city of Lepcis Magna was integral to
its surrounding territory and hinterland (Munzi et al. 2004). Not all resources and items
needed at the city could be obtained locally and from an early date Lepcis Magna
imported artefacts and other goods, through Mediterranean and sub-Sahara trade
(Mattingly 1995: 158). Many of these artefacts and goods were not only distributed at
the urban centre but can also be found spread in the countryside.
In recent decades a great interest in economic archaeology of rural Roman North
Africa has emerged from a number of published survey projects and excavations in the
countryside (Barker 1996; De Vos 2000; Dietz 1995; Fentress 2000; 2001).
Nonetheless, the rural sites have rarely been excavated in contrast with the great
number of excavations that has been concentrated on both large and small urban centres
in Roman North Africa (Stone 1997). The results of these surveys support the dominant
concept of the significant role of the countryside and rural life in the economy of the
Roman Empire; since the majority of its inhabitants were living in the countryside
during the Roman period, whether in villages, hamlets or isolated farms (Barker 1991).
On the other hand most of the evidence from the field surveys indicates that there was
intensive economic growth in the provinces of North Africa, particularly, Africa
Proconsularis (Hitchner 1993). As Mattingly has pointed out:
[These] include: growth in agricultural production and rural production, an increase
in exports of primary products, raised levels of import substitution, large scale units of
production (from farms to oileries, from workshop to manufactory pottery production),
28
the emergence of a society that was pottery involved in risk-taking, economic
calculation, technological innovation, and other ‘rational’ economic behaviour
(Mattingly 1997: 117).
This perspective supports the substantial role of North Africa’s countryside in the
economy of the Roman Empire as a prime resource for the food supply of the city of
Rome.
The importance of the countryside of North Africa in antiquity and its component
cultural features such as hamlets, farmsteads, olive presses and kilns is clearly stated by
Stone: “[I]n the pre-modern societies of Antiquity the countryside played a critical role in the
acquisition of the power, wealth, and resources necessary for the construction of major urban
monuments” (Stone 1997: 2). In considering the importance of the countryside and rural
economy, this research will suggest that the Gebel Tarhuna was one of the best rural
areas in the province of Tripolitania during the Roman era because it included, and still
has, much of the best agricultural land in Tripolitania. It is clear from the density of
olive presses that the territory of Gebel Tarhuna was an agricultural production area that
had the capacity to produce millions of litres of oil in good years and was
supplementary in important ways to the agricultural development of the coastal plain
(Chapter 4), especially the territory of Lepcis Magna (Mattingly 1985: 31; 1988a: 27
1995: 140).
1.5 Background of the study: The archaeological sites of the Tarhuna region were first reported in the nineteenth
century by Heinrich Barth and Edwin Von Bary who visited them in 1849 and 1875
respectively (Barth 1857; Von Bary 1883). Both these pioneers gave a summary
description of some ancient sites on the plateau, which attracted attention of other
29
travellers and scholars in the late nineteenth and early twentieth century. During the
years 1895-6 H. Swainson Cowper visited the Tarhuna plateau and examined in
considerable detail more than 80 ancient sites (Cowper 1897). Although he interpreted
these sites as prehistoric monuments of religious character (senams), his work
constitutes a significant pioneer survey. Once it was revealed that his senams were in
fact olive oil presses, the enduring value of his work was that it revealed the importance
of the Tarhuna region as a zone of intensive olive-cultivation during the classical
period. Gebel Tarhuna was also the area partly surveyed by Goodchild (1951), who
examined a number of ancient sites, specially his excavation in the sanctuary of
Ammon at Ras El-Haddagia, and the villa and pottery-kilns at Ain Scersciara
(Goodchild 1951: 43-77). At the same time (1949-51) Oates carried out more
comprehensive archaeological survey in an area of some 300 square kilometres around
Gasr Ed-Dauun in the eastern part of the Tarhuna Plateau. In three seasons of work, he
revealed a distribution of more than 100 sites, which extended chronologically from the
first century to the fifth century AD (Oates 1953). The sites recorded by Oates consist
for the most part of varying sizes of farms associated with their presses, water control
and supply works. The 130 separate press structures were interpreted as evidence of the
regional specialization in oil production. It is important to note here in relation to the
dating of dateable sites that in the middle of the twentieth century pottery of the first
century BC date was poorly known. The archaeological notes gathered by Oates (1953:
89-104; 1954: 93-110) are very uneven in general, the descriptions of individual sites
are too brief and lack chronological details. Indeed chronological indicators for sites are
rarely provided, though it is clear from the descriptions of archaeological features that
most belonged to the Roman period. Since 1950, knowledge of pottery has improved
greatly, and subsequent to the work of John Hayes (Hayes 1972; 1980) it is now much
30
easier to recognise and date surface pottery assemblages, especially sigillata wares and
Red Slip wares. For example, African Red Slip (ARS) ware was the most important
type of Late Roman pottery distributed all over the empire. It was produced by various
North African workshops over a period of some six centuries from the end of the first
century until the seventh century AD (Hayes 1972:13). In Tripolitania, it was
accompanied by local variants in late antiquity known as Tripolitanian Red Slip (TRS)
ware.
Archaeological evidence for olive presses from the Gebel Tarhuna suggests a
remarkable level of oil production; for example, Mattingly has estimated that the total
potential oil production capacity in good years will have measured millions of litres.
Sites such as Henschir Sidi Hamdan (with 9 presses) and Senam Semana (with 17
presses) could have produced 100,000 and 200,000 litres in peak production years
(Mattingly 1995: 143). Including these massive oil production sites, more than 260
olive presses have been identified on the Gebel Tarhuna. The region overall
unquestionably specialized in olive oil production, contributing significant exports to
Mediterranean markets (Mattingly 1993: 454). There are some factors that restrict our
ability to make a successful comparison between the Roman period and modern
Mediterranean oleoculture. These factors as illustrated by Mattingly are: “Regional
diversity in olive cultivars, planting densities, soils and climate, cultivation techniques,
harvesting and processing methods” (Mattingly 1994: 91). There is no doubt that the
varying differences in the production techniques and climate condition played a
significant role in the density of planting, the yield for individual trees and the potential
production capacity of the olive presses (Mattingly 1996a). In contrast to the modern
system of olive trees farming in the eastern Gebel, where extensive rows of trees
31
facilitate mechanised cultivated (Taylor 1960), the ancient olives were probably grown
intensively in rows or in scattered groves or in association with other crops, and were
worked by human power and animal traction.
As Figure 1.3 shows, some areas of the Tarhuna plateau are favoured for viticulture
and similar local trends are possible for Roman farming also. As we shall see, there is
some evidence to suggest that wine production was a secondary, but significant
component of the Roman economic specialisation.
32
Chapter 2: The Tarhuna Archaeological Survey
2.1 Introduction:
The Tarhuna plateau (Gebel) is very often referred to in general studies of Roman
North Africa, specifically in Roman Tripolitania, as one of the major olive-producing
areas (Goodchild 1952b). This is based on the evidence of numerous olive presses,
oileries, farms, and other olive oil related facilities found in the hinterland of the
wealthy coastal cities Lepcis Magna and Oea (Lloyd 1991). Archaeological surveys
carried out in the Mediterranean world, particularly, in North Africa, are of major
importance in understanding the ancient Tarhuna landscape, and in clarifying problems
pertaining to its proper agricultural use of land and for its interpretation (Mattingly
1995; Mattingly and Hitchner 1995). Nevertheless, except for work carried out by
David Oates during the years 1949-51 in the eastern part of Tarhuna Gebel, around
Gasr ed-Dauun (Oates 1953; 1954) no systematic study of the rural settlement and
economy of any part of this region has been undertaken. Until there is a systematic
survey of the Tarhuna region, many aspects of its settlement patterns and economic
development in antiquity will remain indistinct. The purpose of this chapter is to
explain within a geographical and typological framework the work of Tarhuna
Archaeological Survey which I conducted in 2007. Concentrating on specific areas of
the Tarhuna region, this chapter presents the area of Wadis Turgut-Doga and some of
their tributaries as a key case study of ancient settlement patterns and site distribution
on the landscape. It will also give consideration to economic aspects of the
archaeological remains.
33
2.2 Aims of the survey
The Tarhuna Archaeological Survey 2007 (TAS) aimed to investigate the
organisation of rural agricultural settlement in the region of the Tarhuna plateau. The
region's economy under the Roman Empire was evidently geared to the production of
olive oil for export (Mattingly 1985, 1987b, 1988a, 1988b, 1988c), but the relationship
between rural settlement and agricultural production in general remains unclear. On the
other hand, the range of settlement types (see Chapter 3) recorded by previous works
(Cowper 1897; Goodchild 1976; Oates 1953, 1954) clearly suggests the existence of
sophisticated agricultural organisation of the countryside on the Tarhuna plateau.
Therefore, the TAS has been set up in order to investigate the composition and
organisation of the ancient rural settlement pattern in the Gebel Tarhuna of northwest
Tripolitania and to gain an understanding of the agricultural economy of the area in the
wider context of the Roman economy.
The nature of that organisation can be established through an understanding of the
relationship between the various types of settlement. This is not easily achieved in the
absence of documentary evidence or of the opportunity for full-scale excavation of
certain sites in the various settlement categories. In an attempt to deal with this gap in
the evidence, the TAS carried out a further detailed recording of the landscape and
standing remains in the survey areas. These details of the landscape survey will show
the relative numbers, distributions and the relationship between the different types of
sites, from farmstead to small village (for initial explanation of site definitions, see 2-5:
distribution patterns of site types), integrated within elaborate terracing, cross wadi
walls, and other field irrigation features. In addition, the TAS made targeted higher-
level recording of a selection of these rural settlements, seeking to improve knowledge
34
of the chronology of evolution and change of landscape and settlement pattern during
the Classical period.
The TAS employed two types of survey method, extensive and intensive survey, in
order to enhance current understanding of ancient settlement in the Tarhuna region and
to explain how the rural settlement pattern, in particular olive farms, had developed
during the Classical period. This explanation will be supported by mapping the
landscape in association with investigating the overall shape of rural settlement patterns
in relation to economic and environmental factors. Many of the sites which were
recorded during the late nineteenth and the mid twentieth centuries had not hitherto
been accurately located on map coverage. For this reason, a key goal of the TAS was to
improve the overall mapping of the extensive spread of settlement on the landscape by
using old and new fieldwork records and satellite imagery as a base to assemble
archaeological materials for a first synthesis of the settlement history of the region.
Google Earth provides increasingly high resolution satellite images on which the larger
sites (especially the fortified ones) show up well (Fig. 2.1). However, there is no doubt,
especially from the field survey results, that many smaller and less observable sites also
exist and are at present only recognisable on the ground.
Another goal of the TAS was to look at the density and distribution of surface
artefacts and to shed light on their range and date, with particularly attention to the
ceramic evidence from all sites. Thus a useful comparison can now be made between
this region and the neighbouring pre-desert zone (Dore 1985, 1988; Mattingly 1996d)
or other surveyed areas in North Africa such as the Kasserine region (Hitchner 1990).
Furthermore, the field survey results can be employed as a guide to find archaeological
answers for historical questions: what is the evidence, if any, for pre-Roman (Punic or
35
Numidian) farming activity on the Tarhuna plateau? What was the character and extent
of the earliest inland Roman settlement in this internal zone? What was the nature and
rate of economic activities in this region during the Classical period?
Figure 2.1: An example of using the Google Earth to identify sites (DOG69).
2.3 Methodology Since the 1950s and 60s landscape perspectives and archaeological survey have
together developed as the result of movements away from individual sites as analytical
foci. Although landscape has served as a framework for constructing narratives that
pursue chronological/historical rather than explicitly processual or post-processual
methods (Banning 1996; Barker 1995, 2000), there remains a wide distinction among
scientific versus interpretive approaches to landscapes and survey. Although issues of
landscape and archaeological survey span an enormous range of topics, some of the
most pronounced distinctions between intensive and extensive methods involve
36
conflicting views about sampling. Scientific conceptions of landscapes implicitly rely
on the view that for any region and period, there is an archaeological landscape waiting
to be retrieved and reconstructed, while interpretive methods depend on the contention
that there can be many equally appropriate, archaeologist-specific encounters with
landscapes. Indeed, these views are reflected in disparate survey methods. Faced with
limited time and resources, fieldworkers are often have challenged choices between
systematic against judgmental, opportunistic sampling. While scientific methods have
sought increasing sophisticated sampling methods appropriate for quantitative analyses,
interpretive studies have deliberately eschewed systematic sampling (Tilley 1994).
Non-systematic or non-random samples are inappropriate for many types of
statistical analyses, but for some this is not an analytical goal. Unchanged systematic
sampling can result in substantial amounts of time spent in areas lacking evidence
relevant to issues surveys particularly seek to address. The most powerful approaches
utilise a combination of both systematic and opportunistic survey so they can both
generate data appropriate for quantitative analyses, and take advantage of
archaeologists and locals distinctive knowledge of areas where certain types of remains
are likely to be found. The field work of a given study area is concerned with collection,
management, and analysis of spatial information, which offers a variety of informative
means for investigating archaeological landscapes. It combines a range of closely-
related data generating and analytical tools including satellite imageries (mainly based
in the Google Earth), Global Positioning System (GPS), and Geographic Information
System (GIS) technologies. Applications of each of these technologies (often
independently) have recently become common in archaeology (Barratt et al. 2000;
Ebert 1984); together they offer an even more valuable research triad.
37
Intensive and extensive landscape approaches apply many of the same tools and
information sources noted above, but in substantially different ways. Air photographs
were used for archaeology as early as the late 1800’s (Ebert 1984), but since then have
become far more than just pictures from above. Although aerial and satellite imagery
were initially pursued as a means of identifying undiscovered archaeological sites they
have more recently become an important means of characterising environments and
visualising landscapes. Satellite imagery have proved particularly useful in the Tarhuna
plateau where generally little cloud-cover provides unobstructed views of physical
landscapes. Geographic Information Systems (GIS) are one of the primary
methodological arenas where explicit conflicts between deferent conceptions of
archaeological landscapes have centered (Gillings and Goodrick 1996; Kvamme 1997;
Stoddart 1997). GIS was traditionally lauded as a tool that could allow social scientists
more objective, quantitative, scientific means for spatial analyses (Openshaw 1991).
However, GIS is inherently pre-disposed to the inclusion of data that are more readily
available, and more easily represented as maps. For archaeology, these data are often
environmental (soil cover, elevation, aspect, and slope) and GIS is therefore (to some
degree) bias toward materialist analyses that consider associations between
environmental conditions and ancient human behavior (Gaffney and van Leusen 1995;
Wheatley and Gillings 2002).
The survey was conducted in the hinterland of the Punico-Roman coastal centres of
Lepcis Magna and Oea (mainly focused on the areas seen to have been located within
the territory of Lepcis Magna, Fig. 2.2) and was designed to investigate the nature and
extent of rural settlement and agricultural development in the semi-arid high plateau of
Gebel Tarhuna in the Roman period. The methods of any survey must take account of
38
the survey’s aims and questions with regard to the available data. It is especially
important to provide some discussion of procedure, since among my goals was an
attempt to demonstrate the value of information from surface survey carried out in a
more detailed manner than had been performed previously on this plateau.
Figure 2.2: Location of the Tarhuna plateau and the surveyed area.
As discussed above, a fundamental object of the TAS was to study how rural
settlements and economic activity on the Gebel Tarhuna developed and changed over
the extensive period from the later centuries BC until the 5th century AD. A second aim
was to investigate the chief factors that may have lain behind these developments. The
first significant matter requiring consideration is the choices made concerning the scale
and nature of the area within which the survey was conducted: briefly, the question
concerned the satisfactory definition of boundaries. However, the natural environment
must have had some limiting and directional influence on the site distribution and the
development of rural settlement over this hinterland region. As in most archaeological
µThe Mediterranean Sea
Lepcis Magna
Oea
0 9 18 27 364.5Kilometers
cation of the Tarhuna platea
High: 650m
Low :00
300
250
200
Gefara plain
The pre-desert
The area surveyed by the TAS 2007 Rainfall isohyets (mm).
39
surveys, the typical modern approach to doing a survey is to select specific sample
areas in order to employ effective methods of survey (Barker 1991). From the
beginning, the TAS was designed to investigate only selected and restricted parts of the
Tarhuna plateau, since the total area of the plateau (some c. 3500 km2) far exceeded the
available time and resources. As this limitation did not permit total coverage at a
satisfactory level of intensity, several sample sub-areas were selected for detailed study,
these being representative of recognised topographical units. It was clear from the
outset that these areas had great archaeological potential, and they seem to have been
the focus of important economic activities and water management during the whole
Roman period, especially for olive oil pressing and amphora production.
2.3.1 Intensive survey method
Modern archaeological landscape survey, especially in the Mediterranean world, has
benefited from a long history of applying well-developed methods for site discovery
and investigation both to sites and to their surroundings (Bevan 2002; Gillings 1999;
Pettegrew 2001). Nowadays archaeologists know that the archaeological evidence takes
many forms and its study requires suitable field survey techniques to be developed.
Even within the areas selected for intensive survey on the Tarhuna plateau there was a
number of obstacles restricting observation of the whole of the ground surface. This is
a phenomenon that affects most archaeological surveys (Mattingly 1989b). As has often
been noted by archaeologists, there are two sets of variables that hinder field
observations: geomorphological and vegetational (Velde 2001). However,
archaeologists have attempted to reduce this field effect by repeated visiting of the area
under survey (Bintliff 1988; Kamermans 1995). The TAS found that the most
si
bu
Gu
an
re
sit
or
in
de
th
pl
gnificant o
uilding, pav
Figure 2.3: Tuman; (3) the
In addition
n intensive
egion, a few
te hierarchy
rder to test t
nvironment
efferent top
he Wadi Haj
lateau; area
obstacles w
ved roads, an
The three areWadi Beni M
n, fieldwork
coverage o
w small area
y and to elu
the density
and surro
pographical
ajaj (a small
a (2) covers
which affect
nd land und
eas intensivelyousa.
k has often
of small sel
as were sele
ucidate the r
and diversi
ounding lan
location on
l tributary o
the upper
ted the loc
der cultivati
y surveyed by
combined
lected areas
ected for int
relationship
ity of settlem
ndscapes, t
n the Tarhu
of the Wadi
sector of th
cal archaeo
on.
y the TAS 200
extensive r
s (Keller 19
tensive surv
between th
ments and t
the sample
una plateau
i Doga) on
he Wadi Gu
ological rec
07; (1) the ‘Ha
regional sur
983). As re
vey in order
he largest an
their relation
d areas w
(Fig. 2.3).
the north e
uman close
0 52.5
Wad
cord were
ajaj’ area; (2)
rvey covera
egards the T
r to recover
nd smallest
nship to the
were selecte
Area (1) lo
astern secto
of the edg
10Km
di Targelat
40
modern
the Wadi
ge, with
Tarhuna
r the full
sites. In
e natural
ed from
ocates in
or of the
ge of the
41
higher plateau; area (3) asouthern area of the Gebel Tarhuna and takes its place in the
Wadi Beni Mousa (one of the northern tributaries of the Wadi Targlat).
In order to collect detailed information, the intensive survey method has been
applied in three selected small areas of the Tarhuna plateau. I shall trace the evaluation
of research problems that have been addressed, field methods adopted, and forms of
analysis employed. The first intensive survey was made in the Hajaj area, where there is
a small tributary flowing into the Wadi Doga (Fig. 2.4). This area measured c.82 ha and
is known locally for the visible standing structure - called Gasr Dehmesh (it was named
by Cowper 1897: 237 as Kasr Gharaedamish).
Figure 2.4: The first intensively surveyed area, the ‘Hajaj’ area (the Wadi Doga).
The second area, located in the Wadi Guman, covered c.125 ha (Fig. 3.46). This
valley forms one of largest tributaries of the Wadi Turgut. These two areas are in the
northern sector of the Tarhuna plateau, located north of the Roman Eastern Gebel road
42
(built AD 15/17) that linked Lepcis Magna with the southwest limit of its territory,
approximately three kilometres west of Mesphe (Medina Doga) on the Tarhuna plateau
(Di Vita-Evrara 1979; Di Vita-Evrard 1979). The third intensively surveyed area was a
small sector (10 ha) located in the Wadi Beni Mousa (Brogan 1964), a small wadi that
runs southwards and links with the Wadi Taraglat (Fig. 2.5).
In each of the three areas mentioned above, the boundaries of targeted survey areas
were plotted in relation to natural features such as gullies, pathways, slopes and hills.
Each survey area was divided into squares of standard size of 100 x 100m and every
unit was given a letter code for the East-West and numbers for South-North side inside
its area. Thus HAJ A.1 designates square 1 of unit A in the Hajaj area. In order to be
familiar with the selected areas and their materials, and to get a general impression of
Figure 2.5: The gasr Beni Mousa and its surrounding archaeological features.
43
the boundaries of the intensive survey areas, in a first phase of work the field survey
walkers scouted out different parts of the landscape in these selected areas. In the
second phase, the intensive investigation of the landscape was carried out based on the
basic method of long lines or transects. The survey unit was systematically covered by
fieldwalkers spaced in regular intervals across the field. The field team included five
walkers who covered 2m corridors at a spacing of 20m apart (thus 10 percent coverage
of the landscape). The distance of 20m was chosen as suitable for the Tarhuna region
because very often there is a clear visible view over this distance. Ceramic sherds, kiln
debris, and relevant artefacts were counted, and a representative sample of diagnostics
collected for specialist further analysis. In addition, we made a record of the
topographic features and land use in association with the survey units. All this
information was entered into a database which was then linked to GIS analysis of site
location, artefact distribution and human economic activity across the landscape.
Pottery collection involved the pickup of “diagnostic” artefacts, a category that in
pottery typically includes rim, base, handle, and neck or shoulder, as well as all painted
or decorated sherds. Thus the principal elements of the ceramics typology can be
identified through this process (Arthur 1982; Dore 1989; Hayes 1972; Keay 1984).
Ceramics collected included not only sherds of imported pottery and local amphorae but
also the coarse pottery sherds that contributed in a major way to refinement of the
dating of sites on the Tarhuna plateau (Dore 1984, 1988).
The initial survey was designed primarily as a tool for site discovery within the
survey areas. The main objectives behind using intensive methodology were to evaluate
the varying density of material remains and advance preliminary hypotheses concerning
the presence and absence of human economic activity; to characterise where possible
44
the chronology and functional characteristics of material remains. It was not the
purpose of this method, in this initial phase, to investigate and analyse at a very fine
resolution the structural remains of sites within the survey Unit, but rather to collect
basic data over a broad area of the landscape. The methodology relied, in part, on
relatively standard procedures for intensive survey in the Mediterranean World (Barker
1999; Given 2004). Nevertheless, the traditional concept of the “site” has become
increasingly problematic in theoretical and methodological terms; archaeologists have
generally retreated from terminology that makes transparent associations between
artefacts scatters and traditional functional classes of sites, giving rise to such concepts
as the “place of special interest” (Caraher 2006). In the course of TAS, when the survey
team found an evident archaeological site within an intensive survey unit, such as
pottery scatters around some mud brick walls, or recognisable architectural features that
were visible on the surface, they examined it as a separate entity using the method of
laying out a specific collection area either a square or circle, and counting and
collecting the informative and unusual pieces, and such places were often mapped and
planned if there were any standing structures as a second phase of the work.
2.3.2 Extensive survey method
The TAS extensive survey has focused on the discovery and investigation of “sites”;
however, with particular attention being paid to the broader picture of evidence relating
to olive oil pressing and pottery production sites, which have been identified as more
significant in terms of rural economic sites. The targeted region for extensive survey
was divided into a number of areas: TUT (Wadi Turgut), GUM (Wadi Guman), DOG
(Wadi Doga), TEL (Tella), DUN (Ed-Dauun) and TRG (Taraglat). The most important
reason for choosing an area for field survey was usually to locate and map as many sites
45
in as short a period as possible. Although such extensive survey has been widely
adopted as a first step towards the establishment of a general overall picture of
settlement pattern and site distribution of larger area (Hope-Simpson 1983), I suggest
here that the extensive survey of the valleys (wadis) in which the most important sites
lay will usefully supplement my intensive areas of work on the Tarhuna plateau as well
as helping to place the archaeological sites in their wider topographical context. The
extensive survey was thus intended to provide a basic record of the archaeological (and
economic) sites in the study area with a special focus on the northern part of Tarhuna, a
region where the first four areas mentioned above are located. By the same approach,
many previously known sites have been revisited in order to locate them more
accurately and to enhance the records, particularly, for oil presses and amphorae kilns.
Thus the method also involved driving down every passable road or track, looking at
sections of roadcuts, wadi sides and hill slopes for visual remains.
To a degree the basic strategy adopted in the TAS survey was an extensive rather
than an intensive one. The priority was given to the broad, comprehensive coverage of
the region to create a new baseline of knowledge of overall settlement pattern with
some more systematic efforts to record a sample more intensively and to recover data
on site construction and function. It is important to recognise that the Tarhuna plateau is
a semi-arid North Africa landscape where topographic and environmental factors play a
significant role not only in the visibility but also have affected past and present
settlement dynamics and land-use.
2.3.3 The application of GIS
One of the research methods involved the application of GIS techniques in order to
display the recorded sites, and to execute spatial analysis. Archaeologists have long
46
used GIS as an active tool in archaeology, especially for inventory and mapping matters
(Fisher 1999). Also GIS is a suitable tool for the investigation of broader spatial
structure of rural landscapes; the assessment of the effect of surface visibility on the
recovery of archaeological remains; the examination of site definition and
characterisation; and the analysis of decision making behind site location (Bevan and
Conolly 2004). The GIS analysis was instigated by creating a standard list of site data.
According to the site location on the Tarhuna plateau, the site list was divided into two
spreadsheets. One belongs to the north-eastern sector (TUT, GUM, DOG and TEL) of
the plateau. In this sector, a total of 135 sites has been recorded by the TAS. The second
spreadsheet deals with a lesser number of sites (20) scattered on the south-eastern flank
of the Tarhuna region. The increasing importance of GIS for any archaeological survey
has been demonstrated in most survey projects in the Mediterranean during the last
decade (Gillings 2000; Wheatley 1995). GIS has been used to analyse surface trends in
site density patterning across the survey areas (Lock 2006; Wheatley 2002). This
procedure involved comparing the site distribution and density of a particular area with
the densities present in its immediate vicinity, while also considering the effects of
topographic factors. Integrating GIS with the archaeological data to create a spatial
analysis of site distribution across the landscape can be expected to be influenced by
environmental and topographic factors (Roy 1990). As we shall see, the use of the GIS
in this research was useful for revisiting and reassessing specific locations to do further
investigation.
2.4 Wadis Turgut and Doga as a key study
As mentioned above, this chapter will analyse the distribution of Roman period
farming sites using the results of a systematic survey and GIS applications. It formally
47
presents a preliminary investigation of a current research project which aims to examine
the spatial distribution of the archaeological sites on the Tarhuna plateau and to
investigate the main aspects for the rural economy and settlement patterns in this region
of Tripolitania. Since the second half of the twentieth century archaeologists have
attempted to explain the distribution of Gebel Tarhuna farming sites. Goodchild
examined several different categories of site and noted the high density on the plateau
(Goodchild 1976: 72-113). In a more detailed study of early and late Roman period
settlement in the eastern Gebel Tarhuna, Oates (1953-4) presented the distribution of
olive farm buildings in the Fergian area. More recently, it has been suggested that the
Tarhuna plateau was one of the densest area of olive cultivation in the Tripolitanian
landscape and actively produced a large surplus of oil, particularly in bumper years
(Mattingly 1985; 1988a; 1988b;1995).
The selection of the northern area of the Tarhuna plateau as my key case study was
made because it has not previously been studied in details, unlike the Fergian area. The
best known site is the mausoleum of Gasr Doga (Fig. 2.6) (Barth 1857; Haynes 1959).
The small town of Medina Doga was one of the largest sites, located near the headwater
of the Wadi Doga, being established probably at the beginning of the first century AD
and serving as a road station of the road that linked Lepcis Magna with the Gebel
Tarhuna (Goodchild 1976: 75-79).
By focusing on specific areas (Wadis Turgut and Doga), I shall try to explain the
site distribution pattern using two different scales of analysis. The first scale will
address the general distribution of Roman-period archaeological sites within the survey
area; the second scale involves the detailed investigation of the location of individual
sites. This two-scale approach will hopefully provide a clearer interpretation of the
48
decisions made by the owners/occupiers of these sites regarding their exploitation of the
Tarhuna plateau.
The Tarhuna Archaeological Survey largely focused on the catchment areas of
Wadis Turgut and Doga. Those valleys are located in the north-eastern sector of the
Tarhuna plateau, and they run from south to north (Fig. 2.7). The remains of ancient
sites, mainly farming sites and activities representing the Classical period of agricultural
use, are very often still visible on the landscape. These wadis are still in all probability
the richest in terms of the high quality of preservation and the size of Roman ruins
(senams) as described by Cowper more a century ago:
Although it would appear that the series of senams are to be found almost everywhere
within the limits of the country traversed ….Of these districts, the Tarhuna plateau has
perhaps the most numerous remains, but those observed in the Wadis Doga and Terr'gurt
have upon the whole the most remarkable features, and are perhaps the best preserved
(Cowper 1897:133-4).
The great part of this sector comprises a range of hills separated on the north
from the coastal plain by a district of gently undulating - north or north-east -
falling slopes, reaching some 12 to 15 km in width (Cowper 1897). The Wadi
Turgut runs through the extreme north-east part of Gebel Tarhuna, beginning its
natural course at the northern border of modern-day Gsea village (5 km east of
Gasr Ed-Dauun) and gaining in breadth from several main southern tributaries
(Fig. 2.6); it is the principal drainage of the north-eastern sector of Gebel
Tarhuna. Its northern margin is sharply defined by the Gefara plain which reaches
its high elevation of 150 m at the foothills of Gebel Tarhuna. This location marks
49
Figure 2.6: The Gasr Doga mausoleum, the best known site in the Gebel Tarhuna since the nineteenth century (photo D. Mattingly).
the most north-easterly and eastern edges of the Tarhuna region and comprises the zone
in closest proximity to Lepcis Magna (c.40 km). The TAS has surveyed an area of
about 115 km2 in the Wadis Turgut and Doga (Fig. 2.7). The Wadi Turgut covers
approximately 97 km2 of which 66 km2 or 68.5 percent was extensively surveyed,
locating 70 sites. The Wadi Doga covers about 73 km2, of which about 49 km2 or 67
percent was covered and 45 sites identified. The sites within both wadis vary in their
topographic location; some were situated on hilltops, while others were in low-lying
areas, in particular very close to the wadi courses.
50
The Wadi Doga is physically similar to Wadi Turgut, and can be considered as the
best area of the Gebel Tarhuna for the preservation of significant archaeological
remains. The point is clearly stated by Cowper: “although Kasr Doga, at the southern
end of this important wadi, has been visited by Barth, and perhaps others, the fine wadi
itself, with its wonderful remains, have not apparently been hitherto visited by
Europeans” (Cowper 1896). In addition, the Wadi Doga has a particular importance as
it appears to have, in past at least, marked the ancient boundary demarcated c.AD 75
between Lepcis Magna and Oea after a private dispute between themselves had erupted
into open war in AD 69 (Tacitus, Histories,4,50; Mattingly 1995:71, 140-1).
The epigraphic evidence for the boundary line has been previously studied by Di Vita-
Evrard, drawing upon four inscriptions which were found on the Tarhuna plateau
during the last century, comprising two milestones and two boundary inscriptions.
Indeed, the discovery of these inscriptions gave her the opportunity to consider in
concrete terms the extension of the territory of Lepcis Magna towards the west, where it
bordered on that of Oea (Di Vita-Evrard 1979:67-98). The road built by the Proconsul
Aelius Lamia in AD 16 - 17 linked Lepcis Magna with the Gebel Tarhuna, where the
milestone at Lepcis indicates its destination point as 44 miles ‘in Mediterrano’
(Goodchild 1952; Di Vita-Evrard 1979: 89-91; Mattingly 1995). The road end point
seems to have coincided with the south-western boundary of the territory of Lepcis
Magna (Di Vita-Evrard 1979: 89-91; Mattingly 1995: 140; 1996a).
51
Figure 2.7: The surveyed area of the Wadis Turgut and Doga in the Gebel Tarhuna (measured by the Google Earth software).
In agreement with Di Vita-Evrard’s geographical conclusions, the TAS 2007 has
provided further proof of the line taken by the border between the territories of
Lepcitani and Oeenes in the area of Gebel Tarhuna (Fig. 2.8). This boundary line
evidently did not run perfectly straight, but it was likely affected by the topography of
this rugged terrain. The boundary in fact seems to have followed natural features, such
as valleys and ridges, that ran through the area between Medina Doga and Gasr Doga
on the high plateau in the east, and Ain Scersciara to westward (Goodchild 1976: 76-9).
From this point the boundary appears to have been inclined to northeast and ran along
the watercourse of Wadi Doga. Three boundary stones are now known: two reported on
by Di Vita-Evrard and the third found by me (Fig. 2.8). The most southerly of the
extant boundary inscriptions (Ras el-Halga), lies about 1 km north of the projected
location of the 44 Roman mile mark from Lepcis Magna. Further proof came to light by
the discovery of a new inscription at Ras Abadla in the same area (c.3.5 km north-west
52
of Gasr Doga). Although this new inscription is missing a number of lines, it is clearly
based on the same text as the previously known Flavian boundary makers. Like those
texts (from Ras el-Halga and Gasr Masaud), it related to an operation of boundary
marking between the adjoining territories of Lepcitani and Oeenses (Di Vita-Evrard
1979: 77-82).
Describing the above-mentioned inscriptions in a geographical order from south to
north, the Ras el-Halga inscription was discovered in the bed of small ravine (Wadi
Scafell), below the hill of Ras el-Halga where stand the remains of small gasr (15 x
11m with walls built in opus quadrum). The gasr is located beside of a track, 6 km west
of Gasr Doga and about 3 km south of Gasr Bu Tuil. The preserved lines have been
read by Di Vita Evrard as follows:
Ex [auctoritate] /
[I]mp(eratoris) . Ves[pasiani Cae] /
saris . Aug(usti) . p(atris) . p(atriae) . po[nt(ificis) max(imi) trib(unicia)] /
potest(ate) . V . imp(eratoris) . XIII . c[o(n)s(ulis) V desig(nati) VI] /
Q(uintus) . Iulius . Cordinus . [C(aius)? Rutilius Galli] /
cus . leg(atus) . Aug(usti) . pro [pr(aetore) con(n)s(ul) pont(ifex)] /
limitem . inter . Le[pcitanos et Oeen] / ses . derexit /
Lepcitan[i pub(lice)? Pos(uerunt)?]
The second inscription was discovered at a place called Gasr Masaud on the right
bank of the Wadi Msabha (middle of the Wadi Doga) below a small hill. It has been
read by Di Vita Evrard:
53
Figure 2.8: A new boundary inscription fits within the Di Vita Evrard projected by blue line.
Ex au]ctoritate /[Imp(eratoris) Ve]spasiani . /
[Caes]aris . Aug(usti) . p(atris) p(atriae) /
[po]nt(ificis) . max(imi) . trib(unicia) pot(estate) /
[V im]p(eratoris) . XIII . con(n)s(ulis) V design(nati) . VI /
Q Iulius] C[ord]inus Rutilius
Gallicus leg Aug pro pr] cos pont .
Limitem inter Lep]citanos /
[et Oeenses derexit]
(Di Vita-Evrard 1979: 78-9).
Boundary stone of AD 75.
54
As can be seen from the above inscriptions, the boundary operation was performed
under the supervision of a well-known person, the Consular legate Q. Iulius Cordinus
Rutilius Gallicus, acting in accordance with special powers which were delegated to
him by the emperor Vespasian (Legatus Augusti pro praetor), the auctoritas of which
guaranteed the juridical validity of the act (Di Vita-Evrard 1979: 82). Although the new
inscription found in 2007 in Ras Abadla is missing most of the text, with only a few
words in the first two lines preserved, it is clearly another example from the same
series:
Ex auct[oritate] I[mp]
Vespasian[i Caes]
........
The Wadi Doga’s southern margin is located 8 km north-east of the town of
Tarhuna and is oriented south-north, the wadi bottom varying in elevation from 455 m
down to 150 m. It starts from a number of small tributaries close to Medina Doga and
Gasr Doga, in an area rises to over 470 m from the southern margin of the wadi. In
contrast, the watchtower of Gasr Abdalhadi reaches elevations of 500 m from its south-
east tributary of the Hajaj area (Fig 2.10). Excluding the section running through the
Gefara plain, the Wadi Doga runs through the Tarhuna plateau for 20 km. The wadi in
the most of its middle part, from Ras Abadla to the Wadi Msabha, formed a natural
boundary line which was linked with the higher southern area of the wadi.
The diverse topographic setting and alluvial deposits in these two wadis offered a
varied range of natural resources for exploitation. In comparison with the Tripolitanian
pre-desert area, these Wadis of the Gebel Tarhuna are shorter, but they receive much
higher winter rains before they join the Gefara plain. As fertile, well-watered valleys,
55
they sustained a strong agricultural economy and contributed to the flourishing of the
coastal centres during the Classical era. In modern times, they still provide a good
example of regional agricultural potential and practices because they contain arable
land and have sufficient rainfall to allow dry farming not only for olives but also for
large scale arboriculture of almonds, figs, and other fruits as well as grapes and cereals.
Most of these agricultural products, nowadays, are used for domestic consumption and
local markets.
2.5 Distribution of sites in the landscape
It should be clear from the aims outlined earlier that the essential requirement of the
survey was the collection of detailed information about the location of ancient
economic activity and rural settlement patterns on this hinterland plateau. The study
takes account of existing data from earlier archaeological works, literary evidence and
travellers’ accounts. The TAS has identified that there was a bimodal pattern at the
regional scale, characterised by the existence of both small farm sites with one or two
olive press elements, and of much larger sites with larger number of presses (from three
to seventeen) and usually evidence of long periods of exploitation from late Punic or
early Imperial times to Late Antiquity (Cowper 1897; Oates 1953; 1954; Goodchild
1975; Mattingly 1985; 1988b; 1988c; 1995). The remains of rural settlement on the
Tarhuna plateau comprise a rather wide diversity of phenomena, including sites of open
farm building (oileries, oilery villas, villas with indicators of luxurious elements, other
farms and farmsteads), baths, fortified farm structures, cisterns, wells, dams and terrace
walls (Fig. 3.42). The key question to be addressed here is whether the diversity and site
distribution across the landscape were the result of natural or socio- economic processes
that differentially affected the landscape? It is possible to assess whether the spatial
56
distribution of settlements and their associated field systems followed natural divisions
in the landscape such as wadis, plateaus, slopes and hills. If, for example, the local
ancient economy primarily relied on agricultural production, it is reasonable to suppose
that the actual choice for site location was determined by the availability for suitable
land and management of water for agriculture (Shaw 1984).
2.5.1 Site elevation
In order to investigate the distribution of settlement sites recorded in the surveyed
area, I will analyse their locations in terms of elevation and topography. The locational
characteristics of sites involve a more subjective assessment. The landscape within the
Wadis Turgut and Doga is variable and important local natural features influenced the
choice of particular positions for settlement. The discovered sites in this area show that
there was a clear preference for certain favoured places in the landscape, either at the
wadi side, at the valley-side, break of slope or for dominant positions on hill-tops,
where sites, especially for fortified ones and watchtowers, overlook the surrounding
landscape. Figure 2.9 clearly demonstrates that great numbers of sites were located by
the wadi side or on adjoining slopes. This is most likely due to the obtainable alluvial
fertile soil, access to the seasonal rainfall (via capture in cisterns and wells) and the use
of wadi-courses as routes to communicate, in different directions, with other areas.
Site elevation was considered an important variable since it might be indicative of
patterned subsistence strategies within settlement systems. These patterns might, in
turn, be associated with political and socio-economic changes during late antiquity.
El
tri
lo
de
el
so
Tu
el
El
th
25
sit
levations w
ibutary drai
ocation to b
eeply carve
levation. Th
outhern part
urgut that e
levations ra
levations w
The gener
hat 33 perce
50 and 300
tes located
Fi
were initially
inages wher
be detected.
d in the nor
hus, one wo
t of these w
exhibits the
anging from
were grouped
ral distribut
ent of sites
m.asl, of w
above 500
0
5
10
15
20
25
30
Wafloo
igure 2.9: Site
y examined
re relief wa
. In general
rth than in
ould expect
wadi draina
most bimod
m 135 m to
d into four a
tion of site
were locat
which 86 per
m asl. are
di or
Wadi side
e location in re
for Wadis T
as sufficient
l, these Wa
south wher
more altitud
ages. In fac
dal site distr
o slightly m
altitudinal b
locations in
ted in a rel
rcent came f
relatively
Break of slope
elation to the
Turgut and
tly marked t
adis and th
re the Tarhu
dinal differe
t, however,
ributions (F
more than
bands (Table
n Wadis Tu
atively narr
from the W
rare in the
Plateau Hi
topography.
Doga, and t
to allow for
eir tributari
una plateau
entiation in
, it is the m
Fig. 2.10). T
515 m abo
e 2.1).
urgut and D
row elevati
Wadi Turgut.
survey area
ll‐foot Hill‐t
then within
r difference
ies are muc
reaches its
n site locatio
middle of th
The survey i
ove sea lev
Doga area i
onal band b
On the oth
a. Equally,
top
57
specific
es in site
ch more
s highest
on in the
he Wadi
included
vel (asl).
ndicates
between
her hand,
there is
58
only one site located below 150 m asl (Senam Semana 135 m asl), and this exception
relates to a location below the extreme northern edge of Gebel Tarhuna at the point
where it meets with the Gefara plain.
Elevation level Number of sites W. Turgut W. Doga Elev. Average
135 - 249 m 34 30% 22 12 212.5 m
250 - 299 m 36 32% 31 5 272.5 m
300 - 399 m 20 19% 14 6 347 m
400- 515 m 20 19% 6 14 442 m
Table 2.1: Site location in relation to the elevation level. Key here is that the densest settlement is in lower reaches of the Wadis, closest to the most fertile land and the communication routes that run through wadis.
The main focus here will be to the spatial analysis of the Roman period settlement
data assembled by the TAS in the study area. The first section examines the distribution
pattern of the different types of site. The second section analyses the location of a
number of selected sites in order to understand the factors behind the location choice for
different types of settlement and the apparent preference by the owners for certain
locations.
59
400 – 515m (18%) 300 – 399m (18%) 250 – 299m (32%) 135 – 249m (30%)
¹
Ancient routes
Figure 2.10: Site location and elevation in the north‐eastern sector of the Gebel Tarhuna.
2.5.2 Distribution patterns of site types
The Tarhuna archaeological survey has provisionally identified two main types of
rural settlement on Gebel Tarhuna; the first type is characterised by an agricultural
character and function with marked tendency to specialisation in oleoculture and olive
oil production. The farming sites include small villages, oileries, large and small farms
and fortified farmhouses (gsur). The second type includes a range of non-agricultural
sites which nevertheless often appear to have a kind of relationship with the agricultural
activity, for example, dams, wells, cisterns, watchtowers, mausolea and amphorae kilns.
Site.
60
Although the TAS has applied both intensive and extensive survey methods, the
largest part of the surveyed area was covered by the extensive survey. Sites have been
assigned to types according to the nature of the archaeological material observed on the
surface of the surveyed area. Each site was defined by the extent of archaeological
material and was assigned a single survey number by the TAS. Some numbers
comprised more than one type of feature. For example, the site of Sidi Eysawi
(TUT53), which was a large farm-villa, containing three olive oil presses, was
associated with a pottery kiln, a bath-building, a cistern, an enclosure and quarry (see
Chapter 3).
The specific criteria applying to the descriptive terms currently used to describe the
recorded archaeological sites of the Tarhuna plateau need to be clearly defined and
clarified to avoid later confusion and misunderstanding. Previous studies of the rural
settlements of Gebel Tarhuna have succeeded in providing some interesting insights
about distribution and typology of farming sites in this hinterland region of Roman
Tripolitania (Cowper 1897; Goodchild 1976; Mattingly 1987, 1988a, 1988b, 1995;
Oates 1953, 1954). However, the vast majority of Gebel Tarhuna farming sites have
been described as utilitarian sites constructed in most cases for olive oil production,
which is clearly seen in the press elements associated with them. On the other hand,
most previous typological study has focused almost exclusively on olive oil production
facilities and on press elements as the key characteristics of the sites.
Agriculture was the main basis of the Roman economy and contributed significantly
to the development and maintenance of the Empire through a closely promoted
relationship between urban centres and the rural communities in their hinterlands
(Erdkamp 1999; Greene 1986). A major factor in the process of settlement and
61
agricultural development of the Gebel Tarhuna area seems to have been the input in this
hinterland zone of the expanding wealth of the coastal centres, and of the desire of the
urban elite to invest in and profit from agricultural products (Mattingly 1995).
2.6 Site typology
The overall map of Roman period settlement in the study area is heavily influenced
by where detailed work has been carried out (Fig. 3.43). The Gebel Tarhuna is still far
from totally surveyed. The researches of Cowper, Oates and Goodchild were focused
more on the eastern area of the Tarhuna plateau, and as a result the blank areas of the
map cannot be assumed to be devoid of ancient sites. The apparent absence of recorded
sites is more likely a result of the lack of fieldwork in these sectors. The next section
will describe the principal types of site recognised and present representative examples.
2.6.1 Villages (small towns)
There are two previously known small towns (or large nucleated settlements) in the
Gebel Tarhuna: Gasr Ed-Dauun (Subututtu) and Medina Doga (Mesphe). According to
Oates, Gasr Ed-Dauun functioned as a market village for the surrounding regions such
as Fergian, Gsea and Turgut. At the same time it was the focus of their communications
via a number of tracks in addition to the main Roman road (the Gebel road) that linked
Lepcis Magna with the Tarhuna plateau (Oates 1953: 89-90). The traced remains
indicate that the village structures seem to have been concentrated in ribbon-like
fashion along the road line. Figure 2.11 illustrates the distribution of buildings at Gasr
Ed-Dauun forming a straggling line of structures alongside the main road, though these
remains have been largely demolished by fifteen centuries of erosion and of almost
continuous settlement around the wells here (Oates 1953: 90-1).
62
The second nucleated settlement is called Medina Doga (Mesphe) and was located at
the head of the Wadi Doga at an important meeting-point of five tracks used in the
Roman period. Some of these tracks can still be traced on the Google Earth imagery
Figure 2.11: The Gasr Ed‐Dauun village (from Oates 1953: 85).
(Fig. 2.12); one comes from the east and crosses the town towards probably Ain
Scerscirara (ancient Cercar ?). Another track seems to have led to the north through the
Wadi Doga. Two other tracks traceable on the image run to south south-west, which
perhaps were linked with Ain Wif (Thenadassa) and the pre-desert zone. At
approximately 130 ha it was one of the largest sites in the Gebel Tarhuna (Goodchild
1976: 76). Although the village lies inconspicuously among the modern olive
plantations, Goodchild, in 1949, was able to identify the limits of the site and the
63
different building types which included (Fig. 2.13): a small mausoleum (A), a
colonnaded building with limestone columns of 60 cm. diameter (C), two bath buildings
(D) and (E), another colonnaded building (F), a large enclosure (G), a fortified
farmhouse and necropolis (Goodchild 1976: 78). He believed that Medina Doga, which
Figure 2.12: Location of the Medina Doga, the central meeting of 5 ancient tracks. The polygon shows the approximate limits of surface evidence of ancient structures and pottery spread. (Google Earth background).
he identified as the Mesphe of the Antonine Itinerary, was gradually developed from the
early first century AD and might have functioned as local administrative centre of the
Tarhuna plateau during the Roman period (Goodchild 1976: 78-9).
Two new sites located in the Wadis Turgut and Doga can be added to the nucleated
settlements, though these appear to be agricultural villages. Astail village in the Wadi
Turgut and Gasr Dehmesh village in the Wadi Doga (see Ch. 3) probably functioned as
‘estate villages’ and thus were different to the road stations (small towns) discussed
64
above. Their location seems to be highly related to the availability of land under
cultivation and water resources.
Figure 2.13: Plan of Medina Doga (from Goodchild 1976: 77).
Italian archaeologists working in a 20 km2 area near the Villa Silin have also
observed the phenomenon of small villages in the coastal hinterland of Lepcis Magna
during the Roman period (Munzi 2004). In considering the location of small
agricultural villages of Gebel Tarhuna, two factors seems to have primarily determined
their location. On the one hand, communication links no doubt contributed not only to
their chosen position but also to their development through the Classical period. On the
other hand, these sites tend to coincide with the existence of major water resources, in
particular springs.
65
2.6.2 Oilery farms
Site definition within the TAS depends in part on the overall size of a settlement, in
part on visible functional features and thus, in part on the degree of site preservation.
The term oilery is applied to describe the largest physical traces of sites linked to olive
oil production and includes sites with five or more presses. It can be an agricultural
settlement comprising a complex of standing structures and usually using opus
africanum construction. This is a method of constructing walls (already utilised by the
Phoenicians) in which piers of ashlar blocks were set up at intervals from one another
and the gaps filled in with undressed stone, concrete or other material. It was
technically adopted by the Romans and became exclusively used in their construction
schemes, especially in North Africa (Adam 1999). In Tripolitania, in particular during
the first two centuries AD, it was the characteristic construction method on rural sites
and its use expanded even into the pre-desert area (Barker 2000; Mattingly 1996c).
Most of the oilery sites are characterised by the existence of a large courtyard which is
surrounded by a number of press rooms marked by stone monolithic and other pressing
facilities such as oil settling tanks and millstones. Similar features occur at other
locations, but where sites had less than five certain presses they have been classified
separately as large or small (olive) farms.
The TAS has recorded 16 sites in Wadis Turgut and Doga which could be identified
as oileries (essentially factory farms); 10 sites are distributed along the Wadi Turgut,
whilst the rest (six sites) are concentrated in the northern half of the Wadi Doga (Fig.
2.14). This map indicates that the majority were located very close to the main course
of the valley and at lower altitudes on its side slopes. Large-scale production sites
commonly employed five or six presses, but larger numbers are known in a few
instances; one site (TUT54) included 17 presses in a single building (Cowper 1897:
27
hi
de
th
Li
O
pr
in
th
th
th
32.5
7633
6
Figur
79-82, no.5
igh degree
emonstrates
he region f
ibyphoenice
ates 1953:
roperties. H
ndicators of
he wealth of
he rural settl
he productio
0
#
0
re 2.14: Distrib
7, Senam S
of investm
s that there
for an exp
es elite, bas
112), who
However, o
luxury life
f these settle
lement hiera
on capacity
42
#
#
##
#
2
WadiD
ogabution of oile
Semana). Th
ment in cult
was a large
ort market
sed in the m
o probably
only eight o
(e.g. mosai
ements and
archy durin
y of oileries
8Km
#
#
#
13.754246
13.754246
4
ry sites in Wa
he distribut
tivation of
e-scale expl
t, which m
major coasta
spent som
of the oile
ic, wall pain
their impor
ng the early
s with luxu
#
#
#
#
8Km
Wadi Turgut
dis Turgut and
tion of thes
olive trees
loitation of
most likely
al cities (M
me time of
ries were a
nting, portic
rtance as the
imperial pe
ury element
##
d Doga.
se rural oil
s and produ
f the oil-pro
was on b
Mattingly 19
f the year
associated
co and bath-
e upper eco
eriod. These
ts, are defin
##
factories re
uction of o
oducing pote
ehalf of th
987: 34; 199
in their hi
with one o
-building), a
nomic level
e sites that c
ned here as
µ
##
66
eflects a
oil. This
ential of
he local
95: 140;
nterland
or more
attesting
l sites in
combine
s ‘oilery
13.975084
13.975084
32.5
7633
6
67
villas’. The rest appear to have been purely functional facilities for the bulk production
of olive oil – presumably within large estates.
2.6.3 Large farms
This term is used to describe large farm sites, containing three or four presses,
comprising a courtyard surrounded by five or more rooms and stores showing signs of
having served as centres of fairly substantial estates for production of olive oil (cf.
Hitchner 1988). Their distribution is generally the same as that for oilery farms, that is,
close to the wadi or on its main tributaries. The distinction between oilery and large
farm may be a fine one, determined by our ability to identify positively the exact
number of presses present. Some large farms may in fact originally have possessed five
or more presses and would fall into the oilery class. Three out of 17 large farms have
shown certain characteristics of the rural luxurious villa and are considered in this work
as ‘farm villas’. Other sites could be described as utilitarian villas (as Percival 1976
did). The majority of large farms are located in the Wadi Turgut (Fig. 2.15), though
some scholars have demonstrated that the vast majority of the Gebel Tarhuna villas are
utilitarian ones and have not provided, from the surface, any of evidence luxurious
materials (Mattingly 1987: 37; 1995: 141) with the exception of the Ain Scersciara villa
(Goodchild 1951). The luxurious indicators (e.g. mosaic tesserae, portico elements, and
bath-buildings) found at a number of sites during the TAS, demonstrate that the Ain
Scersciara villa was not a unique indication for luxury (see Chapter 3). These large
farms, like the oileries, were probably central facilities within rural estates owned by
the elite families of Lepcis Magna and Oea. The generally utilitarian character of many
villas may support the view that the Lepcitanian and Oean elites had multiple rural
estates, but only erected rural residential units for themselves at some locations.
Ex
an
ya
th
op
ev
nu
Tw
32.5
7633
6
xcavation c
nd oilery sit
2.6.4 Sma
These are
ard, with ev
he larger far
pus african
vidently det
umber occu
wenty smal
0
#
0
could yield
tes.
Figure 2.15: D
all farms
defined as
vidence of 1
rms (see Ch
num constru
termined in
upied the wa
ll farms hav
42
#
#
42
WadiD
oga
additional e
Distribution of
sites of sma
1 – 2 presse
hapter 3), b
uction. The
n relation to
adi-edge foo
ve been rec
8Km
#
#
#
#
13.754246
13.754246
evidence of
f large farms i
all agricultu
s at most. T
but on a red
spatial dis
o several ty
othills or the
corded by t
#
##
#
8Km
Wadi Turgut
f a luxury c
in Wadis Turg
ural settleme
The architec
duced scale
stribution o
ypes of lan
e margins o
the TAS in
##
omponent a
gut and Doga.
ents compri
cture of thes
. They wer
of this type
ndscape; in
of the alluvia
the survey
#
at other larg
ising a cour
se sites is si
re normally
of settlem
particular,
al soil of th
y area of the
µ
#
#
#
#
68
ge farms
rtyard or
imilar to
built in
ment was
a large
he wadis.
e Wadis
#
13.975084
13.975084
32.5
7633
6
69
Turgut and Doga, in addition to approximately 55 previously recorded by Cowper and
Oates (Cowper 1897; Oates 1953). The map displays a remarkable density of these
small farms on the eastern sector of the Gebel Tarhuna (Fig. 2.16). However,
approximately half of the total number of them is concentrated in the area around the
Ed-Dauun village. On the other hand, it appears that the area of the Turgut and Doga
valleys had a lower density of small farm sites. It could be suggested here that these
areas were dominated to a greater extent by the presence of oileries and large farms
rather than small farms. For example, 23 sites which can be identified as oileries and
large farms have been recorded in the Wadi Turgut, while only nine small farms are
recorded in the same wadi. However, six out of these are concentrated in the middle
sector of the wadi (Fig. 2.17).
Figure 2.16: Distribution of the recorded small farms on the eastern Tarhuna plateau.
0 0.75 1.50.375Km
70
Figure 2.17: The middle sector of the Wadi Turgut.
2.6.5 Fortified farmhouses
This definition is applied to farm structures which were given a fortified appearance
usually by being surrounded by broad ditches or with high walls enclosing the main
building. The ditches were, in most cases, of square or rectangular shape and they
generally enclosed the most defensible position at the site location. Goodchild found
that the great majority of the ancient sites examined in his review on the Tarhuna
plateau were encircled by a broad ditch (Figure 2.18), and they were easily identifiable
on air photographs (Goodchild 1976: 88-9). In agreement with Goodchild’s point of
view that these sites were widely distributed on the Tarhuna region, my survey by
Google Earth program (Fig. 2.19) has shown that there is dense distribution of defended
farmhouses in the districts which are covered by high resolution imagery.
±
71
Figure 2.18: Some of the fortified farms identified on Google Earth imagery of the Gebel Tarhuna.
Distribution of these fortified farms seems to be characterised by two facts: first,
they are mostly located north of the east-west Gebel road; second, they are also more
concentrated in the area south-west of Tarhuna close to the Thenadassa (Ain Wif) road
station (Mattingly 1982), and in the north-eastern district of the Tarhuna plateau.
According to the works conducted by Oates (1953; 1954) and Munzi et al. (2004) in the
areas around Gasr Ed-Dauun and villa Silin respectively, these constructions can be
dated to the fourth and fifth centuries AD, the date also confirmed by the survey
conducted recently in the territory of Lepcis Magna (Wadi Caam-Taraglat), which
indicated that the phenomenon of fortification became more pronounced during the
fourth and fifth centuries AD (Felici et al. 2006). In contrast with these areas, the
changeover from open to fortified farms in the Tripolitanian pre-desert area to the south
occurred more gradually, starting around the end of second century and continuing into
the fourth and fifth centuries and later (Barker 1984; Jones 1985).
TUT13
TUT28 TUT55
TUT17
TUT51
TUT58
72
Figure 2.19: Distribution of defended farms recorded from the high resolution imagery of the Google Earth.
If we judge by the surface evidence alone, we would have to conclude that although
all of the Gebel Tarhuna fortified farmhouses (gsur) seem to have been created in the
later imperial period (fourth and fifth centuries AD), it has now become clear through
the work of the survey that the individual buildings consist of two types. Type 1 sites
seems to have developed from earlier period open settlements, being set on the ends of
spurs overlooking the surrounding area, and very often were constructed by reusing the
masonry materials taken from the pre-existing open farms. The reuse of earlier
materials has been noted also by Oates especially in the upper reaches of the Wadi
Turgut (site 13, Gasr Shāeir). He described the site as originally comprising a large
farm (c.70 x 40 m), with 4 presses, built in ashlar masonry. This was later replaced by a
defended farm superimposed over a great part of the original area (Fig. 2.20) and built
by reusing the original ashlar blocks. (Oates 1953: 105-7).
73
Figure 2.20: The Gasr Shāeir (from Oates 1953: 106).
The TAS found 11 sites of opus africanum construction, ranking from small farm to
oilery in the Wadis Turgut and Doga which were certainly replaced by or partially
incorporated into ditched farms (Fig. 2.21). The second type shows a considerable
decline in the techniques of fine stone-dressing and of mixing strong and durable
concrete with rubble or roughly squared small blocks (Mattingly 1996b).
Type 2 gsur were sited to take maximum benefit of natural topographic features,
generally being constructed on hill-tops and on the highest pieces of land, natural
defences being augmented with broad surrounding ditches. The question that arises here
is why some gsur were built on these carefully selected positions? It could be suggested
that they were built simply for defensive purposes – that is, in order to protect the
dwellings and their estates. The epigraphic evidence, especially from the pre-desert
zone, suggests that the most important goal of the gsur-builders was to protect their
Late antique fortified site
Traces of early oilery villa site
ow
hi
th
ov
to
str
po
(G
Tr
of
32.5
7633
6
Figu
wn families
ighest piece
he hypothes
verlooking
o the defens
ructures co
ower and w
The gasr
Gebel and
ripolitanian
ften built at
’4
’4
ure 2.21: Loca
s and estat
es of land in
sis that the
communica
sive purpos
uld have fu
ealth within
is the mo
pre-desert)
n gsur are k
the expens
’4
’4
’4
0
ation of 11 op
es (Matting
n close pro
ese defende
ation pathw
se an ideolo
unctioned a
n the society
st typical a
) during t
known as sm
e of previou
0
’4
’4
13 .754246
13 .754246
52.5
en farms repl
gly 1995: 1
oximately to
ed sites m
ays along th
ogical interp
s an elite p
y (Mattingly
archaeologi
the Late R
mall fortific
us open con
425
aced by fortif
103). Furth
o the wadi b
might have
he wadi tra
pretation; h
prestige indi
y 1996a: 32
ical sign of
Roman and
cations, som
nstructions (
8Km
’4
10Km
fied ones in th
hermore, the
bed (Fig. 2
functioned
ackways. M
he suggests
icator, and
28-31).
f the Tripo
d post-Rom
metimes fitte
(Brogan 197
’4
’4
he late Roman
eir location
.22) could b
as contro
Mattingly ha
that these
also as sym
olitanian hi
man period
ed with a d
76-77). The
µ
’4’4
74
n period.
n on the
bear out
l points
as added
fortified
mbols of
nterland
ds. The
ditch and
e typical
µ
’4
’4
’4
13
13
.975084
.975084
32.5
7633
6
75
0 Figure 2.22: An example from the middle of Turgut shows how some of the hill‐top gsur here built close to the wadi bed.
fortified farm (gasr) was a tall square, or nearly square, structure with a single entrance
leading into an internal courtyard, onto which faced two or three stories of rooms. In the
pre-desert area, they are usually to be seen along the wadis at intervals of a kilometer or
more, but a number of gsur may be grouped together at the convergence of two or three
tributaries where cultivable land is available (Di Vita 1964; Goodchild 1950; Mattingly
1996c). Many gsur in the pre-desert area have yielded evidence to indicate that the
process of constructing and maintaining this type of settlement continued into the
Islamic period (Barker et al. 1996). There has been much discussion of the question of
whether these gsur were built by official Roman action or whether they were primarily
of indigenous origins. Goodchild argued that the first or earliest gsur were built and
designed by Roman architects for military purpose, while the later gsur were
constructed by local peoples (Goodchild 1976: 30). However, the inscriptions found in
the farms associated with the gsur demonstrate that most were the work of indigenous
76
people (Elmayer 1983, 1984). Modern studies favour the view that these sites were the
standard form of native farm, rather than a sign of paramilitary organisation of rural
society. However, they may indicate that rural conditions were less secure at this date.
Turning to the distribution of fortified farmhouses in my case study area, Figures
2.19, 2.23, and 2.24 show how these sites were distributed in relation to topographic
features. During the fieldwork a number of the Type 2 hill-top gsur were documented.
Most of them were dated to the Late Roman and Byzantine period. They have been
dated in a preliminary way on the basis of the pottery sherds collected in the field. It is
worth noting, here, that many of these hill-top sites were previously unknown, and
within the TAS a total of 17 ditched hill-top sites was recorded, five of them also
having evidence of early Islamic occupation.
In favour of a defensive interpretation of gsur one can see from the overall
distribution of the Type 2 ditched hill-top sites that they too probably had a strong
relationship with pre-existing open farms/estates, which were already located close by –
often at the hillfoot. It appears from the surface evidence that the Type 2 gsur were
mostly created in the Late Roman period. The TAS identified that some hill-top sites
were established on the top of a hill very close to the open farm building (Fig. 2.25),
suggesting that they continued to protect the people and the cultivated land surrounding
the gsur.
Fion
igure 2.23: Chn a hill‐top (W
Figure 2.24:
0
1
2
3
4
5
6
hart showing tWadis Turgut a
Location of t
Plateau
the location ond Doga).
he Type 2 fort
Preak
f Type 1 (open
tified farmhou
k of slope
n/fortified site
uses.
Hill‐top
es), where on
nly one site is l
77
located
78
0
1
2
3
4
5
6
7
< 100m 101‐200m 201‐300m 301‐500m > 500m
0 Figure 2.25: The distance between hill‐top gsur and other farming site. The interesting thing is c. 66% of Type 2 gsur were located within 300m of earlier farm sites.
2.6.6 Dams, Cisterns and Wells
The vast majority of archaeological sites, recorded during the TAS, are associated
with one or more types of water management works (dams, terrace walls, and cisterns
or wells). Cisterns were more common in the surveyed area, and rarely, did settlements
stand isolated from works for water control and supply, in particular with the case of
farming sites such as oileries, large farms and farmsteads. Groundwater wells are less
common than cisterns fed by runoff rain. The former type is found in a few places, such
as in two tributaries of the Wadi Turgut (Guman and Astail), in the upper sector of the
Wadi Doga (below Gasr Doga), in the Wadi Twafga (2 km south of Khadra church),
and the well-known example in Ed-Dauun village (Brehony 1960). These wells were
usually associated with bath-buildings or pottery kilns. Construction of cisterns, wells
and spring catchments illustrate the needs of the ancient community to exploit rainfall
and groundwater facilities for their benefit.
no. of sites
79
Cisterns and wells play an important role in a dry and thirsty land (Graham 1971).
These water reservoirs are associated with the vast majority of the archaeological sites
on the Tarhuna plateau, especially within sites characterised by agricultural or industrial
activities. The Roman period cisterns lined with waterproof cement (tebshemet) are a
remarkable feature of the Tarhuna landscape. No less than 85 sites, out of 111 sites
recorded in the Wadis Turgut and Doga, had visible traces of cisterns and wells. They
indicate that the process of water management and control was very significant in such
an environment. Undoubtedly many further examples are buried underneath the ruins
and soil.
There are two main types of cisterns in the Tarhuna plateau: basins (feskyah) and
deep rock-cut shaft cisterns (majel or majen). The rock-cut type is more abundant than
the basin cisterns in the study area, and some of them continued to serve the nomads
and farmers until the last few decades. The same phenomenon of continuing use of
some cisterns has been observed in the Tripolitanian pre-desert (Reddé 1985). For
instance, Michel Reddé mentioned that many ancient cisterns in Wadi Tlal were re-
lined with cement by the Italians or during recent agricultural works, which
unfortunately, for the archaeologists, does not always allow their initial use to be dated
with certainty (Reddé 1985: 175).
Similar types have been recorded by the UNESCO Libyan Valleys Survey in the pre-
desert zone (Mattingly 1996d: 134). In comparison with the pre-desert area, examples
of the basin type of cisterns found during the TAS were generally smaller in size and/or
lesser capacity: the largest recorded single cistern (TUT43) does not exceed 22 by 4 by
4.5 m deep (about 400 m3). Furthermore, the cisterns on the Tarhuna plateau are always
located close to the buildings, especially on the slope just below the main structure, in
80
contrast with the pre-desert area where they are usually located at the edge of the wadi
floor in the vicinity of ancient settlements (Mattingly 1996d: 134; Reddé 1985). Their
location in Tarhuna could indicate that they were established to collect rainwater from
building-roofs rather than from surface runoff water. The majen type is the most
common type in the study area, with about 70 cisterns of this type recorded and most of
them between 4 to 5 m deep, with 2 or 3 radiating tunnels at the bases of the shaft. They
were filled by run-off water and very often occupied slopes in the vicinity of
settlements (Fig. 2.26).
Dams were commonly distributed along the wadis systems of Roman
Tripolitania (Munzi et al. 2004; Vita-Finzi 1961), though only a small number of dams
established on the Gebel Tarhuna have received the attention they deserve. To elucidate
this point, Oates (1953) mentioned that there were about sixty of these structures of
varying size in Udei el-Me and its tributaries alone (in the vicinity of Gasr Ed-Dauun).
However, he described in details only two barrages, one in Udei el-Me and the second
in the Wadi Turgut. The distribution of these water management works in this arid area
indicates that they were very important and valuable for hydraulic systems designed to
improve soil fertility and agricultural production (Oates 1953: 87-9). The archeological
evidence evidently shows that dams of the Gebel Tarhuna were built as an enhanced
system of soil and water conservation.
Remains of dams are a more common feature in the Wadi Turgut and its tributaries
(Fig. 2.27) than in the Wadi Doga. For example ten dams have been recorded on the
Wadi Turgut, whereas only two dams have been observed on the Wadi Doga. It is
possible that the Doga watercourses were criss-crossed by earth-dams which have been
washed away by water pouring over them. It could be readily assumed perhaps that the
81
rural farming of the hinterland of Lepcis Magna introduced a level of homogeneity into
systems of land organisation and exploitation. The creation of varied sizes of concrete
and earth-dams across wadis was essential to the success of farming - not simply the big
dams on the larger wadis such as the Wadi Taraglat-Qaam. The dams, here, built on a
local system of land-use (Vita-Finzi 1961:15). It is also evident that many of the large
and minor settlements on the Tarhuna plateau, as the Tripolitanian pre-desert, were
situated on or very near wadis (Fig. 2.29) in order to capture the runoff water and to
exploit the fertile soil in the wadis above the dams (Barker et al. 1996:159- 90;
Mattingly 1996c: 170- 71). Thus by maximizing the potential rainfall through run-off
technology, the farmers achieved the highest degree of land exploitation, and benefitted
both from the water catchment potential of the wadis and from the natural topography.
Figure 2.26: The oilery farm of Loud Meghara (TUT43) as an example showing how cisterns were associated with settlements. C= cistern.
The stone and concrete dams were erected in different situations in the wadis. Where
the watercourses were narrow and steep (which is very common in the northern wadis
82
of the Gebel Tarhuna such as Udei el-Me and Turgut) high mortared stone dams were
needed (Fig. 2.27); otherwise the wider earth and concrete dykes were constructed on
the larger southern wadis such Tareglat where the biggest expanses of cultivable area
are found (Vita-Finzi 1961). The apparent discrepancy between the distribution of
different sorts of dams may be explained partly by the spatial development of the rural
settlements, with the focus of their organisation and land-use. Furthermore, dams were
built as a principal means of slowing down the floodwater and of controlling its
capacity for destruction, while retaining as much water as possible to feed the fertile
soil created upstream from them (Oates 1953:88).
Figure 2.27: Dams in the Wadi Turgut.
83
Figure 2.28: A dam in the Wadi Turgut, (TUT24).
As discussed earlier, the differences between the sizes and types of dams found on
different areas of the Tarhuna plateau may have been influenced by the type of
topography and how it was suitable for exploitation and settlement. The type of dam
was evidently adapted in relation to varying circumstances (Oates 1953). The chosen
type was likely influenced by varying locational characteristics. The distribution of
dams in the Wadi Turgut for instance, shows that dams were mostly constructed in the
tributaries rather than in the main wadi course (Fig. 2.27). It seems that the object was
to minimize the risks of most essential the flash floods, such as may have occurred
more commonly in the main valley, and their possible destruction of the dams. For
example, the dam (GUM84) was established in the Wadi Guman (one of largest
tributaries of the Wadi Turgut), a short distance below a bath-house (GUM87, Figure
3.41). Its overall length is 50 m and height 3.50 m; the thickness at the base is 4 m, but
it decreases to 3.20 m at the top. This dam seems never to have increased above its
84
Figure 2.29: DOG 111, large farm associated with cisterns (C), dam (D), kiln (K) and Tank (T).
original height, which makes it different from a number of dams located in Udei el-Me
and Wadi Turgut examined by David Oates (Oates 1953: 87-89). However, most of the
design of the examined dams shows that they were constructed in order to control run-
off water rather than for long-term water storage. It is evident that the erosive force of
the water was always a considerable danger. Whether these dams were built under the
supervision of the central authority or under the initiative of a local community, their
location and distribution show that must have been very valuable in protecting
farmlands, increasing soil fertility, and supplying water.
Settlements in such a semi-arid area were under continuing pressure to collect runoff
water for domestic use or for watering the crops by devising means to capture and
divert water (Kennedy 1995). These varying systems of water harvesting and runoff
capture techniques most likely required a considerable labour and effort to construct
T
D
85
and maintain. The UNESCO Libyan Valley Survey found substantial numbers of dry-
stone walls built in most of the pre-desert wadis. The purpose of the vast majority of
walls was intended to catch, conduct and accumulate soil and floodwaters (Gilbertson
1996). Twenty five types of ancient walls and wall junctions, in the Tripolitanian pre-
desert, have been classified by the ULVS investigators (Gilbertson 1984). A number of
the Wadi Taraglat’s tributaries (e.g. Agoubia, Tahwalat and Beni Mousa wedian) in the
southern districts of the Tarhuna plateau seem to parallel the pre-desert valleys by using
different types of wadi walls and floodwater farming system. One of the most
remarkable systems of floodwater systems I found in the southern of the Tarhuna
plateau, is located in the upper Beni Mousa tributary (Fig. 2.5). Here I recovered a
concentration of types 11, 12, and 19 of wadi-floor walls (Gilbertson 1996) in the
vicinity of an ancient settlement. Not only was the settlement of the Wadi Beni Mousa
located on the wadi, but the vast majority of south Tarhuna remains are found on or
near wadis. This phenomenon ‘demonstrates that the water catchment potential of the
wadis was the primary factor affecting settlement location’ (Mattingly 1996d: 171). In
contrast with the southern Tarhuna plateau wadis, a smaller number of wadi walls have
been found in the Wadis Turgut and Doga. The Wadi Guman presents a good example
of five cross-wadi walls (Type 11), two barrages (Type 12) and two rectangular
complete enclosures (Type 19); they were all created within a 1 km length. The absence
of obvious chronological evidence (literary or epigraphic) means we cannot hope to
provide a certain date. Nonetheless, these constructions could plausibly be dated to the
fourth and fifth centuries AD in relation to the two nearby fortified farmhouses.
86
Chapter 3: Ancient rural settlement on the Tarhuna plateau
3.1. Introduction There are few satisfactorily excavated Roman-period rural sites in the hinterland of
the Tripolitanian coastal centres. There is a similar shortage of systematic
archaeological surveys conducted in the Gebel Tarhuna. Nevertheless, this area appears
to show a remarkable diversity of settlement type across space and time: there are
varying sizes of open and fortified farm-buildings, pottery kilns, baths, mausolea,
watchtowers and water management works. In considering this differentiation, for site
size and type, it is evident that the Roman settlement pattern of the Tarhuna plateau was
a response to the high demand for settlement construction and land exploitation. The
archaeological evidence shows that this demand reached its peak in the period between
the first - third centuries AD, though with quite a large number of fortified farmhouses
(gsur) also producing evidence for their continuous function up to at least the sixth
century AD (Brogan 1976-77; Felici et al. 2006; Goodchild 1951; Mattingly 1983,
1995).
In the surveyed area, there are several sites of different function that certainly
continued into the late Roman period. However, it appears that sites of the early periods
were more numerous and much more widely scattered across the landscape. The
economic structure of the Tarhuna countryside can be investigated in several ways from
the evidence of rural settlements. An important distinction concerns the scale of sites –
that is, whether all buildings or production sites are of the same order of size, or whether
some types (or perhaps just one type) seem larger than the others.
In the Roman economy, agricultural investments became increasingly more
necessary and more profitable, in parallel with a more intensive exploitation of land and
87
a growing demand for agricultural products by the Roman State and urban markets
(Matijašić 1982). Conditions for growing olives in North-West Libya are most
favourable in the hill lands of the Gebel Tarhuna – Msellata and the Gebel Gharian.
Numerous remains of olive oil presses testify to a sophisticated process and large-scale
capacity for the extraction of oil (Mattingly 1988a, 1995). There is almost no farm or
farmstead on the Tarhuna plateau without its own processing facilities for agricultural
products; some of the larger sites were capable of almost industrial scale production.
The significance of the rural potential, the distribution of agricultural resources, and the
organisation of agricultural production were prime considerations in the economy of
pre-industrial societies (Alcock 1989). The rural settlement pattern is undoubtedly the
most familiar aspect of the Roman period in the Gebel Tarhuna, although it has been
subject to fewer archaeological works in comparison with other areas in the North
Africa. Rome’s economy was mainly based on the agriculture, and agricultural
characteristics of the Gebel Tarhuna fit perfectly in this economic direction.
The establishment of farming settlements has dominated the settlement distribution
pattern. The main concern in this chapter is to explain the relationship between this
diffusion and the distribution of agricultural resources. Contrary to the concept that
expansion of rural settlement in the Gebel Tarhuna followed an ecological course in the
most productive lands, the argument can be made that the settlement pattern on the
Tarhuna plateau was an off-shoot of what has taken place in the coastal area, drawing on
capital investment from the Emporia (the coastal cities). The increased demand for
cultivable lands by the coastal cities was thus the main cause for the expansion pattern
to the Tarhuna region (Mattingly 1989: 143-5; 1995: 140-41; 1996: 167-8).
88
In moving beyond the outline of settlement type provided in Chapter 2, I will provide
more detailed answers these questions: what types of rural settlement existed in the
region? Was there a specific model of a typical farm? How were these rural settlements
influenced by environmental and natural resource factors? Does the archaeological
evidence of the Tarhuna rural faming sites reflect a certain type of ownership? What
was the extent of the role played by the urban elite in the emergence and spread of the
rural sites? What is the archaeological evidence of this role? In dealing with the
settlement pattern on the Tarhuna plateau I shall look at the distribution of different
types of site across the landscape, and their relationship to each other, to the
environment and to natural resources. The relationships between people and the
environment have been accepted in archaeology and geography already long ago as a
crucial characteristic for understanding the ancient human settlement and its
developments (Goudie 1981). For farming sites, for instance, their physical
surroundings were necessarily related to the main economic activities undertaken at and
around those sites. More insights can be provided into the economic dimension of
settlements through site location analysis (Van Ossel 2000). The impact of the physical
surroundings on settlement location and land-use has been recognised in many
archaeological surveys in the Mediterranean world (Barker 1995). In addition to this
view, the influence of the natural environment on human settlement has remained a
dominant theme in many archaeological works, especially in regional analysis and
landscape studies. The Gebel Tarhuna situation, however, is a reminder of the
distinction which can be made between observations of landscape and the actual
physical characteristics of the natural environments.
89
In order to gain an insight into the parameter of the settlement patterns in the Gebel
Tarhuna, the examination of farming sites will stand at the heart of this issue, as it is
generally assumed to represent the outcome of a Tripolitanain economy mainly based on
its olive oil production and export (Mattingly 1988a; 1988b; 1988c). The archaeological
evidence, however, highlights some differences between the coastal zone and the
Tarhuna plateau countryside, in particular the north-east sector, in the Roman period.
The Gebel Tarhuna landscape was based, as was the coastal region, on numerous
varying sizes of farm buildings (from farmstead to oilery/villa – see below). However,
the absence of reliable evidence for the emergence of villas before the end of the first
century BC underscores the difference of the Tarhuna plateau landscape from the
coastal villas which have been recorded in Wadis al-Tūra, al-Fäni, Jabrūn and Caam and
which appear to date from the second century BC (Cifani 2002; Munzi et al. 2004).
A major problem regarding the documentation of the archaeological record of the
pre-Roman period (Punic and Numidian) in Tripolitania and elsewhere is its
recognisability. The identification of Carthaginian and Numidian period remains in the
archaeological record is still in its early stages. There is also a gap in terms of
classification and description of Punic pottery in Tripolitania; the consequence of this
lack is, of course, a difficulty of identification. The detailed studies have only focused
on some types of Punic transport amphorae which have been found at a number of sites,
especially funerary sites, in Tripolitania (Anna-Maria 1983; Di Vita-Evrard 1997; Dore
1989). My identification of seven sites (Table 3.8) that yielded second and first century
BC pottery as well as the (2nd century BC?) Jbibina hoard of the Numidian coins, in the
territory of the Gebel Tarhuna, represents the first evidence for the appearance of early
settlement in the Tarhuna region. This evidence can be compared with others that have
90
recently emerged from other areas in the close hinterland of Lepcis Magna and which
indicate a slightly earlier intensification of rural settlement (Cifani and Munzi 2002).
The impact of the Roman economy on rural settlement organisation is very clear in
the archaeological evidence of surface pottery. Without excavation, it is impossible to
assess the scale and organisation of pre-Roman settlement. However, the epigraphical
evidence unquestionably demonstrates that people of Punic culture (Libyco-
Phoenicians) were dominant in the area during the early years of first century AD
(Goodchild 1976; Mattingly 1995). The agricultural improvement of the Roman period
brought new agricultural implements and resulted in a spread of different types of farm
and architecture. The organisation of the regional rural economy reached a much more
developed scale during the early imperial period (first and second centuries AD) with
the establishment of numerous oileries and large farms. The building of these farm
constructions may represent a cultural change in the Gebel Tarhuna countryside. The
builders on the one hand, conformed to some degree of Roman life-style; on other hand,
the establishment of these sites may represent changes to economic targets, new farming
techniques for achieving those objectives, new organisation of the labour force on the
property, and the possibility that all of these changes were a consequence of a basic
change in the land ownership related to new political and socio-economic circumstances
in the urban centres. These affected rural settlement organisation and production, and
they probably already affected the relationship between property owners and the
productive lands, with an increase in large estates. Such changes are evidently of great
importance to the understanding of the economic activity during the Roman period, but
they are very difficult to recognise in the archaeological evidence. In addition, the issue
91
of continuity and change in the ownership of lands can hardly be identified or proved in
the absence of written documentation.
A good sample of data is necessary to identify regularities and deviations in the
available rural settlement data and to distinguish patterning in it. The TAS records thus
provide a new baseline to discuss and evaluate the evidence of rural settlement type and
organisation. For example, site size is a major factor in identifying the function of the
site. Evidence from the TAS and from previous fieldwork in Tripolitania informs my
presentation of the settlement pattern and its typology. It is known from the surveys and
excavations that there were larger, wealthier luxurious site within the coastal area, in
particular the maritime villas between Lepcis Magna and Oea (Di Vita 1995; Munzi
1998; Musso 1998; Sergio 2004), while the smaller and less wealthy sites can be found
in the hinterland. It has been stressed that only a hierarchical society with leisured elite
could have established and occupied such luxury sites (Mattingly 1985; 1995).
However, the archaeological evidence suggests that all rural sites, from both areas, were
related to each other, especially in the kinds of the hierarchical organisation in the
settlement system and how it was related to economic exploitation of the landscape.
The Gebel Tarhuna, in particular the eastern area, is considered to be the most densely
settled zone in the Gebel, forming a deep hinterland for Lepcis Magna and linked with
the city by the Eastern Gebel road for which the earliest dating evidence for its
construction is AD 16-17 under the authority of L. Aelius Lamia, proconsul of Africa
(Goodchild 1976: 75).
3.2 Settlement type and organisation
3.2.1 Site size For the great majority of farm-buildings of the Tarhuna plateau it is relatively easy to
delineate their plans as a wealth of detail is visible at the surface, such as press elements
92
(orthostats, press rooms, press floors, counterweights, mills and tanks), upright blocks of
the opus africanum structure, courtyards surrounded by a number of rooms and water
catchment works. In addition, there is usually a varying density of scattered potsherds
around the farm sites. In numerous cases I have recorded rubbish middens on the
adjacent slopes to farm buildings. The combination of these structural elements in the
rural settlements of the Gebel Tarhuna and the pottery scatters permit a close estimate of
the size of many sites. The degree of precision is much closer than in conventional
plough-zones survey in north Mediterranean countries.
The characteristic features of the settlement pattern on the Tarhuna plateau were
different forms of olive oil farm buildings (Cowper 1897; Goodchild 1951; Mattingly
1985; Oates 1953). With regional variations, these range in size from small farms
(farmsteads) to villages or small towns. These variations appear to have been replicated
in most of the olive oil production regions of Roman North Africa. For example, a
remarkably similar varying of site size and press distribution is to be found in the
territories of Kasserine, Segermes and Caesare (Hitchner 1992-1993; Hitchner 1988;
Hitchner 1990; Leveau 1984; Mattingly D 1995; Ørsted 1992). In this chapter the
assessment of the Roman period settlement variation is based as far as possible on
quantitative measurement in order to evaluate if there was a noticeable peak in site-
numbers at specific period. A second issue concerns the settlement hierarchy (road-
stations, agricultural villages, and farms). I shall discuss whether the shape of the
settlement hierarchy was stable or whether it changed over time.
In this first section, the research is primarily devoted to the development of the
inventory of rural archaeological sites within the chosen geographical and chronological
framework. The different types of rural settlements are used here for quantitative
93
analysis based on an architectural approach. A number of descriptive terms are used for
site typology (see Chapter 2 for the terms and their description) to classify these rural
settlements of the Gebel Tarhuna. Classification of these rural sites is based primarily on
their size and functional characteristics. Those criteria have been applied to 100 rural
sites recorded or re-recorded by the TAS in the Wadis Turgut and Doga. It needs stating
that site definition, estimation of site size and number of presses recorded at each site
have depended on the visibility conditions and degree of preservation. Sites such as
isolated dams are not included in size measurement.
Measurements of site size performed on the ground by the TAS have been confirmed
wherever possible by the Google Earth measurement tools (Fig. 3.1). The estimated size
is without doubt still approximate in many cases especially where subsequent
demolition, cultivation or building over the site have been on a large scale or sustained
across a long period. The estimated size of rural sites is a broad indicator of the level of
the productive and residential unit, and of their continuity of use (Alcock 1989). The
analysis of settlement size on the Tarhuna plateau provides important insights on site
type and function. For example, the type ‘oilery’ concerns a large or very large farm-
building with large scale press capacity, as defined by Mattingly (Mattingly 1985: 35 -
38; 1987). In this survey, as noted already the prime criterion to define an oilery is a
complex of farm buildings with at least five olive oil presses. Sites meeting this
requirement vary considerably between 0.5 and 2.8 hectares in area. Some oileries were
also equipped with signs of luxury facilities (porticoes, mosaics, and bathsuites) or other
kinds reflecting a wealthy life style.
Most rural sites in the Wadis Turgut and Doga range between 0.015 ha. to 2.8 ha.
(excluding agricultural villages) and are marked by the surface presence of architectural
94
material including buildings, walls, cisterns, kilns and dense spreads of pottery and tile
fragments scattered on the surface (it should be noted here that enclosed or attached land
is not included in the site measurement). The smallest sites have been divided into two
groups: first band (less than 0.05 ha.) and the second band (0.05 to 0.1 ha.) constituting
13 percent and six percent respectively (Fig. 3.2). In most cases, they represent types of
structure that are sometimes encountered as sub-elements of larger sites - such as pottery
kilns, bath-buildings, watchtowers and mausolea (except the mausoleum of the Gasr
Doga that is lager size). These types were characterised by non-agricultural function but
still played a role in the settlement pattern and the socio-economic life of the rural
society.
Figure 3.1: Measuring site (TUT54) by the Google Earth (Senam Semana + 17 presses).
A third band concerns sites measuring larger than 0.1 ha. to 0.5 ha. The total number
of this category within the study area is 41 sites. The fortified farmhouses (gsur)
constitute an important number of these compact farms with 18 sites belonging to this
type. This identification attaches, in most cases, to the hill-top gasr type rather than to
the gsur established on top of an original open farm sites (see Chapter 2). The second
m
Sm
po
to
pr
w
FiDo
re
in
ea
di
of
fa
ar
pl
main type, w
mall farms
otential, suc
o the wadi
resence of
watchtower a
igure 3.2: Theoga.
Sites that
ecorded site
ncluded sev
arlier open
ivided betw
f Senam Se
arms from th
rable land a
lateaus or o
0
10
20
30
40
50
within this b
are often
ch as on gen
courses. A
three or fo
also fit with
e chart shows
occupy an
es. The site
ven small fa
farms) and
ween two typ
emana, situa
his band lie
and many o
on terraces
band, repre
located in
ntle slopes
small num
our presses)
hin this size.
bands of 100
n area betw
s of this ca
farms, five
d six oileri
pes: eight o
ated below
close to the
of them wer
s above val
esents small
areas with
and foot of
mber of larg
) and an o
.
sites of know
ween 0.5 -
ategory mos
large olive
es. Thirteen
oileries and
150 m alti
e 375 m con
re situated i
lleys. Thes
l farm build
h good agri
f hills prefer
ge olive far
oilery, a ma
wn size which
1 ha form
stly had a c
e farms, six
n sites occ
five large o
tude, most
ntour line, o
in elevated
e sites ten
dings with
icultural an
rably facing
rm sites (fi
ausoleum (
are located in
m 24 percen
clear agricu
x gsur (ma
cupy areas
olive farms.
of the oiler
often lookin
positions,
d to be hi
15 out of
nd water ca
g east or we
ve sites de
Gasr Doga
n the Wadis T
nt of all th
ultural func
ainly establi
between 1
. Except for
ries and lar
ng out over p
on the top
ghly visibl
95
41 sites.
atchment
est, close
fined by
a), and a
urgut and
he above
ction and
ished on
- 2 ha.,
r the site
rge olive
pieces of
of small
le in the
96
landscape, as an expression of economic and social status, as well as architectural
pretension. Finally, three rural sites comprised particularly large scatters (generally
between 2.1 – 2.8 ha). These are represented by oilery/villa sites and are often
associated with pottery kiln/kilns such as at Henchir Assalha (TUT15) and Sidi Eysawi
(TUT53). This type represents the maximum size of rural settlements below nucleated
villages and small towns category (1 – 2 ha), all these sites were characterised by oil
production facilities.
Figure 3.3: Distribution of oileries and large farms in the Wadis Turgut and Doga.
The common size for oileries and large olive farms in the surveyed wadis of the
Gebel Tarhuna is ranked between 0.4 to 2.85 ha. in extent (Tables 3.1; 3.2). Large olive
farms had evidence of three or four presses attached to a large building, often with
storage rooms placed within the building and revealed by quite dense scatters of
amphora, dolia, and coarse ware sherds with a lesser density of imported fine wares.
H L
97
Sometimes pottery and tile kilns can be identified close to these productive sites.
Oileries are similar, but with five or more presses. On the basis of the survey evidence,
undoubtedly the majority of the oileries and large olive farms of the Gebel Tarhuna
were productive units tending towards amassing large-scale surpluses of olive oil for
export (Mattingly 1985: 31- 38; 1987: 56; 1988b: 25- 27; 1989b: 144- 45: Oates 1953:
87). The amphora kilns established within or adjacent to a number of large estates (Fig.
3.3; Table 3.3) also clearly attest to the density of specialised cultivation and oil
production on these estates, and that the intensive agricultural economy not only aimed
at the production of a surplus destined for wider markets, but was capable of providing
suitable containers for its carriage.
ID Local name Site type No. presses Elevation Size (m2) TUT8 Oilery 5 290 8500 TUT10 Oilery 5 295 6500 TUT16 H.Boshaina Oilery 5 375 15000 TUT20 H.Henash Oilery 6 297 11250 DOG66 Sidi al-Akhder Oilery 6 230 8000 HAJ82 Oilery 5 280 11200 DOG106 Sh'bet asc-Schood Oilery 5 217 5000 DOG107 H.ash-shuaud Oilery 5 232 5200 TUT12 Sidi Buagila(2) Oilery/ villa 8 242 8000 TUT15 H.Assalha Oilery/ villa 5 280 28500 TUT38 Assenam Oilery/ villa 7 227 11300 TUT43 Loud al-Meghara Oilery/ villa 6 219 10200 TUT46 Kerath Oilery/ villa 5 250 21300 TUT52 Sidi Madi Oilery/ villa 7 150 9150 TUT54 Senam Semana Oilery/ villa 17 145 12500 DOG60 Senam Aref Oilery/ villa 6 184 10200
Total: 103
Table 3.1: Oilery sites in the Wadis Turgut / Doga and their sizes.
ID Local name Site type No. presses Elevation Size (m2) TUT3 Large Farm 3 300 14120 TUT5 H.Aziza Large Farm 4 315 8300 TUT11 Large Farm 3 266 4000 TUT14 Bu-Kaala Large Farm 3 333 4000 TUT26 Large Farm 4 254 4500 TUT27 Large Farm 3 280 10100 TUT29 Large Farm 4 255 8000 TUT35 Large Farm 4 275 10100 TUT36 Large Farm 3 226 7200
98
ID Local name Site type No. presses Elevation Size (m2) TUT44 Sidi Yekhlef Large Farm 3 211 9000 TUT57 H.Hmoudat Large Farm 3 250 7500 DOG81 Large Farm 4 215 4500 DOG111 Wadi Mseel Large Farm 3 170 10200 TUT112 Large Farm 4 280 12400 TUT1 Large Farm/villa 3 280 15000 TUT53 Sidi Eysawi Large Farm/villa 3 180 21000 DOG104 Large Farm/villa 3 210 6200 Total:57
Table 3.2: Large farm sites in the Wadis Turgut / Doga and their sizes.
ID Location Loc.name No.
Kilns Stamp Within/near
of Distance from the F-building
TUT12 W. Hwatem Sidi Buagela 2? Yes Oilery 50 m E TUT15 W. Turgut H.Assalha 4? Yes Oilery 70 m W TUT18 W. Turgut Astail 2 Yes Oilery 250 m S-E TUT48 W. Turgut Arbaia 5 Yes Large farm 150 m N TUT53 W. Turgut Sidi Eysawi 1 No Large farm 50 m S GUM86 W. Guman Scegafiat Asray 1 No Small Farm 400 m N-W GUM89 W. Guman Scegafiat Atriq 3? Yes Small Farm 300 m S-W
GUM90 W. Guman Scegafiat Ben Hemad 3 Yes Small Farm 350 m S-W
TEL102 W. Tarabut Hamzia 2 No Small Farm 250 m S TUT108 W. Turgut Armadia 4? Yes Small Farm 400 m W GUM110 W. Guman Scegafiat Maamri 2? No Small Farm 300 m S DOG111 W. Doga Almseel 1 No Large farm 50 m W
DUN6 Fergian Alhlafi 5? Yes Large farm 300 m N-W SRI2 W. Sri W. Sri 2? No Large farm 200 m W
Table 3.3: the amphora kilns identified by the TAS in the Tarhuna plateau.
A peculiarity of these rural sites is the evidence for luxurious amenities: 11 out of 33
oilery/large olive farm sites (within the area of Wadis Turgut and Doga) have produced
surface evidence for luxury elements such as columns, mosaics, wall-paintings and bath-
buildings. Although these features are well represented in maritime villas of Tripolitania
such as the villa Dar Buk Ommera, villa Silin, and Taggiura (Aurigemma 1926;
Blazquez Martinez 1990; Di Vita 1995), the Gebel Tarhuna villas have previously been
described as utilitarian ones, with luxury villas seen as something rare and exceptional
Total: 35+
99
in this hinterland (Percival 1976). Few of the olive farms planned by Oates in eastern
Tarhuna give an impression of having luxury remains. However, he did mention the
existence of porticoes and bath-houses and noted at Henscir Sidi Hamdan: “On the
south-east side a scatter of mosaic tesserae near the east corner may indicate the
presence of a small bath-house just here; the tesserae are of the four common colours of
black, white, brick-red, and yellowish-buff” (Oates 1953: 99).
Figure 3.4: Senam al‐Halafi 1.
100
Figure 3.5: Complex site of Senam Halafi 1.
Figure 3.6: Senam Halafi 1; line of columns at eastern side. Size of the scale (1 m).
101
Closer examination or excavation of other sites now provides additional examples.
For instance, the site of Senam Halafi 1, south of Gasr Ed-Dauun in the Fergian area
(c.3 km north-west of Henscir Sidi Hamdan), provides sufficient evidence about how
some of these sites were very carefully designed to meet both the particular agricultural
needs and to provide comfortable and status – enhancing residential accommodation for
elite owners (Fig. 3.4). The site occupies about 1 ha., today still covered by a huge
amount of ashlar stones, orthostats and columns drums (Fig. 3.5). The columns seem to
have formed a portico or colonnade which ran all along the eastern side of the building
where a column is still standing in situ (Fig. 3.6). Many mosaic tesserae of different
colours mixed with large and small pieces of mortar bedding are scattered on the
northern side very close to a large cistern, surrounded by a dense spread of potsherds
and fragments of tile. This indicates the location of a bath suite. Although it can be
hard to distinguish a building which was luxurious from the surface survey alone, this
type of evidence clearly indicates a greater level of wealth invested in accommodation
on rural estates than the previous records in the Fergian area have indicated. There are
important implications from this for our interpretation of the oilery-villas with regards to
the elite class.
3.2.2 Site type description
Because of the high degree of site preservation characterising most of the farming
sites on the Tarhuna plateau, very important information can be obtained from a study of
details of the buildings. In particular, a detailed typology of the plans of rural sites of the
Roman period in this region can be attempted from the surface remains (see above,
section 2.6 for initial discussion). Agricultural villages, open and fortified farms and
water management works formed a key element of agricultural intensification and
102
specialisation in the Roman period, especially in the early imperial period (first and
second centuries AD). The density of the distribution of sites with indicators of
agricultural specialisation (especially olive presses) has been taken as evidence for the
intensification of agriculture in response to market demand (Mattingly 1987a; 1988a;
1988b; 1995).
3.2.3 Term of villa Although many oilery sites and large farms give an impression that they functioned
as rural agricultural production centres, there is a problematic issue concerning using the
term ‘villa’ to describe these rural sites. The term ‘villa’ remains a critical one and is
difficult to define accurately (Smith 1997). The problem is related to the definition and
description of villae themselves in the literary sources. The most important conclusion,
derived from the literary evidence, is the absence of a particular meaning of the term
villa (Marzano 2007). It is a Latin word which can mean a farm or a country residential
place (Mulvin 2004). The term suggests that it can be a building in the countryside or in
the sub-urban area and had an agricultural purpose or a complex of buildings located in
the heart of a working farm (Varro, De Re Rustica,3.2. 3-6; 6-12). The phenomenon of
Italian luxury villas has strongly influenced the recent understanding of the term ‘villa’
(Dyson 2003). C. Gandini clearly emphasises this terminological problem, and accepts
the following characteristic: “architecturally, it should be seen as the most complex rural
settlement, bringing together a more or less comfortable residential part and a building
part involved in agricultural activity which is clearly distinct from the first” (Gandini
1999). It can be suggested that the rural villa appeared more often in areas that meet
these main two conditions: 1) regions characterised by a coincidence of favourable
factors such as fertility, climate and possibility of a good water supply; 2) locations
close to the transport routes, linking estates with cities for trade and exchange
103
commodities. The villa was a representative sign reflecting a form of economic
organisation centred on the rural estate. In economic terms, the Roman rural villa
functioned as an organiser centre performing management of the property, and held a
central role in coordinating its production to supply urban centres and market needs.
Recent archaeological surveys conducted in North African rural areas have
highlighted several types of agricultural rural sites including villas. For instance, the
Kasserine Survey identified five types of agricultural settlements (ranking from
agrovilles to small structures) in the Tunisian high steppe. Sites typified as villa
occupied the second class in the settlement hierarchy after agroville. Hitchner has
described the Kasserine’s ‘villas’ as large centres of agricultural exploitation comprising
numerous buildings (Hitchner 1989). The focus on agricultural activities, in particular
olive cultivation, at the rural settlements in the Kasserine region led Hitchner to
conclude that the marketing potential of olive oil in the Roman period had encouraged
the development of a hierarchical settlement system (Hitchner 1988; 1989; 1990). He
also suggested that the villas functioned as centres of rural estates, and comprised a
number of dependent residences and buildings (Hitchner 1990; 1995). Villas of the
Kasserine region were large centres of agricultural exploitation. Archaeologically, they
comprised numerous constructions, some of monumental ashlar buildings containing
multiple presses, storage facilities and other rooms, such as in Henchir el Guellali and
Henchir et Touil. Farms were distinguished from the villas by the absence of
monumental buildings and on the basis that commonly they only contained one or two
presses (Hitchner 1989: 392-4).
The archaeological survey, conducted by an Italian-Libyan team, around the Roman
villa of Silin marked the first systematic topographic approach to this hinterland strip
104
near Lepcis Magna (Munzi et al. 2004: 11). The survey evidence from the Silin survey
demonstrates substantial development in this countryside in the vicinity of Lepcis
Magna. This was a settlement hierarchy dominated by rural villas and farms. The early
Roman period (first and second centuries AD) represented the peak period for rural
villas and farms in the region around the Villa Silin (Munzi et al. 2004: 19-24). Table
Period I II III IV Va Vb VIa VIb VII VIII Dates 1st-3rd BC 2nd BC 1st BC 1st AD 1st half
2nd AD 2nd half 2nd
AD 1st half 3nd AD
2nd half 3nd AD
4th-5th
AD 6th AD
Rural villas
4 6 15 15 15 10 8 7 2
Farms 5 7 24 26 26 19 14 12 2 Table 3.4: the Silin Survey, rural villas and farms by periods (after Munzi et al. 2004).
3.4 illustrates that the first appearance of rural villa was in the 2nd century BC; while the
maximum expansion of villas and olive oil farms occurred between the first and second
centuries AD. The villas and farms of the Silin area were both marked by opus
africanum style, but the villas were architecturally defined in addition by existence of
decoration elements such as painted walls, mosaic floors and slabs of marble. However,
they both were equipped with one or more oil presses (Munzi et al. 2004: 26). Parallels
with the typology of the Kasserine Survey are evident, even though it is not possible to
make more explicit comparisons owing to the extensive nature of the Silin project and
the brevity of its single published report. Nonetheless, the high density of rural
settlements in the early Roman period and the distinct hierarchy indicate their
integration into a regional agricultural economy (mainly based on cultivation of olive
oil) and controlled by the Lepcitanian elite.
By way of contrast, De Vos, who directed an intensive topographical and
archaeological reconnaissance in the vicinity of Dougga, the ancient Thugga in northern
Tunisia between 1994-1999, avoids using the term ‘villa’ in her terminologies. She
prefers to define the rural sites as farms instead of villas because of the confusing
105
meaning of villa as derived from literary sources. The villa could be rural and/or urban,
suburban or seaside residence of otium or even a combination of these categories, with a
wide range from the overwhelmingly luxurious cottages or down to urban houses.
Moreover, according to De Vos, the ‘villa’ was transformed in the course of Roman
history (the villa of Cato was different from those described by Varro, Cicero,
Columella and Pliny), and varies according to geographical, climatic and cultural factors
(De Vos 2000: 21).
Most of our previous knowledge of rural building types located in the Gebel Tarhuna
depends on evidence derived from their archaeological remains; the great majority of
these rural structures can be described as utilitarian buildings (Mattingly 1985; 1995;
Oates 1953; 1954; Percival 1976). However, fashions in interpreting the significance of
the settlement data have changed considerably in recent decades, increasingly so with
refined chronological evidence and identification of new categories of sites. On the
other hand, there is an increase in our knowledge of the archaeology and settlement
patterns of the Gebel Tarhuna since Cowper’s day. Parallels for the represented types of
sites of the Tarhuna plateau can be found in other recent rural surveys in the Roman
North Africa (Barker et al. 1996; Carlsen 1989; De Vos 2007; Rebuffat 1988).
Consequently, new conclusions can be drawn from analysis of the material concerning
settlement patterns and economic activity in this region.
The key problem for knowledge of villas of the Gebel Tarhuna is generally the
absence of excavations that provide full information about their main buildings and
other facilities as well as the true scale of production carried out in them. However, a
point to stress here is that the villas of the Tarhuna plateau were rural buildings with
clear signs of Roman influence. This influence appeared in their architectural design
106
or/and in their exploitation of building materials. Features that indicate the presence of a
pars urbana include mosaics, baths, hypocaust installations, porticoes and wall-painting.
By these indicators, 12 sites from the TAS survey can be typified as ‘villa’. Their
interpretation only emerged from my recording of the archaeological surface evidence.
The expansion of archaeological systematic survey on other districts of the Gebel
Tarhuna and excavations certainly will increase the number of sites in this category.
The variability in site size and the evidence concerning the scale of agricultural
production associated with each type have helped me to shed light on the organisation of
these rural settlements. The description of settlement types is, in itself, a big order. I
decided, however, that since there was already a small amount of published data
concerning the different kinds of sites in the Tarhuna region, I shall focus my initial
discussion on sites not yet reported or not well reported. To the extent that, as I have
just emphasised, the examination has focused on rural settlements, it seems important to
submit now the types of rural sites which I will deal with throughout this chapter, and to
clarify their terminology. The validity of the different groups of rural settlements, which
have been chosen and used in this quantitative analyses, is reinforced by comparisons
with a number of other recent rural settlement studies in Roman North Africa, in
particular the UNESCO Libyan Valleys Archaeological Survey and the Kasserine
Archaeological Survey (Barker et al. 1996; Hitchner 1988; 1989; Hitchner 1990).
3.2.4 The agricultural villages
A substantial body of farming sites provides an excellent and datable record of rural
agricultural production in the Roman imperial period. The quantity and density of these
sites recorded in the Gebel Tarhuna reveal the fundamental role of olive oil production
in the Tripolitanian economy during the Roman period. Without doubt the scale and
107
number of presses found at many sites show that they were the centres of large estates
with a high level of surplus production. Although the Gebel Tarhuna landscape was
dominated in antiquity by different types and sizes of rural farm-buildings, there were
few settlements that can be classified as small agricultural villages. The TAS has
identified three sites (Fig. 3.7) of this type (Ain Astail in the Wadi Turgut, Gasr
Dehmesh in the Wadi Doga, and Halafi in the Fergian area).
They comprise different types of buildings, linked together probably to establish a
large unitary settlement. In addition, they have some common characteristics. First, their
archaeological evidence reveals that they were large production centres for both olive
oil and amphorae. Second, they were all located on or very close to the transport routes.
Since there were no towns in their vicinity, they were located in the vicinity of major
transport routes, to trade their products fully. Third, oilery buildings existed in the first
three villages named above. On the other hand, in terms of architectural material, it is
clear similar building techniques were used within these agricultural villages.
Ain Astail: The small agricultural village of Ain Astail is located in the middle
section of the Wadi Turgut and occupied an important position in the area. There were
two significant factors in its location: first, the village most certainly was a meeting-
point of, at least, two important ancient tracks. One ran through the Wadi Turgut and
linked the region of Gasr Ed-Dauun in the south with the Gefara plain to the north. The
second track united the higher plateau areas to west and southwest (where are al-Khadra
and Medina Doga), with the Wadi Turgut via the tributary of Wadi Astail. Second, it
without doubt benefited from the natural spring of Ain Astail (200 m southwest) where
remains of a dam still can be seen in this tributary below the village (Fig. 3. 8).
108
Small town. Agricultural village. Roman road. Track.
Figure 3.7: The agricultural villages and small towns in eastern section of the Tarhuna plateau.
The water from this spring probably was extensively used in the village, in particular
for the bath building and pottery kilns constructed below the main farm-building on
the south and southeast slopes respectively. On the western flank of the village lay an
oilery site with five presses, overlooking the surrounding cultivated land. This was
the largest structure in the village in which some of the smaller buildings may have
been satellites of this larger one. That is, this may have been an estate centre village.
’
’
’
!
!
al-Halafi
G-Dehmesh
Ain Astai
G-edDauun
Mdina Doga
13.754246
13.754246
13.975084
13.975084
32.5
7633
6
32.5
7633
6
0 5 102.5Km
±
109
Figure 3.8: Ain Astail agricultural village.
At the eastern extremity of the site a late Roman period fortified farmhouse (gasr)
occupied the top of a small hill. This gasr was defended by a sub-rectangular broad
ditch and is now in a ruined state (mostly reduced to wall foundations along with the
best preserved northeastern wall which still stands a few courses high). Within the
structure itself, along the western wall, is the remains of a cistern that functioned as a
small reservoir for the site. A notable feature is the quality of the gasr masonry, which
basically employed smaller and semi-coursed blocks rather than the larger and
carefully-coursed blocks that were used in opus africanum style farm construction.
However, the presence of this hilltop defended sites on the Tarhuna plateau was
certainly created in the late imperial period (Goodchild 1951; Oates 1954: 91-3), as it
appears from surface ceramic evidence.
Gasr Dehmesh: In the Hjaj area, on the northern bank of small tributary that runs
north-west and flows in the main water course of the Wadi Doga, there are remains of
Gasr
Kiln
Oilery
Bath
Spring
Dam 0 50 100 m
110
another small agricultural village scattered over an area of c.11 ha (Fig. 3. 9,10,11). The
remains are partially well preserved such as in the gasr (no. 9 Kasr Gharaedamish,
Cowper 1897: 237) which was established on top of a small hill at the east side of the
village. Other structures have suffered many disruptions: ploughing, reclamation of land
and building of new houses. For instance, it seems to be that there was a pottery kiln
located on the bottom of the eastern slope next to the gasr but it is hard to make a
considered decision from the surface evidence due to the site having been demolished
by bulldozer.
To the west of the gasr a group of buildings comprises a large farm with four olive
oil presses, bath-house and other enclosures. The bath-suite probably was revealed in
trenches cut for the foundations of a new house. Further west and north-west, on an area
of rocky relief, there are remains of an oilery with at least five presses. Although this
side of the village has suffered a lot of modern activity, which has damaged a large part
of the surface material, the extension of subsidiary facilities still can be traced. It is
worth pointing out that the village was watered by at least five rock cut shaft (majen)
type of cistern. The better preserved example was examined by the TAS team (1.20 Χ
1.50 Χ 4 m deep), with 2 tunnels dug facing each other at east and west sides of the
shaft (Fig. 3. 11). The eastern subterranean storage tunnel is the longest (16 by 3 by
2.50 m) in comparison with the western tunnel which was 11 by 3 by 2.50 m. However,
this majen had a capacity of about 200 m3 which certainly is less than many cisterns
recorded in the pre-desert area (Reddé 1985; Reddé 1988). Nonetheless the village was
able to keep over 1,000 m3 in its five cisterns. This estimated figure could be increased
if excavations uncovered further buried cisterns.
111
Figure 3.9: The Gasr Dehmesh village. C = cistern. (The Google Earth background).
Another noticeable feature is the remains of a small mausoleum detect at about 150 m
on the opposite side of the wadi. Unfortunately, all surface traces of it have been
levelled by a bulldozer and looting operation. Nevertheless, the small subterranean
funerary room is still visible but completely robbed. It was dug in a hard clayey deposit
and lined by fine ashlar blocks (Fig. 3. 12). Figure 3.13 shows early Italian sigillata
(Conspectus Form 4) and Eastern sigilata A (form type 43= Hayes 1991a, fig. 4, no. 30)
sherds (Fulford 1994) collected at this site, indicating that the mausoleum probably
belonged to the early first century AD. As with Ain Astail, this appears to be a village
serving as an estate centre.
112
Figure 3.10: Gasr Dehmesh occupies a top of small hill nearby a large farm building to the west.
.
Figure 3.11: A majen (cistern) in the Gasr Dehmesh village.
113
Figure 3.12: Plan of subterranean funerary room at Gasr Dehmesh.
5 c
Figure 3.13: The early 1st c. AD fineware collected from a mausoleum in the vicinity of the Gasr Dehmesh village.
Halafi: Halafi village is the only agricultural small village identified by the TAS in the
southern Fergian area. It is located about 3 km north-west of Henscir Sidi Hamdan and
sat at an important junction in the network of tracks, serving as a stopping point on the
track from Subututtu (Gasr Ed-Dauun) in the north to the further interior zones such as
the Wadi Tareglat-Gaam (Cinyps) and the pre-desert (Fig. 3. 14). For transportation and
communication, the Halafi village seems to have been functioned as a collection point
for olive oil in addition it was a centre for amphorae production.
During a reconnaissance survey in autumn 2007 I was lucky to identify a large
pottery production site with five or six kilns (Fig. 3. 15), which is uncommon in the
Fergian area except for a kiln previously noted within the Gasr Ed-Dauun village
(Mattingly 1995: 133; Oates 1953: 90). Although the latter is characterised by the
114
density of olive farming sites (Cowper 1897; Oates 1953; 1954; Mattingly 1985), there
is no comparison between the Fergian area south of the Gebel road and the Wadi Turgut
northward in terms of recorded amphorae kilns (Table 3.1 above).
Halafi village. Track ways. The Roman road. Farms recor-
ded by Oates (1953).
Figure 3.14: Location of Halafi village.
All the sites mentioned so far have aspired to a greater or lesser extent to be
recognised both in scales of production and in site-size terms as small agricultural
villages. These villages and the road stations or small towns (see Chapter 2) were
located within the territory of Lepcis Magna, in which location they could have
accomplished a significant function as local regional centres. Because they are
0 4 82Km
Gasr Ed-Dauun
S. Hamdan
115
considered the largest undefended settlements in the Tarhuna plateau and due to the
factor of distance from the major coastal cities, they probably served as local regional
Figure 3.15: Halafi village.
markets and acted as gathering points for onward transportation of olive oil
consignments from the many farms in their surrounding areas (Mattingly 1995: 133).
116
Like Ain Astail (five presses) and Gasr Dehmesh (nine presses), there was a
significant level of oil production here (nine presses). The main difference at Halafi
concerns an extraordinary central-aisled building (50 x 30 m), close to the pottery kilns.
This most likely related to the filling and storage of amphorae, suggesting that this
village had a broader function in terms of handling oil surpluses from the Fergian
region, rather than simply being a self-contained estate centre.
3.2.5 Oileries and large farms
The oilery farms occupy the second class in the settlement hierarchy of rural Roman
period sites of the Tarhuna plateau. As already noted in Chapter 2 the term ‘oilery’ is
reserved for sites which were clearly substantial oil “factories”. This classification is
based on the number of identified presses at each site, with the “oilery” being a farm
containing five or more olive oil presses. The high degree of preservation of press
elements at many sites makes the identification and counting of presses relatively easy.
Many of them are still in much the same condition as they were when seen and
described by Cowper during the 1890s (Fig. 3. 16).
Cowper 1896 The TAS 2007
Figure 3.16: Senam Aref (DOG 60).
117
The majority of farming sites of the Tarhuna plateau employed the opus africanum
technique in their constructions. Nor is there reason to consider that the distribution of
oilery farms in the Gebel Tarhuna was atypically dense in comparison to other regions
of Tripolitania, and the limited survey of the Tarhuna region leaves the question of the
total number there an open one.
The TAS recorded (or re-recorded) 16 oilery sites in the Wadis Turgut and Doga.
Although, statistically these oileries constitute 30 percent of the total number of farming
sites, their presses form more than the half of total number of presses (Table 3.5 and Fig.
3.17). Clearly the oilery farms worked as the largest olive oil production centres and
were likely established within the largest agricultural estates in this area. It is worth
noting that the Gebel Tarhuna certainly contained more oilery sites in addition to those I
have recorded in the area of Wadis Turgut and Doga. Although the Tarhuna plateau has
not been completely surveyed in order to estimate the total number of oileries, the
previous works (especially Cowper and Oates) identified some press complexes (Fig.
3.18) which may justifiably be described as oileries (Cowper 1897: 254-90; Oates 1953:
89-110). The majority of oilery sites were concentrated in the eastern part of the Gebel
Tarhuna, a point already apparent from Cowper’s work. He recorded only one site
containing five or more presses in the western section of the plateau during his travel to
the Gebel Gharian (Senam el-Megagerah, no, 52, Cowper 1897: 276). This
concentration may indicate that the most intensive exploitation of the Gebel Tarhuna
lands was linked to the to the territory of Lepcis Magna rather than the territory of Oea,
while certainly extended into western Gebel Tarhuna.
118
Site Type no. Sites Percentage no. Presses Percentage Oilery 16 30% 103 52%
Large farm 17 32% 57 29%
Small farm 20 38% 37 19%
Total 53 197
Table 3. 5: Numbers of farms and numbers of presses in Wadi Turgut and Wadi Doga.
Figure 3.17: Farms and presses in Wadi Turgur and Doga.
In terms of the number of presses at each site, it is necessary to take into account that
many of the other farming sites, in particular large farms, could also have served as
oileries. Because the judgement is based only on the visible surface evidence, there is a
likelihood that further presses have either disappeared underneath rubble and soil,
especially when sites are close to wadis-bed, or have been removed and reused in later
constructions. As Oates pointed out, the highly developed Italian settlement of the
Tarhuna plateau, during the colonial period, wiped out many ancient sites (Oates 1953:
110). It must, however, be remarked that most sites of the Tarhuna plateau were
characterised in Cowper’s day by good visibility above the surface of the ground. There
has been some differentiation as a result of subsequent development, starting in the
Italian colonial period (Oates 1953: 85):
Oilery Large farm Small farm
no. Presses 103 57 37
no. Sites 16 17 20
0
20
40
60
80
100
120
140
119
High. Oileries recorded/re-recorded by the TAS. Boundary line.
Low. Oileries recorded by Cowper and/or Oates.
Figure 3. 18: The recorded oilery sites in the Gebel Tarhuna.
“Farms comparable in size to Sidi Hamdan did exist nearer the road, but are usually too badly damaged for direct comparison of their layout. Henscir el-Mohammed in Wadi Gsea, of which Cowper published a description and a rudimentary plan, has been reduced by Italian quarrying to a barren hummock with two standing presses and a few battered blocks”. (Oates 1953: 101).
Sites demolished or quarried during the Italian colonial period were also witnessed
by Goodchild during his investigation of the sanctuary of Ammon at Ras el-Haddagia
(el-Khadra) during 1947:
The inscription, rediscovered by Aurigemma and Beguinot in 1911, was transported to Tripoli Museum ......, and the site of the discovery was soon forgotten, so much so
%
%
%
%
%
%
%
%
##
##
#
#
## #
#
###
#
#
#
7
5
5
5
9
8
5
6
6 7
5 7
6
5
855
5
6
6 5
55
17
13.533408
13.533408
13.754246
13.754246
13.975084
13.975084
32.5
7633
6
32.5
7633
6
±
0 6 123Km
Oea’s land Lepcis’ land
W. Turgut
120
that in 1935 the contractors who built Breviglieri village-centre quarried much of their stone from the ancient walls. (Goodchild 1976: 79).
The oileries vary in their layout and size from one site to another; they are also
differentiated by the numbers of presses identified, though the scale and layout of most
presses suggest that they served the same purposes and employed the same facilities.
They appear to have been built as central facilities on large estates from where their
production was transported to the major urban centres for marketing and export. A
principal distinction in the oilery sites examined by the TAS in the Wadis Turgut and
Doga is that some sites had additional signs of luxury, suggesting elite occupation (at
least periodically). A total of eight oilery buildings (50%. Table 3. 1) can thus be
classified as oilery-villas.
Oilery-villas are characterised by a number of luxurious elements (e.g., mosaics, bath
suites, porticoes and wall-paintings) which distinguish them from other more utilitarian
oileries (Table 3.6). Although the countryside is considered primarily as a place of
agricultural production, the principles of luxury in the Roman period were applied not
only to coastal villas in the vicinity of Lepcis Magna (Fontana et al. 1996), but also
expanded to the productive lands of the Gebel Tarhuna. This seems to have occurred in
parallel with the extension of intensive olive oil production from the coastal plain into
the Gebel. This development may have gathered pace with the construction of the road
that linked Lepcis Magna with the Gebel Tarhuna in AD 16/17 (Mattingly 1995: 140).
121
a) Distribution of oilery‐villas in the Wadis Turgut and Doga.
#
#
#
##
#
#
#60
54
52
4643
38
15
12
13.754246
13.754246
32.5
7633
6
32.5
7633
6
±
0 2 41Km
W.Turgut
W.Doga
Oilery‐villa. Luxury villa. Known Roman road.
b) The oilery‐villas in the Wadis Turgut and Doga and the luxury villas in the area between Lepcis Magna and Oea (after Mattingly 1995).
Figure 3.19.
122
ID Local name no.press Location Lux. Signs TUT12 Sidi Buagila(2) 8 hillslope bath-suite TUT15 H.Assalha 5 break of slope bath-suite TUT38 Assenam 6 wadi side bath-suite and pieces of columns.
TUT43 Loud al-Meghara 7 wadi side bath-suite, capital of column and pieces of columns.
DOG46 Kerath 5 break of slope portico and bath-suite TUT52 Sidi Madi 7 wadi side portico and bath-suite TUT54 Senam Semana 17 wadi side mosaic, portico and bath-suite DOG60 S.Aref 6 wadi side portico and bath-suite
Table 3. 6: The recorded oilery‐villas in the Wadis Turgut and Doga.
These oilery-villas indicate that there was both large-scale investment in productive
facilities and lavish spending on materials displaying the wealth and prestige urban elite
outside their towns (Mattingly 1995: 141; 1988b: 27). Figure 3.19a shows that the
location of these oilery-villas was mostly linked to the main wadi valleys on the
northern flank of the Gebel Tarhuna, placing them on the shortest routes down to the
coastal road and the major coastal cities (Fig. 3.19b). Mattingly argued that some of the
coastal luxury villas were also located on estates and potentially related to an area of
intense agricultural activity (Mattingly 1995: 141). Indeed the recent archaeological
survey in the Silin area has confirmed Mattingly’s argument: several oil pressing
elements have been recorded in the vicinity of a number of luxury villas (Munzi et al.
2004). For instance, the excavation carried out in 1996-7 by the Department of
Antiquity of Lepcis Magna at ez-Zeita (Wadi Zennad, c. 3 km south-west of Lepcis
Magna), and also the cooperative Libyco-Italian work on the recording and planning of
the settlement complex near the Wadi al-Fani (3.5 km west of Lepcis Magna) indicate
that these villas were also highly involved in oil production, and taking part in the
region’s successful olive oil economy (Ben Rabha 1997; El-Nemsi 1997). Although the
oilery-villas of the Gebel Tarhuna were located further inland, their position close to the
main wadis offered them favourable ways to communicate with the coastal villas and
cities. Figure 3.20 illustrates that most of these oilery-villas looked out towards the
123
Mediterranean coast. For example the most northerly oilery-villa (no. 54, Senam
Semana) is only about 15 km from the sea. From the early Roman period this part of the
Tarhuna plateau had been intensively and efficiently developed. The valleys Turgut and
Doga and their surrounding lands, located between mountains and hills, were probably
the most fertile olive-cultivation area in the whole of the Gebel. As a result, wealthy
Libyphoenices had their estates in this region. As Mattingly wrote: “Much of the best
agricultural land in the region (Tripolitania) is in fact to be found in the foothills and on
the plateaux of the Gebel” (Mattingly 1995: 140). Thus, economically this part of the
Gebel Tarhuna became more distinct agriculturally once Lepcis Magna came to
dominate the region. This is clearly stated by Mattingly: “We know that by the Early
Principate, Lepcis Magna and Oea had carved up the best olive growing lands of the
Gebel Msellata and Tarhuna between them, with Lepcis certainly controlling the better
share... It is clear that the higher quality of Lepcis’ territory and its closer proximity to
the coast will have given her considerable advantages in developing it” (Mattingly
1988b: 23- 24). The location of most of the utilitarian oileries and oilery villas in the
zone that can be assigned to the territory of Lepcis offers further confirmation of this
suggestion.
It is suggested here that the eastern and north-eastern zone of the Tarhuna plateau
was inextricably tied to the properties of the Lepcitanians. It is also possible that some
estates were developed and owned by local Libyan residents, who were able to pursue
independent relations with main the urban centres. For example, ‘NKSF (or TKSF) son
of Shasidwasan (or Shasidwasat) son of Namrar (or Tamrar) of the sons of Masinkaw’
built the Ammonium of Ras el-Haddagia (al- Khadra) in AD 16/17 at his own expenses
(Levi Della Vida 1951).
124
Nonetheless, scholars have taken the absence of large numbers of mausolea in the
Gebel Tarhuna as a primary indicator for the majority of the most substantial estates
being owned by wealthy urban citizens (Mattingly 1985; 1987a; 1995; Oates 1953). For
instance, one can cite the case of Aemilia Pudentilla, the Oean woman who married
Apuleius and had large country estates and invested her fortune in land, houses, animals
and slaves (Apuleius, Apol. 44. 6; 71.6; Mattingly 1995: 143). Apuleius stated that
many of the elite Oean possessed multiple estates, during his time, distributed
throughout Oea’s territory and managed on behalf of them by bailiffs or slaves. As
regards the wealth of Pudentilla, Mattingly believes that she was not the only
millionaire at Oea and the aristocratic elite at Lepcis Magna were even more wealthy
(Mattingly 1995: 143). This point is supported by the higher density of oileries in the
territory of Lepcis
The concept has been supported by two factors. First, the investment in these oileries
and large farms, constructing them in ashlar masonry with large scale presses facilities
is undoubtely equivalent to conspicuous consumption of profit on site. Second, the
utilitarian character of the majority of the farming sites suggests that their owners did
not reside on their estates.
.
125
Figure 3.20: A Google Earth im
age shows the distribution of oilery‐villas in the w
adis Turgut and Doga and m
easuring the distance between site no. 54
(Senam Sem
ana) and the Mediterranean Sea.
126
3.2.6 Associated mausolea and other tombs Indeed, there are few recorded mausolea in the Gebel Tarhuna in comparison with
the number of known farming sites (Fig. 3. 21), and a considerably smaller number of
farming sites associated with mausoleum (Oates 1953). While further to the south in the
Figure 3.21: Locations of the eight mausolea that have been recorded in the Gebel Tarhuna.
pre-desert area, the area was dominated by elite farms, very often associated with
mausolea - more than 70 mausolea have been recorded in that area (Mattingly 1996c). A
total of only eight mausolea was recorded on the Tarhuna plateau; three were recorded
by Oates in the early 1950s (Oates 1953: 104-5). With the notable exception of the
massive mausoleum of Gasr Doga, the other seven known mausolea of the Tarhuna
plateau are of small or medium size and lesser architectural decoration. The largest of
the three mausolea described by Oates in the Ed-Dauun-Gsea area measured 3.40 by
3.10 m, with a podium 1.50 m high. He considered them as an exceptional mode of
burial (Oates 1953: 104-5).
!!
!
!
’
’
’
’
’
’
’
’
DauunKhadra
Tarhuna
Souk aj-joumah
Msid
Khadra Zituna
G.Doga
Sonama
Khabesh
Dehmesh
Medina Doga
13.754246
13.754246
13.975084
13.975084
32.5
7633
6
32.5
7633
6±
’ Mausoleum.
! Modern centrs.
Dranage0 8 164 Km
127
A mausoleum examined by the TAS called as-Sonama (4.5 km north of Ain
Scersciara and 7 km north-west of Gasr Doga) is located in the vicinity of a small farm
(TEL 95, Fig. 3. 22). On a foundation course, it measured 2.50 x 3 m, with two
surviving courses of fine dressed limestone blocks standing to c. 2 m high above a
podium of 0.75 m high. From the collapsed architectural elements, it appeared to be
carefully decorated, with angle pilasters surmounted by a good Corinthian capital and
frieze at each corner.
Figure 3.22: Corner Corinthian capitals from es‐Sonama mausoleum.
The archaeological evidence shows that mausolea were not the only impressive
burial type employed in the Tarhuna plateau; two hypogeal (subterranean) tombs have
recently been discovered at Zwitina and Wadi es-Sri. The ceramic evidence found in
these tombs reveals that they can be dated to the second century AD (Faraj et al. 1997).
There are another nine hypogeal tombs discovered in the Gebel Tarhuna during
1970s and 1980s (Table 2. Appendix). Their finds indicate the predominance of the
cremation rite. Tombs of the eastern sector of the Tarhuna plateau seem to be earlier
than the western Tarhuna plateau; only one tomb out of five in el-Zagadna and the Sidi
Asid areas has been dated to AD 1-50, and the others are dated to the fourth and fifth
128
centuries AD. On the other hand, the tombs of eastern Tarhuna are dated between the
beginning of the first to the end of the third centuries AD.
3.2.7 Layout of the presses
During the Early Imperial period the distribution of oilery sites reveals a dense
cluster in the north-eastern part of the Tarhuna plateau, especially in the Wadi Turgut
(Fig. 3.18). The character of these oileries, moreover, expresses a high potential output
of olive oil. Despite the lack of epigraphic and literary records on land exploitation of
the Gebel under Roman rule, the area seems to have been largely dominated by estates
of oileries and large farms (Mattingly 1996a). The organisation of the oileries and large
farms reflects on the nature of economic exploitation and the important role they played
in sustaining the development of coastal cities and luxury coastal villas. Of particular
note are the serial nature of pressing facilities and the use of architectural elements such
as fine dressed masonry, square pillars and cylindrical columns and opus siginum floors,
which reflect both a high level of investment on the part of the owners and the fact that
this was targeted to achieve a large scale of production and income.
The vast majority of investigated oileries, in the Gebel Tarhuna, are similar to large
farms; both are dominated by a monumental ashlar courtyard building with attached
press facilities, associated cisterns, and sometimes pottery kilns. The range of potsherds
at these sites includes imported fine wares (mainly Italian sigillata and ARS), local
amphorae, and local coarse wares. Sometimes the presses were located on one side of
the building only, but in most cases they stood on two or more sides. Sites such Senam
Semana and Sidi Buagela 2 (TUT12 and TUT54) are examples of oileries with a linear
arrangement of presses (Fig. 3.23a and 3.24). At Sidi Buagela 2, there is continuous
arrangement of eight presses occupying the west side of the main oilery-villa building. It
129
is quite similar to the oilery site of Henscir el-Bagar 2 in Tunisia (Fig. 3.23), which can
be positively identified as the centre of a senatorial estate called the Saltus Beguensis to
the north of Kasserine (Cillium) in the Tunisian high steppe (Sehili 2008). At Senam
Semana (Fig. 3.23b), in a line nearly north to south, a total of 17 olive presses form a
row 73 m in length. Orthostats of the presses still stand 2.60 m (below lintels), showing
a potentially massive height for the beam operation (see further, Chapter 4). Opposite
the line of presses are two rows of square columns (0.50 x 0.58 by 3.00 m high); the
latter are segregated by a long corridor 3.20 m wide. Each row, originally, seems to have
contained 17 columns, of which 12 survive in the western one and 14 in the eastern row
are (with varying height) still visible on the ground. A capital of trapezium shape
appears to have been set on the top of each column to support the roof (Fig. 3. 25).
Again the total number of 17 presses and the other architectural elements employed in
the site reveal the high level of investment by the owner to build and maintain such a
huge oilery-villa.
Although they served similar purposes and used the same building materials, the
other utilitarian oileries and oilery villas examined (e.g., TUT38, TUT42, TUT52,
DOG60, and DOG106) show a different distribution of presses generally around two or
more sides of the construction. As can be seen, the majority of oileries of the Tarhuna
plateau had varied sizes of rooms and storage areas surrounding different sizes of
courtyards (Fig. 3.26). Their plans reflect traditional rural building styles influenced by
a functional typology, in order to provide higher productive potential. As regards these
oileries, therefore, the same diversity of plan has been recognised with the large farms
type. One of these farming sites is the large farm-villa of Sidi Eysawi (TUT53, Fig.
130
Figure 3.23: a) Plan of Henscir Sidi Buagela (TUT12).
Figure 3.23: b) Plan of Senam Semana (TUT54).
Bath suite
131
Figure 3.24: Plan of Henscir el‐Begar, Tunisia (after S. Sehili).
Figure 3.25: Orthostats and columns with trapezium capitals at Senam Semana (TUT54).
Cistern
132
Figure 3.26: Plan of the oilery‐villa of Henscir es‐Senam (TUT38).
3.27). The site can be classified as a Roman rural working villa with three olive oil
presses (the site originally could have had five or more presses and thus equate with an
oilery villa). It appears to be based on a similarity with the coastal villas in such
elements of architectural decorative designs. Here both the luxury aspects and farming
establishment were on the one hand incorporated into the architectural layout, and on
the other hand, the site represented a continuation of the urban lifestyle in such rural
places. The accommodation part appears in the eastern part combining a number of
rooms lined behind a portico, and its roof was supported by probably eight large
columns of which some of their bases (Fig. 3.28) are still in situ.
133
Figure 3.27: Sidi Eysawi, a large farm‐villa with pottery kiln.
Figure 3.28: A base of column at Sidi Eysawi.
134
3.2.8 Small farms
I use this term to refer to small farming buildings (Fig. 3.29). These almost always
began as open farms, though many of them became fortified during the late third or the
fourth centuries AD. They are differentiated from oileries and large farms by the fact
that they had one or two presses at most, normally attached to courtyard buildings. In
the absence of any epigraphic and literary records, the main problem is to identify what
kind of relationship existed between these small farms and the oileries and/or large
farms. Can they be seen as isolated and independent farms? Or were they related in
some form of dependence to other lager farms? In terms of property, these questions
lead me to state, while the oileries and large farms belonged to the coastal urban elite
such as Aemilia Pudentilla, the small farms could have been either attached to the larger
farms and owned by the same large landowners, or they could have belonged to a group
of less wealthy independent farmers who looked forward to make fortunes through
exploitation of the land. These farmers are the sort of people described by inscriptions of
Roman Africa as ‘agricola bonus’, ‘diligens agricola’ (Stone 1997). One of those was
the initially landless reaper from Mactar, who after acquiring a piece of land and
increasing his holdings, made a considerable fortune and obtained a high social position
(C.I.L., VIII, 11814).
A further step in understanding the relationship between the small farms and the
oileries/large farms can be achieved by detailed analysis of their distribution within the
surveyed region. Figure 3.42 shows that some small farms, especially in the northern
part, were most likely to have been managed and operated by the nearby oilery or large
farms. The oileries and large farms can be argued to have functioned as central estates
which were probably owned by the elite families of the coastal cities. It is possible that
135
small farms in close proximity to the larger facilities formed component parts of a villa
estate; equally some villa-owners may have held several agricultural estates, on only one
of which they built a villa residence.
Thus, in the middle of the Wadi Turgut, for example, the small farms (TUT39,
TUT41) seem most likely to have been operated by owners of the oilery-villa of Henscir
es-Senam (TUT38), because they are all located a few hundred metres apart from one
another (Fig. 3.30). The location of many small farms in close proximity of larger ones
raises the possibility that agricultural settlement of the Gebel Tarhuna was largely
formed around large estates centring on oileries and large farms, and also involving
smaller dependent farms, perhaps leased to tenants.
In the southern and south-western parts of the surveyed area on the north-east side of
the Tarhuna plateau, some of the small farms appear to be isolated and it is possible that
these sites did not belong to an oilery or a large farm. In a number of locations, small
farms have been noted, though without any evidence for an oilery or large farm at the
heartland. This may be a case of independent farmers who probably cultivated small
pieces of land and produced olive oil for their own consumption and the market. The
architecture of the sites and investment in presses suggests above subsistence level
production and engagement in the Roman market economy (and perhaps the emergence
of the rural elite). However, this hypothesis is difficult to prove archaeologically with
the lack of the epigraphic or literary evidence. For instance, we lack exhibit references
to periodic rural markets (nundinae) which were normally run in other parts of the
Roman Africa (Shaw 1981).
As in the case of oileries and large farms, most of the small farms of the Gebel
Tarhuna were built in the opus africanum style and characterised, during the early
136
Figure 3.29: Comparative plan of some small farms (the upper plans from Oates 1953‐4).
imperial period, by the absence of formal defences that dominated the later Roman
period farming sites on the plateau. A common feature of the buildings was the use of
concrete walls, sometimes faced with small coursed blocks, and supported at intervals of
137
Figure 3.30: Location of an oilery‐villa with the nearby two small farms (the Google Earth background).
2 to 3 metres by dressed limestone orthostats (Oates 1953). The analysis of small farm
plans (Fig. 3.29) indicates that they generally had a small number of rooms (8-12?) and
pressing facilities designed around a small courtyard, and were rarely associated with
luxurious elements. Only one small farm contains a bathsuite; at Henscir Aulad Ali
(DOG 105), this was established on the southern slope 25 m below the farm-building.
The construction of this type of farming sites has been attested in a vast geographical
area extending from the coast to the pre-desert zone. In the Silin area, the vast majority
of identified farms could be typified as small farms with one press (25 examples
recorded) compared to only 4 farms equipped with two presses (farms 4, 19, 60, 61). A
total of 15 rich farm-villas, in the Silin area, exhibited luxury markers including wall
paintings, mosaic floors and slabs of marble (Munzi et al. 2004: 13, 26). Sites with olive
presses have also been recorded in the Tripolitanian pre-desert where they are
138
considered as “the most obvious archaeological feature attesting a specific agricultural
activity” (Mattingly 1985). These small olive farms occupied the top of the settlement
hierarchy of the pre-desert area; principally, they began as undefended sites, often of
opus africanum construction, during the first and second centuries AD before being
increasingly replaced from the third century onwards by fortified farms (gsur)
(Mattingly 1996: 168). The archaeological remains and pottery sherds, together with the
wadi farming systems, demonstrate how agricultural, especially olive oil, production
developed economically in the early centuries AD, despite this being a marginal zone
(Barker et al. 1996; Mattingly 1985; 1995; 1996). Their importance as the upper echelon
sites in the rural settlement hierarchy is emphasised by their association with mausolea
(Mattingly 1996c). Mattingly believes that the pattern of pre-desert settlement was also
based on estates holding quite large pieces of land, rather than a widespread network of
small individual free-holdings (Mattingly 1996: 178). In view of the fact that the pre-
desert small farms seen to have functioned as centres of independent farms/estates, then
we could accept that some at least of the Gebel farms were also probably in independent
ownership, and not all controlled by bigger estates.
3.3 Settlement construction and organisation The large amounts of data gained by the TAS add important information to what may
be gleaned from literary sources relating to the organisation of Tripolitanian rural
production and economy, such as the urban-rural links, specialist crafts production and
local settlement hierarchy. Previous studies of Roman Tripolitania have not addressed
whether or not the high level of standardisation in manufacture of pressing
paraphernalia involved specialist craftsmen. The evidence from the Tarhuna plateau
suggests that press building and press/mill elements were created or overseen by trained
specialists who possessed a detailed knowledge of measurements and function. On
139
current available evidence, it is difficult to determine whether these press and mill
suppliers dwelt in the major cities and brought the pre-fabricated elements to the
countryside or whether there were specialised outfits based permanently in rural areas
or in small towns/villages on the main roads of the Gebel hinterland. The areas closest
to Lepcis Magna (Figure 2.2) and to the coastal maritime villas (Figure 3.20) may have
been supplied by urban based units utilising the two main communication ways - the
coastal road and the Gebel road (Goodchild 1976). However, at greater distance into the
Gebel, it is likely that subsidiary workshops were established, perhaps based on some of
the larger estates or in the small towns on the road, with specialist craftsmen able to
fashion standardised stone press elements, timber beams, mills and their wood and
metal fittings, pulleys and ropes, baskets and tanks and vats. The evidence for the
quarrying of stone orthostats at some sites, suggests that where possible, stone elements
were produced on site to limit transport problems. It is likely that the workshops
supplying the skilled personnel for this and other elements of the presses and mills were
based close by.
The impressive level of olive oil production of Roman Tripolitania raises the question
whether the development of the Liby-Phoenician/Roman urban centres engendered
growth in the economy beyond their own consumption needs. The evident investment
in specialised production, the scale of the oilery facilities, and the evidence of amphora
production all support the view that the production went far beyond regional market
requirements. This appears to be a case of growth well beyond the expected
demographic trend in the region. It is thus a clear example of genuine economic growth
in this region of the Roman empire, probably dated primarily to the early centuries AD.
Economic growth on this scale has implications for the size and organisation of the
labour force, both in agriculture and manufacturing, and must have entailed a
140
substantial investment of capital (Hopkins 1995; Kehoe 2007). To what extent did the
exploitation of the countryside reflect on the wealth of urban elite? It can be suggested
that the surplus agricultural production, especially olive oil, played a significant role in
the wealth of the urban elite (Mattingly 1995). Through the early Roman empire era,
agriculture dominated the economy and characterised the empire system. For the urban
elite, land represented a resource providing not only their income but also make
protection from unexpected risks, in addition supporting their social and political
position (Kehoe 1997). The main risk facing agriculture could be come from the
irregularity of rainfall. However, both large landowners and small farmers were
dependent for their livelihoods on the agricultural production. Thus, they made efforts
to conserve rainwater and land fertility by creating water management system based on
dams, wadi-walls and cisterns. On the other hand, even though the olive oil was the
main crop, in order to reduce risks, they might have practised ‘polyculture’ that is,
mixed plantation of olives, vines, vegetables and cereals which were harvested in
different seasons (Kehoe 2007: 551). The engagement of the urban elite in rural
production most likely was reflected in the employment of a large number of people
from both urban and rural communities, and also affected investment in the
development of new technologies that might have increased productivity. Settlement
sites of the Roman period were widely distributed in the landscape. The archaeological
evidence reveals that this landscape was mainly engaged the olive oil production.
However, livestock-rising was also of high importance to rural economies in antiquity
and the issue needs consideration in relation to the TAS evidence. There were often
close associations between animal husbandry and more specialised cultivation in rural
communities; exchange of produce (or labour) was commonly established within rural
societies between the farmers engaged in agriculture and people who practised animal
141
husbandry (Whittaker 1988). The natural environmental conditions of the
Mediterranean lands played a central role in defining the relationship between
traditional transhumance and the regional agricultural/arboricultural specialisation
(Halstead 1987: 78). At this stage of investigation, it is difficult to judge whether the
Gebel Tarhuna had the same traditional North Mediterranean transhumance pattern of
lowland winter grazing areas and highland summer pastures (Barker 1975; Frayn 1984).
However, the existence of the pre-desert zone to the south of the Gebel is a factor that
has favoured seasonal transhumance within this region in the past (Mattingly 1995: 37-
8).
Agricultural production and land use were integrally related to the traditional nucleated
pattern of farming sites that clustered near to the best arable intensive farming lands
(Keller and Rupp 1983). There are good reasons why sedentary farmers and pastoral
groups have commonly enjoyed symbiotic relations. Production at rural sites was
considerably dependent on working animals whether for ploughing or carrying
equipment, workers and products, especially in the case of large estates involved in
large-scale surplus production (Foxhall 1990). Furthermore, the use of animals for a
wide range of key functions will have had wider implications for estate management,
productivity and costs. Specialised estates had a need for extra labour and animals at
certain times of the year and draught animals lost through death, injury or illness, will
have needed regular replacement. Manuring of orchards and fields could also benefit
from the seasonal grazing of flocks and herds that were located elsewhere for much of
the year. Finally, agricultural products could be traded by the estates for meat and other
animal products with pastoralists, meeting subsistence needs on either side. All these
factors favour close and positive relations between sedentary farmers and pastoralists.
142
3.3.1 Farming sites It seems that most of the utilitarian farming sites on the Tarhuna plateau are a type of
building which combines in an integrated structure the functional requirements both of a
working farm and a residential place for workers. The type of farm building is well
represented by recorded examples (Cowper 1897: 254-93; Goodchild: 1976: 88-93;
Mattingly 1985: 34-8; Oates 1953: 89-110), though the layout of the different farms
varies considerably, according to altitude, topographic location and the scale of
agriculture practiced within it. At many sites, occupation and activity is likely to have
varied considerably across the year, peaking in the olive season, when the pressing
operation was entirely employed. Some sites which show luxurious elements suggest
that these farms were occupied, at least periodically, by their owners who ran them as
central estates. The capital outlaid on the pressing facilities of the farms supports the
idea that the owners of these were in general wealthy citizens, and it seems reasonable
to suppose that their prosperity was mainly derived from the production of these rural
farms.
What were the architectural traditions behind the design of these types of farm and
how did this rural architectural of the Roman period correspond to the productive
purposes of these farms? Firstly, there were characteristic features in the design of the
oileries and large farms, and indeed in the case of some of the small farms. The
Tripolitanian examples are analogous of certain features encountered in the rural
architecture of Roman Africa more broadly. The arrangement of courtyard, pressing
rooms, storage areas and the use of opus africanum or opus quadratum styles can be
paralleled in many of the rural farming sites in other parts of Africa Proconsularis (De
Vos 2000; 2007; Hitchner 1988; 1989; 1990). Like the farm (TUT11), Henscir es-
Senam (TUT38), Henscir Sidi Madi (TUT52), and Senam Aref (DOG60), the
143
employment of opus quadratum generally appears to have been used in the outer walls
(Figs. 3.31a,b , 32). The use of opus africanum at these same sites was primarily in the
partition of pressing rooms. At the oilery-villa of Henscir Sidi Madi, with 7 olive oil
presses, there is evidence of fine ashlar masonry, opus signinum, a bathsuite and an
arched gateway.
Figure 3.31: a) Use of opus quadratum and an arched gate at Sidi Madi (TUT52).
Figure 3.31: b) Fine ashlar masonry at Sidi Madi (TUT52).
144
Figure 3.32: Use of the opus quadratum at Senam Aref (DOG60).
A second feature of the farming sites of the Gebel Tarhuna is their use of local
limestone, quarried from nearby scrap-foot and hill slope locations. Some quarrying
sites have been examined by the TAS, identifying several examples of unfinished
limestone blocks left on the ground of the quarry. At a quarry found on the eastern slope
facing the main course of the Wadi Turgut, and about 130 m above a small farm
building (TUT45), two unfinished olive press uprights and the base of a column had
probably been left from the last quarrying activity (Fig. 3.33).
Figure 3.33: Some architectural elements left in a quarry close to TUT45.
145
The large farm (villa) of Sidi Eysawi (TUT53) was built in limestone blocks cut from
the immediate locality, the quarry visible as a stepped rock face with traces of wedge
marks still visible. Another interesting feature at this site concerns a number of
rectangular blocks marked by symbols which most likely are New-Punic letters (Fig.
3.34). These were possibly engraved by the quarry workmen.
Figure 3.34: Symbols (probably New‐Punic letters) mark limestone blocks at the large farm‐villa at Sidi Eysawi (TUT53).
In the Wadi Guman, another quarry was identified on a hill edge just 100 m south-
west of a villa with mosaic and bath. Two crushing stones (three crushing stones of the
same type were recorded in TUT16) have been found in this quarry. The different kinds
of architectural elements used at the olive farms such as orthostats, press floors,
counterweights, columns, capitals, bases and the well-cut holes, incisions and channels
give an impression that the quarry workers were skilled in cutting and dressing with
perfect accuracy all these types of stone masonry.
Most of the farming sites described here seem (from surface evidence) to have
comprised two main parts: a working part and storage facilities part. These two elements
were, of course, common to most farming sites of whatever size in the Gebel Tarhuna,
although their design varied considerably from one site to another. The two parts were
sometimes combined into a single architectural building, especially in a group of these
146
farms built on a larger scale such as at Senam Aref (DOG60, Fig. 3.35a), Sidi al-Akhder
(DOG 66, Fig. 3. 35b) and TUT3 (Fig. 3.36).
The vast majority of farming sites in the Tarhuna plateau are generally assumed to
have been focused on producing a single agricultural product, olive oil. However, with
the absence of excavation work and flotation analysis of the pressing deposits, we
Figure 3.35: a) Senam Aref (DOG60).
Figure 3.35: b) Sidi al‐Akhder (DOG66). 0 25m
147
Figure 3.36: TUT3.
cannot reject the possibility that some of presses might also have been worked as wine
presses. This hypothesis is especially valid in relation to sites that have not provided
evidence of olive mills or where there were large vats and presses. Wine pressing
elements are definitely similar to those at olive farms and can be hard to differentiate
without exceptional preservation at the surface, or excavation. The impression of the
organisation and function of these farming sites is dominated by the quality of ashlar
building. These sites were evidently constructed in order to produce large surpluses of
olive oil (and wine?). Thus, the archaeological records show that the pressing facilities
occupy a large percentage of the total site area. Although no site can be fully recorded
148
without an excavation, the pressing facilities in a number of sites apparently extended to
about a third out of the total surface area of the site.
3.3.2. Rural baths
It is commonly known that many Roman villas contained bathing facilities. Baths
associated with villas have been considered as an indicator of luxurious living practiced
at these sites. The archaeological evidence of the rural settlement of the Tarhuna plateau
reveals that bath-houses were among the most important ‘palatial’ characteristics of villa
sites. A small number of baths had been recorded by previous work in the Gebel
Tarhuna. Goodchild partly uncovered the frigidarium room from a bath-building
located at the Medina Doga (E in Figure 2.16, Goodchild 1976: 78). A small bath-house
was identified by Oates alongside the Udei el-Me (Fig. 2.14) in the Gasr Ed-Dauun
village (Oates 1953: 90). The two examples were associated with road-side
settlements/small towns. In addition, he mentioned two other small bath-buildings, with
mosaics, one associated with the Sidi Hamdan oilery (Oates 1953: 99), and the other
located with an oilery (Gasr Shāeir) in the upper Wadi Turgut and overlaid by a fortified
construction of the late Roman period (site no. 13, Oates 1953: 105-6).
In addition to these baths, the other bath-buildings recorded on the Tarhuna plateau
can be divided into two types. The first type is characterised by its association with
farming sites and usually was close to the main farm building (e.g. TUT1, TUT19,
TUT53, TUT54, DOG80, DOG105 and Halafi1). The second type were isolated
structures very often sited at the location of perennial springs (in Arabic ain, prl. aioun),
as at Ain Wif, Ain Doga, Ain Guman, Ain Hamzia and Ain Tarabout. There is also one
isolated bath-house recorded near a well, which is known by the name Bir Attwafga (in
the Wadi Garāah, 5 km south-west of Gasr Ed-Dauun). The baths/springs type appears
149
to be larger and more complex, and it is safe, perhaps, to say that they benefited from
the water provided by the springs where there may also have been shrines or nymphaea.
A total of 15 baths of both types has been recorded in the Gebel Tarhuna; of which
nine baths were found in association with agricultural villages (such as Astail and Gasr
Dehmesh) or with the other farming sites of varying sizes. Six baths were located where
there were springs (Fig. 3.37). Some of these rural baths still have a good preservation
that allowed me to draw their plans, as at Ain Guman, Bir Twafga and Astail village. At
Bir Twafga, the site had completely disappeared under a thick layer of soil from, most
likely, ancient times until autumn 1996, when a great flood uncovered part of the site.
Figure 3.37: Distribution of rural baths in the Gebel Tarhuna.
The original burial of the bath had occurred probably due to its location at the meeting
point of the main course of the Wadi Garāah with one of its tributaries. The gully
erosion in 1996 exposed a number of rooms, in particular at the eastern side where the
internal walls still survive to 2 m or more in height. A preliminary observation of the
0 6 123 Km
Rural baths
XY Ain bath
% Farm bath
! Modrn centers.
±Modern centres
150
eastern room revealed that the walls were painted in green, brown and yellow colours
(Fig. 3.38). At the east side also there is a hypocaust system to heat the tepidarium or
caldarium room which opens onto another room placed in the centre of the bath-
building and crossed by an opus signinum channel. At least three further rooms are
partially exposed underneath the alluvium in the northern side of the gully.
Figure 3.38: A bath‐house at Bir Twafga.
Figure 3.39: A general view of Ain Guman villa/bath showing the opus africanum structure of its eastern wall.
151
In the Wadi Guman, above the spring, there are remains of a probable villa with bath
suite. The outer walls were built in the opus africanum style as still can be seen along
the eastern side of the building (Fig. 3.39). The bath rooms appear to have extended
onto the northern side, where at least two rooms were paved with mosaics and fragme-
a
b
Figure 3.40: Mosaic (a) and tile, bronze pipe (b) at the Ain Guman villa/bath.
nts of marble are still plastered on the wall. A second feature of the bath comprises
pieces of a c. 2.5 cm diameter bronze pipe found on the surface and which likely formed
a pipe providing the hot water to the caldarium room on the east side. There is also a
152
huge quantity of tile (including specialised bath tiles) scattered in and around the site
(Fig. 3.40).
The location of the villa was carefully chosen in the upper part of the wadi close to a
dam where a large area benefited from the enhanced fertile soil (Fig. 3.41). From the
landscape perspective, the site seems to have been placed within gardens which
contained different kinds of fruit trees. Some of these trees such as dates, olives,
pomegranates and figs are still planted in the wadi. The architectural elements of the
above two examples indicate a large size of rural baths located (635 m2 and 950 m2
respectively). They show some prestige signs indicating the involvement of the urban or
local elite who had the wealth and power to invest in such sites in the rural hinterland.
The size and ornamentation of the bath suites including mosaic pavements, marble slabs
and painted walls, make them comparable with baths related to villas in the coastal area.
They provide evidence that the elite families of Tripolitania made a significant
investment not only in pressing equipment, but also spent part of their profit on rural
structures related to display and conspicuous consumption.
Figure 3.41: Location of villa/bath and the dam at Ain Guman.
153
Some rural baths in the Gebel Tarhuna can be considered as indicators of wealth. The
size and quality of the mosaic tesserrae in Ain Guman, Ain Doga, Sidi Eysawi and
Halafi reveal a similarity with the Ain Scersciara villa mosaic (I rediscovered this
mosaic in 1997 after its disappearance since Goodchild’s days). Goodchild suggested
that the mosaic of Ain Scersciara was a type that can be dated to the second century AD
(Goodchild 1976: 85). The ceramic evidence scattered on the surface of these sites
indicates their establishment in the early imperial period. However, some late ARS and
TRS sherds and fourth century AD coins were also collected (in association with earlier
material) from Ain Guman and Bir Twafga. The later Roman period evidence clearly
indicates that there was some continuity of use of baths and villas until late antiquity.
3.4 Settlement density and diversity
The term ‘settlement’ is applied, in most cases, to any site that has material culture
remains in close association with architectural features spread over an observable space,
or including several distinct structures. The Roman period rural settlement in the
Tarhuna plateau evidently had a distinctive economic predilection and this had
significant consequences in shaping the general relationships between the settlement
distribution and the landscape.
The density of rural settlements in the Gebel Tarhuna seems to have depended on
their ability to benefit from the available lands for cultivation in order to derive income
from olive oil (and possibly also wine) production. Most of the varying types of
settlement were concentrated along the wadis. Wadis and their adjacent hills and slopes
presented preferable places for settlement location; the best soils were in these valleys,
and these soils retained moisture better in drying years. Similar patterns are visible in
many other hinterland areas of Roman North Africa (Barker et al.1996; Barker and
154
Gilberston 2000; Carlsen 1989; De Vos 2000; Felici 2006; Hitchner 1988; 1990;
Mattingly 1989b; Vita-Finzi 1961). As has been discussed in Chapter 2, the preferential
areas for settlement were the main wadis, where the most fertile were lands were
concentrated and along which ran the main communication routes.
Approximately 115 km2 have been surveyed by the TAS in the Wadis Turgut and
Doga (Fig. 2.7) and 111 sites have been recorded. A remarkable density of rural
settlements has been identified. The distribution of rural sites in the Wadis Turgut and
Doga (Fig. 3.42) shows a formidable density of settlement, entirely consonant with the
agricultural development of the Roman period. However, there are other criteria for
defining the characteristics of rural space. Dealing with the rural settlement patterns, it is
necessary to understand two principal factors which played a fundamental role: the
settlements’ geographical distribution, and the density of settlement sites. It is definitely
clear from the survey results that the settlement pattern was governed by these agents.
As already noted, the wadi areas were more favourable for settlement than other
geographical features. Settlements needed to be close to cultivable land and, because of
the low and irregular rainfall, in locations where they could exploit nearby water
catchment surfaces. As a consequence it has been found that most of the settlement sites
in the survey area, in particular the oileries and large farms, concentrated in or in the
vicinity of the main wadi courses.
Although the chosen area for survey has only been extensively surveyed, the state of
the countryside and its exploitation levels in the Roman period can be evaluated. The
data from the survey provide the ability to produce maps that show settlement
distribution, and distinguish settlements that potentially served as central foci of estates.
On the other hand, it is helpful to make a judgement whether the Tarhuna plateau was
155
characterised by nucleation or dispersal of settlement. The ceramic evidence collected
from most of the sites, moreover, allows me to show a chronological dimension to the
settlement pattern on the Tarhuna plateau.
The well-documented distribution map of the Wadis Turgut and Doga illustrates that
the rural settlement pattern was a dispersed one. Although the evidence reveals that
some settlements were nucleated around attractive poles such as in the Msabha area in
the Wadi Doga and around Henscir Assalha in the Wadi Turgut, in general, the
distribution pattern could be described as scattered even though some settlements were
highly concentrated on a broad scale near the wadi courses. The important
communication tracks through the wadis appear to have attracted settlement, especially
as regards the large olive oil pressing centres and pottery production sites. Some rural
settlements benefited from their location near or by the Roman Gebel road or other
tracks and some eventually became small towns and agricultural villages.
A large number of Roman period sites (over 200 sites) are known from previous
works in many parts of the Gebel Tarhuna. However, they also indicate that the areas of
the plateau that were most densely occupied were the fertile soils of the wadis, and this
pattern is also paralleled by the system recorded by the UNESCO Libyan Valley Survey
in the pre-desert zone to the south (Barker et al. 1996). As noted earlier (Figures 2.19;
3.13; 3.16), although the wadi beds of the north-eastern plateau and their adjacent slopes
were naturally the most favourable in terms of site location, settlements also occupied
extensive areas around the upper northerly tributaries which led into the Wadi Targelat
(Cowper 1897; Oates 1953; 1954). The density of rural settlements in the Wadis Turgut
and Doga can be examined from the evidence of the field survey, to see if there was a
concentration of sites within specific areas in the Roman period. The middle of the Wadi
156
Turgut appears to have had the highest density for settlement (Fig. 3.43). A total of 27
rural sites were noted in an area of about 30 km2. This sector of the Wadi Turgut seems
to have been dominated by four oileries and five large farms during the early imperial
period. Three sites of pottery kilns, four small farms and five dams probably were all
constituent elements of large estates governed by the oileries and large farms. Except
perhaps the dams, all these sites declined and might have been abandoned and/or
replaced by the six fortified farms (gsur) probably from the late third century AD.
Agriculture appears to have been somehow specialised, based in particular on
cultivation of olive groves alongside the wadis, as traced by the distribution of olive
farms. Sites, especially olive farms, reveal the significance of the wadis as both a
terrestrial and fluvial network that facilitated communication between the coastal area to
the north and the interior of the region. In the Wadi Turgut, the western side of the
valley was evidently more favourable than the eastern one, indicated by the fact the
majority of ancient sites were established in the west. This might have happened as a
result of there being larger tributaries such as the Wadis Astail, Guman, Tershan and
Hwatem, which most likely formed principal communication routes to the south, north
and west of the Gebel Tarhuna. In relation to the Roman period rural sites distributed
along these wadis, the distance between each settlement ranges between about 100 m
and a few kilometers (Figs. 3.42; 3.43). For example, considerable new evidence has
emerged from the TAS regarding the site Henscir Assalha (TUT15) at the Wadi Turgut.
The oilery farm can now be seen in clear relation to other archaeological features
located just to the west, where there are traces of four possible pottery kilns and three
dams (Fig. 3.44).
Figure 3.42: Distribution of the rural sites in the W
adis Turgut and Doga.
#Sm
all Farm
!M
odern centre
#O
ilery
Pottery kiln
"Fortified farm
#Large Farm
XM
ausoleum
¢Baths
157
Figure 3.43: Density of rural sites in the m
iddle of the Wadi Turgut.
158
159
Figure 3.44: The complex site of Henscir Assalha (TUT15) in the Wadi Turgut.
The evidence emerging from the field survey shows clustering of rural settlements of
varying types. The findings have led to an overall acceptance of some characteristics of
settlement patterns of the Gebel Tarhuna during the Roman period. A total of 113 rural
sites (from the 115 known sites) in the Wadis Turgut and Doga have been classified and
presented in Figure 3.45 and Table 3.9. The categories of sites characterised by an
agricultural function represent 78 percent (excluding the pottery kilns), and 12 percent
of them have produced luxury signs. This high overall percentage is a normal
consequence of the fact that the agricultural economy, mainly based on olive oil
production, was the factor behind the settlement patterns in this area during the Roman
period. The fortified farmhouses represent 24 percent of the total number of sites, and
also 31 percent of the agricultural sites, suggesting continuity of some estates into late
Roman/late antique times. There was a small number of recorded mausolea. However,
FigDo
ev
fe
et
su
w
Tr
fe
re
19
gure 3.45: Pieoga.
ven much c
ew recorded
t al., 2004:
urveyed its l
were no mau
ripolitanian
eature and d
ecorded by t
996: 145). T
7%
12%
24%
e charts show
closer to Le
d rural maus
13). Even in
left bank du
usolea recor
n pre-desert
does not see
the UNESC
These seem
24%
%8%
5%3% 3%
%13%
ing the settle
pcis Magna
solea with o
n the centra
uring two se
rded (Felici
t, south of
em compara
CO Libyan V
m to suggest
14%
3%
ement diversit
a, the Silin
only two ma
al Wadi Tar
easons 1999
i et al. 2006
the Tarhun
able. In a mi
Valleys Sur
t a much str
15%
18%
87%
ty and their p
area has sh
ausolea (no.
raglat (wher
9-2000) out
6). In sharp
na plateau,
inimal coun
rvey in such
ronger deve
Oile
L‐F
S‐F
Gas
Dam
P‐K
Bat
Ma
AgricuLuxuriGasrNon-ag
ercentages in
hown a simi
. 17, 49) ou
re I joined t
t of about 5
p contrast to
demonstrat
nt, over 70 m
h a margina
elopment of
ery
arm
arm
sr
m
Kiln
th
ausoleum
ulturalious
gricultural
n the Wadis T
ilar phenom
ut of 63 sites
the Italian t
0 ancient si
o this evide
tes a very
mausolea ha
al area (Bark
f a rural eli
160
urgut and
menon of
s (Munzi
team that
ites there
ence, the
different
ave been
ker et al.
ite in the
161
latter area than was the case on the Tarhuna plateau, where the influence of large estates
controlled by the urban elite was perhaps more the norm.
The archaeological evidence from the TAS has shown varying types of rural sites,
but the olive farm buildings with their oil presses were the main settlement element (see
Chapter 4). In the intensively surveyed area of upper Wadi Guman, the surface evidence
reveals a quite dense pattern of various rural sites (Fig. 3.46). Site density, in this area,
appears to have evolved over the course of time. The archaeological evidence reveals
that sedentary farmers had settled and exploited the area from probably the second
century BC to the seventh AD. A hoard (373 coins) of Massinissa, king of Numidia (238
– 149 BC), discovered in 1995 (Fig. 3.47), can be compared with other evidence from
the Wadi es-Sri. Although the wadi has not yet produced a farming site of large scale (it
is difficult to judge from the surface evidence in this area because ancient building
materials have mostly been removed from the surface and reused in modern
constructions), the Wadi has produced the highest density of pottery production sites in
the Gebel Tarhuna. These pottery kilns were specialized in the Tripolitanian amphorae
(see Chapter 5), and they concentrated in an area which offered the key factors for this
economic activity: water, clay sources and communication. The first factor provided by
the existence of a spring of Ain Guman, with the second aspect served by many clay
bands in the wadi-sides. The third factor related to two major tracks or routes which
have been used until modern days: the first runs from north-west to south-east, and is
known by the name attariq atrablsia (the Tripoli road), and the second runs east-west
and is known as attriq msellatia (the Msellata road).
Although it was already well known that the Gebel Tarhuna’s archaeology was
characterised by olive-presses, the data from the TAS have increased the total numbers
of
kn
TA
w
Sc
ad
di
Fo Opus Figure 3.46:
f the rural
nown from
AS. Perhap
which has in
cersciara an
dding the 14
iscussed in C
ortified site (gass siginum. Q
The density o
settlements
works of C
ps the most
ncreased re
nd Sidi Sai
4 new sites
Chapter 5).
sr). DaQuarry. T
of sites in the
s. Table 3.8
Cowper, Goo
interesting
emarkably
id (Arthur
identified b
am. EnclosuTerrace wall.
intensively su
8 gives est
odchild and
value in thi
from three
1982; Good
by the TAS
ure. Pottery Bath. H
urveyed area o
timation for
d Oates and
is table is th
e recorded
dchild 1976
S (these new
y kiln. FarmHoard.
of the upper W
r numbers
the new tot
he number o
sites at Ga
6; Oates 19
w recorded
stead. Track.
Wadi Guman.
of sites pr
tal number
of pottery k
asr Ed-Dau
953) to 17
sites will b
162
reviously
after the
kiln sites,
uun, Ain
sites by
e further
163
Figure 3.47: A sample of the hoard of Numidia coins found in the upper Wadi Guman.
Before TAS After TAS Sites 205 297 Presses 262 415 P-kilns 3 17
Table 3.7: Some of the estimated figures of the rural settlements before and after the TAS in the Tarhuna plateau.
3.5 Evaluation of settlement pattern over time
The state of the Gebel Tarhuna countryside and its settlement levels in the Roman
period can now be evaluated in the light of the available evidence. Although the
available data do not cover the whole of the Tarhuna plateau, the settlement features and
chronology can be stated. The archaeological data are sufficient as regards the east and
north-east of the Gebel Tarhuna which allow me to produce a map with evident
chronological data (Table 3.8).
Intensive collection strategies and the study of surface material can reveal the
location of sites, even where their surface remains have been entirely destroyed by
164
construction work or deep ploughing (such as in many places in the high Tarhuna
plateau where the Italian established new farms in the colonial period). The peaks and
troughs of landscape exploitation can be identified chronologically (Dunnell 1983;
Lewarch 1981). The collection and examination of imported and local products may
reflect the extent to which the settlements were involved in local, regional, provincial
and empire-wide economic systems (Barker 1991: 6; Lloyd 1991: 238). This type of
interaction is only very rarely perceivable in the literary texts (Barker et al. 1995). Many
of recent archaeological surveys have mapped the distribution of potsherds around the
dense concentrations of building materials identified as settlement sites, and interpreted
the assemblages as evidence for the occupation period of these settlements (Bintliff
1985).
It seems clear that the beginning stage of farming the Tarhuna plateau, which was
tentatively dated by previous studies to the early first century AD, can be preceded into
the second and first century BC. The associated pottery sherds allowed me to find out an
approximate date for the vast majority of the sites that have been recorded or rerecorded
by the TAS. The gathered data show the development of settlement trends (Table 3.8),
providing a preliminary chronological overview of the settlement patterns on the
Tarhuna plateau.
Many of the best agricultural lands in the Tripolitania were, in fact, located in the
Gebel. The ancient exploitation of the Gebel lands was linked to the agricultural
development of the coastal area (Mattingly 1995: 140). The Gebel Tarhuna-Msellata
occupies the central and eastern limit of Lepcis’ territory, while the eastern boundary
between Lepcis and its subsidiary town of Thubactis is not definable in terms of
epigraphic evidence, but it was likely that the Wadi Taraglat-Gaam formed the extreme
165
eastern extension of the Lepcis’ territory related to the Gebel (Felici et.al., 2004: 634).
Mattingly argues that the territory of Lepcis Magna was very extensive, perhaps as
much as 3000-4000 km2 (De Vita Evrard 1976; Mattingly 1996: 167). He also believes
that this expansion to the Gebel Tarhuna-Msellata reflected the view of the
Libyphoenices that these lands formed one of their heartlands (Mattingly 1988: 21-41;
1995: 140; 1996: 167-8). A measurement by the Google Earth software (3.530 km2, fig.
3.48) has confirmed Mattingly’s estimation of the territory of Lepcis, taking in account
Felici’s judgment that the Wadi Taraglat-Gaam was its eastern limit.
Figure 3.48: The measurement of territory of Lepcis by the Google Earth (3.530 km2) which is equivalent to Mattingly’s estimation.
The new archeological evidence provided by the Silin and Wadi Gaam-Taraglat
surveys has shown that some rural settlements, in particular farms and villas, took their
place in the landscape as early as the fourth - third century BC (Cifani and Munzi 2002).
The evidence from Silin area indicate that a total of 28 out of 63 sites has been dated to
a period between the fourth and the first centuries BC: only one site certified as from the
166
No.
Site Type Period II. BC I. BC I. AD II III IV. V VI VII
1 TUT1 L. farm/villa, qasr 2 TUT2 S.farm, qasr 3 TUT3 L. farm, qasr 4 TUT4 S. farm 5 TUT5 L. farm, qasr 6 TUT6 Hill-top qasr 7 TUT7 L. farm, qasr 8 TUT8 Oilery 9 TUT9 S. farm 10 TUT10 Oilery 11 TUT11 L. farm 12 TUT12 Oilery/villa 13 TUT13 Hill-top qasr 14 TUT14 L. farm, qasr 15 TUT15 Oilery/villa, pottery kilns 16 TUT16 Oilery, gasr 17 TUT17 Hill-top qasr 18 TUT18 Pottery kilns 19 TUT19 Bath 20 TUT20 Oilery 21 TUT26 L. farm 22 TUT27 L. farm 23 TUT28 Hill-top qasr 24 TUT29 L. farm 25 TUT30 Gasr 26 TUT31 S. farm 27 TUT32 S. farm 28 TUT33 Gasr 29 TUT34 Hill-top qasr 30 TUT35 L. farm 31 TUT36 L. farm 32 TUT37 Hill-top qasr 33 TUT38 Oilery/villa 34 TUT39 S. farm 35 TUT40 S. farm, qasr 36 TUT41 S. farm 37 TUT42 Hill-top qasr 38 TUT43 Oilery/villa 39 TUT44 L. farm 40 TUT45 S. farm 41 TUT46 Oilery/villa 42 TUT47 S. farm 43 TUT48 Pottery kilns 44 TUT50 Gasr 45 TUT51 Hill-top qasr 46 TUT52 Oilery/villa 47 TUT53 L. farm/villa 48 TUT54 Oilery/villa 49 TUT55 Hill-top gasr 50 TUT56 L. farm, gasr 51 DOG57 L. farm 52 DOG58 Gasr 53 DOG59 Hill-top gasr 54 DOG60 Oilery/villa 55 DOG61 Quarry 56 DOG62 Hill-top gasr 57 DOG63 Hill-top gasr 58 DOG64 S. farm 59 DOG65 Hill-top gasr 60 DOG66 Oilery 61 DOG67 S. farm, gasr 62 DOG68 S. farm, gasr 63 DOG69 Gasr 64 DOG70 Hill-top gasr 65 DOG71 Inscription
167
No. Site Type Period II. BC I. BC I. AD II III IV. V VI VII
66 DOG72 Mausoleum 67 DOG73 Bath 68 DOG74 S. farm 69 DOG75 Gasr 70 HAJ76 Hill-top qasr 72 HAJ77 Watchtower 73 HAJ78 Mausoleum 74 HAJ79 Gasr 75 HAJ80 Bath 76 HAJ81 L. farm/villa 77 HAJ82 Oilery 78 GUM83 Hill-top qasr 79 GUM85 Quarry 80 GUM86 Pottery kiln 81 GUM87 Bath/villa 82 GUM88 S. farm 83 GUM89 Pottery kilns 84 GUM90 Pottery kilns 85 TEL91 Mausoleum 86 TEL92 Gasr 87 TEL93 Watchtower 88 TEL94 Gasr 89 TEL95 S. farm 90 TEL96 S. farm 91 TEL97 S. farm 92 TEL98 Gasr 93 TEL99 S. farm 95 TEL100 Bath 96 TEL101 Bath 97 TEL102 Pottery kilns 98 DOG103 S. farm 99 DOG104 L. farm/villa 100 DOG105 S. farm/villa 101 DOG106 Oilery 102 DOG107 Oilery 103 TUT108 Pottery kilns 105 TUT109 S. farm 106 GUM110 Pottery kilns 107 DOG111 L. farm, quarry, kiln 108 TUT112 L. farm 109 SRI113 Bath 110 SRI114 Pottery kilns 111 SRI115 L. farm, cemetery 112 SRI116 Hill-top gasr 113 TRG117 S. farm 114 TRG118 S. farm 115 TRG119 S. farm, quarry 116 TRG120 Gasr 117 TRG121 S. farm 118 TRG122 Gasr 119 TRG123 Gasr 120 TRG124 Gasr 121 DUN128 Oilery 122 DUN129 L. farm 124 DUN130 L. farm 125 DUN131 Oilery 126 DUN132 S. farm 127 DUN133 Bath
Table 3.8: A synoptic table of the rural archaeological sites (excluding dams) they have been recorded/rerecorded by the TAS.
Fig
fo
20
in
co
se
sit
Fi
w
an
H
id
‘C
Te
gure 3.49: The
ourth - third
004: 19). Th
n such areas
ollected by
ettlement is
te (SRI 115
igure 1 in A
ware, produc
nd small am
Hddad and
dentified. M
Campana A
erra sigillat
0
10
20
30
40
50
60
70
80
90
100
2 BC
e number of r
d, 11 sites o
he increase
s close to Le
the TAS at
located 12
5), has prod
Appendix). T
ction ‘Camp
mount of fra
Asmia for
Mainly their
’ bowls and
a.
C 1 BC 1 A
rural sites hav
of second a
of rural set
epcis seems
ttest the sit
2 km west o
duced the ea
The collecte
pana A’ ass
agments of
rthcoming).
r dating ev
d dishes and
AD 2 AD 3
e recorded/re
and 16 sites
ttlements du
s to have ex
tes’ chronol
of Tarhuna
arliest evide
ed material
sociated wit
the Van De
From the
vidence is b
d the mid-fi
AD 4 AD 5
erecorded by
s from the f
uring the se
xpanded to
logy from t
town in a s
ence on the
s include m
th the secon
er Werff (ty
first centu
based on la
first century
5 AD 6 AD
the TAS and t
first century
cond and th
the Tarhun
the second c
small valley
e Tarhuna p
many sherds
nd century B
ype 7c) and
ury BC si
ate Punic a
BC Arretin
7 AD
their dates.
y BC (Mun
he first centu
na plateau. M
century BC
y (the Wadi
plateau (Tab
of the blac
BC Numidi
d Dressel 1-
ix rural sit
amphora, s
ne ware and
no. Sites
168
nzi et.al.,
uries BC
Materials
C: a rural
i es-Sri),
ble 1 and
ck glazed
ian coins
-type (al-
tes were
herds of
d eastern
169
This evidence demonstrates that the Gebel Tarhuna witnessed a degree of
exploitation from the pre-Roman period (late Punic and early Numidian periods, the
second and the first centuries BC). However, the previous works of Goodchild and
Oates did not record any evidence dated before the first century AD (Goodchild 1976;
Oates 1953). It might be assumed that the pre-Roman settlements and agricultural
practices on the Tarhuna plateau were influenced by the policy of the Numidian kings.
Ancient historians mentioned how Massinissa ‘turned’, said Strabo “nomads into
farmers and welded them into a state” (xvii.13.15). Polybius also described him as a
great cultivator encouraging his people to be farmers and settlers (xxxvi.16.7-8).
However, the archaeological records indicate that agriculture was practised in at least
some parts of North Africa long before Massinissa’s reign (Cherry 1998; MacKendrick
1980; Mattingly 1995; Whittaker 1980). It is difficult to judge the significance of the
large number of Massinissa’s coins that have been found in some sites in the Gebel
Tarhuna. It is recorded that Massinissa briefly held the area south and west of the
Emporia in 204 BC (Livy xxix.3.9; Mattingly 1995: 51). The economy of the Emporia
grew after the defeat of Carthaginians at Zama by Scipio Africanus in the second Punic
war in 201 BC, because they became free from the one Talent daily tribute they had to
pay to Carthage (Di Vita 1982). The increase in the number of rural settlements
established especially in Leptic territory, after the end of the second Punic war, reflects
the agricultural development, particularly for olive oil, in the region.
Compared to the small number of the pre-Roman settlements, the early Imperial
period registers a great development of rural sites in the Gebel Tarhuna. The TAS found
about 43 percent of the total recorded sites existed at this period. This increase
especially was applied to open farms (from small farm to oilery) and pottery kilns
170
(Table 3.8 and Fig. 3.49). During the Roman period, the region experienced a new phase
of development, especially from the first century AD onwards. Settlement intensified
and agricultural settlements flourished. Mattingly suggests that the Gebel Tarhuna-
Msellata was a desirable region due to market and investment demands (Mattingly
1985; 1987; 1989a; 1995). The expansion of settlement reached the marginal lands in
the pre-desert area. This period was also characterised by a greatly increased density of
settlement, and evidence for highly intensive commercial farming. The flourishing of
settlement patterns and the maximum economic expansion appear to have continued into
the end of the third century AD, but probably not as strong as previously after the
Severan era. Ceramic surface materials of the first three centuries AD are considerably
more abundant and more widely distributed throughout the region than those from
earlier or later periods.
The first massive decrease in the open farms seems to have started in the fourth
century AD, though this was also a period of emergence of the defended sites (gsur)
which dominated the landscape from this time and forwards. Overall settlement
numbers thus remained relatively constant. The TAS has identified that about a half of
open farming sites had been replaced by the fortified ones from the fourth century AD.
Ceramic evidence collected from these gsur reveal that some of them, especially the
hill-top type, continued in use until the early Islamic period.
Rural fortification does seem to have been a very widespread phenomenon in the late
antique period. The settlement pattern of the Gebel Tarhuna appears to have been
dominated by the gasr type in much the same way as the pre-desert area when this kind
of structure functioned as an agricultural installation. Mattingly suggests that from the
fourth century the gsur became increasingly important, replacing unfortified farms as
171
the dominant form of settlement (Mattingly 1987b; 1989b; 1995). Survey evidence for
the later Roman period from the Tarhuna plateau indicates increased construction of the
fortified farms, and that these expanded into areas which had not been settled very
densely in earlier centuries. This implies increased using of high lands and hill-tops that
had not been chosen for buildings earlier, when the preference was for activity in the
wadi beds and foothills. This change in settlement patterns probably relates to an
increased demand for defence and security. This trend was caused by a growth in the
risk posed by raiders, who probably came from the interior. Archaeological studies from
the Tripolitanian pre-desert reveal that many of these fortified farms were built by the
landowners who were stimulated by local tribes’ incursions to take measures for their
security (Barker et al. 1996; Brogan 1977; Filici et al. 2004; Mattingly 1995). O.
Brogan from her studies of some ancient sites in eastern Tripolitania added that the
careful sighting of many of later Roman buildings indicates that one of their main
functions was to guard the routes and signal the approach of strangers (Brogan 1977).
The rise of gsur was accompanied by progressive decline in overall site numbers, to
judge from the data for sites with pottery of fifth to seventh centuries (Fig. 3.49).
Goodchild argued that the fortified farmhouses (gsur) of the Gebel Tarhuna had an
uneven domination; they formed the great majority of the ancient farm sites and
occupied the upper rank of the settlement hierarchy in the late Antiquity (Goodchild
1951: 88 – 89). Certain gsur were built to replace the previous open farms within the
boundaries of the previous estates, and their positions suggest that they were erected to
guard the approaches to the main farms. Epigraphic evidence from a number of fortified
farms shows that some of these gsur were private property built by the landowners in
selected positions on their lands to provide security and protect the boundaries of their
172
farms. As in the pre-desert as in the Gebel, a number of inscriptions confirm that the
gsur were erected on behalf the landowners. For instance, an inscription from the gasr
at the Bir Scedua records that a certain ‘Flavius Dasama and his son Macrinus, the
landowners, built the gasr to protect their own estate (Elmayer 1983). Another example
from Sidi Ali ben Zaid in the Gebel Tarhuna mentions that a certain Thiana Marcius
Cecilius constructed a centenarium and a small altar, and that he lived in a state of grace
(IRT 877). The inscription from Sidi Sames (Sidi Assid, Tarhuna) indicates how
perhaps some of these gsur were constructed on family estates to protect the inhabitants
against an expected danger from the gentiles and barbari (Goodchild 1976: 112;
Mattingly 1995: 195). Although I have judged from the surface evidence alone, I would
have to conclude that this chronological statement and settlement evaluation could be
modified by the further survey in other parts of the plateau and/or by excavation.
173
Chapter 4: Olive oil pressing facilities and pressing process
4.1 Introduction
The history of the Mediterranean world is linked to the cultivation of the olive.
Geographically, the distribution of olive tree corresponds to the Mediterranean climate
zone where it has been densely cultivated since ancient times, in particular during the
Roman period (Amouretti 1986; Brun 1986b; Foxhall 2007; Mattingly 1996a). For
many people in antiquity the production of olive oil and wine for the markets marked the
difference between the civilized provinces and the barbarian lands (Lanfranchi 2009).
Under the Roman Empire a number of regions or provinces, such as Italy, Baetica, and
Africa developed agriculturally and became specialised in the production of wine and
oil. Oil and wine were important commodities and in high demand not only in Rome, the
capital, but also in the other large cities such as Alexandria, Carthage, Antioch and, in
late antiquity, Constantinople (Brun 2003). Tripolitanian amphorae carring Tripolitanian
olive oil (and wine) have been inter-regionally and broadly recorded (Mattingly 1988a).
Particularly the Tripolitanian amphora III has been identified at many other centres,
especially in the western Mediterranean (Carandini 1970; Marquez Villora 1999;
Remesal Rodriguez 2004).
The olive tree is one of the three components of the Mediterranean triad (wheat, vines
and olives). It occupied a prime position in the life of the Mediterranean peoples in their
beliefs and religious rites and mythology, and also was one essential component in
nutrition, lighting and body care (Mattingly 1988b, 1996b). In Roman times, the
cultivation of the olive tree in Africa was encouraged by the imperial power, enhancing
lands which were once grazing lands or forests (Kehoe 1984). These incentives were
174
reflected in the well known agrarian tenancy law, the lex manciana, a text specific to
Africa and relating to management of imperial estates that emerged there under the
Flavians and more specifically under Vespasian. Clauses in the Henchir Mettich text and
lex Hadriana text allowed tenants to acquire specific rights over orchards and vineyards
on land they leased. While the lands remained the emperor’s property, tenants
effectively owned what they did to enhance the land and could transmit to their heirs
(Carcopino 1906; Kehoe 1988). It is likely that similar provision applied to imperial
estates in Tripolitania, though no epigraphic proof survives.
Ancient Tripolitanian production of olive oil has been studied increasingly in recent
years (Mattingly 1993b). One of the main realisations to emerge from such efforts is that
the agriculture practised in the Gebel Tarhuna was central to the development of the
Tripolitanaian coastal centres, especially Lepcis Magna. As Mattingly has formulated it
‘The existence of well-developed agricultural lands extending so far into the Gebel
certainly helps to explain the wealth of many of the Lepcitanian aristocracy from the
Augustan period. The primary cash crop of the Gebel farming was olive oil, as is made
clear by the abundant evidence for olive presses, though no doubt a far wider range of
produce was grown’ (Mattingly 1995: 141). Rural settlements in Roman Tripolitania are
predominantly seen from the point of view of agricultural production. More than 150
farming sites with presses have been identified, and the number of presses varies
between 1 to 17. The archaeological evidence has indicated that the olive cultivation and
olive oil production were the defining characteristics of the Gebel Tarhuna landscape.
Furthermore, the numerous remains of olive oil production during the Roman period
support its significance as the most important economic resource in Roman Tripolitania,
175
echoing Columella; description of the olive as ‘first among all trees’ (De re rustica,
5.8.1).
As we shall see in this chapter there is some evidence to suggest that not all
Tripolitanian presses were for olive oil production, some were used to produce wine as
well (how many is difficult to evaluate accurately, without excavation). While I follow
the consensus view that most presses were used for oil, the possible extent of wine
production will be discussed later in the chapter.
The production of olive oil has undergone transmutations through incessant
innovations in techniques since ancient times. However, some primitive methods have
been preserved especially in traditional societies, still only partially integrated in the
industrial revolution. The archaeological attribution of ancient processing facilities
specifically for olive oil, wine and other liquids is a fundamental and recurrent issue.
Significantly, technological development in the Roman period had comparatively little
effect on the quantity or quality of wine or oil produced per volume of fruit, but rather
more impact on the volume of oil or wine processed per press per hour. The well-
preserved evidence, in particular for the large presses of the Tripolitanian Gebel and
some other parts of the North Africa, has allowed scholars to estimate the productive
potential of these presses and to speculate on the major capital investment required in
such large presses to produce a large scale of output. Citing the size and density of these
presses, and the volume of associated ceramic production, Hitchner (1993: 499-508)
argues that the scale of olive oil production and export in the Roman empire exhibited
‘real economic growth’ sufficient to justify the rejection of the Finley orthodoxy that
sees the Roman economy as underdeveloped and thus capable, at best, of only modest
growth (Finley 1985).
176
In a series of articles, Mattingly has surveyed the production of olive oil and
endeavoured to assess the scale of a number of well-preserved lever and windlass type
olive presses in Roman Libya and Tunisia (Mattingly 1993b; Hitchner and Mattingly
1991). His study provides the most detailed example of how quantitative research can be
utilised effectively to interpret scales of regional production and of how results would be
related to the agricultural economy. Mattingly combines textual, archaeological and
ethnographic evidence in a detailed and convincingly argued estimate of the potential
output of olive oil presses in North Africa. He suggests that ‘peak olive oil production in
Tripolitania for one of the larger presses could have been 9,000-10,000 kg, in Tunisia
for the Kasserine presses 5,000-10,000 kg; in the Libyan pre-desert for small presses
2,500-5,000 (Mattingly 1993b). Mattingly (1988a: 21) justifies the use of quantification,
because “... to talk simply of ‘huge production’ and ‘large exports’ will invite different
readers to reach widely divergent quantitative conclusions”.
Mattingly points to the numerous olive oil presses in the Tripolitanian Gebel and pre-
desert and from assumptions as to their potential productivity suggests that olive oil was
being produced for export. The profit from this export trade would have been a source of
wealth for the elite and would have helped them maintain and enhance their political,
social and economic status (Mattingly 1989a).
A major body of evidence is supplied by the remains of mills and presses. This
archaeological evidence for olive presses from the Gebel Tarhuna suggests a remarkable
level of oil production: for example, Mattingly has estimated that the total potential oil
production capacity in good years will have measured millions of litres; sites such as
Henschir Sidi Hamdan (with nine presses) and Senam Semana (with 17 presses) could
have produced 100,000 and 200,000 litres in peak production years (Mattingly 1995:
177
143). Including these massive oil production sites, about 415 olive presses have been
identified on the Gebel Tarhuna, and there is other evidence to indicate that the area was
cultivated as part of extensive territory linked to the main coastal centres. Mattingly has
argued that the region overall specialised in olive oil production, contributing significant
exports to Mediterranean markets (Mattingly and Hitchner 1993: 454). There is no doubt
that varying differences in the production techniques and climatic conditions across the
Roman empire play a significant role in the density of planting, the yield of individual
trees and the potential production capacity of the individual olive presses (Mattingly
1996a). In contrast to the modern system of olive farming in the Gebel Tarhuna, where
extensive rows of trees facilitate mechanised cultivation (Taylor 1960: 91), the ancient
olives were probably grown intensively in rows or in scattered groves or in association
with other crops, and were worked by human power and animal traction.
4.2 Type of the press in the Tarhuna plateau
All olive presses known from the surveys in the Gebel Tarhuna are of the lever and
windlass type. This result is not unique to this region but also conforms with the picture
from most other surveyed areas in Roman Africa, in particular, the Kasserine survey
(Hitchner et al., 1990: 231-60; Hitchner and Mattingly 1991: 36-55). Hitchner and
Mattingly have presented a useful comparison between three North African zones with
large numbers of well-preserved presses of Roman date: the Tripolitanian Gebel (the
Tarhuna plateau), Kasserine zone (central Tunisia) and the Tripolitanian pre-desert.
They were dealing with the physical survivals of a particular form of press (Fig. 4.1). In
the African lever press the pressure is exerted by a long horizontal timber beam or tree
trunk (A), pivoted at one end (B) so as to be free to move in a vertical plane, being fixed
either between two upright heavy stones (C) or supported in a wall or some other form
of
ol
th
at
w
do
w
of
pr
th
co
Fi
f vertical su
live pulp (D
he beam. En
t the free en
windlass (G)
ownward pr
weight of var
f baskets (H
re-desert pr
hough techn
ounterparts
igure 4.1: Sch
upport. The
D) were pla
nhanced pre
nd (F). The
mounted o
ressure on
rious eleme
Hitchner and
resses of T
nologically
in the Gebe
ematic drawin
woven bask
aced in a pil
essure was e
e counterwe
n its top and
the mats in
ents, the len
d Mattingly
Tripolitania
y the same
el region (H
ng of the Tripo
kets or mats
le on the so
exerted by
eight was li
d connected
n a press us
ngth of the b
y 1993: 439
, Mattingly
e, had a
Hitchner and
olitanian olive
s (fiscinae)
olid bed of
lifting a co
ifted and pr
d to the free
sing a wind
beam and t
9-40). In com
y concluded
smaller pr
d Mattingly
e press (from
containing
the press (E
unterweigh
ressure exe
e end of the
dlass (G) is
the height a
mparing the
d that the
rocessing c
1993: 458).
Oates 1953).
the already
E), which l
ht stone off
erted by me
beam by ro
proportion
and size of t
e Gebel pre
pre-desert
capacity th
.
178
y crushed
ay under
the floor
eans of a
opes. The
nal to the
the stack
esses and
presses,
han their
179
As noted above, the lever and windlass press type was widespread in North Africa
and was based on the lever principle with downward pressure exerted by a long timber
beam (prelum) connected to a windlass (sucula) secured to a counterweight (Camps-
Fabrer 1953). The pressure was exerted with the help of the windlass, set between two
vertical uprights (stipites) whose lower ends were anchored to a parallelepiped stone
(counterweight) by two dovetail joints. A bar placed in a special channel cut into the top
face of the counterweight united the two stipites to reinforce the windlass (Fig. 4.26).
A number of other studies has focused on the mechanical efficiency of the presses
and their productive capacity (Sounni 1982). Scholars in general are in agreement with
Mattingly’s estimations (Mattingly and Hitchner 1993: 483-498) concerning both the
literary and archaeological evidence, including size of press elements, tanks and vats,
storage jars, mills and press density. These studies conclude that the ancient presses
were worked with lower pressures in comparison to modern presses. However, the
difference in pressure primarily involves a longer cycle in the process of pressing rather
than a reduction in yield of the extracted oil from the olive fruits.
4.3 Production process
Mills, the orthostats, the press beds, and the counterweights are all basic and
fundamental elements of the processing operation and their existence is a clear
indication of the presence of press(es) within the site.
4.3.1 Mills: Milling is the first step in the olive oil pressing process after the harvest. The
presence of mills is one of ways to distinguish between olive oil and wine presses.
Although, Columella mentioned four different types of ancient olive crushing apparatus,
180
the two best known types of mill, archaeologically, are the mola olearia and the
trapetum (Frankel 1993). As regards the mola olearia type, Columella (xii, 52, 6-7) did
not give any details about its measurements or what were the characteristics that define
its production capacity or time of milling. The mola olearia (Fig.4.2) is identified with a
circular crushing basin with a flat crushing surface and one or two wheel-shaped
crushing stones (Drachmann 1932).
Figure 4.2: trapetum and mola olearia mill types (Frankel 1993).
Citing Ben Baaziz, Mattingly has indicated that the mola olearia type was commonly
known and the predominant mill in the olive production areas in Africa Proconsularis
and Tripolitania (Fig. 4.3), except in the Cap Bon region which used the trapetum type
(Ben Baaziz 1985; Mattingly 1996a).
The quantity of identified mills found by the TAS in the Gebel Tarhuna (11 mills
from 63 sites) is far lower than the other press elements such as orthostats, bases of
orthostats, arae, counterweights and vats (Fig. 4.4 and table 4.1). For comparison, there
are more than 400 presses recorded in the Tarhuna region. However the shortfall in the
number of recorded mills is paralleled in other areas in the Roman world. For example,
181
Hitchner found only about seven mills against more than 350 presses in the high steppe
region of central Tunisia (Hitchner 1990; Mattingly and Hitchner 1993: 443-4). In the
Wadi el-Htab (the Tunisian Tell) Ben Baaziz identified nine mills compared with 97
Figure 4.3: Distribution of trapetum and mola olearia mill types in North Africa (after Mattingly 1996a).
olive presses (Ben Baaziz 1985: 209-15); while in the Caesarea survey there were only
three mills from 55 sites (Leveau 1984: 427-39). In the territory of Dougga, an area of
150 km2, only four mills have been detected against 196 counterweights (Lanfranchi
2009: 274). Also during the period of ancient Greece olive mills of olive oil were few or
absent from farming sites (Foxhall 1993). Because of their expense, mills are often not
used for domestic production of oil, where other methods of crushing olives may be
used. Mills are thus, in themselves, an indicator of surplus market-oriented production.
The absence of olive mill mortars and mill-stones from the majority of olive oil
production sites is most likely explained by the continuing phenomenon of the
preferential removal and reuse of mill-stones and mill basins in the surveyed areas of the
North Africa (Mattingly 1993a). The stone mill elements were expensive items and
182
required considerable technical skills to manufacture them. A further factor is the lack of
excavation at rural sites. Mills were usually located in the countryside and therefore, are
rarely to be found in urban excavation.
Sites Pres. Elements Mills Arbores Arbores-bases arae Counterweights vats 64 370 11 95 98 48 38 15
Figure 4.4: The chart shows the number of different press elements have been recorded by the TAS within a number of farming sites in the Tarhuna plateau.
Three broad types of the mola olearia were recorded by the Kasserine survey (Fig.
4.5), with no evidence for the use of the trapetum type (Mattingly and Hitchner 1993:
444). These three types were described as following:
• Type 1: Large shallow, generally flat bottomed, stone basin with integral central pier (with
square socket hole). The outer diameters vary in the four examples recorded from 1.19 to 1.58 (in the
Kasserine survey). The working surface of the basin, between outer lip and central column varies between
0.30 and 0.42 m (bottom) and between 0.37 and 0.50 m (top), with a depth between 0.12 and 0.20 m.
• Type 2: Large shallow, flat-bottomed basin with central area cut away to leave a pierced ring of
stone. The central pivot in this case was either a separate stone column anchored to the mill substructure
or conceivably was made of wood and regularly replaced.
• Type 3: Shallow, flat-bottomed basin, with broad central depression. This type is similar to type
2, but the bottom is not pierced through. The central column for the pivot was presumably detachable and
replaceable (Hitchner and Mattingly 1991: 45).
0
50
100
150
200
250
300
350
400
Fi(H
th
ce
re
tim
su
fr
igure 4.5: MiHitchner and M
All of the
he Kasserine
Type 1a:
entral pier (t
Type 1b:
ecorded mill
mes). This t
urrounding
om flowing
ills of the thMattingly 1991
mills recor
e type 1, wit
Large sh
the recorded
Shallow, fl
l at TUT16
type is diffe
incision pro
g out.
ree types of 1).
rded on the
th two furth
hallow, gene
d examples
flat bottome
lacks a cen
erentiated by
obably for
mola olearia
Tarhuna pl
her sub-type
erally flat
show both
ed stone ba
ntral pier. T
y the provis
fixing a fil
a have been
lateau by th
es (Fig. 4.6
bottomed,
square and
sin, with ce
This seems t
sion of a rec
tration grill
recorded in
he TAS are
and Table 4
stone basin
circular soc
entral pivot
to have been
ctangular or
l to prevent
the Kasserin
the mola o
4.1).
n with a d
cket holes).
t pier (altho
n broken of
r square dra
t the crushe
183
ne survey
learia of
depressed
ough the
ff in later
ain with a
ed olives
Figure 4.6
Ty
6: Types of mil
Type 1
ype 1a
ll recorded by
Type
y the TAS in th
e 1b
he Gebel Tarhuuna.
184
185
Site name no. mills Type Exter. Diam. Intern.Diam Pivot Capacity (m3)
V1
TUT4 1 1a 180 150 x 30 20 x 18 0.52469
TUT5 1 1 135 105 x ? 27 x 10
TUT16 1 1b 132 100 x 38 0.29857
TUT20 1 1 145 130 x ? 22 x 15
TUT38 2 1b 145 125 x 40 35 x 40 0.44691
DUN128 2 1 210 185 x 50 50 x 45 1.24166
DUN130 2 1 150 120 x 40 40 x 18 0.42994
DUN132 1 1 180 155 x 41 35 x 25 0.74128
Table 4.1: The Gebel Tarhuna olive mills.
The calculation of capacity figures is a result of the following progress: (1) the total volume = V = π r2 h. For example, the mill of site TUT4 can be estimated:
3.14159 x (0.75)2 x 0.30 = 0.53036 m3 (2) Excluding the pivot volume = 3.14159 x 0.102 x 18 = 0.00565 (3) Gross volume (capacity) = 0.53036 – 0.00565= 0.52471 0.52422 m3.
The refining of the classification of mola olearia mills of the Gebel Tarhuna shows
important variants in the mode of mounting the millstones. The thickness of the outer
rim is on average 15 cm, the range of the external diameter varies between 132 - 210
cm, and the external height is between 30 – 50 cm. Some of the recorded mills of the
Tarhuna plateau (TUT5, TUT16, TUT20, TUT38 and DUN130) are quite similar in
their internal diameter to mills recorded in eastern Algeria; here Lanfranchi was able to
measure 60 mills, their inner diameter varying between 100 – 153 cm (Lanfranchi 2009:
273). Some of those from the Gebel Tarhuna can be classified with the largest known
Roman mills (Table 4.2) which have been listed by Brun (1986: 77) and also cited by
Lanfranchi (2009: 273). Four mills from the Tarhuna plateau (TUT4, DUN128 (2) and
DUN 132), and two mills identified by Ben Baaziz (Ben Baaziz 2000) in the upper
valley of the Wadi el-Htab (Tunisia) can be added to the mills mentioned by Brun,
which were distributed in the Middle East and the North Africa. These mills are larger
186
than six mills observed and studied by Mouna Hermassi at sites called Slougia and
Dakhlet Zmit in the Tunisian high steppe; their external diameter varies between 0.70 –
1.55 m (Hermassi 2004).
Mill location Diameter (cm) Reference
Sbeitla (Tunisia) 150 Duval and Baratte 1973
Wadi Sebt (Algeria) 150 Leveau 1979
Kafr Nabo 1 (Syria) 160 Callot 1979
Taqle (Syria) 165 Callot 1979
El Arba (Algeria) 170 Leveau 1979
Khorazin (Palestine) 180 Yeivin 1966
Aghrem (Algeria) 190 Leveau 1979
Tirat Yehuda (Palestine) 190 Yeivin 1966
Khirbet Yajuz (Palestine) 200 Thomson 1979
El Kfeir (Syria) 205 Callot 1979
Kafr Nabo 2 (Syria) 210 Callot 1979
Amman (Jordan) 210 Zayadine 1977-8
Table 4.2: The large olive mills listed by Brun 1986.
Finds of millstones are rarer than mill mortars. The surface archaeological evidence
found at a very small number of sites (three sites, Tables 4.3, 4.4) suggest that the olives
were crushed in the mills by solid cylindrical stones bored through the centre (Fig. 4.7).
There is a difference between the examples which are illustrated in Figure 4.6; the type
in the bottom image had a central hole fitted within a square of 5 cm depth. This was
probably for an axle fitting around a free-rotating horizontal bar. The millstone was
rotated around another fixed element set in the pivot hole and probably connected to the
ceiling or overhead beam (Fig. 4. 8).
187
Site Diam. Stone Length Diam. hole
TUT16-1 55 ? 15
TUT16-2 50 45 15
TUT16-3 50 ? 15
TUT18 45 47 15
GUM85 45 45 15
Table 4.3: Dimensions (cm) of some millstones.
Figure 4.7: Two types of crushing stones found in the Wadi Turgut.
TUT16
TUT18
188
Figure 4.8: A reconstruction elevation showing the possible milling process of the Gebel Tarhuna mills.
Site name Capacity (m3) V1
Vol. of engaged millstone (Vm)
Capacity (m3) V2
Capacity (m3) V3
TUT4 0.524 0.029 0.495 (c.175 kg)
0.465 (c.171 kg)
TUT38 0.446 0.029 0.417 (c.163 kg)
0.387 (c.157 kg)
DUN128 1.24116 0.029 1.211 (c.380 kg)
1.182 (c.375 kg)
DUN130 0.42977 0.029 0.400 (c.168 kg)
0.370 (c.162 kg)
DUN132 0.60845 0.029 0.578 (c.188kg)
0.549 (c.184 kg)
Table 4.4: Estimation of processing capacity for the recorded mills by the TAS.
V2= V1-Vm. V3= V1-2V. (V2) = a mill with one millstone. (V3)= a mill with two millstones. It should be noted that these figures are showing unmilled olives, so the processing capacity would be higher by supplying more unmilled olives during the milling process.
4.3.2 Arbores As already mentioned all Roman period olive presses in the Tarhuna plateau were of
the lever type, with the head of the lever anchored between a pair of limestone uprights
(arbores). These orthostats form the most obvious archaeological features of the
Ceiling or overhead beam
189
pressing sites. In view of the fact that these massive limestone uprights blocks are still
standing at many farming sites (Fig. 4.9), their remains in the Gebel Tarhuna attracted
the attention of early European travellers and scholars who initially mistook them for
remains of megalithic structures (Barth 1857; Cowper 1897). Nonetheless, the research
of men like Cowper, for instance, on the senams of the Tarhuna plateau is
extraordinarily valuable, especially since he described over 70 sites (with many
photographs, measurements and sketches). Some of these sites are nowadays destroyed
or completely disappeared. However, he misinterpreted them completely because he
assumed that they were religious monuments of pre-Roman date (Cowper 1897: 131;
Mattingly 1988a).
Figure 4.9: An olive press locates within the ancient oilery farm (TUT43) in the Wadi Turgut.
The data derived from Cowper’s records for the northern Tripolitanian presses “are
unquestionably less reliable than those from the Kasserine survey” (Mattingly and
Hitchner 1993: 454). The information on the Gebel Tarhuna presses presented here is
190
mainly drawn from the TAS evidence. The TAS records and measurements improve on
the data previously published by Mattingly (largely based on Cowper’s records). This
makes the Tripolitanian evidence more reliable (see below for examples where
Cowper’s measurements have been corrected) in comparison with other North African
areas.
The size of the recorded ancient olive presses evidently varied from one area to
another. Mattingly argued that there were evident differences in size and scale between
the elements of olive presses that have been recorded in the Tripolitanian Gebel, the
Kasserine region and the Tripolitanian pre-desert (Mattingly 1993: 485). With regard to
the differences and the regional variation in the size and scale of pressing facilities, the
size of the Gebel Tarhuna press-orthostats has been considered to be the largest of the
three studied areas (Fig. 4.10 and Table 4.5) in Roman North Africa; the Tripolitanian
Gebel (Gebel Tarhuna), the Kasserine region and the Tripolitanian pre-desert (Mattingly
1988b: 188; Mattingly and Hitchner 1993: 456-7). The data recorded by the TAS allow
these views to be re-evaluated.
In respect of press elements, exact measurements were a priority for the TAS survey
work ‘since the size of the arbores and the manner in which they were bonded into the
structure of the building can help elucidate the potential force generated by the press,
and this can help to distinguish in broad terms between presses of high and low
capacity’ (Mattingly 1988b: 187). Table 4.3 and Figure 4.10 – 4.11 illustrate the
measurements of a total of 40 press-uprights examined by the TAS. It must be noted
that, firstly, I have selected only this number of presses because they are considered the
best preserved uprights in the Gebel Tarhuna, particularly in the Wadis Turgut and Doga
(including 37 press-orthostats in addition to three from south of Gasr Ed-Dauun).
191
Secondly, only a small number has been chosen from the latter area because many of its
presses have been examined by Cowper, Oates and Mattingly (Cowper 1897; Mattingly
1985; Oates 1953; 1954). Thirdly, these measurements are only applied to the orthostats
without the addition of basal blocks and lintels that added extra height and weight. The
Figure 4.10: A selection of press orthostats from the Tarhuna plateau.
m
m
192
Site H W T H.1 H.2 H.3 Weight (t) Type TUT1 230 45 50 70 140 1.381 1 TUT3 310 50 50 95 175 220 2.069 4 TUT5 265 50 50 90 150 1.768 1 TUT7 240 50 50 85 140 1.602 3c TUT8 285 45 55 100 145 1.883 2a TUT9 360 50 50 100 155 2.403 2a TUT11 310 55 65 100 160 2.959 5 TUT14 340 50 50 70 120 195 2.269 5 TUT15 265 45 50 80 135 1.591 2 TUT16 315 50 60 70 115 180 2.523 3 TUT20 330 50 60 75 130 185 2.643 3 TUT26 300 50 55 90 170 2.403 2 TUT27 290 55 55 100 155 2.342 3 TUT29 245 50 50 100 150 1.635 1 TUT35 285 50 55 95 170 2.092 2 TUT36 220 45 50 100 160 1.321 2a TUT38 240 50 55 100 165 1.762 2 TUT40 290 50 50 60 120 170 1.935 1 TUT42 310 55 60 100 160 2.731 2 TUT43 335 55 65 140 185 3.197 2 TUT44 315 50 55 110 175 2.312 1 TUT46 220 50 50 130 180 1.468 2 TUT52 255 50 55 110 180 1.872 2 TUT53 230 50 50 100 175 1.535 2 TUT54 320 55 60 115 165 2.819 1 TUT57 300 50 50 80 155 2 2a DOG60 295 50 50 100 170 1.969 2a DOG66 305 55 60 80 130 175 2.687 2 DOG67 290 50 50 120 180 1.935 2a DOG68 310 50 55 85 135 180 2.276 1 HAG82 300 55 55 130 180 2.423 2 TEL95 290 50 50 80 130 185 1.935 2b TEL96 280 50 50 120 170 1.869 2 DOG104 290 50 50 125 180 1.935 1 DOG106 320 55 60 150 195 2.819 2 DOG107 300 50 55 135 180 2.202 2a TUT109 240 50 50 120 175 1.602 2 DUN128 380 55 60 150 205 3.348 2a DUN129 350 55 60 150 220 3.083 2a DUN130 335 55 55 145 210 2.705 2 Average 292 51 54 103 162 188 2.182
Table 4.5: The dimensions of selected the Tarhuna plateau press orthostats.
All measurements are in cm, and the weight is in tonnes and applying to a single press upright. The weight of the press upright is defined by applying the following formula:
193
Weight= H x W x T x 2.67. The limestone of the Tripolitanian Gebel has an approximate density 2.67 g/cm3 (Donahue et. al., 1971; Shishov et. al., 1980). For example the weight of the orthostat from site TUT1 is calculated as follows: 230 x 45 x 50 x 2.67= 1.381 tonnes. Key to heading: H= height of the orthostat from its lower edge above the base stone up to the lintel; W= width; T= thickness; H.1= the height of lower edge of the lower hole; H.2= the height of lower edge of the 2nd hole; H.3= the height of lower edge of the 3rd hole. The majority of holes are square shape measuring between 15 – 20 cm.
Figure 4.11: Height (in cm) of the selected press orthostats from the Tarhuna plateau arranged in descending height order.
height of the upright means its length from its bottom above the base to its higher edge
under the lintel. Finally, only two press uprights shown in the table (TUT9 and TUT43)
appeared in Mattingly’s tables of Tripolitanian presses (Mattingly and Hitchner 1993:
458, nos. 44 El-Gharabah and 59 Senam Terr’gurt).
One particularly striking aspect of the Gebel Tarhuna press orthostats (Tables 4,5 and
4.6), is that they generally were characterised by massive sizes and extraordinary
weights; 50 percent were over 3 m in height to lintel alone, with individual orthostats
generally weighing 2-3 metric tonnes apiece; three exceptional press orthostats (TUT43,
0
50
100
150
200
250
300
350
400
DU
N12
8TU
T9D
UN
129
TUT1
4TU
T43
DU
N13
0TU
T20
TUT5
4D
OG
106
TUT1
6TU
T44
TUT3
TUT1
1TU
T42
DO
G68
DO
G66
TUT2
6TU
T57
HA
G82
DO
G10
7D
OG
60TU
T27
TUT4
0D
OG
67TE
L95
DO
G10
4TU
T8TU
T35
TEL9
6TU
T5TU
T15
TUT5
2TU
T29
TUT7
TUT3
8TU
T109
TUT1
TUT5
3TU
T36
TUT4
6
194
Count Average Max. Min.
Height 40 292cm 380 cm 220 cm
H.1 40 104 cm 150 cm 60 cm
H.2 40 162 cm 220 cm 115 cm
H.3 8 188 cm 235 cm 170 cm
Weight 40 2.182 t 3.348 t 1.321 t
Table 4.6: Average, maximum and minimum measurements of the Tarhuna press orthostats recorded by the TAS.
DUN128 and DUN129) weighed 3.197, 3.083 and 3.348 tonnes respectively (the figures
should be doubled for each pair of orthostats). The arbores of site DUN 128 constitute
the tallest uprights recorded by the TAS, reaching 3.80 m height and abutted by a wall
built of ashlar blocks that rises to more than 5 m height (Fig.4.12a). Because his
measurements were based on information recorded by Cowper 1897, Mattingly has
commented that these measurements must in all cases be seen as approximate. For this
reason I re-measured the press-uprights of site TUT9 (no. 44 El-Gharabah, Cowper
1897) in 2007.
Figure 4.12: a) Images of the highest press‐uprights of sites DUN128 and TUT9;
TUT9
DUN128
195
Figure 4.12: b) Unusual press orthostats of site TUT3 (Type T6).
The results of the new records are as following: its height was corrected from 3.30 to
3.60 m, and the measurement of the height of its top hole was 1.75 instead of 1.65 m,
and the base of its lower hole 1.10 instead of 0.75 m above the base of the orthostat
(Table 9, Mattingly and Hitchner 1993: 458).
Apart from Tripolitania, the Tunisian High Steppe and other areas of North Africa
have well-preserved arbores. The features of the Tarhuna plateau press orthostats can be
compared with a number of press orthostats from the Thelepte region in Tunisia (Table
4.7). By reviewing their dimensions in Table 4.7, it is evident that the press orthostats at
Mguismet were the highest (2.43 m); while the press orthostats at Henchir Boudhaif
were the lowest (1.7 m) in the examined area (Hermassi 2004: 120). Indeed it is clear
that the size and weight of the Tarhuna plateau press orthostats were far of greater than
the Thelepte examples. In addition, from about 38 press orthostats examined by the
196
Kasserine survey (again close to Thelepte), there were only three press arbores (Ksar el-
Guelal: KS031.p13, KS031.p20 and KS031.p7) that reached 2.5 m height (Hitchner
1990).
Site name Height Width Thickness Weight (tons)
Zâati 2.29 0.42 0.67 1.720 Touil 2 0.42 0.75 1.682 Oum Debban 2.3 0.45 0.8 2.210 Mguismet 2.43 0.5 0.8 2.595 Ksar Touil 2.25 0.38 0.75 1.712 Khimet Gharsallah 2.15 0.38 0.83 1.810 Herbouk 2.28 0.38 0.75 1.735 Guetib 2.36 0.47 0.73 2.162 Es Sdid 2 0.3 0.35 0.561 El Mlez 2.17 0.37 0.74 1.586 El Khmira 2 0.45 0.45 1.081 El Khima Darraouia 2 0.35 0.5 0.934 El Khangua 2 0.5 0.73 1.949 El Kamour 2.2 0.42 0.72 1.776 Dekhlet Zmit 2.3 0.45 0.75 2.072 Dalia 1.85 0.53 0.7 1.832 Betoum 1.93 0.33 0.78 1.326 Henchir Boudhiaf 1.7 0.5 0.45 1.021 Henchir Abacha 1.9 0.37 0.73 1.370 Average 2.11 0.42 0.68 1.638
Table 4.7: Dimensions (in m) of some press orthostats recorded in Thelepte region in Tunisia (after Hermassi 2004).
The selected press orthosats recorded by the TAS have been classified into six types.
Only a single new type (Type T6) has been added to the Tripolitanian press orthostats
typology made by Mattingly:
• Type T1: Two pairs of lateral holes (normally pierced only in one of the two orthostats and recessed c. 15-20 cm into the other), no angle cut slots.
• Type T2: Two pairs of lateral holes, one pair of angle cut slots (almost invariably located above the top pair of holes).
• Type T2a: Two pairs of lateral holes and two pairs of angle cuts. • Type T2b: Two pairs of lateral holes and three pairs of angle cuts. The angle cuts either coincide
with the holes or are interspersed with them. • Type T3: Three pairs of lateral holes and one pair of angle cut holes (normally positioned above
the top pair of lateral holes). • Type T3a: Three pairs of lateral holes and a single pair of angle cuts located between the top and
middle pairs of lateral holes. This is possibly a type T2 press which has been adapted by the insertion of a third pairs of holes to increase its capacity.
197
• Type T3b: Three pairs of lateral holes and two pairs of angle cuts. Once again, perhaps this is the result of the conversion of a type T2 press to enlarge processing capacity.
• Type T3c: Three pairs of lateral holes and three pairs of angle cuts. • Type T4: Four pairs of lateral holes. • Type T5: No pairs of lateral holes, but four pairs of angle cuts (Mattingly and Hitchner 1993:
460). • Type 6: The orthostats had no pairs of lateral holes cut into their inside faces, but
had three angle cuts in the right orthostat and three longer cuts along the width of the left
orthostat. The uppermost of the long cuts extended only part way across the front of the
left orthostat and does not provide an angle cut to match the right orthostat’s top angle
cut. A single site is recorded by the TAS with this arrangement and illustrated in Figures
4.12b and 4.13.
Figure 4.13: An illustration of the Tarhuna plateau press orthostats types.
198
4.3.3 The press floors (arae)
Press floors or beds (arae) often consisted of monolithic slabs of limestone set in
front of the orthostats (Fig. 4.14). These press beds were used in a middle stage, between
milling and refining, to extract the oil; when the paste was obtained after milling, the
pulped olives were inserted in baskets (fiscinae) and placed on the arae for pressing
(Brun 1986: 47; Mattingly 1988b: 187; Mattingly and Hitchner 1933: 451). The press
floor is usually defined by a circular or square channel carved in its surface. This
channel had a function of collecting the liquid from pressed olives and directing its flow
into the adjacent vat or vats. The archaeological evidence from around the
Mediterranean world indicate that press floors can be a monolithic slab or a pavement of
opus siginum, but the latter must be very strong to bear the large forces on it during the
pressing operation (Vismara 2007).
Figure 4.14: Schematic view of the main press elements: a base of the orthostats (A); press bed (B); counterweight (C). They have been drawn as found in situ.
Mattingly and Hitchner have described three types of press floor found in the
Kasserine survey: Type 1, was a monolithic slab with a circular channel; Type 2, a
TUT12
199
circular channel cut across a series of slabs; and Type 3 seems to have used opus
siginum floors, though without excavation it has not been established whether floors of
this type had a circular channel built into them or not (Hitchner and Mattingly 1991: 46;
Mattingly and Hitchner 1993: 451). However, the archaeological evidence from the
Tarhuna plateau has produced only one type of large monolithic slab (Fig. 4.15), with
two variations: the first with a circular, the second with a square channel (Table 4.8)
with sub-classification (see below). It is worth noting that a few opus siginum floors
have been recorded by Oates (1935) and the TAS, but it is hard again to judge without
excavation if these floors were used for pressing. I believe these opus siginum floors did
not function as press beds because they were not located in front of surviving press
orthostats, and they were probably used as a waiting area for the filled baskets with
pulped olives or as treading floors for wine production.
Figure 4.15: A line of seven in situ press beds at Sidi Aboageala (TUT12).
200
Site Type ext. Dimensions int. Dimensions approx. Weight
TUT14 C 200X194 142 5.179 TUT27 C 195X192 148 4.998 TUT29 C 197X193 145 5.075 TUT38 C 200X195 145 5.233 TUT38 C 193? TUT45 C ? TUT44 C 198X195 140 5.028 TUT46 C 200X192 140 5.128 TUT54 C 198X195 142 5.154 TUT54 C 195X195 144 5.076 TUT54 C 195X193 140 5.024 TUT56 C 200X198 135 5.286 TUT56 C 205X197 137 5.391 HAJ81 C 200X200 138 5.034 DUN128 C 205X202 145 5.528 DOG106 C 193x188 135 4.867 HAJ82 C 198x175 133 4.023 HAJ82 C 202x198 144 5.333 DOG107 C 197x185 133 5.128 TUT10 S 200X192 175X163 5.126 TUT10 S 200X193 168X165 5.153 TUT12 S 203X198 155X150 5.365 TUT12 S 200X200 150X150 5.034 TUT12 S 202X198 152X148 5.286 TUT12 S 200200 150X150 5.034 TUT12 S 198X195 145X142 5.154 TUT12 S 202X200 170X168 5.393 TUT12 S 205X197 172X170 5.393 TUT14 S 195X193 162X160 5.024 TUT15 S 205X200 166X163 5.473 TUT15 S 200X197 168X165 5.259 TUT15 S 193x190 153x148 5.164 TUT16 S 190X? TUT26 S 197X? TUT29 S 202? TUT38 S 200X196 162X148 5.18 TUT38 S 198X196 165X150 5.206 TUT54 S 193X190 153X148 5.164 DOG106 S 205X200 163X158 4.998 DOG107 S 200X197 162X158 5.247 DOG66 S 203X200 145X145 5.042 HAJ81 S 207X202 160X157 5.582 HAJ81 S 198? DUN128 S 195? DUN130 S 210x200 155X150 5.607
Table 4.8: Some of the Gebel Tarhuna press beds recorded by the TAS and selected for measurement. All measurements are in cm. Key to headings: Type S = press bed with square
201
channel; Type C = press bed with circular channel; Ext.Dims = Exterior dimensions of the press bed; Int.Dims = Interior dimensions of the square or interior diameter of circle defined by channel; the weight is approximate because I was able to measure the thickness of only a few examples, while the majority of the Tarhuna plateau press beds are partly buried. Their thicknesses were between 45 and 55 cm. I have therefore used an average thickness of 0,50 cm when calculating weight.
Table 4.8 illustrates how the Gebel Tarhuna arae were characterised by large sizes
and heavy weights; their total surface area often reaches 4 m2 and their total weights are
roughly between 4.23 and 5.607 tonnes. Numbers of square channels were slightly more
than circular channels; 26 press beds had a square channel and 21 a circular one. The
number of examples of the two types of channels reveals that both were in common use
in the Gebel Tarhuna during the Roman period. In view of fact that the internal diameter
of the circular channel and the largest circle that could fit within the square or
rectangular one reveal a maximum dimension for the size of the flat fiscinae, the
Tarhuna plateau press beds seem to have been intended for placing baskets of very large
diameters (1 m or more) “and this has major implications for the quantity of olives
which could be pressed at one time” (Mattingly 1988b: 187). Knowing the diameter of
fiscinae is an important factor for the calculation of the productive capacity of the
presses. There have been no archaeological discoveries to demonstrate the actual size of
fiscinae; they appear to have been organic material, which modern parallels would
suggest could have been halfa grass or palm leaves. Mattingly believes that there is
close relation between the internal diameter of the channel cut in the press bed surface
and the size of baskets employed, and he emphasises the following point: “I do not
believe that such large monolithic slabs would have been quarried and transported if
baskets of much smaller diameter (say 0.60 m) were in use” (Mattingly 1993: 489).
202
The diameter of the channel thus defines the maximum size of the flat baskets
(fiscinae) used to contain the olive pulp (or possibly grapes) (Mattingly 1988b: 187).
The liquid would begin to ooze out from the stacked baskets even before applying the
pressure by the beam (prelum) and windlass. Some of the Tarhuna plateau press beds I
have examined have a ring of eroded notches just inside the circular channel which were
probably created by the acids in olive oil during long use (Brun 1986a, 2004). These
grooves probably formed through the flow of olive oil on the exposed stone surface
outside the stack of baskets. Press bed figures such as TUT27 illustrate how the inner
diameters of the eroded areas are not much smaller than the diameter of channels,
providing further support for Mattingly’s view of the large size of the fiscinae. For
instance, a press bed from site TUT27 (Fig. 4.16) has a 7 cm wide circular channel with
an inner diameter of 148 cm. There is a clear pattern of acidic erosion grooves on the
inner side of the channel and taking the inner edge of these grooves as marking the outer
limit of the baskets, the maximum size of baskets used in this press appears to be 125
cm. The variance between the two dimensions (23 cm) is occupied by the wavy eroded
meanders. A few press floors with similar eroded meanders (Fig. 4.17) have been
recorded at the site of Kef Lahmar (sito 93, de Vos 2007: 50-1), in the oil farm of the
Oued R’mel in the region of Segerme, Tunisia, and in a rural villa at Madaure, Algeria
(Brun 2004: 211, 219).
Turning again to the Gebel Tarhuna, evidence from some other examples provides
suggestive indicators that have allowed me to assess fiscinae diameters (Fig. 4.18). In
particular, regarding press floors with a square channel from oilery farms (TUT15,
TUT38) and an ara with circular channel from TUT14, examination of the possible
203
TUT27
Figure 4.16: A circular channel of press slab with internal meanders. The circle indicates hypothetical diameter of stacked baskets.
diameters from these examples indicate that, although their diameter may well have
varied from one press to another, they generally, at least from the available evidence,
point to the use of baskets of diameter larger than 1.00 m. This confirms the argument of
Mattingly that Tripolitanian presses had very large fiscinae compared to more recent
traditional presses (where fiscinae rarely exceed 60 cm diameter) and support this aspect
of his calculation of productive capacity of Tarhuna presses (Mattingly 1993).
Kef Lahmar (from de Vos 2007). The Oued R’mel (from Brun 2004). Figure 4.17: Examples of press beds with eroded meanders recorded in Tunisia.
204
The fiscinae used to contain the pulped olives had to be carefully washed in order to
separate the paste after pressing, an operation that would reduce the danger of such
residues giving an unpleasant taste to the oil obtained from the next pressing load (Cato,
agr., 67.2; Columella, XII, 52.22; Pliny, nat., XV,22).
TUT14 TUT38
Figure 4.18: Some of press beds with traces of stacked baskets recorded by the TAS in the Wadi Turgut.
However, there are additional factor to be considered in relation to the Tripolitanian
presses not considered by Mattingly, especially the possibility that not all presses
employed baskets. In the Digesta it is stated that there were two ways to do pressing, a
method with the use of regulae and another method without (Digesta 19.2). Pliny wrote
TUT 15
205
that the regulae represents a next step in technological innovation “ut nuper inuentum”
after the fiscinae (Pliny, N. Hist., XV, 5; Vismara 2007: 459). The regulae was a kind of
container made of wooden slats “exilibus regulis”, which was built up directly on the
press bed around the load of pressed olives or grapes. Hero described two types of
containers (made from thin laths) called galeagra (Hero, Mech., 3. 16-17); these
containers have been reconstructed (Fig. 4.19) by Drachmann. An important factor to
bear in mind is that this sort of container may have been better suited to wine than oil
production.
Figure 4.19: A reconstruction of exilibus regulis (by Drachmann 1932: 150).
It is possible that press floors with square or rectangular channels indicate the use of
regulae rather than fiscinae. Square channels could thus indicate wine rather than oil
production. Moreover, the available archaeological evidence from the Tarhuna plateau
raises the possibility that some of these press slabs were even used for pressing both
olive oil and wine (as explained below). It has to be admitted that no resolution of the
wine/oil issue can be definitely reached without excavations in some of these rural
farms. However, as a working hypothesis, it is suggested here that circular press bed
channels generally indicate pressing of olives in fiscinae and that square channels may
indicate wine production using regulae. Baskets (fiscinae) were most likely used for
pulped olives, while wooden slabs “exilibus regulis” probably functioned for pressing
206
grapes. It is conceivable that some presses with square channels were used both with
regulae and fiscinae and there is evidence of circular channels converted for use with
square regulae.
Figure 4.20: Two press beds with circular channels with angle cuts.
The latter press beds appear to belong to a hybrid type, designed for use with either
fiscinae or regulae. A number of press beds recorded by the TAS (TUT12, TUT27,
TUT29, TUT38, TUT44, and DOG60) have traces of angle cuts which are perpendicular
TUT27
TUT12
207
or semicircular cut outs adjacent to the external edge of the circular channel (Figs. 4.20;
4.21). Close examination of these corner cuts reveals traces of grooves at right angles
extending beyond the circular channel – almost certainly employed to fix the wooden
slats of regulae (Fig. 4.19 and 4.21). Figure 4.20 (TUT29) illustrates how “exilibus
regulis” can be fitted on top of this type of hybrid press bed. It is not clear whether the
superimposed square and circular features reflect contemporaneous use of a press for
both wine and oil production (in the different pressing seasons in a single year) or
subsequent conversion of oil presses into wine presses at some point. The possibility
that a press could be used for both wine and oil is further suggested by the fact that some
press beds had more than one channel allowing liquid to be directed into different
systems of tanks and vats. It is difficult to identify the full distribution of this sort of
press floor in the Gebel Tarhuna because many of them have disappeared underneath
soil and collapsed walls. On the other hand, this innovative pressing technology seems
to have been less used in the pressing operation in Tripolitania and also in the other
pressing sites in the North Africa. The previous archaeological surveys in the other parts
of the North Africa have not recorded such modified press beds, although a few eroded
meanders on circular arae have been identified (Brun 2004; de Vos 2007). An
excavation (started in 2007 and still in operation) inside of Lebda Cement Factory (12
km south-west of Lepcis Magna) has uncovered a large rural villa which is about 9.400
m2 (by Google Earth measurement, Fig. 4.22). A press bed discovered during the 2007
season in the western side of the villa, also looks to have been converted from use with
circular fiscinae to a square base for regulae. In this case the press size is smaller in
terms of beam length than those used on the Tarhuna plateau (Fig. 4.23).
208
Figure 4.21: An illustration of the press bed (TUT29) shows the fitting of wooden slabs on top of it. Here the eroded grooves and the circular channel certainly indicate oil production, while the square frame press relate to wine production.
Figure 4.22: A Google Earth image demonstrates the location of Lebda Cement Factory rural villa.
In this section on press beds recorded on the Tarhuna plateau, I have tried to examine
a number of aspects and I conclude here that until we have done systematic excavation
in association with botanical analyses we still deal with hypothetical conclusions about
the commodity processed in individual Tripolitanian presses. Some press beds
209
functioned with large circular baskets and seem to have related to olive oil (though wine
cannot be excluded in all cases). Eroded channels and olive mills certainly support the
Figure 4.23: A press bed with angle cuts discovered in the Lebda Cement Factory.
(The scale is 1 m, photo J. Dore, Oct. 2007).
identification of oil production at some sites. Square channels and indication of
conversion/adaptation of some press beds to support square wooden structures (regulae)
could possibly be related to wine production. Some press beds seem to show use for
both wine and oil, but whether contemporaneously or consequently modified is not
certain. In any case the proportion of wine production in the region was certainly larger
than previously recognised.
210
4.3.3.1 Classification of the Gebel Tarhuna press floors A total of 44 press floors have been examined by me in the Tarhuna plateau; it is
worth noting that some of these beds were incompletely preserved or exposed (and thus
difficult to classify); these are marked by (?) in the Table 4.6. They are illustrated in
Figure 4. 24 and categorised as following:
• Type 1: monolithic stone slab with a square channel cut on the bed surface.
• Type 1a: monolithic stone slab with a square channel and one corner cut.
• Type 1b: monolithic stone with a square channel and two corner cuts (only one press has been
recorded of this type).
• Type 1c: monolithic stone slab with a square channel, but with eroded meanders along side the
inner edge of the channel.
• Type 2: monolithic stone slab with a circular channel cut in its surface.
• Type 2a: monolithic stone slab with a circular channel and one corner cut.
• Type 2b: monolithic stone slab with a circular channel and one angle cut.
• Type 2c: monolithic stone slab with a circular channel and four angle cuts.
• Type 2d: monolithic stone slab with a circular channel, but with eroded meanders alongside the
inner edge of the channel.
Site
S. channel type
Corner cut
No. Corner cut Place of C. Cut Angle cut
No. A. Cut
Added channel Meanders
TUT10 1 TUT10 1 TUT12 1 TUT12 1 TUT12 1 TUT12 1 TUT12 1 TUT12 1 TUT12 1a Yes 1 (L) back corner TUT14 1 Yes TUT15 1a Yes 1 (R) back corner TUT15 1 TUT15 1 TUT16 1 ? ? ? ? TUT26 1a 1 (L) back corner TUT29 1c ? ? ? ? Yes TUT38 1 TUT38 1b Yes 2 (L) + (R) iner corners Yes TUT54 1 DOG66 1a Yes 1 (L) back corner DOG106 1c Yes 1 (R) back corner Yes DOG107 1a Yes 1 (L) back corner HAJ81 1 HAJ81 1 DUN128 1 DUN130 1 Yes
211
Site
C. channel type
Corner cut
No. Corner cut Place of C. Cut Angle cut No. A. Cut
Added channel Meanders
DOG106 2 DOG107 2 DUN128 2b Yes 1 HAJ82 2 HAJ81 2a Yes 1 (R) back corner HAJ82 2 TUT14 2d Yes TUT27 2c yes 4 Yes TUT29 2c yes 4 TUT38 2c Yes 4 ? ? TUT38 2 TUT44 2c Yes 4 ? TUT46 2 Yes TUT54 2 TUT54 2 TUT54 2 TUT56 2c Yes 4 Yes TUT56 2c Yes 4 Yes Table 4.9: The 44 classified press floors (recorded by the TAS). S= square, C= circular. (L) = left, (R) = right.
Type 1 Type 1a
Type 1b Type 1c
212
Type 2 Type 2a
Type 2b Type 2c
Figure 4.24: Types of Tarhuna press floors.
4.3.4 Counterweight blocks
The counterweight is a large rectangular block of stone and it was employed to draw
down the free end of the press beam by means of a windlass (Brun 1987: 96-113;
Mattingly 1988b: 182; Mattingly and Hitchner 1993: 452). Their distance from the front
of the orthostats indicates the minimum length of the pressing beam (Fig. 4.25). Most of
the Gebel Tarhuna counterweights were located at a length of between 8.5 – 9.5 m from
the external limit of the orthostats making the anchored end of the beam. Cowper
named theses blocks the “Semana” type (= Brun type 11, Fig. 4.26), because he first
observed them at Gasr Semana in the Wadi Doga. Although he described them as being
commonly found items at sites of the Tarhuna plateau, he did not understand their
function (Cowper 1897: 149-50).
213
The recorded counterweights in the Gebel Tarhuna were dominated by the “Semana”
type with a continuous groove in its upper surface linking two dovetail cut-outs into
which the uprights of the windlass were set (Fig. 4.26).
Only one counterweight out of a total of 38 counterweights (recorded by the TAS)
can be distinguished and typified as Brun type 30 – which employed butterfly shaped
clamps in place of the longitudinal groove (Fig. 4.26). The “Semana” type is
characterised by a mortise joint formed by interlocking tenons and mortises that do not
extend to the full height of the block (Mattingly and Hitchner 1993: 453). The Semana
type was widely used in Tripolitania, and also identified recently in a late Roman
imperial period site at Hendek Kale (Figs. 4.27, 4.28) in Turkey (Bennett and Coockson
2009). This type probably was initiated in the Aegean world, even though it was more
commonly known in the western Roman world, particularly in southern France, North
Africa and Spain (Frankel 1997: 77).
Figure 4.25: An over view of the distribution of the main pressing elements in the site TUT14.
214
Figure 4.26: Types of counterweight recorded by the TAS in the Gebel Tarhuna.
Figure 4.27: Reconstruction of a Tripolitanian lever and weights press with perforated piers and “Semana” type counterweight (from Frankel 1997).
Figure 4.28: A “Semana” type of press counterweight found at Hendek Kale in Turkey.
215
The vast majority of counterweights are still partially buried (Fig. 4.29), making it
impossible to measure their height without digging around them. Only one
counterweight at oilery farm (DOG82, Fig. 4.30) was available to me to take its
complete measurements because it had been pulled out some time ago (210 x 125 x 115
cm). Nonetheless, the surface traces of other counterweights suggest they were of
similar size (Table 4.10). In order to reach the highest pressure and also to match the
weight of the other pressing elements, the Tarhuna plateau’s counterweights were cut
from solid limestone with large dimensions and their approximate weights may have
been as high as 7 tonnes (height x width x thickness x 2.67 [density of the limestone] =
weight). For example, from the recorded dimensions the weight of the DOG82
counterweight is: 2.10 x 1.25 x 1.15 x 2.67= 8.60 tonnes, excluding the weight of the
dovetails cut-outs (0.1193) = 7.415 tonnes.
Figure 4.29: An example of the imbedded in situ counterweight from the oilery TUT43.
216
Figure 4.30: A press counterweight recorded at oilery farm (DOG82).
Site Length Width Approx. height
Weight/ tonnes
TUT3 2.10 1.15 1.15 7.295 TUT5 2.15 1.20 1.15 7.802 TUT14 2.10 1.05 1.15 6.651 TUT15 2.10 1.10 1.15 6.973 TUT16 1.85 1.10 1.15 6.129 TUT26 2.00 1.10 1.15 6.635 TUT27 2.10 1.10 1.15 6.973 TUT31 2.00 1.15 1.15 6.942 TUT35 2.10 1.05 1.15 6.651 TUT38 2.10 1.15 1.15 7.295 TUT43 1.95 0.95 1.15 5.568 TUT54 2.00 0.95 1.15 5.714 DOG60 2.10 1.20 1.15 7.618 DOG64 2.00 1.05 1.15 6.328 DOG66 2.00 1.10 1.15 6.635 DOG68 2.05 1.00 1.15 6.157 HAJ81 2.10 1.20 1.15 7.618 HAJ82 2.10 1.15 1.15 7.295 GUM88 2.05 0.95 1.15 7.173 DOG104 1.95 0.90 1.15 5.269 DOG106 2.10 1.00 1.15 6.328 DOG107 2.10 0.95 1.15 6.000
Table 4.10: The visible surface size (in cm) of selected counterweight blocks recorded by the TAS.
217
By way of comparison, Mattingly showed that the press facilities of a pre-desert press
at el-Amud were smaller in size, weight and potential production than these of the Gebel
Tarhuna. For instance, the counterweight was smaller in size, perhaps 2 tonnes
maximum (Mattingly 1984). By contrast, in the Kasserine region few counterweight
blocks were visible in the lower press rooms (Mattingly and Hitchner 1993: 452). Their
categories mostly belong to types Brun 11, 30, and 32 (Fig. 4.31). Table 4.11, again in
terms of size and weight, indicates that the recorded examples from the Kasserine area
were much smaller than those recorded on the Tarhuna plateau.
Site name Con.Dims Dovetails weight (tones)
KS010 124x86x27+ 22x16x? [1.100 KS013. p2 155x64x58 23/30x14/20x58 1.14 KS041 100x71x30 19x10x30+ 0.8
Table 4.11: Some of the Kasserine counterweights (from Mattingly and Hitchner 1993: 453).
Figure 4.31: Types of press counterweights (Brun 1987).
218
The size and weight of counterweights of the Tarhuna plateau presses (as other
elements of their pressing facilities) indicate that the pressing operations there were
highly organised and requiring considerable labour, technical skill and capital
investment. The large-scale presses are among the very largest known anywhere in the
Roman empire and must have been expensive to install and maintain. These massive
pressing facilities also support Mattingly’s point of view that there had been “a high
level of local innovation and experimentation with the basic lever and windlass system”
(Mattingly 1996a: 590).
4.3.5 Tanks and vats Vats were usually set in the ground and have disappeared at many pressing sites
underneath soil and collapsed walls. However, even without excavation, the TAS was
able to recognise partly or completely a total of 13 vats (Table 4.12). The size of vats is
another significant factor, as well as other press elements, to determine the scale of
production (Mattingly 1988b: 187-8; 1993: 493), but without excavation it is impossible
to obtain a complete measurement of vats, in particular their depth. Most of the recorded
tanks were located close to the press floors and often below on the slope (Fig. 4.25). The
recorded lengths and widths suggest that the Gebel Tarhuna vats were also characterised
by a large scale of capacity which can be another indicator of high productivity and may
also relate to the issue of oil or wine production (wine tanks were generally larger than
oil vats).
The oily liquid produced from pressing is an emulsion of oil and sludge and might be
mixed with suspended solid fragments. In terms of weight, the liquid corresponds
roughly to 60 percent of the weight of the processed olive, but it still comprises two
219
Site Vat.Dims Site Vat.Dims TUT3 180x115x? TUT43 185x100x?
TUT8 200x120x? DOG60 188x105x?
TUT14 250x115x? TEL97 155X100x?
TUT115 155x110x? TEL99 170X105x?
TUT16 175x115x? DOG107 210X115x?
TUT35 165x115x? DUN128 280x100x?
TUT41 150x50x? Table 4.12: Dimensions of some press vats recorded by the TAS.
Figure 4.32: A press vat cut from a piece of limestone and installed in front of one of press beds at DUN128.
thirds water and olive (residues) and one third oil (Vismara 2007: 468). These
components have to be separated by means of decantation in special basins or tanks.
This separation process refines oil from the other components, and is required to be done
quickly since the olive residues (lees) ferment promptly in contact with the air. The
virgin oil obtained from the first pressing of olives cannot be separated at a later stage
220
from the less pure oil of a second pressing if allowed to accumulate in the same vat
(Brun 2003: 156-8; Vismara 2007: 468).
In order to prevent leakage of liquid from vats, they were coated with a thick layer of
tebshemt (a mixed mortar of lime and crushed pottery), and the archaeological evidence
reveals that this operation was often repeated two or more times. For the same reason,
some of vats in the Tarhuna plateau were cut from a monolithic slab of stone (Fig. 4.32).
The press vat discovered in the Lebda Cement Factory (Fig. 4.33) had a capacity of
2000+ liters with a hollowed out sediment trap set into the floor.
Figure 4.33: A press vat discovered in 2007 at the Lebda Cement Factory (phot. J. Dore).
221
4.4 Standardization
From the end of the third century BC Roman Italy had access to a profuse supply of
servile labour that encouraged major property owners to develop the manufactures of
construction materials such as bricks or lime. Trained labour in making construction
materials could easily become quite specialised and capable of producing materials to
more precise standardisation (Adam 1999: 259). Economic specialisation and product
standardisation became features of the more developed sector of the Roman economy.
Similar processes occurred outside Italy, even where there was less abundant slave
labour. A degree of ‘standardisation’ within the Roman North Africa rural production
sites has already been suggested by Mattingly and Hitchner through their investigation
of the oileries in the Kasserine region. The type layout of some oilery sites indicates
that they had been arranged in standard pattern, as though following a blue-print design
(Hitchner et al 1990: 251-2, with figure showing almost identical plans).
This pioneering observation can now be taken a stage further with the TAS evidence.
My calculation of measurements of the press equipment of the Gebel Tarhuna
demonstrates that many of these were produced as standardised components. This
feature is clearly seen in the recorded mills, orthostats, press beds and counterweights.
Tables 4.1, 4.5 and Figure 4.11 illustrate the close similarity of the press material
measurements; the data in Table 4.1 shows that the outer diameter of these mills varies
between 1.32 and 2.10 m, but 63% were between 1.32 and 1.50 m. The diameters of
these mills are modest compared to those studied by Ben Baaziz in the upper valley of
Oued el-Htab. There they vary between 1.85 and 1.95 m at sites such as Henchir el-
Hammam and Henchir Gouzzah (Ben Baaziz 2000: 193, 198). However, what is
striking about the Tarhuna material is the regional consistency of the size of mills.
222
In order to investigate further the significance of the closely similar dimensions of
many press elements, I have followed up a suggestion by Lin Foxhall that I look at
some preliminary standard deviation data on my presses and those from the Methana
survey in Greece (See Appendix, Table 7). Looking at mill diameters, press orthostat
heights and press bed internal diameters yields mixed results, though the Libyan
material looks different to that from Greece. The orthostat heights vary quite a bit and
have a standard deviation of c.0.4, which does not look particularly significant, but a
glance at Fig. 4.11 reveals that the vast majority of the orthostats fall between 2.80-3.20
m high. If the outliers were excluded, the degree of variation from a norm of c.3 m is
fairly small. For greater statistical reliability it would be better to have a larger sample
of press elements still. However, the initial results are more encouraging for circular
press beds. A low standard deviation is clear, for instance, in the case of the diameter of
Type 2 press beds (=0.045821, See Appendix, Table 7). Here the contrast with the very
varied dimensions of the Methana presses is particularly marked. Of course, one of the
implications of the rather standardised internal diameters of the press bed channels is
that it implies an associated industry producing standard sized baskets (fiscinae) to use
on the presses. Overall, the press elements and related materials (mills, fiscinae, etc.)
from the Tarhuna plateau show far clearer indications of standardised sizes
predominating (see Figure 39, appendix) than press elements from Methana, Greece,
where dimensions and standing variation shows greater randomness (Foxhall 1997).
This would seem to confirm that the production of presses as well as the overall layout
of press buildings was a quite specialised and standardised process in Tripolitania. If
press elements were rather standardised, that implies also a specialised ‘industry’
supplying these needs for the olive farmers.
223
4.5 Capacity production of the Gebel Tarhuna olive presses
A further objective of my work on the remains of presses in the Tarhuna plateau is to
focus more attention on the potential productivity of this agricultural area. This element
builds on (and in general supports) Mattingly’s several studies about olive oil farming
and pressing in Roman Africa, especially Roman Tripolitania (Mattingly 1985, 1987b,
1988a, 1988b, 1988c, 1993a; 1994). As regards the data employed by Mattingly, these
were mainly based on the work of Cowper (1897) and Oates (1953) and relate to the
well-preserved olive presses, consisting of measurements of several different elements
of these presses (Mattingly 1993a: 483). The significance of these archaeological works,
in addition to Goodchild’s work (1951), remains central to any new attempt to discuss
how the presses functioned in the Tripolitanian Gebel. The details of these early surveys
allowed Mattingly to made an overview of the production of olive oil, a formulation of
press capacities, and an estimation of the annual production in this region.
By using the TAS data, and especially the press dimensions taken in the field, in this
section I shall first present a large sample of measured presses of the Gebel Tarhuna and
secondly attempt to develop arguments based on the most certain data. Thirdly, on the
basis of this data, I will re-examine the maximum and minimum processing capacity of
Roman period olive presses in the Gebel Tarhuna. Fourthly, I shall attempt to estimate
the annual amount of the region’s olive oil productivity; this is fundamentally based on
the details provided by the TAS and will be compared with the olive cultivars and the
traditional olive oil presses in the Mesllata region (north-east of the Tarhuna plateau)
operated during the late Ottoman period.
224
The evaluation of the performance of any ancient press is not easy. This performance
was dependent on several factors that must be taken into account: the quality and degree
of ripeness of the olives, the milling, the type of machinery, the amplitude of the press
and the dimensions of filled baskets, the force exerted by the timber beam, the time
taken for each pressing load, the number of subsequent loads for pressing and the
duration of the harvest season (Mattingly 1993; 1996b; Vismara 2009: 445). It is worth
remembering Mattingly’s assessment of the question:
despite the relative frequency with which olive presses are encountered in
Mediterranean archaeology, there has been remarkably little curiosity as to their
processing capacity. More attention has focussed on the efficiency of mechanical
presses (Mattingly 1993: 483).
The collected data confirms Mattingly’s point of view that “There are clear
differences in size and scale between the elements of olive presses I have recorded from
Africa at both inter- and intra- regional levels” (Mattingly 1993: 485). Scholars are in
agreement with the estimation of a ratio of 15 – 25 percent between the weight of olives
and the oil obtained from them. On the other hand, some have argued that pressing
facilities were established on the basis of production in bumper harvest years, which
normally alternated with ones of poor harvest (Brun 1993). Mattingly has discussed in
several articles the production capacity of the presses, offering maximum and minimum
values, particularly those well-known from surveys in areas of the Kasserine, the
Tripolitanian Gebel and the Libyan pre-desert. He also addressed the issue of scale in a
comparative way with an ethnographic study of traditional lever presses in southern
Tunisia. Mattingly has suggested two potential operational strategies. The first of which
is also reflected in the ethnography, involved a long pressing process (up to 24 hours)
for a large load, supposing that it was employed for great properties as they were
225
requiring a robust press. The second strategy was based on many short pressing
processes containing small loads on behalf of small owners (Mattingly 1993: 494-6).
The larger presses, however, have greater aggregate extraction and capacity in both
cases.
It seems to me that Mattingly’s calculations of the Gebel Tarhuna olive presses, have,
in most cases, tended to under-estimate the production. This minimisation, as explained
by him, partly relates to the estimated height of the beam, which was reconstructed and
mostly based on measurements taken by Cowper.
Although Cowper took detailed measurements of the senam, he did not always
specify whether he was measuring centre to centre or edge to edge, nor did he
invariably list all the necessary measurements in his published accounts. My
reconstructions have been facilitated by personal observation and measurement at
some sites (nos 11 and 41) and by scaling off from Cowper’s excellent
photographs some measurements he omitted (Mattingly 1988b: 190).
4.5.1 Calculation of capacity production
The archaeological evidence allows me to calculate the size of a press load of crushed
olives. Although a number of important press elements are missing from the
archaeological record relating to many regions in the Roman world because they were
made of wooden or organic materials, the presses of North Africa (areas of Kasserine
and Tripolitanian Gebel and pre-desert) form an exception of this respect. They offer
important evidence that reveals the fixing points of the beam head where it was placed
between two orthostats and locked in special holes at two or three different operating
heights (Mattingly 1993; 1996a).
226
The previous reconstruction by Mattingly (1993) was based on examination of the
orthostats found in the regions mentioned above. He assumed that the several pairs of
holes indicate the various operational heights of the press beam. Mattingly has been able
to calculate the size of the crushed olive load from these values:
1. The maximum height of the filled baskets stacked beneath the press beam
closely corresponded with the highest pair of holes in the press uprights.
2. The diameter of baskets holding the crushed olives during the pressing process.
3. The minimum operating height of the press beam (necessary to maintain on
operational horizontal angle of the beam) corresponded to the lower pair of holes in the
orthostats.
For the first and third factors, Mattingly’s measurements of the maximum and
minimum operating heights are shown in the Table 4. 13. The highest measurement for
the lower base of a top hole was at site Kom es-las reaching 2.10 m, whereas it dropped
down to 1.30 m at site Senam el-Jereh which was the only one below 1.50 m high
though their average height was c.1.67 m. The height of the base of the bottom hole
relates to the lower operating height of the press beam and also may give indication of
the possible lowest stack volume.
Table 4.13: Mattingly’s measurements (in m) of maximum and minimum operating heights of selected Tripolitanian presses as suggested by positioning of holes for securing the beam end. All figures are approximations based on his interpretation of Cowper’s measurements and photographs or on his observation at some of the sites (Mattingly 1988b: 191).
Site no Site name
Height of base of bottom hole
height of base of top hole
11 Gasr Doga 1.05 1.65 20 Kom es‐las 0.65 2.1 24 Kom Nasr 0.80 1.5 26 Senam el‐Jereh 0.60 1.3 36 Senam Ferjana1 0.60 1.8 41 Senam el‐Nejm 0.60 2 44 El‐Gharabah 0.90 1.65 45 Hr. El‐Mohammed 1.10 1.65 49 Bu Mateereh 0.90 1.55 59 Senam Terr'gurt 0.90 1.55 Average 0.81 m 1.67 m
227
Site Height of base of bottom hole
Height of base of top hole Note
TUT1 0.70 1.40 TUT3 0.95 2.20 3rd hole TUT5 0.90 1.50 TUT7 0.85 1.40 TUT8 1.00 1.45 TUT9 1.00 1.55 44 El‐Gharabah TUT11 1.00 1.60 3rd hole TUT14 0.70 1.95 3rd hole TUT15 0.80 1.35 TUT16 0.70 1.80 3rd hole TUT20 0.75 1.85 3rd hole TUT26 0.90 1.70 TUT27 1.00 1.55 TUT29 1.00 1.50 TUT35 0.95 1.70 TUT36 1.00 1.60 TUT38 1.00 1.65 59 Senam Terr'gurt TUT40 0.60 1.70 3rd hole TUT42 1.00 1.60 TUT43 1.40 1.85 TUT44 1.10 1.75 TUT46 1.30 1.80 TUT52 1.10 1.80 TUT53 1.00 1.75 TUT54 1.15 1.65 TUT57 0.80 1.55 DOG60 1.00 1.70 DOG66 0.80 1.75 3rd hole DOG68 0.85 1.80 3rd hole HAG82 1.30 1.80 TEL95 0.80 1.85 3rd hole TEL96 1.20 1.70 DOG104 1.25 1.80 DOG106 1.50 1.95 DOG107 1.35 1.80 TUT109 1.20 1.75 DUN128 1.50 2.05 DUN129 1.50 2.20 DUN130 1.45 2.10 Average 1.03 1.62Max 1.50 2.35Min 0.60 1.35
Table 4.14: Maximum and minimum operating height (in m) of selected Gebel Tarhuna presses as recorded by the TAS.
As can be seen in Table 4.14 the maximum height of the base of the top hole from a
total reaches 2.35 m which are slightly higher than ones in Table 4.13 (2.10 m). In
228
contrast their average value is a bit lower (- 5 cm). The same minimum value of the base
of the bottom hole is recorded in both tables (0.60 m) though the average is quite a bit
higher in the TAS’ table.
In Table 4.13 the bottom hole was commonly at least 0.80 m above the base block,
though three holes (30%) did not exceed 0.60 m (Mattingly 1988b: 191), but in my list
the vast majority were over 0.80 m (35 out of 40, 87.5%) and 62.5 percent over 1.00 m.
The distance between the upper and lower holes was in almost all cases in the range 0.45
– 0.70 m which is nearly the same as Mattingly’s measurements (0.45 – 0.65 m). The
combined figures from the two tables demonstrate that the large presses of the Gebel
Tarhuna could accommodate a very substantial pile of fiscinae of olive pulp below their
press beam. In some cases such as at TUT3, DUN128, DUN129, and DUN 130 the
maximum space for stacking could exceed 2 m tall. It is interesting to mention here that
the three largest presses were recorded at sites located near Senam el-Nejm in the east-
southern part of the Tarhuna plateau near Gasr Ed-Dauun village (Fig. 4. 34).
The stacked baskets did not necessarily stand right up the height of the top hole. A
space has to be allowed for the bulk of the beam and perhaps a stone or metal board set
on the top of the ordered baskets. The weight of the stack by itself will have been
enough to start compression. Some liquid should already have started to ooze out during
loading, even from the upper baskets, before the pressing operation began. Mattingly
states that “if the stack height is 1.20 m and it is comprised of individual baskets of
pulped olives each of which stand 0.04 m high prior to loading, the column would likely
comprise more than 30 baskets. Compression of the lower part of the stack during
229
Figure 4.34: Location of sites with the highest orthostat holes south of Gasr Ed‐Dauun.
loading of the press would allow perhaps a third more baskets to be accommodated”
(Mattingly 1993: 489). For a stable pressing process I agree with Mattingly’s
consideration that with the large-scale presses the maximum and minimum heights of
stacked baskets were in an average 1.40 m and 0.70 m respectively.
The diameter of the fiscinae is a second significant and necessary measurement in
order to calculate the productive capacity of the presses, but as I have already mentioned
there has been no discovery of the baskets themselves to prove their actual size. The
only way to estimate their diameter is through the archaeological evidence of circular
channels in the stone press beds (arae) (Table 4.15 and Fig. 4.35). In addition there are a
larger number (26) of press floors with a square channels which could also have framed
stacks of large baskets (though square wooden structures - regulae – are also possible)
(Table 4.8). The archaeological evidence reveals the use of very large ones.
Ta
FigTa
be
be
di
ha
Site
TUT14
HAJ82
TUT54
TUT56
HAJ81
DUN128
able 4.15: Diam
gure 4.35: Chable 4.11.
The diame
ed surface.
eds of circu
iameter at le
ave been sli
Channel Dims
142
140
142
135
142
145
meter (in cm)
art showing m
eter of fiscin
The averag
ular channel
east 0.80 - 1
ightly bigge
Tutal
18
Total
18
Total
Site TUT27
TUT29
DUN128
TUT54
DOG60
HAJ82
of 18 circular
maximum, min
nae must be
ge of the inn
s. Mattingly
1.00 m (Ma
er at 1 – 1.2
Average
1.40
Average
140
Channel Dims S137 TU
133 TU
139 TU
144 TU
133 DO
143 TU
r channels of t
nimum and av
e lesser than
ner diamete
y suggested
attingly 1993
25 m. Based
eMax
m 1.4
Max
145
ite ChDim
UT38 145
UT56 137
UT46 140
UT54 140
OG107 133
UT38 135
the press bed
verage of the
n the circula
er is c.1.40 m
d such figur
3: 490) thou
d on these d
xM
5m1
Min
133
annel ms
5
7
0
0
3
5
s recorded by
e circular chan
ar channel c
m for the G
es would in
ugh my data
ata and on o
Min
1.33m
y the TAS.
nnel diameter
cut into on t
Gebel Tarhu
ndicate fisci
a suggest th
other consid
230
rs listed in
the press
una press
inae with
hey could
derations
231
relating to the volume and quantity of pulped olives in them (estimated at 50 percent of
the total volume of stacked baskets and pulp), Mattingly has estimated that the large
Tripolitanian presses (the Gebel Tarhuna) could process a load equivalent to one tonne
of pulped olives, those of the pre-desert had a capacity 0.25 – 0.33 tonnes, while those in
the Kasserine region occupied a middle position between them. These African presses
would therefore have had a production capacity for a single load of 250 to 1.000 kg of
crushed olives, corresponding to 50 to 250 kg of oil output, with processing requiring 24
hours for the load.
Building on the above estimations, Mattingly proposed an assessment of an annual
production for these presses:
- The Tripolitanian Gebel (the Tarhuna plateau) presses reached 9 -10 tonnes of oil.
- The Kasserine region presses ranged between 5 - 10 tonnes.
- The Tripolitanian pre-desert presses had a production capacity of 2.5 - 5 tonnes.
These figures are related to the bumper years, when processing work might last for
three months (Mattingly 1988b; 1993; 1996b). Based on these calculations and the large
number of presses that existed in the territory of Lepcis Magna, Mattingly suggests that
“in a peak production year ... Lepcis would have had the theoretical capacity of
manufacturing 15 million litres of olive oil, though in years of dearth the level could
have been a fraction of this” (Mattingly 1988a: 37).
My calculation of the press capacity in the Gebel Tarhuna is built on these factors:
- Maximum height of uncompressed stack = 1.60 m (perhaps 1.40 m when
stacked).
- Diameter of fiscinae = 1.25 m.
232
- Volume of stacked baskets/pulped olives = π r2 h = 3.142 x 0.625 2 x 1.60 = 1.96375.
- Volume of olive pulp = 50% of stack volume = 1.964 /2 = 0.982 m3 of pulp =1 m3
(probably = 1 tonne of pulped olive?) = 200 kg oil/per day (perhaps = 3 loads
milled olives?).
In the surveyed area of Wadis Turgut and Doga (c.115 km2), the TAS has recorded
more than 200 large-scale presses (see Chapter 3). Taking into account the production
capacity of 1,000 kg/24h during the bumper production seasons, the total number of
presses (if all in operation for oil production) could have had a production capacity of
c.200.000 kg/24h, which let us say yielded c. 40,000 litres of oil/day. Taking
Mattingly’s assumption that based on press loads ranging from 250 - 1000 kg olives,
with an average of 20 percent oil extracted (Mattingly 1993: 492), the following
hypothetical oil yields can be accepted (Table 4.16).
No. Presses
Load size (kg olives
Daily yield (kg oil)
30 Days yield
60 Days yield
90 Days yield
200 250 (small press)
10,000kg 10,869LT
300,000Kg 326,086Lt
600,000Kg 652,173Lt
900,000Kg 978,260Lt
200 600 (medium press)
24,000Kg 26,000Lt
720,000Kg 782,688Lt
1,440,000Kg 1,565,217Lt
2,160,000Kg 2,347,826Lt
200 1.000 (large press)
40,000Kg 43,478Lt
1,200.000Kg 1,304.347Lt
2,400,000Kg 2,608,695Lt
3,600,000Kg 3,913,843Lt
* 1 kg olive oil = 0.92 litre.
Table 4.16: Hypothetical oil yields from the Wadis Turgut and Doga presses of small, medium and large capacity.
The above figures have built on Mattingly’s proposition that the maximum capacity
for the large African presses was c.10,000 kg. Note that my figures suggest one of the
largest presses used every day at full capacity for 60 days could have yielded 12,000 kg.
His lower proposal was based on several factors. “First, the larger the press the greater
the likelihood of reduced efficiency through friction, periods of enforced inactivity
233
through mechanical failure (breakage of wooden parts/ropes) or through under-capacity
loading. Second, the largest presses often occur in multiple banks of presses, where it is
possible that there was some separation of the different stages of process (one press
being used only for the first pressing, another for second pressing etc). Such practices
could have reduced the aggregate output of the plant, though have facilitated the
separate collection of the better quality oils” (Mattingly 1993: 492). Moreover, from the
new archaeological evidence relating to the press beds of the Gebel Tarhuna, I can
suggest here that more or less 10 percent may have been used for wine production,
though some of them probably were functioned for production of both wine and olive
oil.
If the production of the territory of Lepcis Magna reached 10 million litres in peak
years (Mattingly 1988a: 38), this would correspond to the production of 400,000 to 3
million trees (an adult tree gives 20 – 100 kg olives). This number of trees would have
covered an area c. 400 km2 which is about 1/10 of the total extent of Lepcis’ lands, so
quite feasible. By taking the above hypothetical yields of 60 days, the 200 TAS presses
could have produced between 652,000 – 2,608,000 Litres (and 3.5 – 7.5 million from
750 presses). The estimated number of 1500 presses distributed in the whole territory of
Lepcis Magna during the Roman period could have resulted in a production potential
between 7 – 15 million litres. The new data thus seem to fully support the original
calculation made by Mattingly.
4.5.2 The production capacity of traditional lever presses in the Msellata region during the late Ottoman period:
The Msellata region occupies an area of 10,050 km2 on the north-eastern sector of the
Tripolitanian Gebel. Approximately 7,350 km2 (70%) are exploited as agricultural land
234
(Alarabi 2006: 22). It formed with the northern sector of the Tarhuna plateau the core of
the Lepcitanean cultivated hinterland during antiquity, though unlike parts of Tarhuna
Msellata maintained its traditional farming practices until modern days. Olives are still
considered the most important plants in the socio-economic life of Msellata’s local
people and they have paid a great attention to olive oil production since ancient times. It
also has the oldest planted olive trees in the territory of Lepcis Magna; many people
believe that they go back to the Punic and Roman periods (Abdassadq 2003).
By contrast, investigation of a traditional olive press (Alarabi family press) in the
Msellata region reveals a slightly different picture from the Gebel Tarhuna ancient
presses. The press elements here were smaller in size and correspondingly lower in
production capacity. The press facilities were established inside underground rooms.
The main central room occupied by a semi-concave mill of 2.50 m diameter applying a
cylinder millstone of 1.00 x 0.90 m (Fig. 4.36). At the left side of the room there is 7 m
length of thick tree trunk which formed the press beam. This press beam was anchored
to a single fixed point of a shorter vertical trunk (c. 2 m height) which was built into
floor and ceiling.
235
Figure 4.36: The mill and the millstone at Alarabi family olive oil press.
The mill and press load capacity was c. 30 keala; each keala weighted about 14 kg
olives (420 kg total). Each keala had a potential capacity between 3.5 – 4.5 litres of oil.
Thus a single pressing load could yield 105 – 135 litres of oil from a 24 hours
processing. The estimated annual production for 60 days can thus be estimated between
6.300 – 8.100 litres in good years. I chose 60 days as medium indicator, but in some
productive years the press would have worked for more than three months. However, it
must be noted that the annual olive oil production generally fluctuates from one year to
another and was principally affected by rainfall average and summer dry winds
(Abdassadq 2003: 36-7). As this press is smaller than the Roman presses, it encourages
confidence in the higher estimate for annual production given above for the latter.
Economically, Msellata during the Ottoman period was related to olives. It seems
there was a specific significance to plant good sorts of olive, a policy not only followed
by the local authority but also encouraged by the provincial government in Tripoli. For
example, a letter relating to the administration of province issued in 10 November 1880
236
ordered 1000 olive transplants to be planted in a number of specific farms (waqef), with
a note mentioned the care that should be taken through the cutting process (JAG 1880,
447: 1).
The documentary evidence also reveals that the Ottoman taxation system imposed a
tithe tax on the agricultural production. However, the system itself gave an exemption
for quite a large number of landlords such as tribal chiefs, elites, administrators and
religious men (M.T.T. Fig 4.37). For instance, a document 64/M/Ch (Fig. 4.38) contains
a decree that came from an administrative official to the governor of Msellata in 1843
regarding an exemption that had been given to heirs of Abi Atabel for their 600 olive
trees after they had paid 750 Turkish piaster.
A calculation can be made here from another manuscript document about the olive
oil annual production in the Msellata region. The document (D/M/T/T 1863) related to a
committee assigned the task of accounting for the productive olive trees liable to the
tithe tax. Their result was a total of 711,592 olive trees. By excluding 5,525 as exempted
trees, 656,342 trees were charged for the tax (Fig. 4.39).
Another document mentions the existence of 92 olive oil presses in Msellata working
during the season 1875 (Alarabi 2006). These presses probably were able to produce an
annual production between 580,000 – 750,000 litres during a 60 day pressing process.
Clearly, it can be noted that figures of the Ottoman Msellata press capacity were
much lower than estimated Roman period production capacity. However, they support in
general terms the estimates made above. With larger and far more numerous presses and
with more extensive olive orchards in Msellata and Tarhuna, the idea that Lepcis Magna
could have olive oil production in millions of litres appears feasible.
237
238
Translation
As it known in all provinces the authority is announcing for the collection of the
tithe tax which is sold to an interested party. In the present case relating to the
district of Msellata for the year of 1888, the announcement was issued by the local
council via a public auction according to these conditions:
- The collection of the tithe amount must be in kind as it has been legally
stated, that is, part 1 out of 10.
- An exemption from tax that has been given for decades by the governmental
decree dobyourldi to members of councils and tribes chiefs (shayoukh) for an
amount of 88 aogtt olive oil (c. 80 Lt). Those (of exempt status) who have got more
than this amount will be charged for the addition.
- The value of the contracted tax amount must be paid in cash by the Turkish
piaster and sent to the provincial treasury by four regular instalments starting in
December and ending in March.
According to these conditions the auction has been convened on Msellata’s olive
tithe tax and knocked down to the merchant Livardo Csar for a price of four hundred
and fourteen thousand and five hundred (414500) piaster divided into four
instalments of 103625 each.
Issued by the provincial government of Tripoli of the West.
Figure 4.37: The document M.T.T 1888.
239
To the honourable Mohammed, regarded our son, the general
director of Msellata. You know that the herirs of al-Sheikh Abi
Tabel came to us and paid to the prosperous treasury seven
240
hundred and fifty Turkish piasters (750) regarding their six
hundred (600) planted trees in Msellata. So, our son, they are
exempted to pay more and you have to respect that as indicated.
Written in 7 Moharam (1st month in Islamic calendar) 1259
(1843). Issued by the provincial officer of Tripoli the West
Mohammed Amin Basha.
Figure 4.38: The document 64/M/Ch 1843.
241
242
A dobyourldi issued by the provincial officer stating the
(tax) accounting of Msellata olive trees by a provincial
committee in association with local individuals who have a
good experience about Msellata’s farms. As usual they do the
job by counting the number of trees and this year they found a
total of 71,159. By excluding the trees belonging to exempted
individuals (5,525), the outstanding number was 65,634
olive trees.
Written in 11 Rabeea the first 1285 (1863).
Figure 4.39: The document D/M/T/T 1863.
243
244
Name of olive grove Value (gafeez) Value (litres)
1 Gaream 300 2,100
2 Selama 725 5,075
3 Momen 725 5,075
4 Khalafoun 400 2,800
5 Karartha 455 3,185
6 Shāafeyeen 225 1,785
7 Shehaāni 225 1,785
8 Ghawain 775 5,425
9 Galeil 1900 13,300
10 Zawiat Samah 1450 10,150
11 Bni Leath 925 6,475
12 Jadidd 750 5,250
13 Lawata 1175 8,225
14 Wadna 2850 19,950
15 Esh-Shaāba and Ghaba 2000 14,000
16 Waārr 3000 21,000
17 Khorma 1600 11,200
18 Agnool 900 6,300
19 Aulad Sulaiman 850 5,950
20 Razagna 1300 9,100
21 Messed Audan 725 5,075
22 Bni Yekhlef 200 1,400
23 Zāfaran 1200 8,400
24 Tellan 600 4,200
25 Amamra 225 1,785
26 Zerad 825 5,775
Total 26300 184,765
Table 4.17: Calculation of the olive oil produced from the Msellata olive forests in 1874.
245
The calculation process of the Msellata olive forests in 1874
confirmed by the reporters and accountants. Indeed the amount of the
tithe tax (miry) was twenty two thousand gafeez [each gafeez = 7 litres of olive
oil ]. A dispatch has been submitted to Ahmed Bek (the Homes region
Kaimmakam) who came to Msellata this year accompanied by a clerk to
verify the calculation results accompanied by some local council
members. After their careful examination of the yield they found an
increase of four thousand and one hundred (gafeez) over the previous
year. So the miry of this year is twenty six thousand and three hundred
gafeez olive oil as it is shown above each amount relating to each olive
grove. Edited at Msellata Council 29/01/1875.
Figure 4.40: The document (D/M/T/T 1863 and D/M/T/T 1874).
246
Chapter 5: Amphora production sites on the Tarhuna plateau
5.1 Introduction
Previous evidence for amphora production in Tripolitania is relatively meagre and
relates to two district areas: firstly, to the Gebel Tarhuna and secondly to the coastal
zone in the immediate hinterland of the main cities (and exports harbours). Although
the Gebel Tarhuna is characterised by a widespread distribution of different sizes of
Roman period farming sites, mainly concerned with olive oil production, only three
pottery production sites were previously known in the region. Two kiln-sites were
discovered in the years 1947-8 by Richard Goodchild. The first site contained two kilns
located c. 200 m north of a luxury villa at Ain Scersciara, and 100 m north of the
waterfall at the spring head (Goodchild 1951: 96-7). The situation of these kilns,
located close to the villa and to the Roman period road linking Medina Doga with Oea
through the Wadi Reml, and the local availability of water and clay, suggest that this
production site formed a workshop belonging to a wealthy estate (Arthur 1982;
Goodchild 1951; Mattingly 1995).
The second kiln site was identified adjacent to the gasr of Sidi es-Sid (Tazzoli)
which lies some 5 km west of Tazzoli village on the western part of the Tarhuna plateau
(Fig. 5.1). A collection from this ceramic production site was examined and illustrated
by P. Arthur (1982: 61-72). The illustrated figures reveal that these kilns were mainly
producing amphorae of forms Tripolitana I and III. However, the evidence indicates
that coarse wares and tile were also produced. The third kiln site, relating to Oates
survey around the Gasr Ed-Dauun, is located at a short distance westward of the eastern
three dams constructed in the Udei el-Me and was recognised by its circular shape and
247
burnt brick. Oates believed that this pottery kiln was of similar size to those at Ain
Scersciara (Oates 1953: 90).
In addition to the above pottery production sites on the Tarhuna plateau, a few other
kiln sites have been identified in the Tripolitanian coastal area. Although the earliest
identification of a pottery kiln was in 1925 (Bartoccini 1928-9: 93-5), found within
modern Tripoli (outside the walls of ancient Oea), reported discoveries have been few.
Goodchild also recorded a pottery kiln located on the north side of the main Tripoli-
Khoms road (Km 102) at the head of the Wadi Giabrun (Goodchild 1976: 96-7).
As I already mentioned, the Ain Scersciara kiln site was characterised by a good
water supply from the spring of Scersciara, while the other two were most likely
dependent on water kept in cisterns and wells. Although the production of these
Tripolitanian pottery kilns was mainly utilitarian, it was recognised early on that their
products would help to investigate the long term occupation of site in this hinterland,
where the problem of establishing an absolute chronology is always acute (Goodchild
1976: 99).
Some other kiln sites have been recently discovered in Tripolitania such as the Hai
al-Andalus (Tripoli) kilns (Shakshuki and Shebani 1998: 279) and at Sidi Andulasi
(Tagiura) kilns (Gatansh, forthcoming). During the late 1990s, an archaeological survey
conducted by a mission of the University of “Roma Tre” in the Wadi Caam-Taraglat
identified four ceramic production sites (Felici and Pentiricci 2002: 1875-1900). Three
sites were specialised in production of amphorae and coarse wares (sites: 47, 67, 106),
while the fourth (site 91) was used to produce Tripolitanian Red Slip wares. In order to
obtain good supplies of water and clay, all these sites (Fig. 5.1) were situated beside the
wadis and in the vicinity of several types of rural settlements. For example, the surface
248
evidence from sites 47 and 91 shows huge concentrations of potsherds and wasters,
many of which are blackened, deformed and partly vitrified. The former lies
approximately 4 km from the coast on the west-side of the Wadi Caam, with the
evidence of ceramic production spread over a large area. Fragments of amphorae and
coarse wares have been recorded in extremely high concentrations distributed over an
area of approximately 6 hectares (Felici and Pentiricci 2002: 1879-80). These new
pottery production sites lay in the immediate hinterland of Lepcis Magna and should be
examined in relation to the other economic activities practised in its vicinity.
Comparisons should be made to examine their production in relation to the large
assemblages that have come out of excavation in the city.
5.2 Distribution pattern of amphora kiln sites on the Tarhuna plateau
From the above, it will be clear that research into the pottery production sites and
their relation with farming sites, in particular olive oil pressing, in the Gebel Tarhuna is
still in its infancy. There is almost nothing available in the literary sources dealing with
economic phenomena in this region. Archaeologically, data for the Roman period
economic activity are very imbalanced, with lots of evidence for olive oil (and wine)
production, extensively scattered and, on an enormous scale. However, hitherto the
archaeological evidence for pottery production has been less impressive. My PhD
project thus also set out to expand knowledge of amphora production sites on the
Tarhuna plateau. I have recorded many new sites and have mapped their distribution in
relation to farming sites and communication routes. This pioneering study develops our
understanding of the Roman period rural economic organisation. Although the results
Figure 5.1: A m
ap of pottery kilns in Tripolitania.
249
250
are still of a preliminary nature, the new data enable us to reinterpret our sources and
advance our knowledge about this aspect of economic activity.
In the light of the new evidence identified by the TAS, the study of olive oil
production can be complemented by an enlarged knowledge of Tripolitanian amphorae
and their associated epigraphy. Moreover, these data could help:
to overcome the old debate about ‘primitivism’ and ‘modernism’ in the Roman
economy. Earlier historians, including M. I. Finley, were unaware of a mass of
archaeological data with which Rostovtzeff had been familiar (Remesal Rodriguez
1998: 183).
The absence of secure data regarding the non-agricultural rural economic activity
resembles many other regions in Roman Africa. Producing several types of pottery
within rural estates, villas and farms would necessarily create additional labour needs
inside and outside these rural properties. This diversity in rural economic activity,
therefore, would provide estates and estate workers with work and income whether in
goods or money. One possibility is that pottery production may have employed estate
labour outside the seasons of harvest and pressing, thus turning casual workers into full-
time employees. The concept could be applied not only to rural labour but it also has
important consequences for other involved individuals, such as owners of draught
animals which were used to transport goods and on the suppliers of other requirements
of the rural community whether from the countryside or urban centres. As already
mentioned, the TAS has mainly concentrated on the Wadis Turgut and Doga in the
north-eastern sector of the Tarhuna plateau (see Chapter 2). During this survey 12 new
pottery production sites were identified within these two valleys and two other sites
were recorded in Halafi village (south of Gasr Ed-Dauun) and in the Wadi es-Sri (12
251
km west Tarhuna town). By adding the three kiln sites previously recorded (by
Goodchild in Ain Scersciara and Tazzoli and Oates at Gasr Ed-Dauun) the total for the
Tarhuna plateau is 17 ceramic production sites. Most of these new sites came from the
survey of the Wadi Turgut with one of its largest tributaries (the Wadi Guman), as was
the case for other categories rural settlement sites (62.5 percent, Fig. 5.2 and Table 5.1).
Six sites were identified in its main course (TUT12, TUT15, TUT18, TUT48, TUT53
and TUT108) and four sites in the Wadi Guman (GUM86, GUM89, GUM90 and
GUM110). In contrast, only one pottery kiln has been recorded in the Wadi Doga
(DOG111). It should be considered here that any future expansion of systematic
archaeological survey, especially within the Wadis located north and north-west of Ain
Scersciara, certainly will increase the number of pottery production sites in the Tarhuna
Gebel.
These new ceramic production sites, especially for the Tripolitanian amphorae,
represented a considerable advance in knowledge over previously known amphora kiln-
sites in the Tarhuna Gebel, for a number of reasons:
(1) They have been systematically surveyed, revealing at least 35 amphora kilns
(Table 5.1). Thus, the increased number confirms Goodchild’s point of view that
“further archaeological survey will undoubtedly bring to light additional kiln-sites along
the whole length of the Gebel” (Goodchild 1976: 98).
(2) Amphorae sherds from these kilns exhibited more epigraphic information than
had been found at the previously discovered kiln-sites; only one amphora stamp is
known from the Tazzoli pottery kilns and was read by Arthur as SPNS (Arthur 1982:
64; Goodchild 1976: 98). The amphora kiln-sites recorded by the TAS have produced
16 amphora stamps (Table 5.4 below and Fig. 5.3).
252
(3) These kilns were distributed within or nearby other rural farming sites,
particularly oileries and large farms, and were not isolated production sites.
Attention has focussed hitherto upon the extent of the relationship between these
kiln-sites and their surrounding landscape. It can be suggested that the establishment of
the Tarhuna plateau pottery kilns in the landscape generally met the following
conditions. First, they were often attached to properties characterised by a high scale of
olive oil (or wine) production capacity. This is indicated by a large number of pressing
facilities such as at TUT12, which was an oilery with eight presses located very close to
TUT56, a farm with at least two further presses. At TUT15 (Henscir Assalha) the kilns
were located about 100 m below an oilery with five presses. The pottery workshops in
Halafi village (DUN131) were situated midway between Halafi luxury villa with four
presses and Halafi large farm with four presses (Fig. 3.14). Second, some pottery
production sites appear to have been more affected by the availability of water springs
rather than other landscape factors. This factor seems to have played a greater role with
the location of kiln-sites of TUT18, TUT53, GUM86, GUM89, GUM90, GUM110,
TEL102 and SRI132. Their favourable situation is comparable to the Ain Scersciara
pottery kilns, which exploited the spring and also the clay beds along the valley. By
contrast, a lack of spring water at the other kiln-sites was offset by digging a number of
different sizes of cisterns. For instance, at Sid Buagela (TUT12), the oilery, bath-suite
and probably two kilns were supplied by the water captured in two large shaft-cut
cisterns (Fig. 3.21). Third, in order to facilitate the movement of their products to
markets, all of the recorded pottery production sites in the Gebel Tarhuna were located
either close to the main Gebel road (built AD 15/16) or by secondary routes linking the
plateau with the northern coastal area and the southern pre-desert zone (Fig. 5.2).
253
µ
05
102.5
Km
&
&&
&
&
&&
&
&
&
&
&
&
&
&
&
&
"
"
"
"
GU
M9
0G
UM
89G
UM
86
TU
T53
TU
T48
TU
T1
8T
UT
15
TU
T1
2
S RI13
2
DU
N13
1
DO
G111
GU
M110
TU
T108
TE
L10
2
Tazzoli
ed-D
auu
nS
cersciaraD
auu
nK
had
raTa
rhuna
So
uk aj-jouma
h
1313
1313
1313
32
Kiln site
Modern centres
Rom
an roadA
ncien t caravan track.
Figure 5.2: A m
ap shows the distribution of pottery production sites on the Tarhuna plateau and their relationship to the R
oman eastern G
ebel
road and other ancient tracks. 0 3 6 12 Km
.
254
ID Location Loc.name No. Kilns Stamp
Within/near of
Distance from the F-building
TUT12 w. Hwatem Sidi Buagela 2? Yes Oilery 50 m E TUT15 W. Turgut H.Assalha 4? Yes Oilery 1000 m W TUT18 W. Turgut Astail 2 Yes Oilery 250 m S-E TUT48 W. Turgut Arbaia 5 Yes Large farm 150 m N TUT53 W. Turgut Sidi Eysawi 1 No Large farm 50 m S GUM86 W. Guman Scegafiat Asray 1 No Small Farm 400 m N-W GUM89 W. Guman Scegafiat Atriq 3? Yes Small Farm 300 m S-W
GUM90 W. Guman Scegafiat Ben Hemad 3 Yes Small Farm 350 m S-W
TEL102 W. Tarabut Hamzia 2 Yes Small Farm 250 m S TUT108 W. Turgut Armadia 4? Yes Small Farm 400 m W GUM110 W. Guman Scegafiat Maamri 2? Yes Small Farm 300 m S DOG 111 W. Doga Almseel 1 No Large farm 50 m W DUN131 Fergian Halafi 5? Yes Large farm 300 m N-W SRI132 W. Sri W. Sri 2? No Large farm 200 m W
Table 5.1: The amphora kilns identified by the TAS in the Tarhuna plateau. F= farm site.
TUT15 GUM89 TUT48
TUT15 GUM89 GUM110
Total: 35+ kilns
255
GUM90 GUM89 GUM110
TEL102 GUM89 TUT12
DUN131 TUT18
TUT108 TUT18
GUM90
Figure 5.3: The 16 amphora stamps collected by the writer in the last two decades from pottery production sites in the Gebel Tarhuna.
256
The discovery of the large number of Tripolitanian kiln-sites in the Tarhuna region
strongly suggests that this was a rural economic activity located predominantly
alongside the oil/wine production facilities, rather than close to the port cities. Many
rural estates incorporated pottery workshops and functioned as centres of this sort of
production. For that reason, I would argue that this pattern of the location of the pottery
kilns in the countryside provides further indications regarding the economic orientation
of Roman-period rural landscape, especially considering that amphora production sites
enjoyed a quite strong relationship to olive oil farms and oil pressing sites. Moreover, it
is the pottery evidence that defines the chronology of occupation of sites and building.
It is clear that there was a significant relationship between agricultural production,
particularly olive oil and wine, and sites of pottery production. This phenomenon, in
Tripolitania, is comparable with most other areas in Roman Africa, especially in Africa
Proconsularis; where the archaeological evidence very often reveals that this activity
was part of the rural economy (Ben Mousa 2007: 225). In fact, there is at present a
complete absence of evidence of such economic activity within the city of Lepcis
Magna. Lepcis appears to have depended on rural ceramic production centres as well as
imported wares for its pottery needs.
Two amphora stamps support these proposals; In 1994 I collected a sherd of
Tripolitana II amphora rim from the kiln-site GUM90 carrying a stamp which can be
read as ARHC (Fig. 5.4). The same stamp was found during my participation in the
excavation of a maritime villa at the Wadi er-Rsaf (in the western suburbs of Lepcis
Magna). It appeared on the same type of amphora as the example from GUM90 and
came from a context dated to the second half of the second century AD (Munzi and
Pentiricci 1997: 272-6). Another example concerns amphorae produced in Henscir
257
Assalha (TUT15), where a waster amphora sherd was found with a stamp (Fig. 5.5a). It
offers a perfect example of regional and empire-wide exchange. This stamp belongs to
an estate production centre (an oilery and four kilns), with examples of amphorae
carrying the stamp itself found at the 3rd century AD Roman fort of Gholaia (Bu Njem)
(Rebuffat 1976-7) and at the Laurons II wreck near Marseille (Fig. 5.5b). This amphora,
categorised by Bonifay as type 19 (Bonifay 2004: 105), must have been produced in the
above site.
Figure 5.4: An amphora stamp recorded in 1994 at GUM90.
258
Figure 5.5: a) A stamp on Tripolitana I amphora found by the TAS at Henscir Assalha (TUT15).
b) The same stamp on a Tripolitania I amphora found at the Laurons II wreck in Marseille (from Bonifay 2004).
259
5.3 Types of amphorae produced by the Tarhuna plateau kilns
It is widely known among archaeologists and historians that there are three main
types of Tripolitanian amphorae: I, II and III (Figure 5.6). These amphorae are
considered among the imperial Roman-African style, perhaps derived from the Punic
amphorae equivalents. Related to Dressel type 26, the two main forms are known as
Tripolitana 1 and III (Bonifay 2004: 105; Bonifay and Garnier 2007; Panella 1973, 560-
562 and 564-571). Classification of the Tripolitanian amphorae can inform us about
their contents.
a b
Figure 5.6: The Tripolitanian amphorae I, II and III. (a): Lepcis Magna muesum, (b): stores of the Department of Antiquities (Tripoli).
Examples of both Tripolitana I and III have been found at Monte Testaccio, with a
date range from the Augustan period to the mid-third century and almost certainly
I
II
III
260
contained olive oil. On the other hand, there was also a regional distribution of the
Tripolitanian amphorae, as attested at Bu Njem (Rebuffat 1973a; 1973b; 1973c; 1975;
1977). These amphorae were present at this fort, held by the Roman army in the mid-
third century AD (Bonifay 2004: 20; Rebuffat 1973a; 1973b).
Scholars believe that the distribution of African and Tripolitanian amphorae in many
Mediterranean ports and centres reflects the expansion of the export of agricultural
products (particularly olive oil and wine) and other commodities such as salt fish which
were involved in commercial mobilisation of surplus production (Carandini 1970; 1983;
Keay 1984; Mattingly 1988a; 1995; Reynolds 1995). In their peak period of production
especially during the Severan period, there is some evidence of state interaction in their
supply for both Roman civilian and military markets, in some cases via the exploitation
of the imperial properties (Kehoe 1988; Vera 1988), in others by indirectly encouraging
certain policies by cultivation by private owners or transport by merchants, or perhaps
through imperial agents who directed this trade as a part of the annona system
(Reynolds 1995: 108-11). However, further detailed studies need to be conducted on the
Tripolitanian amphorae in order to develop our knowledge and understanding of their
morphology, reduction and chronology.
5.3.1 Tripolitanian amphora I Although this type was first described by Zevi (Zevi and Tchernia 1969), Panella
was the first to characterise it (Panella 1973). This type is a cylindrical vessel with a
thickened, turned over rim and concave outer face (Fig. 5.7 for examples recorded by
the TAS at the kiln-sites in the Gebel Tarhuna). Its neck is also cylindrical or very
slightly tapered at an angle to the junction with the body (Bonifay 2004). Two short
attached handles link the neck with the shoulders and the body ends in a solid or hollow
261
cone-shape point (Fig. 5.6 and 5.8). The fabric is commonly a mixture of red orange
colour and has a hard and compact texture with angular limestone inclusions. The
archaeological and tituli picti evidence reveal that this type mainly contained olive oil.
However, there is sufficient variation in rim forms to suggest that some sub-types could
have been used for other products such as wine.
TUT48 TUT48 TUT18
GUM89 GUM110 GUM89
TUT15 TUT48
0 5 c m
GUM89 GUM90
Figure 5.7: Examples of The Tripolitanian amphora I produced in kilns recorded by the TAS.
262
Figure 5.8: The Tripolitanian amphorae I, II and III (from http://ads.ahds.ac.uk/ catalogue/archive/ amphora_ahrb_2005/details.cfm?id=304. 04/12/2009).
5.3.2 Tripolitanian amphora II This is a cylindrical amphora topped by a short neck with a thickened everted rim
with two overlying steps (Fig. 5.9). The two ear-shaped handles are placed on the body
below the shoulders. The base is tapered and hollow or filled by clay (Panella 1973:
564). Its form suggests it was derived from earlier Neo-Punic amphorae (Bonifay 2004:
88; Van der Werff 1977: 78, 184). At Ostia the Tripolitanian amphora II were recorded
in levels dated from the first until the middle of the third centuries AD, with a peak
occurrence during the Antonine period (Bonifay 2004: 92). In Tripolitania, it is attested
from the first half of the first century until the late fourth centuries AD (Keay 1989: 43;
Panella 1973: 563). Bonifay remarks that the morphological evolution of amphorae of
this type, during these centuries, is still unclear (Bonifay 2004; 2005). He suggests that
the form of the first century AD is clearly a transitional form of the Neo-Punic type Van
der Werff III. Forms of third century date can be distinguished from the earlier
examples by their more elongated neck, while the neck edge of forms of the fourth
263
century were more atrophied (Bonifay 2004: 92). The fabric is generally either red with
streaks of gray or brown in the core, covered with gray surface, with abundant
limestone inclusion, or a red-orange mixture rather hard and compact with white
angular granules. The contents carried by this type of amphora are contradictory. While
a complete profile was discovered in situ (in an oil press?) in the hinterland of Leptis
Magna (Barker et al. 1996: 279-280 and fig. 9.11), another fragment collected from the
ancient port of Toulon has revealed waxed sediment. The amphora found at Pupput
appears to be coated with remains of salted fish (Bonifay 2004: 89, 92 and table: 4, 474-
5). Figure 5.9 shows some examples from the Tarhuna plateau kiln-sites again with
quite a lot of variation in the rim shape, especially forms produced in the Wadi Guman
sites. It is difficult to identify the reason behind this variation without finding complete
examples which was impossible during my survey when the surface finds were found in
fragmentary remains.
0 5 c m
Figure 5.9: Examples of Tripolitanian amphora II recorded by the TAS.
TUT48
TUT15 TUT48
TUT48
GUM110
GUM86GUM89
264
5.3.3 Tripolitanian amphora III This type is characterised by a tall cylindrical body with an everted rim, less massive
than the Tripolitanian I (Fig. 5.10a); the neck is very often short and connected with the
shoulder in a continuous line (Bonifay 2004). The body ends with a curved base and
conical foot usually filled with clay. In most examples there are two ear-shaped handles
fixed below the rim. The Tripolitanian amphora III succeeded type I in the second half
of the second century AD and dominated the distribution process of the Tripolitanian
amphorae during the third century AD (Bonifay 2004: 105). The archaeological records
demonstrate a continuous production into the fourth century (Panella 2001: 211); during
this century, it was probably characterised by a hypertrophy of the upper edge such as
with the examples found at Ostia and Lepcis Magna (Fig. 5.10b, Bonifay 2004: 105).
Both Tripolitanian I and III were very often used to transport olive oil (Bonifay 2004:
470-5).
TUT108 TUT12 TUT53 0 5 c m
TEL102 TUT15 TUT18
Figure 5.10:a) Examples of Tripolitanian III of the Gebel Tarhuna recorded by the TAS;
265
b) Some examples of Tripolitanian III of the fourth century AD (from Bonifay 2004).
5.4 Construction of amphora kilns on the Tarhuna plateau
In terms of kiln construction, the Tarhuna plateau pottery kilns seem to be paralleled
by ones excavated and examined by Goodchild at Ain Scersciara (Fig. 5.11a) and
Tazzoli (Goodchild 1976). Regarding their furnaces, they mainly can be divided into
two types: Type 1, with furnace opening directly into the low-level compartment of the
oven; Type 2, characterised by a dome supported on a central pillar and with its circular
interior divided into two levels. An example of this type has been found outside the wall
of ancient Tripoli and excavated by Bartoccini (Bartoccini 1928-9: 93-5). This kiln was
described by Goodchild in the following terms: “the lower level, elliptical, was close to
the furnace, while the upper level, ‘three-quarter-moon’- shaped, lay more distant from
it” (Goodchild 1976: 96). In contrast, the design of Hai al-Andulas pottery kilns (Fig.
5.11b) appears to be similar to Ain Scersciara and Tazzoli kilns (Shakshuki and Shebani
266
1998). Thus, the archaeological evidence suggests that the Tripolitanian pottery kilns
were dominated by two main types regarding the level and situation of the furnace.
(a) (b)
Figure 5.11: Sketch of Ain Scersciara and Hai al‐Andulas pottery kilns (after Bonifay 2004).
It was not possible during my survey to do plans of all the pottery kilns I identified
in the Gebel Tarhuna. The principal reason behind that was the poor preservation of
many kilns. Most of the kiln-sites were recognised as mounds of collapsed brick
structures, ash and numerous waster potsherds. Nevertheless, in a few cases I was able
to trace the remains of their walls and measure their dimensions. In the case of
destroyed kilns, an estimation of the number of kilns at each site was made by counting
the mounds of kiln debris on the surface. The last kiln load was still in situ in some
cases, as suggested by test excavation of one of the Arbaia pottery kilns during
February 2007 (Fig. 5.12).
Figure 5.12: Sketch view of kiln no.1 in Arbaia (TUT48).
2 m
267
5.4.1 The excavation of Arbaia pottery kiln (TUT48)
The Arbaia pottery workshop contained at least five kilns located on a slope, close to
the wadi bed, at the meeting point of the Wadi Turgut, running from south to north, and
the Wadi Hawatm Bo Salma, running west to east. The surrounding landscape was
occupied by different types of rural settlements (Fig. 5.13). The excavated kiln was
damaged by the cutting for a modern track cut which had removed its eastern half (Fig.
5.14). However, this track presented me with a cross-section view of the kiln during
reconnaissance survey in October 2004. The short distance of the slope between the
kilns and the wadi bed was occupied by a huge amount of amphora and tile sherds and
remains of clay-built structures.
Fortified farm
Dam
P. kilns
Figure 5.13: A Google Earth image shows location of Arbaia pottery kilns within the surrounding landscape.
268
Figure 5.14: A line of three below kilns cut by a modern track at Arbaia pottery kilns (TUT 48).
A trial excavation was conducted in the kiln no. 1 and brought to light a large
circular pottery kiln in a quite good preservation of Type 1. The walls of the kiln were
built of small rectangular clay blocks which had been unevenly fired in situ. They
enclosed two floors; the upper floor was the oven with its elevated floor supported on a
central circular pillar. The lower firing chamber connected with a stoke-hole opening
westward, which is similar to the Ain Scersciara stoke-hole. The kiln diameter of 4.30
m makes it among the largest pottery kilns yet known in Tripolitania (Table 5.2)
though, from a comparative perspective, it is smaller than the Ain Scersciara kiln (over
6.00 m: Fig. 5.15). The latter was described by Goodchild as “one of the largest Roman
circular kilns yet brought to light” (Goodchild 1976: 88). The Arbaia kiln was full of
waster sherds of amphorae of Tripolitana II and III sherds – possibly its final
(unsuccessful) firing still in situ, rather than rubbish disposal. An interesting collection
269
Figure 5.15: Ain Scersciara pottery kiln which probably was able to fire hundreds of amphorae during one load.
Figure 5.16: Sample from the complete uncrushed olive stones found inside the stoke‐hole of kiln no.1 (TUT48).
270
was made by the TAS team, including four amphora stamps, rims and bases of
amphorae and samples of fuel used. The fuel contained a huge number of complete
uncrushed olive stones (Fig. 5.16) mixed with ash found inside the stoke-hole.
The evidence of using uncrushed olives as fuel has also been found from an
excavation at the site of Tria Platania in Pieria, southern Macedonia (Margaritis and
Jones 2008). These could represent uncrushed olives that had been used as fuel or the
residue of pressed olive. In addition, Brun (1986) found intact olive stones in two
furnaces he excavated in France. The intact olive pits probably represent the solid
residue of pulped olives after pressing. This ‘press cake’ contains residues of oil that
make it an excellent solid fuel once dried. The fleshy pulp and small fragments of olive
stone will generally be thoroughly consumed in the burn, but the intact stones will
sometimes be recognisable in carbonised form. A future careful botanic examination,
perhaps, could cast more light on this debate. Another hypothesis that can be addressed
here is harvested olives in the peak productive years probably waited for days or weeks
to be crushed and pressed. Some olives in stockage would become rotten, and after
drying may have been transferred to pottery kilns.
The recorded diameter of some amphorae kilns on the Tarhuna plateau (Table 5.2)
reveals that most of these kilns can be classified among the largest known pottery kilns
in Roman Tripolitania. This character is comparable to other evidence relating to the
size of farm buildings and pressing facilities (see above Chapters 3 and 4). Capital
investment in kilns, oileries, large farms, water management and baths in this hinterland
was not within the means of poor owners but must be related to rich landlords.
271
Site Kiln no. Inter. Diameter
TUT48
1 4.30 m
2 4.15 m
3 4.85 m
4 2.90 m
5 5.25 m
TUT53 1 2.75 m
GUM86 1 2.80 m
GUM90 1 3.75 m
TEL102 1 3.35 m
2 3.10 m
TUT108 3 3.35 m
DOG111 1 2.50 m
SRI132 16 3.80 m
18 4.20 m
Total 14 kilns
Average 3.65 m
Max. 5.25 m
Min. 2.50 m
Table 5.2: Diameter of some amphorae kilns in the Gebel Tarhuna recorded by the TAS.
5.5 The Tarhuna plateau amphora stamps
The detailed studies of amphorae are significant as these vessels can be considered
valuable evidence of economic activity which is not comparable to most other classes
of pottery. They therefore provide direct and important evidence for the transportation
of a number of valued commodities which played a great role in the Roman economy
272
and life. Amphorae were among the most common ceramics encountered in the Roman
World, certifying to the scale of trade and transport between provinces and to Rome
itself.
The three main types of Tripolitanian amphorae were all produced in the kilns of the
Tarhuna plateau amphorae. The primary classification for these amphorae started forty
years ago through studying the African ceramic containers from the excavation of
Ostia. F. Zevi and A. Tchernia were the first scholars who characterised the
Tripolitanian amphorae in 1969 (Zevi and Tchernia 1969: 193-5). In the early 1970s, C.
Panella described their characteristics, specified sources, refined and defined the
typology and tried to interpret their economic phenomenon (Panella 1973). She initiated
the investigation of names on stamps starting with interpretation of an amphora stamp
CAELEST on a Tripolitana amphora III found at Lepcis Magna (Panella 1977: 135-
149). D. Manacorda produced the first synthesis of the Tripolitanian amphorae stamps
in his article published in “Dialogi di Archeologia” (Manacorda 1976-77: 542-601). In
fact, it should be noted that Tripolitanian amphorae were rarely stamped; perhaps the
stamped Tripolitanian amphorae did not exceed 1 percent of the total produced
amphorae. The most prolific period for stamping seems to have been c.AD 200-230
though some evidence for earlier stamps has been recorded, especially by the TAS for
amphorae Tripolitana I and II (Fig. 5.3 and Table 5.4).
All studies of amphorae stamps build on the material that emerged from H. Dressel’s
work on the inscriptions and stamps on amphorae found at Rome, published in CIL XV.
Recently, this work has been expanded by E. Rodriguez Almeida, focused on the
amphora sherds at Monte Testaccio. This has multiplied the numbers of stamped
amphorae, especially Baetican (Dressel 20) amphora (Rodriguez Almeida 1975; 1978;
273
1984). Based on the detailed analysis of the Tripolitanian amphora stamps of
Manacorda (1973; 1983) and Di Vita-Evrard (1985), in 1988 D. Mattingly listed a total
of 62 known Tripolitanian amphora stamps (Table 5.3). He mentioned that they mostly
related to the Tripolitana III and were datable to the Severan period (Mattingly 1995:
153-55). Most of this evidence for the type III came from Ostia and Rome, particularly
Monte Testaccio (Mattingly 1988b). Recent Spanish work at Monte Testaccio has
added further examples and incorporated all the known Tripolitanian stamps in an
online database (http://ceipac.gh.ub.es/ 09/12/2009).
There are many difficulties with interpreting the abbreviated information on these
stamps. In most cases they appear to relate to abbreviated names (tria nomina), though
other formulations occur (Bonifay 2004; Di Vita Evrard 1985; Mattingly 1988b; 1995).
One group of exceptional stamps seem to relate to imperial estates (Table 5.3, no. 1-4).
For example, four stamps with variants letters IMPANT/AVG, F AVG or IMPANT have
been collected from TUT48 pottery kilns (no. 3 in Mattingly’s table). These evidently
represent the titles of the Emperor Caracalla (Mattingly 1995: 154-5). These imperial
stamps can help localise some of the known imperial estates in the region. Di Vita
Evrard suggested that these stamps related to imperial estates located in the territory of
Lepcis Magna (Di Vita Evrard 1985: 149-50). Some stamps end with the letters ‘CV’
(clarisimus vir or senator) allowing us to relate them to amphorae stamped on behalf of
the leading Lepcitanian families such as the Septimii, Fulvii, Plautii, Marcii, Ulpii,
Vibii, Cornelii, Servilii, Pompeii, Cassii, Granii, Calpurnii and Verginii. These families
are well known in many inscriptions from Lepcis Magna, allowing quite precise
suggestions for identifying some of the individuals involved (see Table 5.3; also Di Vita
Evrard 1985; IRT; Mattingly 1988b; 1995). This evidence is a clear indicator of the
274
relationship between estate owners, the surplus production of olive oil (and wine) and
manufacturing amphorae within the estate as containers.
Figure 5.17: Two Tripolitanian amphora stamps found in Bu Njem.
For example, the letters of the stamp LSACV (2 stamps found in TUT108 by the
TAS) are interpreted as L. Septimius Aper or L. Silius Amicus Haterianus (IRT 542).
While the amphora stamped by the initials MVC (recorded in GUM110) probably is
related to either M. Ulpius Cerialis (IRT 388) or a member of the family of M. Vibii
(IRT 578). It should be noted that some Tripoltanian amphora stamps found at the Bu
Njem fort have been certainly found within the pottery kilns in the Gebel Tarhuna. For
examples, two stamps found there, and published in the CEIPAC database (ns. 18813
and 18808), have been recorded by me at TUT108 and DUN131 (Fig. 5.17)
Stamp Identification Stamp Identification
1. AVGG Septimius Severus & Caracalla 32. L.S.PLH/BVR
L. Silius Plautius Haterianus Blaesilianus (IRT 635)
2. AVGGG Septimius Severus, Caracalla & Geta
33. L.S.PLH/MYC
L. Silius Plautius Haterianus Blaesilianus (IRT 635)
3. IMPANT/AVG Caracalla (or Elagabalus?) 34. [L?]SAHCV L. Silius Amicus Haterianus? (IRT 542)
4. [...]DAVG Severus Alexander 35. CSM/BAICI (?)
C. Servilius Marsus? (AE 1959, 271)
5. LAS L. Avillius[...] or L. Appius [...]? 36. CSMCV C. Servilius Marsus? (AE 1959, 271)
275
Stamp Identification Stamp Identification
6. S.A.BCV/+++ 37. LVTM L. Volusius [...] or L. Verginius Tiro Marcianus?
7. SAB/ACMV 38. MVC M. Ulpius Cerialis (IRT 388) or family of M. Vibii (IRT 578)
8. CBSVR 39. MVM M. Ulpius [...] or family of M. Vibii?
9. PBAV 40. [...]FCV
10. LBAI 41. ACVCF
11. LCS[...] 42. AC[...]
12. PCAGCV/STID Family of P.Cornelii (cf IRT 263, 592)
43. ADYRMP
13. PCAG[...] (retro)
Family of P.Cornelii (cf IRT 263, 592)
44. ARAP Asinius Rogatianus APLL or Adelfius (IRT 539)
14. PCBSCV Kinsman of M. Cornelius Bassus Servianus (IRT 443)
45. BINOMI[...]
15. PCRSSV (retro) 46. CEI
16. PCSSCV/
MARIA[...]
47. CR
17. OCHO Family of Calpurnii Honesti? (IRT 370-71)
48. CRCA
18. QCL Q. Cassius Longinus? (IRT 601) 49. FYN
19. OCLCV Q. Cassius Longinus? (IRT 601) 50. IVI[...]
20. QCV Q. Cornelius Valens? (IRT 594) 51. KATA*
21. CPFCV C. Fulvius Plautianus (cos 293, Praetorian Prefect, father-in-law of Caracalla, executed 205 (PIR 2 F554)
52. MD[...]
22. CFPPP C. Fulvius Plautianus 53. ONII (?)
23.CFPPPCV C. Fulvius Plautianus 54. PC
24. CAELESTIN Q. Granius Caelestinus? (IRT 532)
55. SA[...]
25. LMPP++ 56. SIAP
26. QMD (retro) Q. Marcius Dioga (IRT 401) 57. THER
27 L.PCR cf. Q. Pompeius Cerealis Felix (IRT 444); L.Pompeius Cerialis Salvianus (IRT 602)
58. VAR
276
Stamp Identification Stamp Identification
28. MPF Family of Pompeii 59. VIC
29. POMBAL 60. QPGAT (?)
30. L.APRI L. Septimius Aper, cos 207, executed 212 (IG 12.7.397.28)
61. [...]FC
31. L.S.A.CV L.Septimius Aper or L. Silius Amicus Haterianus (IRT 542)
62 SNPS (retro)
Table 5.3: List of Tripolitania III amphora stamps and suggested identifications with individuals or families known from Lepcitanian epigraphy (after Mattingly 1988b, Table 1, from di Vita‐Evrard 1985; Manacorda 1977, 1983).
No. P. kiln site Stamp Total Position Amp. Type Identification
1 TUT12 HIX 1 Handle ?
2 TUT15 PM? 3 Neck I & II Marci Pompei
Gaetulici or Marcus
Pompeius Geta, (IRT
649) ?.
3 TUT18 Q.P.HANIVLIANV.
.
2 Handle ?
4 TUT48 IMPANT/AVG 4 2 handle
& 2 neck
II & III Caracalla (or
Elagabalus?), no.3 in
Mattingly’s table
5 GUM89 MAF 1 Handle ? Marcus Aemilius? ,
(IRT 714).
6 GUM89 AIM 10 8 rim &
2 neck
I Aemilia Iou[ina?on]
..?, (IRT 363).
7 GUM89 AƗNI 2 Rim I
8 GUM89 MICA 1 Neck II
9 GUM90 ARHC 2 Rim II
10 GUM90 KAVL or KAVC 2 Rim I
277
No. P. kiln site Stamp Total Position Amp. Type Identification
11 TEL102 IS.. 1 Rim II?
12 TUT108 LSACV 2 Neck III L.S(eptimi)A(pri)C(lar
issimi)V(iri), IRT 542,
No. 31 in Mattingly’s
table.
13 GUM110 MVC 1 Rim I M. Ulpius Cerialis
(IRT 388) or family of
M. Vibii (IRT 578). No
38 in Mattingly’s
table.
14 GUM110 IVM.. 1 Rim II
15 DUN131 BOYTO... 1 Rim I?
Total: 34
Table 5.4: List of amphora stamps identified at kiln‐sites by the TAS in the Gebel Tarhuna. This table adds 13 new stamps to the previous list of 62 stamps (= 75 total).
respectively (Rebuffat 1997; http://ceipac.gh.ub.es/corpus/index.php?moftah= 8a48517
cbbef682077e68855513b58b2). Five examples of the later stamp (LSACV) have also
been found in Hortis Torlonia (Rome) and two examples from Monte Testaccio
(http://ceipac.gh.ub.es/corpus/ index.php? moftah=8a48517cbbef 682077e68855513b
58b2. 04/12/2009).
It is important to note here that some of the pottery kilns in the Tarhuna Gebel
certainly produced more than one stamp with different initials. This phenomenon has
been identified especially at the Wadi Guman kiln-sites (GUM89, GUM90 and
GUM110). The reason for the prominence of these sites must relate to the fact that since
1995 I have made repeated visits to these sites as they are all located very close to my
house. Although the kiln site GUM89 has produced four different stamps, the stamp
278
with the letters AIM occurred in 10 examples. The total of 14 stamped sherds was
collected from a sample of 100s of rims and handles examined around the site during
that period (Asmia and al-Haddad 1997: 218).
Remesal Rodriguez has studied in detail the amphora kilns located alongside the
Guadalquivir (Spain) and has observed that the relationship between local land-owners
and the process of the sale of products of the land (olive oil) is hard to establish. The
case is the same with the organisation and exploitation of Tripolitanian amphora kilns.
He established a model explaining their employment and divided them into number of
possible functions:
“a. Kilns located on private estates a-l) exploited by the owner of the estate for the packaging of his own oil alone; a-2) producing containers for the estate where it was located and for neighbouring
properties; a-3) unconnected to its own estate and producing containers for others; either directly exploited by the owner, an actor, or leased to a conductor. b. Kilns situated on public land b-1) leased to a conductor; b-2) managed by a procurator working for the public administration” (Remesal Rodriguez
1998). If we accept the Rodriguez model, Tripolitanian amphora kiln-sites presenting more
than one stamp could have functioned as production sites supplying amphorae for
several different estates or individuals (model a-2, a-3). In fact, the Wadi Guman area
was particularly well provided with the ceramic raw material, wells and water springs,
stone quarries for buildings and easy access to the most important communication
routes (see Fig. 3.46 above).
However, the spatial relationship of many of the kilns and pressing facilities suggests
that the majority of amphorae were likely manufactured on the estates which also
279
produced their liquid contents (whether oil or wine). They were used to transport
products designated for export markets. The detailed study of amphora stamps throws
light on the identification of producing areas with the respective contributions made by
commercial movement of amphorae to different places. It is possible to trace the
contribution made by some of the production centres in the Gebel Tarhuna, as well as
the period of greatest intensity of that contribution. By contrast, some of amphora
stamps produced in the Gebel Tarhuna have not yet been recognised among the main
amphora assemblages such as Ostia and Monte Testaccio. Further research on amphora
stamp find spots and further excavations may shed more light on this debate.
From the new evidence recorded by the TAS, it can be stated with reasonable
certainty that the Tarhuna plateau was the chief producer of amphora-borne goods in
Tripolitania during, at least, the first three centuries AD. Mattingly has already
discussed the evidence of Tripolitanian amphorae as a main factor for identifying a high
level of olive oil exports to Rome and its provinces (Mattingly 1988; 1995). He
commented upon and added to the evidence brought forward by Di Vita Evrard, who
was somewhat cautious about certain aspects of this trade. They both emphasise that
certain stamps, especially of the late 2nd and early 3rd centuries AD, referred to high
ranking landowners, either members of the imperial family, senators or other leading
families.
Stamps on Tripolitanian amphorae found on Monte Testaccio and other locations in
the western Mediterranean indicate that the owners of producing estates were utilising
vessels not simply for their own consumption (as private commodities) but also in
support of export trade. If they were intended for purely private consumption, why was
it necessary to put stamps on at all? The enormous extent of the olive oil and wine
280
trade, particularly from Spain and North Africa, is attested by very large number of
amphorae, both stamped and unstamped, recorded in many sites in the Roman world
(Bonifay 2004, 2007; Keay 1984; Panella 1973; Panella 1982; Panella 1977; Reynolds
1993, 1995, 1997).
Although the amphorae appear to have been made roughly and cheaply, they clearly
contained valuable commodities (oil and wine), with the largest surplus capacity
coming from the elite estates whose names the stamps represented. To judge by stamps
of the Tripolitanian amphorae found on Monte Testaccio (Table 5.5 and Fig. 5.12), the
peak exports of Tripolitanian amphorae are related to type III and dated to the late 2nd
and the first half of the third century AD (though earlier Tripolitanian products may be
hidden in the heart of Monte Testaccio). On the basis of the distribution of Tripolitanian
amphora stamps recorded outside the province, it is clear that many of these amphorae
were filled with the Lepcitanian products and then transported from the Gebel Tarhuna.
CEIPAC no
Mattingly no
Stamp Type Position
Reference Comment
1257 43 ADYRMP & ADYRMF
Trip III Neck CALLENDER M. H. (1965), no 29
1602
QCCC Trip III Neck CALLENDER M. H. (1965), no 1428
1646&1647
18, 19 QCLCV
Trip III Neck CALLENDER M. H. (1965), no 1436(b)
1731 47 CR
Trip III Handle CALLENDER M. H. (1965), no 449
2342 21 CFPCV & CFRCV
Trip III Handle CALLENDER M. H. (1965), no 319
2343 22, 23 CFPPPCV
Trip III Neck CALLENDER M. H. (1965), no 319
3671,3672&3673
28 MPF
Trip III ? CALLENDER M. H. (1965), no 1158(a)
Probably is the same stamp MΛF recorded in GUM89 by the TAS.
4073 32, 33 LSPLHBVR
Trip III Neck CALLENDER M. H. (1965), no 941(b)
281
CEIPAC no
Mattingly no
Stamp Type Position
Reference Comment
4379 38 MVC Trip III Neck CALLENDER M. H. (1965), no 1188(a)
6361 4 ...DAVG Trip III Neck RODRíGUEZ ALMEIDA E. (1981)
6362
52 MD... Trip III Neck RODRíGUEZ ALMEIDA E. (1981)
6363 3 IMPANTAVG? Trip III Neck RODRíGUEZ ALMEIDA E. (1981)
Found by the TAS in site TUT48
6364 7 SABCVACMV...?
Trip III Handle RODRíGUEZ ALMEIDA E. (1981)
6366 45 BINOM…
Trip III Neck RODRíGUEZ ALMEIDA E. (1981)
6371 47 CR
Trip III Handle RODRíGUEZ ALMEIDA E. (1981)
6373 15,16,17 PCSSCV MARIA…?
Trip III Neck RODRíGUEZ ALMEIDA E. (1981)
6375 13 PCAQ...?
Trip III Neck RODRíGUEZ ALMEIDA E. (1981)
6376 11 LCS
Trip III Neck RODRíGUEZ ALMEIDA E. (1981)
6377 61 ...FC
Trip III Neck RODRíGUEZ ALMEIDA E. (1981)
6378&6379 60 OPGAT? Trip III Neck RODRíGUEZ ALMEIDA E. (1981)
6381&6382 31 LSACV Trip III Neck RODRíGUEZ ALMEIDA E. (1981)
Recorded in TUT108 by the TAS.
6383-7 35,36 CSM / BAICI
Trip III Neck RODRíGUEZ ALMEIDA E. (1981)
6389 32 LSPLHBVR Trip III Neck RODRíGUEZ ALMEIDA E. (1981)
6390 49 FYN
Trip III Handle RODRíGUEZ ALMEIDA E. (1981)
1646 18 QCL Trip III? Neck CALLENDER M. H. (1965), no 1436(a)
15375 MIVSCA?
Trip ? Handle J. M. Blázquez Martínez, J. Remesal Rodríguez (Eds) 2001
18484 PSDL Trip ? J. M. Blázquez Martínez, J. Remesal Rodríguez (Eds) 2003
5899
OPHNNAEAES
Trip I Rim Blazquez Martinez et al. 1994
Probably the same stamp identified in TUT18 by the TAS.
18485 MVACGAL
Trip I Rim J. M. Blázquez Martínez, J. Remesal Rodríguez (Eds) 2003
Table 5.5: The main Tripolitanian stamps from Monte Testaccio.
282
18816 LY
18801 LY
06361 IT 06361 IT
18151 MA
06377 IT
18803 LY
18811 LY
18806 LY
15373 IT
05806 IT 15374 IT
15372 IT 06357 IT
06357 IT
06358 IT
06358 IT
06360 IT
06360 IT 05800 IT 18812 LY
18148 MA
06387 IT
06366 IT
06366 IT
18813 LY
08919 MA
08919 MA
26124 IT
06371 IT
06371 IT
25281 IT
06385 IT
06383 IT
06387 IT
25294 IT
06390 IT
18149 MA
18150 MA
06363 IT
08627 IT
18808 LY
18809 LY
06382 IT
06381 IT
06389 IT
06389 IT
05808 IT
18485 IT
283
06373 IT 06373 IT 26122 IT 06362 IT 06362 IT 15375 IT 18485 IT
15376 IT
06378 IT
06378 IT
05899 IT
05899 IT
06375 IT
06375 IT
06373 IT
06373 IT
18484 IT
26121 IT
07232 IT
07232 IT
08920 MA
08920 MA
06363 IT
18807 LY
18819 ES
06376 IT
18810 LY
Figure 5.18: Shows the Tripolitanian amphorae stamps recorded at Monte Testaccio (Rome) and published in CEIPAC database. (http://ceipac.ub.edu/corpus/call.php? moftah=30b932c4790e 3802be2517db8f71d031&COM=M2pvQUFBPT0jcHhNSUpFL1BJTDJsZ1FjSElRPT0=&QUERY_INF&CP=1&VLST=DAT, 02/12/2009). Five new stamps were not listed it previous tables (5.3 and 5.4) = 79 total.
From the Augustan period to the third century AD the major part of the olive oil
required by the Roman heartland was imported from Spain. Many kiln sites have been
identified along the Guadalquivir River in Baetica and these mainly produced Dressel
20 amphorae which were used to transport the Baetican olive oil to Rome (Remesal
Rodriguez 1998). The competition from the North African provinces increased in the
second century and, more dramatically, under the Severans in the early third century
AD. Olive oil from Tripolitania and Africa Proconsularis was carried to Rome and
abroad in several cylindrical amphorae. Whereas during the first two centuries amphora
of Dressel 20 dominated olive oil imported to Rome, from the end of the second and
284
early third century AD Tripolitanian amphorae, with Tunisian Africana I and II (Fig.
5.18), acquired a solid foothold in the greater Mediterranean markets.
During the Severan period the Tripolitanian amphora III became widely
distributed throughout the Mediterranean and the form is represented in large numbers
in Rome (http://ads.ahds.ac.uk/catalogue/archive/amphora_ahrb_2008/details.cfm?id=306).
Excavations conducted in the Monte Testaccio during 1995, 1996 and 1997 revealed
that sherds of the Tripolitanian amphora III were close to 30 percent of the North
African amphorae found, and, in fact, formed 100 percent of the Tripolitanian
amphorae represented in the three seasons of excavation. Meanwhile the African
amphora (Type IB), constituted 60 percent of the total; the other 10 percent comprised
much lower percentages of other Tunisian forms, such as Africana IA, Africana II,
Ostia LIX and XXIII (Aguilera Martín 2004; Remesal Rodríguez 2004). The same
discovery of the Tripolitanian III amphora has occurred in many excavated sites
especially in the western Mediterranean (Blázquez Martínez 2001). A conclusion can be
reached here that the amphora kilns of the Gebel Tarhuna produced the three
Tripolitanian amphora types in workshops certainly related to agricultural production
estates that mainly produced olive oil (types I and III) and also wine (type II?). Thus the
amphorae evidence corresponds with the recorded pressing facilities evidence though it
suggests there was a predominance of oil, but a sizeable minority of production of wine
also occurred during the Roman period.
285
Figure 5.19: The main African amphorae types recorded at the Monte Testaccio.
Main African amphorae
recorded at Monte Testaccio.
286
Chapter 6: Conclusion 6.1 Overview
From the late nineteenth century when H. S. Cowper recorded many rural sites on
the Tarhuna plateau, the significance of its rural landscape and its olive oil productivity
has exerted an important influence over subsequent debate about the nature and scale of
the region’s economy in Roman times. As discussed in Chapter 2, archaeological
fieldwork conducted since Cowper’s pioneer study in the Gebel Tarhuna by Goodchild,
Oates, Di Vita-Evrard and Mattingly, added detail about the density of farming sites,
their potential for surplus production and the role played by this region in the economy
of the main coastal centres especially Lepcis Magna. These works have highlighted the
existence of a sophisticated agricultural organisation of the Tarhuna plateau’s
countryside. This area (and the neighbouring Msellata district) acquired a reputation as
an exceedingly good agricultural district within Tripolitania and this certainly formed
the main reason that members of the Lepcitanian and Oean elites invested in
agricultural estates and amphora production there. Other reasons such as its close
location to the coast and the communication ways crossing and linking the plateau with
other areas in four directions encouraged investors to exploit the area. This perception
of the high economic potential of the Tarhuna landscape during the Roman period was
an important starting point for this study.
In order to investigate the composition and organisation of the ancient rural
settlement pattern in the Gebel Tarhuna and to gain an understanding of the economic
activity of the area in the wider context of the Roman economy, this study has
attempted to focus on a new source of data, the results of the Tarhuna Archaeological
Survey carried out and directed by me and mainly concentrated on two main wadis
287
(Turgut and Doga). This intensive study also built on my previous more extensive
research throughout the Tarhuna region.
Another goal of this research has been to make an assessment of the relationship
between rural settlement, olive oil (and wine) production across a broad timeframe
from the pre-Roman to late Roman periods, seeking to elucidate the pre-Roman (Neo-
Punic and Numidian) origins of production and combining periods in a way that earlier
studies have not done. These data have shown that, as is common in imperial settings,
Roman rule brought major changes not only to the coastal centres but also to their
hinterland. The following sections will review the important changes I have identified
and the improvements in knowledge that my study has made across several main
categories: rural settlement patterns, olive oil production and pressing facilities,
amphora production and economic aspects of the recorded sites on the Tarhuna plateau.
Rural settlement patterns: The TAS aimed to achieve higher level recording of
selected areas of the Gebel Tarhuna to examine different types of rural settlement, and
to attempt to improve knowledge of the chronology and the nature of change in the
ancient landscape and settlement pattern. The TAS has found that the landscape
changed dramatically as early as the beginning of the first century AD when the
Tarhuna plateau became more densely exploited than it had been in the pre-Roman
period (though seven sites have revealed at least first century BC evidence). The second
and third centuries AD represent the peak settlement period; farming settlements
(mainly pressing sites) came to characterise the rural landscape throughout the plateau.
The overall mapping of the surveyed area demonstrated an extensive spread of
settlement on the landscape and that this was even larger than previously known
(Chapter 2). The settlement pattern of the investigated areas was affected to some
288
degree by the natural environment that played a clear role in the distribution and
development of rural settlement over this hinterland region. The TAS has shown clearly
that this area has great archaeological potential, and has confirmed that the area was a
focus of important economic activity growth (in particular for olive oil pressing and
amphora production) during the Roman period.
Wadis and hill-slopes provided suitable conditions for cultivation and were key
structures for shaping the rural landscape of the Roman period. The spectacular growth
of the Tripolitanian coastal urban centres during the Roman period (Mattingly 1995:
140-44) was matched by an increase in the production of agricultural surpluses that
occurred in the rural landscape. The archaeological evidence reveals that the rural
production was likely dominated by a series of major rural estates, which had links with
the urban elite and with small towns and agricultural villages. These rural estates acted
as the productive centres for elite investment (including the imperial family, senators
and other local notables).
The TAS has demonstrated that the Gebel Tarhuna contains evidence for a huge
surplus production of olive oil. The large-scale production was mainly directed to the
wide Roman consumption markets. In order to achieve and sustain high production
capacity, the estate activity required huge processing equipment and infrastructure
(roads, tracks, ports, containers, pressing schemes and facilities). The pressing sites,
hydraulic features for capturing and storing water, and amphora kilns dominate the
archaeological evidence and also give clear indications of the characteristics of
economic activity practised in the Gebel Tarhuna during the Roman period. The
archaeological evidence shows that many sites (notably the oileries and large farms)
were designed for a scale of production beyond local subsistence needs, with a clear
289
export potential. Investing in the rural facilities of the Tarhuna plateau appears to have
been comparable to the economic transformation of the rural landscapes of some other
areas of Roman North Africa, especially the Kasserine region (Hitchner 1988; 1989;
1992-3; Hitchner et al 1990).
However (as demonstrated in Chapter 1), the Tarhuna plateau did not provide a
naturally conducive environment for agricultural development by the highest standards
of the Mediterranean; it is a zone of poor rainfall neither high nor regular enough for
large scale cerealculture. However, it is particularly suitable for olive cultivation (and
most likely grapes in some districts). The investors in this rural landscape could achieve
high rates of production only through extensive efforts to maximise their return on the
land. Therefore, they built cisterns, wells and dams to ensure that necessary amounts of
water reached their arboreal land and conserved its fertility. They provided different
sizes of pressing sites and facilities (oileries, large farms and small farms, Chapters 2
and 3) which I have calculated could produce millions of litres of olive oil in the “peak”
years (Chapter 4).
The TAS has served here as the starting point for analysis of the rural landscape and
site distribution in this study. The collected details regarding site location in relation to
topographic features have identified two specific regional types of farming sites. The
first group comprises small farm sites with one or two press elements. The second
group employed larger numbers of presses (from 3 – 17). For each type of production
site (small farm, large farm and oilery) there were three further variables in
classification categories: open farm building, fortified farm structure and open and
fortified buildings at the same site. In many cases these farm buildings and villa-farms
290
were found in association with cisterns, wells, dams, amphora kilns, mausolea and baths
(Chapter 3).
The natural and topographic divisions of the Tarhuna plateau landscape appear to
have played a key role in the distribution pattern of sites. Because of their prime
agricultural function, the vast majority of farming sites (in particular large farms and
oileries) were located close to the most fertile alluvial soil available and were in
locations where they could maximise rainfall capture and the communication routes
along the main wadis and their tributaries. For these reasons the TAS has specified that
62% of recorded sites in the Wadis Turgut and Doga are situated between 135 – 299 m
(asl.) and 38% are located in the higher reaches of the landscape between 300 – 515
(asl.).
So far it seems that the distribution pattern and choice of location for each site type
were affected also by the terrain altitude. The observation is supported by the
hierarchical organisation of sites and the variance between the size, elevation and
wealth of the agricultural and non-agricultural sites. The TAS has identified an increase
in the number of sites that possessed evidence of luxury elements (what are
conventionally known as villas). Previous work has emphasised the utilitarian
characteristics of the vast majority of Gebel Tarhuna farming sites (Oates 1953;
Percival 1975), but the new evidence requires revision of this judgement. Clearly many
of the larger estates possessed central facilities of some comfort (12 out of 68 farm-
building [18%] have shown luxury evidence).
One of the most important conclusions from the TAS survey about settlement
continuity and change relates to the identification for the first time of the pre-Roman
(second – first centuries BC) evidence on the Tarhuna plateau. However, the pre-
291
Roman settlement pattern cannot be fully evaluated through this limited surface
evidence and without excavation. Future work will unquestionably confirm and add
detail to the evidence presented here. The physical remnants of farm buildings and the
surface sherd-density of the principal Roman-period pottery types indicate that rural
settlement and economic activity (oileries, large and small farms, water management
and amphora production) reached a much more developed scale during the early Roman
period. However, the Gebel Tarhuna appear to have faced changing conditions from the
4th century AD, in line with similar with circumstances attested elsewhere in the
Tripolitanian hinterland regions (Felici et al. 2006; Gűnther 1994). The TAS has found
an increasing emphasises on fortified structures starting from the late Roman period
coupled with a decline in open farming sites and their replacement by the fortified
farms (gsur). This phenomenon is paralleled by evidence recorded in the Tripolitanian
pre-desert by the UNESCO Libyan Valley Survey (Barker et al. 1996; Mattingly 1995).
Measurement of the Tarhuna Gebel farming sites has revealed more information
about sites size, but the TAS has also prompted a re-evaluation of much of what was
known about rural settlement size and complexity in Tripolitania. It is possible to
contrast more utilitarian constructions with the development of surplus productive
centres. Examination of the larger settlement reveals an extraordinary rich range of
building materials with complex technologies of use, although dominated in many cases
by the opus africanum style.
Farming sites of the Tarhuna plateau as described by previous works (Cowper 1897;
Goodchild 1951; Mattingly 1985; Oates 1953;1954) or more systematically by the TAS,
reveal a regional variation in terms of form and size, ranging from small farms
(farmsteads) to agricultural villages. This variation, in type and size, is again
292
comparable to other rural farming areas in Roman North Africa such as the Kasserine,
Segermes and Caesarea (Hitchner 1988; 1990; 1992-1993; Leveau 1984; Ørsted 1992).
Olive oil production and pressing facilities: It is credible to describe the olive oil
production in the Gebel Tarhuna by the word ‘industry’, because of its characteristics of
scale and mass production. The olive oil industry of the Tarhuna plateau was
undoubtedly on a very large scale. Moreover, as we have seen, the installation of
pressing equipment and facilities required accompanying investment in land cultivation,
farm buildings, water management systems, and amphorae production. Furthermore, it
involved also a sizeable direct workforce of varying skills and negotiations with other
indirect sources of labour, which dealt with the several stages of the production and
manufacturing processes and transportation.
The Roman-period agriculture in the Gebel Tarhuna was central to the development
of the Tripolitanian coastal centres, especially Lepcis Magna. The TAS has confirmed
Mattingly’s conclusion that the intensification of production on well-developed
agricultural lands to the Tarhuna plateau from the Augustan period played a significant
role in the increase of wealth of many of the Lepcitanian aristocracy (Mattingly 1987;
1995). More than 200 presses have been recorded or re-recorded by the TAS in the
Wadis Turgut and Doga. This archaeological evidence reveals that the olive (and
vineyard) cultivation and olive oil (and wine) production were a major economic
activity and the defining characteristic of the Tarhuna plateau.
The well-preserved pressing evidence has allowed me to make an examination to
evaluate the production potential. There is mounting evidence that the elite made a
major capital investment in such large presses in order to produce a huge level of
output. Furthermore, the size of pressing elements indicates a large scale of production
293
for export, which clearly indicates that there was a real economic growth during the
Roman period (Hitchner 1993). The TAS has confirmed Mattingly’s estimation that the
Gebel Tarhuna total potential olive oil production capacity (in good years) could have
reached millions of litres (Mattingly 1988b).
Earlier works on the Tarhuna plateau, by H. S. Cowper, and D. Oates, recorded large
number of presses, mainly identified by the uprights that are still standing and visible in
the landscape (Cowper 1897; Oates 1953). The TAS extended recording to the full
planning of buildings and of pressing elements at selected farm buildings and
established a detailed typology for their interpretation. As a pioneering undertaking, the
TAS has transformed our knowledge and understanding of rural settlement and
economic activity in the Gebel Tarhuna. The wealth of pressing evidence of facilities
has been carefully examined and allowed me to establish a new classification for mills
and millstones, orthostats, press floors and counterweights. All of these elements are
typologised and illustrated by analysing, measuring and estimation of weight.
Examination of the press elements of the Tarhuna plateau has increased our
admiration for the technical sophistication of the Tripolitanian olive oil production and
emphasised their extraordinary sizes and dimensions. The density of rural pressing sites
in the Tarhuna region’s landscape indicates how the investment in producing olive oil
(and wine) was a major demand and a key economic resource for many wealthy elites
(Di Vita Evrard 1979; Mattingly 1985; 1987b; 1988a; 1991; 1995). By analysing the
press beds recorded by the TAS, it can be concluded that there were two main types
based on the channel shape (circular and square), but a number of sub-divisions has
been identified. A particularly interesting result that has come to light concerns the fact
that some of these press beds appear to have been used not only for producing olive oil
294
but also for wine (Chapter 4) being adapted for use with circular baskets and square
wooden containers (regulae). The former appear to indicate oil production, the latter
wine. However, excavations in some rural sites are needed to confirm this result.
The quantity of olive oil production from 200 presses has been estimated by
extrapolation from the volume of press facilities, height of orthostat holes, diameter of
fiscinae, milling capacity and pressing season period, at some 3.5 million kg oil over 90
days pressing work (Chapter 4). By contrast, a comparative study has revealed that the
capacity of some Ottoman period presses in the Msellata region (evidenced by a
number of manuscripts) were much smaller in terms of pressing facilities, production
capacity and density of pressing sites than their Roman predecessor.
The initial evaluation of wine production evidence indicates that this production
activity also can be traced to some extent in the farming sites on the Tarhuna plateau.
Wine appears to have been produced by these techniques, but at this early stage of
investigation it is difficult to evaluate the capacity of wine production, though it does
not seem on present evidence to equal olive oil production. For this reason, further
fieldwork on farming sites and wine amphora is necessary to develop this concept.
Amphora production: It is already known that there is strong relationship between
olive oil and wine production and amphora kiln sites. The other main olive oil
production areas in the Mediterranean, such as Baetica and Africa Proconsularis, have
been shown to be served by large numbers of amphora kiln sites (Bonifay 2004; 2007;
Fentress 2001; Keay 1984; Manacorda 1977; Panella 1973; Peacock and Williams
1986). Previous studies have considered Tripolitania as one of the major olive oil
exports (Mattingly 1985; 1988b; 1988c). Recent studies of the Monte Testaccio and
other amphora assemblages in the west Mediterranean have recorded a good percentage
295
of Tripolitanian amphorae evidence (Aguilera Martín and Revilla Calvo 2004; Blázquez
Martínez 1992; Remesal Rodríguez 2004). Nevertheless, Tripolitanian amphora studies
are still in their initial stage. This study of amphora kiln sites in the Tarhuna plateau has
set out to improve our knowledge about the amphora production in a hinterland area
which was already famous for remains of pressing facilities. Only three amphora kiln
sites were previously recorded on the plateau in the middle of the twentieth century
(Arthur 1982; Goodchild 1976; Oates 1953), while the TAS has identified 14 new
amphora production sites. There is no doubt future surveys would be able to increase
this number still further.
This pioneer study has not only increased knowledge about the Roman period rural
economic organisation but also revealed that there was a secure relationship between
the production of olive oil and wine for export and amphora production (Mattingly
1988a; 1995). Epigraphic evidence in the form of many new amphora stamps collected
by the TAS has shed more light not only on the location of amphora kiln sites but also
on the involvement of many of the region’s elite in this economic activity. The
identification of a number of imperial estates and aristocratic estates in the Gebel
Tarhuna made a major advance in knowledge. The TAS has shown from the presence
of the amphora stamps that these families, especially from Lepcis Magna, were firmly
engaged in management and ownership of estates in this hinterland zone.
Amphorae produced in the Tarhuna Gebel are clearly attested in exterior and
regional olive oil (and wine) trade. With respect to established amphora stamp
catalogues (CIPAC; Di Vita-Evrard 1979; Mattingly 1988b; 1995), the TAS has added
five new stamps to the previous Tripolitanian amphora stamps list and has identified the
location of production sites of a number of previously attested stamps. By the study of
296
amphora kiln-sites and amphora stamps from the Gebel Tarhuna, I hope to have
highlighted the need for the development and systematic revision of all data and for the
adoption of a methodology that integrates complementary data for this economic
activity.
The further study of the distribution of the amphora products is essential. Present
results are inconclusive because they are highly conditioned by the state of
archaeological research. The expansion of survey areas in the Tripolitanian coastal and
Gebel areas is needed if we are to develop more fully ceramic studies. Study of the
Tripolitanian amphora finds (types, fabric, stamps etc.) has made sure progress, but can
be misleading when the identification of production sites is so incomplete. In particular,
I believe a great contribution will be established in the Tripolitanian I, II, III types when
proper excavations are carried out at kiln sites and perhaps allowing greater
discrimination between oil and wine containers. However, such research would require
much time and effort.
There is also a need to update epigraphic data on the Tripolitanian amphora
products. Furthermore, there is scope for further studies of the local and wide external
distribution patterns and to explore stamp data to eliminate different elements of
consumption. The linkage of amphora kilns and individual estates needs further
consideration, as data are fragmented and scarce, but at a few sites prosopographic data
seem to relate to numerous producers, with some kilns serving multiple clients. The
possibility of establishing clearer links between the amphorae, olive oil and wine
production and the investment of Tripolitanian elite in these economic activities should
serve as a reminder that there is much more work that needs to be done, particularly
297
when we see that the social and economic organisation of Tripolitanian urban and rural
communities is still a matter of debate.
6.2 Economic aspects of the archaeological sites
Most of the recorded archaeological remains illustrate aspects of ancient economic
practice. The transformation of the region’s economic processes in terms of agriculture,
amphora production and trade and their impact on the settlement patterns are discussed
in this thesis. It is clear that great development in agricultural practices occurred during
the early imperial era. A radical shift occurred towards a market-oriented approach in
agricultural products, especially with specialisation in olive oil in this hinterland zone.
This was achieved by a remarkable level of capital investment – in crops, pressing
facilities and architecture at rural farms, and in dams and cisterns. This led to dramatic
growth of the scale of production in relation to Roman export markets, and also
improved the social and political situation of the regional elite class.
The area south-west of Lepcis Magna is an agriculturally rich region, the potential of
which was clearly exploited in the Roman period. The material, whatever else it may
indicate, makes it perfectly clear that at no point in the Roman period are we dealing
with a peasant subsistence economy. As Mattingly has amply demonstrated, concepts
like “economic growth” and “surplus economy” are quite relevant to the early Roman
period of the region (Mattingly 1993a; Mattingly 1991). By the late second century AD
the agricultural economy of Tripolitania had been developed by its Libo-Phoenician
aristocracy into a great source of wealth and economic complexity. This development
allowed the Lepcitanian and Oean elites to exploit most of the best soil, where most of
the agricultural lands were divided into estates of varying sizes marked by the
foundation of different sizes of farm building which associated with a number of press
298
elements and facilities. These varying farm buildings represent an important new
feature in this rural land during the Roman period, and demonstrate the significance of
Tarhuna’s productive countryside. The capital intensive nature of the oil pressing
facilities, but the lack of evidence of more luxurious elements at many of these sites,
fits in with the idea that these were rural estates run for the main benefit of absentee
landowner based at the coastal cities (Mattingly 1985; 1995; Oates 1953).
Urban elite landowners were looking to make the best use of the produce of their
directly managed estates and any rents collected in kind. There are various indicators
that the Tripolitanian oil production was directed well beyond local subsistence needs.
The extraordinary growth and ornamentation of the Tripolitanian coastal cities
demonstrates a healthy balance of trade with other parts of the Roman world. The
extraordinary scale of investment in olive farms is a strong clue to the key locally
available commodity of trade (as Mattingly 1988c argued). The benefits of the Roman
peace encouraged the efficient inter-regional and external distribution of regional
production and goods. The widespread distribution of Tripolitanian amphorae shows
the quite large scale of exports of Tripolitanian olive oil (and perhaps wine too)
(Peacock and Williams 1986; Bonifay 2004). The Tripolitanian elite may have
marketed their produce themselves or through their dependents (such as freedmen) and
some certainly owned ships for this purpose (Morley 2007: 582).
The increased number of rural sites, presses and amphora kilns provided by the TAS
clearly indicates that the Tarhuna plateau was a major centre of agricultural production,
particularly olives. The Tarhuna plateau rural villas can be added to the figure of about
1,000 villas which have been systematically surveyed in other Roman western
provinces such as Gaul, Germany, Spain and Britain (Leveau 2007: 652-3). Any
299
significant measure of economic growth in this hinterland could have been achieved
only by improving agricultural efficiency or by expanding the amount of land under
cultivation. As in other North African areas, the TAS has now demonstrated
conclusively that production was increased, with the principal agent of expansion a
profound regional specialisation in oleoculture. It is clear that the Roman rule
contributed to the expansion of olive cultivation by providing new export markets that
encouraged the introduction or spread of new or better methods of farming and water
management.
It can be concluded here that the collected archaeological data from the Gebel
Tarhuna makes it a remarkably interesting area of study of ancient economic activity
and an appropriate area for investigating several economic concepts relating to the
larger debates of the Roman economy. The estate organisation of the Tripolitanian rural
economy, the signs of economic growth, the clear evidence of olive oil specialisation
and of standardisation of a wide range of components of the rural production (from
presses, to baskets, to amphorae), can all be linked to the urban elites – who led and
oversaw the economic process and increased the strength of links between the city and
the countryside. The importance of olive oil in the economy of Tripolitania was
comparable to the other identified main olive oil exporting areas in the Mediterranean
basin (Brun 2004; Mattingly 1988c). In Spain, particularly in the Baetican region, the
archaeological evidence indicates that ‘the known number of villas and presses are very
impressive and the original total of presses could have been well in excess of 1000’
(Mattingly 1988c: 41). The evidence from Tunisia shows a very similar picture to the
evidence from Tripolitania, with increasing specialisation of oleoculture and massive
capital investment in multiple press facilities (Mattingly and Hitchner 1993). However,
with the exception of a few facilities in Croatia (Brun 2004: 62: 70), there is a lack of
300
evidence elsewhere in the Roman empire for these massive oilery facilities. Olive
presses were common features in many areas, such as Italy (Brun 2004: 7-60) and
Southern Gaul (Brun, 2004: 37-9), but these often occurred in relatively small units of
1-2 presses and the absolute scale of the presses appears to have been much smaller
than the North African examples to judge from the measured elements (Foxhall and
Forbes 1978; Brun 2004). For example, Foxhall’s detailed study of the evidence from
Greece considers the whole range of evidence, from literary and epigraphic material to
archaeological excavations and surveys. She identified that the majority of pressing
facilities recorded in the Greek rural landscape dated to the Roman period; regions such
as Methana and Southern Argolid were evidently substantial oil-producing areas in the
Roman world (Foxhall 2007: 202). However, the absolute numbers of presses and the
sizes of press elements were consistently smaller than those recorded in Tripolitania,
emphasising the unusual character of the Tripolitanian economic development.
The main contribution of this thesis in the ancient economy debate is thus about
growth and the disposal of surplus capacity of Tripolitanian olive oil production within
the Roman economy. The key to this debate is the level of agricultural surplus,
something for which, the TAS has provided new direct evidence. It is clear enough that,
from the first century AD, Tripolitania was producing a sufficiently large and reliable
olive oil surplus for export use. The total amount of wealth displayed in the coastal
cities, especially Lepcis Magna, was clearly considerable; the evidence suggests
significant increase in income alongside extensive building works, both public and
private, and the expansion of both urban and rural activity. As already mentioned future
efforts are needed to move forwards this debate. Urgent research priorities include the
excavation of an oilery site and a major amphora kiln site (with associated improvement
in knowledge of the Tripolitanian amphora typological series). However, the work
301
achieved to date remains an important contribution to the ongoing debate about the
organisation and specialisation of agricultural production and the investment of urban
elite in the countryside in classical antiquity.
Archaeological data recovered and synthesised by this thesis have demonstrated that
the estates of the Tarhuna plateau were involved in large scale olive oil production
during the Roman period, in particular in the first three centuries AD. Olive oil was the
main Tripolitanian commodity implicated in the long-distance trade in the Roman
world. It is considered as a relatively low-cost commodity (that is, it was not a luxury
item) and like a number of other basic goods, such as salted fish, grain, and table
pottery, its transport around the Roman world is best understood as trade rather than
personal equipment or gift exchange (Wilson 2009; but cf. Whittaker 1985; 1989;).
Archaeological evidence indicates that olive oil (and wine) production was managed by
estates. These estates, such as TUT12, TUT15, TUT38, TUT43, TUT46, TUT54,
DOG60, HAJ81, Halafi and Sidi Hamdan were relatively large in size and controlling
perhaps several hundred ha. They regularly included in addition to a pars rustica, a pars
urbana, an often luxuriously ornamented farmhouse that was periodically used by the
landowner during his or her visits on the estate. The intensive cultivation and large
scale production most likely emerged through a careful management of land and labour
(possibly including transhumant pastoral groups, seasonal harvesters and specialist
press builders alongside the fixed labour of the estates). These estates seem to have
been served by both slaves and tenants (the evidence from survey cannot conclusively
resolve this), as well as hired labour at the busiest times, such as the harvest and
pressing.
The olive oil export trade is also illustrated by the increasing archaeological
evidence for kilns making olive oil amphorae. The evidence that olive oil containers
302
predominated over those for wine supports the view that olive oil production in the
Gebel Tarhuna was organised on a massive scale in the Roman period and was more
significant overall than wine production (Arthur 1982; Blázquez Martínez 2001;
Bonifay 2004; Peacock and Williams 1986). The large number of amphora kilns
recorded by the TAS with the increased number of amphora stamps supports
Mattingly’s view that the hinterland of Lepcis Magna and Oea was intensively
exploited by the urban elite for olive oil production for export and that this made
Tripolitania one of the key areas for olive oil exports in the Roman word (Mattingly
1988c). The aristocratic major landowners benefited from the commercial opportunities
offered by the Roman empire through their investment in the specialised and large scale
cultivation of such a cash crop for targeted export markets. This helped them to
generate huge wealth and then support their social and political positions (Hitchner
1993; Kehoe 2007; Mattingly 1988b).
The wealth generated from agriculture by the urban elite also contributed to an
expansion of production in non-agricultural production. Much of the income gained
from rural estates was most likely spent in the cities and the needs of urban life; the
urban elite, who owned the rural estates, invested in other non-agricultural production
such as ceramic containers and shipping. This stimulus clearly can be traced in the
manufacture of specialised pressing equipment and in the amphora industries that
supported agriculture. Interestingly, these activities appear to have been largely
associated with rural rather than urban landscapes. The scale of amphora production in
the Gebel Tarhuna was obviously related to the scale of olive oil (and wine) production.
It seems to have increased substantially in the rural area where the olive oil was
produced. It is now certain from the TAS evidence that some at least of the urban-based
landowners who engaged in olive oil production also produced their own amphorae (as
303
attested by stamps). The evidence for the organisation of manufacturing on the Tarhuna
plateau is possible to place within the broader context of the Roman economy (Peacock
1982).
The Tarhuna plateau’s rural landscape contains extensive evidence for surplus
production of agricultural produce. Tripolitania is similar to other areas in North Africa,
in particular the region around the Roman towns of Cillium (Kassrine) Thelepe and
Safetula (Sbeitla) in south-central Tunisia (Hermassi 2004; Hitchner 1988; 1990; 1993;
Sehili 2008). In examining archaeological evidence for each of the Tarhuna settlement
categories, I have tried to establish the type of agricultural economy. In the first place,
the evidence suggested that lands of the Gebel Tarhuna drew the attention of the coastal
urban elite to the possibility of investing their capital in the exploitation of this interior
territory. The large amount of capital invested in the agricultural practices and oil press
facilities is clear from the massive rural constructions, particularly with regard to oilery
and large farm sites with approximately 210 presses recorded or re-recorded by the
TAS in the Wadis Turgut and Doga. The general distribution of olive oil pressing
elements in the Tarhuna countryside and their presence in considerable quantities at a
large number of settlements suggests that the region witnessed a high agricultural
exploitation, and it also was one of the most important economic resources in
Tripolitania during the Classical period. In the study area alone, there were at least 35
sites (from large farms with three to four presses to oileries with five or more presses)
that could be described as villas within the agriculturally productive area of the Wadis
Turgut and Doga, including the major sites at Senam Semana (17 presses), Sidi Buagila
(eight presses), Sidi Madi (seven presses) and es-Senam (six presses). The majority of
these farms featured an elaborate pars rustica which no doubt functioned as the central
304
facility of an agricultural estate. The structures were utilitarian farm facilities that were
used for practical purposes as rooms for pressing, storage crops and tools, and a
partition for housing labourers/servants and animals. Within these 35 farming sites,
evidence of some sort of Roman period luxurious rural settlements occurs at only about
12 sites, suggesting that only around a third of these estates had a pars urbana. This
sub-group of villas had both a high productive potential and elite residential facilities.
Within the Roman world, such patterns of rural investment and wealth generate were
comparatively rare – this enhances the importance of the present study.
The settlement and agricultural development of the Gebel Tarhuna seems to have
achieved its peak during the second and the first half of the third centuries AD, a pattern
that was reflected in the development of the countryside by the elite who sought to
make their profit in such interior territories, and extended/echoed the growing
prosperity of the coastal zone (Mattingly 1987b). Though many of the major
agricultural sites on the Tarhuna plateau appear to have been owned by the aristocrat
elite of the coastal cities, these rural production centres were economically linked with
the towns. The rural districts supplied the towns and towns facilitated trade and
exchange with external markets. Indeed the growing importance of exploiting the
countryside put the rural settlement within the Roman imperial exchange network,
where its long term success was influenced by fluctuations in market demand, transport
costs, and peace.
305
µ
06
12
18
24
3K
m
Tne northeastern Tarhuna plateau.
UC
owper's sites.
#O
ilery.
#Large farm
!M
odren centres
Tne northeastern Tarhuna plateau.
UC
owper's sites.
#O
ilery.
#Large farm
!M
odren centres
µ
06
1218
243
Km
!!
!
!
!
!
#
#
#
#
##
##
#
##
#
#
#
#
#
#
# #
##
#
## # ##
#
##
#
#
#
#
U
U
UU
UUU
U
U
U
U
U
U
U
U
UUU
UU UU
U
U
U
UU
UU
U
U
UU
U
UU
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
UU
U
U
U
U
Olive and alm
ond
Olive and alm
ond
Olive and alm
ond
Olive and alm
ond
Olive, alm
ond and fig
Olive-grape-fig-alm
ond
Meji
Dauun
Khadra
Tarhuna
Sidi es-Said
Souk aj-joumah
13.533408
13.533408
13.754246
13.754246
13.975084
13.975084
32.576336
32.576336
Figure 6.1: The coloured polygons show the m
odern cultivated areas on the northern part ofthe Tarhuna plateau.
306
Bibliography:
Abdassadq, A. 2003. Economic activity of the Mesallata region during the Late Ottoman
period. Thesis submitted in requirement for the Degree of Master in History, University of Al‐
Fatah, Tripoli (In Arabic).
Adam, J and Mathews A. 1999. Roman building: materials and techniques. London: Routledge.
Aguilera Martín, A. and Revilla Calvo, V. 2004. Novedades de epigrafía anfórica norteafricana
en el Monte Testaccio. L'Africa romana 15:1445‐1471.
Al‐Hddad, M. and Asmia, M. forthcoming. The excavation at the Wadi es‐Seri, Tarhuna. Libya
Antiqua 5.
Alarabi, A. 2006. Msellata during the Ottoman period. Thesis submitted in requirement for the
Degree of Master in History. University of Al‐Mergab. Khoms (In Arabic).
Alcock, S.E. 1989. Roman imperialism in the Greek landscape. Journal of Roman Archaeology
2:5‐35.
Amouretti, M. C. 1986. Le pain et l'huile dans la Grece antique. Centre de recherche d'histoire
ancienne 67. Annales Litteraires de l'Universite de Besancon 328. Paris
Amouretti, M. C. and Brun, J. P. 1993. La production du vin et de l’huile en Méditerranée, Bulletin de correspondance hellénique, Supplement, 26 (Athènes ‐ Paris 1993).
Anna‐Maria, B. 1983. Le commerce des amphores puniques en Tripolitaine: quelques
remarques a propose des decouvertes de Mellita (Sabratha). Bulletin archeologique du Comite
des travaux historiques et scientifiques. 19: 3‐16.
Arthur, P. 1982. Amphora Production in the Tripolitanian Gebel. Libyan Studies 13: 61‐72.
Aurigemma, S. 1926. I mosaici di Zliten. Rome and Milan: AfrIt monograph.
Barker, G. 1991. Approaches to archaeological survey. In G. Barker, and J. Lloyd (eds) 1991: 1‐
9.
Banning, E. B. 1996 Highland and Lowlands: Problems and Survey Frameworks for Rural
Archaeology in the Near East. Bulletin, American Schools of Oriental Research
301:25‐45.
Barker, G. 1995. A Mediterranean Valley: Landscape Archaeology and Annales History in the
Biferno Valley. London.
Barker, G. 2000. Farmers, herders, and miners in the Wadi Faynan, southern Jordan: a 10,000‐
year landscape archaeology. In The Archaeology of Drylands: Living at the margin, edited by D.
Gilbertson, pp. 63‐85. Routledge, London.
307
Barker, G., Gilbertson, D., Jones, B. and Mattingly, D. 1996. Farming the Desert: The UNESCO
Libyan Valleys Archaeological Survey, Vol. 1, Synthesis; Vol. 2, Gazetteer and Pottery. G. Barker
(ed). London: Society for Libyan Studies; Paris: UNESCO.
Barker, G. and Gilbertson, D. 2000. Living at the margin: themes in the archaeology of
drylands. In The Archaeology of Drylands: Living at the Margin. G. Barker and D. Gilbertson
(eds). London: Routledge, One World Archaeology 39: 3‐18.
Barker, G W & Jones G D. 1984. The UNESCO Libyan Valleys Survey VI: Investigations of a
Romano‐Libyan farm, part 1. Libyan Studies 15: 1‐45.
Barker, G. and J. Lloyd (eds) 1991. Roman Landscapes: Archaeological Survey in the
Mediterranean Region, London.
Barratt, G., Gaffney V., Goodchild H. and Wilkes S. 2000 Survey at Wroxeter Using Carrier
Phase, Differential GPS Surveying Techniques. Archaeological Prospection 7:133‐143.
Barth, H. 1857. Travels and Discoveries in North and Central Africa. London (reprint 1965).
Ben Baaziz, S. 1985. Les huileries de la haute vallée de l’Oued el Htab. Africa 9: 209‐215.
Ben Baaziz, S. 2000. Rohia et le Sraa Ouertane dans l'an quité, Tunisie. République tunisienne,
Ministère de la culture, Institut national du patrimoine (Tunis).
Bennett, J. and Coockson, B. 2009. Hendek Kale: a Late Roman multiple lever press site in
western Asia Minor. Antiquity, 83 (319): 259‐60.
Ben Rabha, K and Masturzo, N. 1997. Wadi al‐Fani (Khoms): villa, mausoleum and gasr. Libya
Antiqua 3: 214‐215.
Bevan, A. and Conolly, J. 2004 .GIS, Archaeological Survey and Landscape Archaeology on the
Island of Kythera, Greece. Journal of Field Archaeology 29: 123‐138.
Bevan, A. and Conolly, J. 2002. GIS, Archaeological Survey and Landscape Archaeology on the
Island of Kythera, Greece. Journal of Field Archaeology 29: 123‐138.
Bintliff, J & Snodgrass, A. 1988. Off‐site pottery distributions: a regional and interregional
perspective. Current Anthropology 29(3): 506‐513.
Bintliff, J. and Snodgrass, A. 1985. The Cambridge/Bradford Boeotian Expedition: the first four
years. Journal of Field Archaeology 12: 123‐161.
Blazquez Martinez, J. M, Lopez Monteagudo, G., Neira Jimenez, M. L. and San Nicolas Pedraz,
M, P. 1990. Pavimentos africanos com espectáculos de toros. Estudio comparativo a propósito
del mosaico de Silin (Tripolitania). Antiquités Africaines. 26: 155‐204.
Blázquez Martínez, J.M. and Remesal Rodríguez, J. 2001. Estudios sobre el Monte Testaccio
(Roma). II. Collecció Instrumenta, vol. 10. Proyecto Amphorae, bajo los auspicios de la Real
308
Academia de la Historia. Universitat de Barcelona. Barcelona. Universitat de Barcelona.
Barcelona.
Bonifay, M. 2004. Etudes sur la céramique romaine tardive d'Afrique. Oxford.
Bonifay B. and Garnier N. (2007). Que transportaient donc les amphores africaines ? Analyse
du contenu d’amphores africaines (annexe). In: Papi, E. e Scardigli, B. (eds). Supplying Rome
and the Roman Empire. Actes du séminaire international (Certosa di Pontignano, 3‐4 mai
2004), Porthmouth. Journal of Roman Archaeology, Supplement, 117‐122.
Brehony, J A. 1960. Semi‐nomadism in the Jebel Tarhuna. In Willimot and Clarke (ed). Field
Studies in Libya. Department of Geography Research Paper 4, 1960: 60‐69: University of
Durham.
Brogan, O. 1976‐77. Some ancient sites in eastern Tripolitania. Libya Antiqua 13‐14(1984): 93‐
120.
Brogan, O and Reynolds, J M. 1964. Inscriptions from the Tripolitanian hinterland. Libya
Antiqua 1: 43‐46.
Brun, J. P. 1986. L’oléiculture antique en Provence. Les huileries du département du Var. Paris:
Revue archéologique du Narbonnaise, Supplément, 15.
Brun, J. P. 1993. L’oléiculture et la viticulture antiques en Gaule. Instruments et installations de
production. In Amouretti M. C. and J.‐P. Brun (eds). 1993: 307‐41.
Brun, J. P. 2004. Archéologie du vin et de l’huile dans l’Empire romain. Paris.
Brun, J. P. 2003. Le vin et l’huile dans la Méditerranée antique. Paris: Editions Errance.
Buck, D. J. and Mattingly, D. J. 1985. Town and country in Roman Tripolitania. Papers in honour
of Olwen Hackett. BAR International Series, 274, Oxford.
Camps‐Fabrer, H. 1953. L’olivier et l’huile de l’Afrique romaine. Alger: Gouvernement Géneral
de l’Algerie. Direction de l’interieur et des Beaux Arts. Service des Antiquités. Missions
Archeologiques.
Caraher, W. R., Nakassis, D., and Pettegrew, D. K. 2006. Siteless survey and intensive data
collection in an artifact‐rich environment: Case studies from the eastern Corinthia, Greece.
Journal of Mediterranean Archaeology 19: 7– 43.
Carandini, A. 1970. Produzione agricola e produzione ceramica nell'Africa di eta imperiale.
Appunti sull'economia della Zeugitana e della Byzacaena. Studi Miscellanei, 15: 95‐119.
Carcopino, J. 1906. La inscription de Ain‐el‐Djemala. Contribution à l'histoire des saltus
africains et du colonat partiaire. Mélanges d'Archéologie et d'Histoire de l'Ecole francaise de
Rome: 366‐481.
309
Carlsen, J. & Tvarno, H. 1989. The Segermes Valley Archaeological Survey (region of
Zaghouan). An Interim Report. L'Africa romana 7: 803‐813.
Cherry, D. 1998. Frontier and Society in Roman North Africa. Oxford University press, Oxford.
Cifani, G. and Munzi, M. 2002. Fonti letterarie e archeologiche per la storia del Kinyps. L'Africa
romana 14: 1901‐1918.
Cowper, H. S. 1896. The senam or megalithic temples of Tripoli. The Antiquary 32: 37‐74.
Cowper, H. S. 1897. The Hill of the Graces. Glasgow.
De Vos, M. 2000. Rus Africum: terra, acqua, olio nell'Africa settentrionale : scavo e ricognizione
nei dintorni di Dougga (Alto Tell Tunisino). Trento.
De Vos, M. 2007. Olio d’oliva per Roma e per il mercato intraregionale. In Supplying Rome and
the Empire: the Proceedings of an International Seminar Held at Siena‐Certosa di Pontignano
on May 2‐4, 2004, (ed). Emanuele Papi: JRA, Supplementary series no. 69.Portsmouth, R.I.
Desio, A. 1971. Outlines and problems in Geomorphological Evolution of Libya from Rertiary to
the Present Day. In Symposium on the Geology of Libya. (ed). C. Gray. Tripoli.
Di Vita‐Evrarad, G. 1979. Quatre inscriptions du Djebel Tarhuna: le territoire de Lepcis Magna.
Quaderni di Archeologia della Libia 10: 67‐98.
Di Vita‐Evrard, G., Fontana, S. and Munzi, M. 1997. Une tombe hypogée de la nécropole
occidentale: Laurentii ou Claudii?. Libya Antiqua 3: 119‐138.
Di Vita, A. 1964. Il "limes" Romano di Tripolitania nella sua concretezza archaelogica e nella
sua realta storica. Libya Antiqua 1: 65‐98.
Di Vita, A. 1982. Gli emporia di Tripolitania dall'età di Massinissa a Diocleziano; un profilo
storico‐istituzionale. In: ANRW, II, Principat 10/2. Berlin 1982: 515‐595.
Di Vita, A. 1995. Leptis Magna. Karthago 23: 71‐77.
Dietz, S., Ladjimi Sebai, L. and Ben Hassen, H. 1995. Africa Proconsularis: Regional Studies in
the Segermes Valley of Northern Tunisia. Copenhagen.
Dore, J. N. 1984. The pottery from the Lm 4 complex. In Barker and Jones, 1984: 22‐31.
Dore, J. N. 1985. Settlement chronology in the pre‐desert zone: the evidence of the fineware.
In Buck and Mattingly 1985: 107‐26.
Dore, J. N. 1988. Pottery and the history of Roman Tripolitania: evidence from Sabratha and
the UNESCO Libyan Valleys Survey. Libyan Studies 19: 61‐85.
Dore, J. N. and Keay, N. 1989. Excavations at Sabratha 1948‐1951, Volume II The Finds, Part 1:
The Amphorae, Coarse Pottery and Building Materials. London: Society for Libyan Studies.
Drachmann, A. G. 1932. Ancient Oil Mills and Presses. Archaeologisk‐kunsthistoriske
Meddelelser, I.1 Kóbenhavn.
310
Dunnell, R and Dancey, W. 1983. The siteless survey: a regional scale data collection strategy.
Advances in Archaeological Method and Theory, 6: 267‐287.
Dyson, S. 2003. The Roman Countryside. London: Duckworth.
Ebert, J. I. 1984 Remote Sensing Applications in Archaeology. Advances in
Archaeological Method and Theory 7:293‐353.
El‐Hennawy M. and Cheshiter, G. 1975. Geological Map of Libya 1: 250 000, Sheet NI 33‐13,
Tarabulus. General report and explanatory booklet. Ind. Res. Centre. Tripoli.
El‐Nemsi, M. 1997. Leptis Magna, excavation at ez‐Zeita (Wadi Zennad) 1996‐1997. Libya
Antiqua 3: 207‐210.
Elmayer, A. 1983. The reinterpretation of Latino‐Punic inscriptions from Roman Tripolitania.
Libyan Studies 14: 86‐95.
Elmayer, A. 1984. The reinterpretation of Latino‐Punic inscriptions from Roman Tripolitania.
Libyan Studies 15: 93‐105.
Erdkamp, P. 1999. Agriculture, underemployment and the cost of rural labour in the Roman
world. Classical Quarterly 49(2): 556‐572.
Faraj, M., Asmia, M., and Al‐Hadad, M. 1997. Tarhuna, Zwitina cave: hypogean tomb. Libya
Antiqua 3: 217‐218.
Felici, F., Munzi, M., and Tantillo, I., 2006. Austuriani e Laguatan in Tripolitania. L'Africa
romana 16(1): 591 – 688.
Fentress, E. B. 2000. The Jerba Survey: Settlement in the Punic and Roman periods. L'Africa
romana 13: 73‐85.
Fentress, E. B. 2001. Villas, wine and kilns: the landscape of late Hellenistic Jerba. Journal of
Roman Archaeology 14: 249‐268.
Finley, M. I. 1985. Ancient history: evidence and models. London: Chatto and Windus.
Fisher, P F. 1999. Models of uncertainty in spatial data. In P. Longley, M.F. Goodchild, D.J.
Maguire and D.W. Rhind, Geographical information systems . Volume 1, Principles and
technical issues. New York: 191‐205.
Fontana, S., Munzi, M. and Ricci, G. 1996. Insediamenti agricoli di età ellenistica e romana
nell'area dell'uadi Bendar (Leptis Magna). Libya Antiqua 2: 67‐72.
Foxhall, L. 1993. Oil extraction and processing equipment in Classical Greece. In Amouretti M.
C. and J.‐P. Brun 1993: 183‐199.
Foxhall, L. 2007. Olive Cultivation in Ancient Greece. Seeking the Ancient Economy. Oxford:
Oxford University Press.
Foxhall, L and Forbes, H. 1978. The queen of all trees preliminary notes on the
311
archaeology of the olive. E expedition 21.1: 37‐47.
Frankel, R. 1993. The trapetum and mola olearia. In Amouretti M. C. and J.‐P. Brun 1993: 477‐
81.
Frankel, R., Avitsur, S. and Ayalon, E. 1993. History and Technology of Olive Oil in the Holy
Land. (Oléarius Editions and Eretz Israel Museum, Arlington, Va., and Tel Aviv.
Gaffney, V. and P. M. van Leusen. 1995. Postscript – GIS, environmental determinism and
archaeology. In Archaeology and Geographic Information Systems: A European Perspective,
edited by G.R. Lock and Z. Stancic, pp. 367‐382. Taylor and Francis, London.
Gandini, C. 1999. Les formes de l’habitat rural dans la cité des Bituriges Cubi, typologie et
hiérarchie: mémoire de maitrise. Paris.
Frayn, J. M. 1984. Sheep‐rearing and the wool trade in Italy during the Roman period.
Liverpool .
Fulford, M. and Tomber, R. (eds). 1994. Excavations at Sabratha 1948–1951. Vol. 2. London:
Society for Libyan Studies.
Gilbertson, D. D and Hunt, C. O. 1996. Romano‐Libyan agriculture: walls and floodwater
farming. (eds). G. W. Barker et al. Vol. I. 1996: 191‐262.
Gilbertson, D. D, Hayes, P. P, Barker, G. and Hunt, C. O. 1984. The UNESCO Libyan Valleys
Survey VII: an interim classification and functional analysis of ancient wall technology and land
use. Libyan Studies 15: 45‐70.
Gillings, M. and G. Goodrick. 1996 Sensuous and Reflexive GIS: Exploring Visualization and
VRML. Internet Archaeology 1.
Gillings, M., Mattingly, D. and van Dalen, J. (eds) 1999. Geographical Information Systems and
Landscape Archaeology. Mediterranean Landscape Archaeology 3. Oxford: Oxbow.
Gillings, M. and Sbonias, K. 1999. Regional survey and GIS: The Boeotia Project. In Gillings et al.
1999: 35‐54.
Given, M. 2004. From density counts to ideational landscapes: intensive survey,
phenomenology, and the Sydney Cyprus Survey Project. In E. F. Athanassopoulos and L.
Wandsnider (eds). Mediterranean Archaeological Landscapes: Current Issues. Philadelphia.
Goodchild, R. G. 1950. The Limes Tripolitanvs II. The Journal of Roman Studies 40: 30‐38.
Goodchild, R G. 1951. Roman sites on the Tarhuna plateau of Tripolitania. Papers of the British
School at Rome 19: 43‐65.
Goodchild, R. G. 1952a. Mapping Roman Libya. The geographical Journal. 118 (2): 142‐52.
Goodchild, R. G. 1952b. Farming in Roman Libya. Geog. Mag. 52: 70‐80.
312
Goodchild, R. G. 1976. Libyan studies : Select Papers of the Late R. G. Goodchild. M. Reynolds
(ed). London: P. Elek.
Goudie, A. 1981. The Human Impact. Man's Role in Environmental Change. Oxford: Basil
Blackwell.
Graham, A. 1971. Roman Africa: Ayer Publishing,.
Grahame, M. 1998. Rome without Romanization: Cultural change in the pre‐desert of
Tripolitania (first‐third centuries A.D). Oxford Journal of Archaeology 17(1): 93‐111.
Greene, K. 1986. The Archaeology of the Roman Economy. London: Batsford.
Gűnther, L. 1994. Die Austorianer als Belagerer tripolitanischer Städte (um 365 n. Chr.)?
L'Africa romana 11: 1643‐1650.
Halstead, P, 1987. Traditional and Ancient Rural Economy in Mediterranean Europe: Plus ça
Change?. The Journal of Hellenic Studies, Vol. 107 (1987), pp. 77‐87.
Hayes, J. W. 1972. Late Roman Pottery. London: British School at Rome.
Hayes, J. W. 1980. A Supplement to Late Roman Pottery. London: British School at Rome.
Haynes, D. E. 1959. The Antiquities of Tripolitania. Tripoli.
Hermassi, M. 2004. L'huile et l'olivier dans la region de Thelepte durant l'antiquite: étude des
installations oleicoles. Tunis: Unpublished PhD thesis. Faculté de Sciences Humaines et
Sociales.
Hitchner, R. B. 1988. The Kasserine Archaeological Survey, 1982‐1986. Antiquités africaines 24:
7‐41.
Hitchner, R. B. 1989. The organization of rural settlement in the Cillium‐Thelepte region
(Kasserine, Central Tunisia). L'Africa romana 6: 387‐402.
Hitchner, R. B. 1992‐1993. The Kasserine Archaeological Survey 1982‐1985. Africa 11‐12: 158‐
198.
Hitchner, B. R. 1993. Olive production and the Roman economy: the case for intensive growth
in the Roman empire. In Amouretti and Brun 1993: 499‐508.
Hitchner, B. R. with contributions by S. Ellis, A. Graham, D. Mattingly and L. Neuru. 1990. The
Kasserine Archaeological Survey ‐ 1987. Antiquites Africaines 26: 231‐260.
Hope‐Simpson, R. 1983. The limitations of surface surveys. In D.R. Keller and D.W. Rupp (eds).
Archaeological Survey in the Mediterranean Area (BAR Int. 155): 45‐47. Oxford.
Hornby, A. J. W. 1945. Northern Tripolitania: a dry Mediterranean coastal region. Economic
Geography 21(4): 231‐251.
Jones, G. D. B. 1985. The Libyan Valleys Survey: the development of settlement survey. In
Buck and Mattingly 1985: 263‐89.
313
Jones, J. R. 1971. Ground water provines of the Libya Arab Republic." In Symposium on
Geology of Libya. Tripoli: 67‐88.
Kamermans, H. 1995. Survey sampling, right or wrong? In Proceedings of the Computer
Applications and Quantitative Methods in Archaeology 1994. J. Huggett and N. Ryan (eds),
123‐126. Oxford: BAR International Series 600: 123‐126.
Keay, S. J. 1984. Late Roman Amphorae in the Western Mediterranean. A typology and
economic study: the Catalan evidence. BAR Int. Ser. 196, Oxford.
Kehoe, D. 1984. Private and Imperial Management of Roman Estates in North Africa. Law and
History Review, 2(2): 241‐263.
Kehoe, D. 1988. The Economics of Agriculture on the Roman Imperia Estates in North Africa.
Gottingen: Hypomnemata 89.
Keller, D. and Rupp, D. (eds). 1983. Archaeological Survey in the Mediterranean Area.BAR
International Series 155. Oxford.
Kennedy, D. 1995. Water supply and use in the Southern Hauran, Jordan. Journal of Field
Archaeology 22(3): 275‐290.
Kvamme, K. 1997 Ranters corner: bringing the camps together: GIS and ED. Archaeological
Computing Newsletter 47:1‐4.
Lanfranchi, R. E. 2009. I torchi per la pressatura delle olive. I ritrovamenti nel territorio di
Dougga. Unpublished PhD thesis, Facoltà di Lettere e Filosofia, Corso di Laurea in Lettere
Classiche: Universtà degli Studi di Trento.
Leveau, P. 1984. Caesarea de Mauretanie et ses Campagnes. Rome: Ecole Française.
Levi Della Vida, G. 1951. The neo‐Punic dedication of the Ammonium at Ras el‐Haddagia.
Papers of the British School at Rome 19: 65‐68.
Lewarch, D and O’Brien, M. 1981. The expanding role of surface assemblages in archaeological
research. Advances in Archaeological Method and Theory, 4: 297‐342.
Lloyd, J. A. 1991. Forms of rural settlement in the early Roman empire. In Barker and Lloyd:
233‐40.
Lock, G R. and Molyneaux, B. L (eds). 2006. Confronting Scale in Archaeology: Issues of Theory
And Practice. Oxford, Oxbow books.
MacKendrick, P. 1980. The North African Stones Speak. London: UNC Press.
Manacorda, D. 1976‐77. Testimonianze sulla produzione e il consumo dell'olio Tripolitano nel
III secolo. Dialoghi di Archeologia 9‐10: 542‐601.
314
Margaritis E. and Jones, M. 2008. Crop processing of Olea europaea L.: an experimental
approach for the interpretation of archaeobotanical olive remains. Veg Hist Archaeobot 17:
381–392.
Marquez Villora, J. 1999. El comercio en el Portus Ilicitanus. El abastecimiento exterior de
productos alimentarios. Alicante: Universidad de Alicante.
Marzano, A. 2007. Roman Villas in Central Italy: A Social and Econlomic History. Leiden‐Boston.
Matijašić, R. 1982. Roman rural architecture in the territory of Colonia Iulia Pola. American
Journal of Archaeology 86(1) :53–64.
Mattingly, D. J 1982. The Roman road‐station at Thenadassa (Ain Wif). Libyan Studies 13: 73‐
80.
Mattingly, D. J. 1983. The Laguatan: a Libyan tribal confederation in the late Roman Empire.
Libyan Studies 14: 96‐108.
Mattingly, D. J. 1985. Olive oil production in Roman Tripolitania. In Buck and Mattingly 1985:
27‐ 46.
Mattingly, D. J. 1987. New perspectives on the agricultural development of Gebel and pre‐
desert in Roman Tripolitania. Revue de l'Occident Musulman et de la Mediterranee 41‐2: 45‐
65.
Mattingly, D. J. 1988a. Megalithic madness and measurement. Or how many olives could an
olive press press?. Oxford Journal of Archeology 7(2): 177‐195.
Mattingly, D. J. 1988b. The olive boom. Oil surpluses, wealth and power in Roman Tripolitania.
Libyan Studies 19: 21‐41.
Mattingly, D. J. 1988c. Oil for Export? A comparison of Libyan, Spanish and Tunisian olive oil
production. Journal of Roman Archaeology(1): 33‐56.
Mattingly, D. J. 1989a. Farmer and frontiers: exploring and defending the countryside of
Roman Tripolitania. Libyan Studies 20: 135‐153.
Mattingly, D. J. 1989b. Field survey in the Libyan Valleys. Journal of Roman Archaeology 2:
275‐280.
Mattingly, D. J. 1993. Maximum figures and maximizing strategies of oil production? Further
thoughts on the processing capacity of Roman olive presses. Amouretti and Brun 1993: 483‐
498.
Mattingly, D. J. 1994. Regional variation in Roman oleoculture: some problems of
comparability. In J. Carlsen, P. Ørsted and J.R. Skydsgaard (eds), Land Use in the Roman World
(Analecta Romanum Supp. 22, 1994): 91‐106.
315
Mattingly, D. J. 1995. Tripolitania. London: Batsford.
Mattingly, D. J. 1996a. Olive presses in Roman Africa: technical evolution or stagnation?
L'Africa romana 11: 577‐595.
Mattingly, D. J. 1996b. Explanations: people as Agency. In Barker et al. 1996: 319‐42.
Mattingly, D. J. 1997. Imperialism and Territory: Africa, a landscape of opportunity? In D.J.
Mattingly (ed), Dialogues in Roman Imperialism 1997: 115‐38.
Mattingly, D. J. with Dore, J. 1996. Romano‐Libyan Settlement: Typology and Chronology. In
Graeme Barker et al.1996: 159‐225.
Mattingly D. J., Dore, J., and Lahr, M. (with contributions by others). 2008. DMP II: 2008
fieldwork on burials and identity in the Wadi al‐Ajal. Libyan Studies 39: 223‐262.
Mattingly, D. J and Flower, C. 1996. Romano‐Libyan Settlement: Site distributions and trends.
In Barker 1996: 159‐225.
Mattingly, D. J. and Hitchner, B. R. 1991. Ancient Agriculture. Fruits of Empire‐The production
of olive oil in Roman Africa. National Geographic Research and Exploration 7(1): 36‐55.
Mattingly, D. J and Hitchner, R. B 1993. Technical specifications of some North African olive
presses of Roman date. In Amouretti and Brun: 439‐62.
Mattingly, D. J and Hitchner, R. B. 1995. Roman Africa: An Archaeological Review. The Journal
of Roman Studies 85: 165‐213.
Mattingly D. J., Lahr, M., Armitage, S., Barton, H., Dore, J., Foley, R., Merlo, S., Salem, M.,
Stock, J, and White, K. 2007. Desert Migrations: peopole, environment and culture in the
Libyan Sahara. Libyan Studies 38: 115‐156.
Mattingly D. J., Lahr, M., and Wilson, A. (with contributions by others). 2009. DMP V:
Investigations in 2009 of cemeteries and related Sites on the west side of the Taqallit
Promontory. Libyan Studies 40: 95‐132.
Mattingly, D. J and Zenati, M. 1984. The excavation of building Lm 4E: the olive press. Libyan
Studies 15:13‐18.
Mulvin, L. 2002. Late Roman villa plans: the Danube‐Balkan region. BAR International Series
1064. Oxford.
Munzi, M. and El‐Nemsi, A. 1998. Leptis Magna‐Khoms, villa romana al porto: un contesto
monetale di eta giulianea. Libya Antiqua 4: 99‐127.
Munzi, M., Felici, F., Cifani, G., Cirelli, E., Gaudiosi, E., Lucarini, G. and Matug, J. 2004. A
topographic research sample in the territory of Lepcis Magna: Silin. Libyan Studies 35: 11‐66.
Musso, L. 1998. Archaeological Missions in Libya Reports. Missione archeologica
dell'Universita Rome Tre a Leptis Magna, 1997. Libya Antiqua 4: 176‐178.
316
Oates, D. 1953. The Tripolitanian Gebel: settlement of the Roman period around Gasr ed‐
Daun. Papers of the British School at Rome 21: 81‐117.
Oates, D. 1954. Ancient Settlement in the Tripolitanian Gebel, II: the Berber period. Papers of
the British School at Rome 22: 91‐117.
Openshaw, S. 1991 A view on the GIS crisis in geography, or, using GIS to put Humpty‐Dumpty
back together again. Environment and Planning ,23:621‐628.
Ørsted, P. 1992. Town and countryside in Roman Tunisia: a preliminary report on the Tuniso‐
Danish survey project in the Oued R'mel basin in and around ancient Segermes. Journal of
Roman Archaeology 5: 69‐96.
Panella, C. 1973. Appunti su un gruppo di anfore della prima, media e tarda età imperiale.
Ostia 3: Le terme del Nuotatore: scavo dell'ambiente V et di un saggio dell'area. Studi
miscellanei. 21: 460‐633.
Panella, C. 1982. Le anfore africane della prima, media e tarda età imperiale, tipologia e
problemi. Actes du colloque sur la céramique antique (Carthage, 23‐24 juin 1980). CEDAC.
Dossier. 1: 171‐186.
Panella, C. 1977. "Anfore tripolitane a Pompei." In L'instrumentum domesticum di Ercolano e
Pompei nella prima età imperiale. Rome.
Peacock, D. P. S. 1982. Pottery in the Roman world: An ethnoarchaeological approach. New
York: Longman.
Peacock, D. P. S. and Williams, D. F. 1986. Amphorae and the Roman Economy. London:
Longman.
Percival, J. 1976. The Roman Villa. London.
Pettegrew, D K. 2001. Chasing the classical farmstead: assessing the formation and signature
of rural settlement in Greek landscape archaeology. Journal of Mediterranean Archaeology
14(2): 189‐209.
Rebuffat, R. 1988. Les fermiers du desert. L'Africa romana 5: 33‐68.
Reddé, M. 1985. Occupation humaine et mise en valeur économique dans les vallées du nord
de la Libye : l'exemple du wadi Tlal." BCTH n.s. 19: 173‐182.
Reddé, R. 1988. Prospections des vallées du nord de la Libye (1979‐1980). La région de Syrte à
l’époque romaine. Paris.
Remesal Rodríguez, J. 2004. L’Afrique au Testaccio. L'Africa romana 15: 1077‐1090.
Reynolds, P. 1993. Settlement and Pottery in the Vinalopo Valley (Alicante, Spain): AD 400‐700.
BAR Series 588. Oxford.
317
Reynolds, P. 1995. Trade in the Western Mediterranean, AD 400‐700: The ceramic evidence.
Oxford: BAR International Series 604.
Reynolds, P. 1997. A 1st century AD pottery assemblage from Lepcis Magna. In H. Walda et al.
'The 1996 excavations at Lepcis Magna. Libyan Studies 28: 49‐63.
Romanelli, P. 1929. L’ economia della Tripolitania Romana sulla base delle scoperte
archaeologiche. Riv. del. Col. Ital. 7: 537‐551.
Roy, R. S and Decker, D. T. 1990. Archaeology, data integration and GIS. In K Allen, S. Green
and E. Zubrow (eds). Interpreting space: GIS and archaeology. London: Taylor and Francis: 134
‐140.
Sehili, S. 2008. Henchir el Begar, centre du Saltus Beguensis, étude archéologique et
historique. In F. Bejaoui (ed). Actes du 6 eme Colloque International sur l'Histoire des Steppes
Tunisiennes. Sbeitla: Institut National du Patrimoine, Tunis: 85‐107.
Sergio, R. 2004. Il dio El a Leptis Magna: note comparative. L'Africa romana 3: 1565‐1566.
Shaw, B. D. 1981. Rural markets in North Africa and the political economy of the Roman
Empire. Antiquites Africaines 17: 37‐83.
Shaw, B D. 1984. Water and society in the ancient Maghrib. Antiquites Africaines 20: 121‐173.
Smith, J. T. 1997. Roman Villas: A Study in Social Structure. London: Routledge.
Sounni, A. 1982. Etude mécanique d’un pressoir de Volubilis. BAM 14: 121‐133.
Stoddart, S. 1997. GIS: a useful research technique, not an end in itself. Archaeological
Dialogues 4(1):65‐70.
Stone, D. 1997. The development of an imperial territory: Romans, Africans, and the
transformation of the rural landscape of Tunisia. Unpublished PhD thesis, University of
Michigan.
Taylor, A. R. 1960. Regional variations in olive cultivation in north Tripolitania. In . I. J. Clarke
and S. G Willimot (eds), Field studies in Libya: Department of Geography, University of
Durham: 70‐82.
Tilley, C. 1994 The Phenomenology of Landscape. Berg, Oxford.
Von Bary, E. 1883. Senams et Tumuli de la Chaine de Montagnes de le Côte Tripolitaine. Revue
d'Ethnographie, 2: 426‐33.
Van Ossel, P. and Quzoulias, P. 2000. Rural settlement economy in northern Gaul in late
empire: an overview. Journal of Roman Archaeology 13: 133‐160.
Velde, P. 2001. An extensive alternative to intensive survey: point sampling in the Riu Mannu
Survey Project, Sardinia. Journal of the Mediterranean Archaeology 14(1): 24‐52.
Vismara, C. 2007. Uchi Maius 3. Il Dipartimento di Storia dell'Università di Sassari.
318
Vita‐Finzi, C. 1961. Roman Dams in Tripolitania. Antiquity 35: 14‐23.
Walda, H. 1997. Excavations at Lepcis Magna. Libya Antiqua 3: 294‐296.
Wheatley, D 1995. Cumulative viewshed analysis: a GIS‐based method for investigating
intervisibility, and its archaeological application. In G. Lock and Z. Stancic (eds), Archaeology
and Geographical Information Systems, London: Taylor & Francis: 5‐13.
Wheatley, D. and Gillings, M. 2002. Spatial Technology and Archaeology. The Spatial
Applications of GIS. London.
Whittaker, C. R. 1980. Rural labour in three Roman provinces. In P. Garnsey (ed), Non‐Slave
labour in the Greco‐Roman World. Cambridge: 73‐99.
Whittaker C. R. (ed.) 1988. Pastoral Economies in Classical Antiquity. The Cam‐ bridge
Philological Society Supplementary Volume no. 14, Cambridge 1988.
Zevi, F. and Tchernia, A. 1969. Amphores de Byzacène au Bas‐Empire. Antiquites Africaines, 3:
173‐214.
319
Appendix 321
Table 1: the archaeological sites recorded by the TAS 2007 in the Wadis Turgut and Doga. 329
Figure 1: Samples of the TAS ceramic evidence. 335 Table 2: The Hypogeal tombs discovered during 1970s and 1980s in the Gebel Tarhuna. All their finds transferred to stores of Lepcis Magna and have catalogued by Italian archaeologists. 337
Figure 2: Three dimensions of the Tarhuna plateau topographic map. 338
Table 3: The sites are located and recorded by Cowper during years 1895-6. 341
Table 4: The gsur (ditched sites) are located on the Gebel Tarhuna and identified by the Google Earth. 342
Table 5: Sites recorded by the intensive survey in the upper Wadi Guman. 343
Table 6: Sites recorded by Oates in the area around Gasr ed-Dauun 1953-4. 346
Figure 3: TUT 38. 347
Figure 4: DOG 66. 348
Figure 5: DOG 60. 349
Figure 6: TUT 5. 350
Figure 7: TUT 1. 351
Figure 8: TUT 3. 351
Figure 9: TUT4. 352
Figure 10: TUT5. 353
Figure 11: TUT6, TUT7. 354
Figure 12: TUT 8. 354
Figure 13: TUT 9. 355
Figure 14: TUT 10. 355
Figure 15: TUT 11. 356
Figure 16: TUT 12. 355
Figure 17: TUT 13. 357
Figure 18: TUT 14. 357
Figure 19: TUT 15. 358
Figure 20: TUT 16. 358
Figure 21: TUT 17. 359
Figure 22: TUT 18. 360
Figure 23: TUT 19. 360
Figure 24: TUT 20. 361
Figure 25: Dams in the Wadi Turgut. 361
Figure 26: TUT 26. 362
Figure 27: TUT 27. 362
Figure 28: TUT 28. 363
Figure 29: TUT 29. 364
Figure 30: TUT 30. 364
Figure 31: TUT33, TUT34, TUT35 365
Figure 32: TUT 38. 365
Figure 33: TUT41, TUT42. 366
Figure 34: TUT 43. 366
Figure 35: TUT44, TUT46. 367
320
Figure 36: TUT48, TUT50. 367
Figure 37: TUT 51. 368
Figure 38: TUT 52. 369
Figure 39: TUT 53. 369 Figure 40: A comparative standard deviation between some press elements of Methana (Greece) and Tarhuna (Done by Lin Foxhall). 370 Table 7: The standard deviation of some press elements recorded in Methana (Greece) and Tarhuna (Libya). 371
321
Site Code X Y Z Site type
Secondary type
No. Pres
s Location Local name Luxury element Other features
Period Evidence
TUT1 13.95416 32.49605 280 Large farm (villa)
gasr
3 Plateau
Traces of bath building in north side.
A cistern on southern slope.
1‐5 AD Eastern sigillata A, Ritt.8, Dr.27 Tetta Sigillata and Trip.I amphora. ARS Hayes 94 and TRS Forms 2, 4C.
TUT2 13.94896 32.50078 295 Farmstead
gasr
1 Plateau
A small dam in the north western gulley.
2‐7 AD Coarseware T. 42,46, Keay 8b amphora, building material.
TUT3 13.9481 32.5105 300 Large farm
gasr
3 Break of slope
Two cisterns and late building
1‐6 AD Curle 15 Tetta Sigillata, CW49, Trip. amphorae I,III, ARS Hayes 94.
TUT4 13.92294 32.50016 280 Small farm
2 Wadi side A mill
2‐4 AD Trip. amphorae I,II and III.
TUT5 13.91784 32.49309 315 Large farm gasr
4 Break of slope
Henscir Aziza
A mill and cistern
1‐6 AD Eastern sigillata A, Dressel 2‐4, Trip. II amphora, ARS Hayes 87b.
TUT6 13.91024 32.48583 374 Gasr
Hill-top 4‐7 AD Hayes 67, TRS
TUT7 13.92287 32.48839 305 Large farm gasr
3 Plateau Ben Hayb Two cisterns 1 BC‐6 AD Gaulish Tetta Sigillata, Trip. amphorae
I,III, TRS Forms 1, 10.
TUT8 13.9271 32.48679 290 Oilery (villa)
5 Plateau
Traces of bath elements, tile and large piece of column.
1‐3 AD Italian Tetta Sigillata, Trip. amphorae I
TUT9 13.92998 32.48258 280 Small Farm
1 Wadi side
2‐3 AD Trip. amphora I.
TUT10 13.92605 32.4914 295 Oilery
5 Plateau
Two wells in the north and north‐east sides.
1‐4 AD Dr 18, Trip. amphorae I,II and III, Hayes 45a and 58.
TUT11 13.92997 32.50511 266 Large Farm
3 Wadi side
1 BC‐3 AD CW 35,70, Arretine cub, Trip. I,II.
TUT12 13.75801 32.59669 242 Oilery villa
8 Hill slope Sidi Buagila
Traces of small bath (north east side) and 2? pottery kilns.
Stamped amphora handle. Cistern
1‐3 AD Eastern sigillata A, Italian Tetta Sigillata, Dressel 2‐4 amphora, Trip. I,II,III amphorae.
TUT13 13.88975 32.5293 318 Gasr
Hill-top 4‐7 AD ARS Hayes 67, Hayes 87c, TRS Forms 8B,
10.
TUT14 13.89032 32.53144 333 Large Farm gasr
3 Plateau Bu-Kaala 1‐7 AD Trip. I,II, ARS Hayes 71, TRS
TUT15 13.87324 32.52065 280 Oilery Villa
5 Break of slope
Henscir.Assalha
Traces of bath building, 5 ? pottery kilns.
Amphora stamps
1 BC‐3 AD Campana A, Eastern sigillata A, Trip. I,II,III amphorae.
322
Site Code X Y Z Site type
Secondary type No.
Press Location
Local name Luxury element Other features
Period Evidence
TUT16 13.85546 32.5195 375 Oilery
gasr
5 Break of slope
Henscir Boshaina
A mill, 3 millstones and 3 cisterns
1‐7 AD Trip. I,II,III amphorae. ARS Hayes 87b.
TUT17 13.83649 32.51971 290 Gasr
Hill-top 6‐7 AD Late R. coarseware, Hayes Type 2 Lamp,
building material.
TUT18 13.82918 32.51762 277 Pottery kiln
Wadi side
Two amphora stamps and cistern
1‐3 AD Trip. I,II,III amphorae.
TUT19 13.82623 32.51857 293 Bath Break of
slope A mosaic floors 2‐5 AD Trip. I,II,III amphorae, mosaics, ARS
Stamped Floor, Style A.
TUT20 13.8264 32.51961 297 Oilery
6 Plateau Two cisterns. 2‐4 AD Trip. I,II,III amphorae.
TUT21 13.82338 32.51862 266 Dam Wadi
floor
TUT22 13.8738 32.51722 265 Dam Wadi
floor
TUT23 13.87036 32.51789 264 Dam Wadi
floor
TUT24 13.86562 32.51982 260 Dam Wadi
floor
TUT25 13.85233 32.51916 253 Dam Wadi
floor
TUT26 13.81845 32.52084 254 Large farm
4 Wadi side
1 ‐4 AD Italian Tetta Sigillata, Trip. I,II,III amphorae.
TUT27 13.82572 32.52801 280 Large farm
3 Hill slope Two cisterns 1 BC‐3 AD Numidian coin, Eastern sigillata A, Tip. II
amphora.
TUT28 13.81337 32.52581 290 Gasr
Hill-top 4‐7 AD TRS Forms 1, 4C, 5, building material.
TUT29 13.821 32.52886 255 Large farm
4 Hill slope Two cisterns 1‐4 AD Trip. I,II,III amphorae. CW Type 133.
TUT30 13.81496 32.53659 228 Gasr Wadi
side 4‐7 AD Building material.
TUT31 13.80145 32.53545 272 Small farm
2 Plateau 2‐5 AD Trip. I,II,III amphorae, Hayes form 71
ARS.
TUT32 13.79842 32.54169 222 Small farm
2 Hill slope A cistern 1‐5 AD Gualish sigillata, Trip. I,II,III amphorae,
TRS Form 3.
TUT33 13.78673 32.51909 268 Gasr Wadi
side Gasr Al-Atresh A cistern
5‐7 AD Hayes 74 TRS, ARS Forms 2, 5.
TUT34 13.77508 32.51848 302 Gasr
Hill-top Ras Al-Assal
5‐7 AD Building material.
TUT35 13.77444 32.52088 275 Large farm
4 Wadi side
1‐4 AD Trip. I,II,III amphorae, Eastern sigillata A, CW Types 43,47.
323
Site Code X Y Z Site type
Secondary type
No. Pres
s Location Local name Luxury element Other features
Period Evidence
TUT36 13.81368 32.52854 226 Large farm
3 Break of slope Three cisterns
2‐4 AD Trip.II, III amphorae, Hayes 197, CW Types 58,62.
TUT37 13.77342 32.54203 241 Gasr
Hill-top Gsair al-Atshan
A dam in the southern gulley
5‐7 AD Hayes 87c, building material.
TUT38 13.78007 32.55392 227 Oilery (villa)
7 Wadi side Assenam
Traces of bath elements, tile and a large piece of column
Two mills and two cisterns
1‐4 AD Eastern sigillata A, Dr. 45 Italian Sigillata, Trip. I,II,III amphorae
TUT39 13.77511 32.55447 242 Small farm
2 Break of slope .
2‐4 AD Trip. I,II,III amphorae. CW Type 70.
TUT40 13.74008 32.57075 251 Gasr
2 Break of slope Kerath A cistern
2‐6 AD Trip. II,III amphorae, building material.
TUT41 13.77651 32.56103 222 Small Farm
1 Wadi side
1‐4 AD Trip. I,II,III amphorae, CW Type 59.
TUT42 13.78141 32.5654 248 Gasr
1 Hill-top 4‐7 AD TRS, building material.
TUT43 13.78664 32.56657 219 Oilery (villa)
6 Wadi side
Loud al-Meghara
Traces of bath elements, tile and a piece of capital.
1‐3 AD Trip. I,II,III amphorae. Eastern sigillata A, ARS forms 193, 199.
TUT44 13.78449 32.57413 211 Large farm
3 Break of slope
Sidi Yekhlef A cistern
1‐4 AD Trip. I,II,III amphorae., Italian Sigillata.
TUT45 13.78409 32.57956 203 Small farm
2 Hillslope 2‐5 AD Trip. II amphora, Hayes 58.
TUT46 13.74235 32.56502 250 Oilery (villa)
gasr 5
Break of slope Kerath
Two columns and capital
1 BC‐3 AD Campana A, Numidian coin, Trip. I,II,III amphorae. CW Types 37, 241.
TUT47 13.7886 32.59335 168 Small farm
2 Wadi side Arrebaia
1‐4 AD Trip. I,II,III amphorae..
TUT48 13.78984 32.59531 150 Pottery kiln
Wadi side Arrebaia
Five kilns and four amphora stamps have been collected
2‐4 AD Trip. I,II,III amphorae, Dressel 2‐4.
TUT49 13.78659 32.59558 200 Dam Wadi
floor Arrebaia
TUT50 13.79342 32.59361 200 Gasr
1 Plateau Two cisterns
3‐6 AD Trip. III amphora, TRS Forms 1. 9.
TUT51 13.78441 32.59186 217 Gasr
Hill-top 4‐7 AD Building material.
324
Site Code X Y Z Site type
Secondary type
No. Pres
s Location Local name Luxury element Other features
Period Evidence
TUT52 13.79937 32.61133 150 Oilery (villa)
7 Wadi side Sidi Madi
Arched gate and column‐pieces
2‐4 AD Trip. I,II,III amphorae. Hayes 58 ARS, CW Types 254, 147, 324.
TUT53 13.77526 32.62271 180 Large farm (villa)
3 Break of slope Sidi Eysawi
The villa contains portico, and it is associated with a pottery kiln and a bath (with mosaic). Two cisterns.
1‐5 AD Mosaic, Italian Sigillata, Trip. I,II,III amphorae, Hayes form 71.
TUT54 13.78341 32.64893 145 Oilery (villa)
17 Wadi side
Senam Semana
Bath‐building, mosaic, cisterns and hypogeal tombs.
1‐4 AD Eastern sigillata A, Dr. 45 Italian Sigillata, Trip. I,II,III amphorae.
TUT55 13.78868 32.64187 219 Gasr
Hill-top 4‐7 AD TRS, building material.
TUT56 13.75662 32.59811 240 Gasr
Earlier open farm 3 Plateau
Sidi Buagila Two cisterns
1‐6 AD Eastern sigillata A, Trip. I,II,III amphorae.
TUT57 13.73877 32.58254 250 Large farm
3 Break of slope
Henscir Hmoudat A cistern
1‐4 AD Building material.
TUT58 13.7325 32.58233 250 Gasr
Earlier open farm 1
Break of slope
4‐6 AD Building material.
DOG59 13.74378 32.58309 266 Gasr Hill-top 5‐7 AD TRS, Hayes Form 87b.
DOG60 13.68954 32.6081 184 Oilery (villa)
6 Wadi side Senam Aref
Traces of bath building in eastern side, capital Large cistern
1‐4 AD Curle 15 Tetta Sigillata, Trip. I,II,III amphorae, CW Type 49.
DOG61 13.69099 32.60569 225 Quarry Break of
slope
DOG62 13.69484 32.60528 235 Gasr
Hill-top 5‐7 AD Building material.
DOG63 13.69968 32.6003 243 Gasr
Hill-top 5‐7 AD Building material.
DOG64 13.70622 32.60406 200 Small farm
2 Wadi side
1‐4 AD Trip. I,II,III amphorae. CW Type 133.
DOG65 13.71082 32.59023 261 Gasr
Hill-top 4‐6 AD Building material.
DOG66 13.71008 32.59188 230 Oilery
6 Break of slope
Sidi al-Akhder
Large cistern on the eastern slope
1‐3 AD Trip. I,II,III amphorae.
325
Site Code X Y Z Site type
Secondary type
No. Pres
s Location Local name Luxury element Other features
Period Evidence
DOG67 13.70269 32.58102 275 Small farm
2 Break of slope
2‐6 AD Trip. I,II,III amphorae. ARS Stamped Floor, Style A(ii), ARS Hayes 91c.
DOG68 13.70177 32.57049 264 Small farm
2 Break of slope
2‐6 AD Building material.
DOG70 13.68687 32.55076 305 Gasr
2 Plateau 4‐7 AD ARS Hayes 105, TRS 3, 4C.
DOG71 13.66637 32.51523 450 Border inscription
Hillslope
1 AD
DOG72 13.69839 32.48787 420 Mausoleum Wadi
side Gasr Doga 1 AD New Punic inscription.
DOG73 13.69999 32.48989 405 Bath Wadi
floor Many scattered mosaic tessary
1‐4 AD Mosaic, Trip. I,III amphorae.
DOG74 13.70738 32.49523 474 Small farm 2 Plateau 1‐4 AD Trip. I,III amphorae. Cw Type 70.
DOG75 13.69739 32.48041 445 Road station
Plateau Mesphy
1‐4 AD
HAJ76 13.72202 32.53628 457 Watchtower
Hill-top
Gasr al-Ash
5‐7 AD TRS Form 10, building material.
HAJ77 13.71188 32.53391 477 Watchtower
Hill-top
Gasr Abdalhadi
4‐7 AD Building material.
HAJ78 13.72621 32.54913 325 Mausoleum Break of
slope 1 BC‐1 AD Conspectus Form 4 Italian Sigillata,
Eastern sigilata A.
HAJ79 13.71947 32.5548 284 Gasr
Plateau Gasr Dehmesh
4‐6 AD ARS Hayes 71, 94.
HAJ80 13.71783 32.55601 270 Bath Break of
slope Two cisterns 2‐4 AD Trip. I,III amphorae.
HAJ81 13.71757 32.55536 215 Large farm
4 Break of slope Three cisterns.
1‐3 AD Eastern sigilata A. Ritt. 12 Italian Sigillata.
HAJ82 13.71551 32.55581 280 Oilery
5 Plateau 1‐3 AD Trip.amph. I, II, III. CW. Type 184.
GUM83 13.77401 32.49535 407 Gasr
Quarry
Hill-top Ras Deiseer
Two unfinished columns and orthostat
4‐7 AD Hayes 68, TRS Forms 2,3.
GUM84 13.77655 32.49497 384 Dam Wadi
floor
GUM85 13.77638 32.49361 397 Quarry
Hill slope Two milling stones
2‐5 AD
GUM86 13.77941 32.49077 404 Pottery kiln Wadi
side 1‐3 AD Trip.amph. I, III.
GUM87 13.77681 32.49409 386 Bath (villa) Wadi
floor Mosaic floors, tile and pipes
Spring and 5 wells
2‐5 AD Mosaic, Hayes 62, Trip.amph. I, II, III, 5th century coins.
GUM88 13.78437 32.49035 423 Small farm (villa?)
2 Plateau Gaytna
Traces of bath‐building and kiln? Large cave
1‐4 AD Eastern sigilata A, Gualish Sigillata, Hayes Form 45a, Hayes 58.
326
Site Code X Y Z Site type
Secondary type
No. Pres
s Location Local name Luxury element Other features
Period Evidence
GUM89 13.78418 32.48644 413 Pottery kiln Wadi
side Amphora stamps
1‐4 AD Trip.amph. I, II, III.
GUM90 13.79228 32.49046 417 Pottery kiln
Plateau . Amphora stamps
1‐4 AD Trip.amph. I, II, III.
TEL91 31.62406 32.5089 455 Mausoleum Break of
slope Sonoma ar-Ragda
1‐2 AD Dr 35 Sigilata. CW Type 66.
TEL92 13.62855 32.50894 471 Mausoleum Break of
slope 4‐6 AD ARS Hayes 94, ARS Hayes 181, TRS Forms
2,3.
TEL93 13.63722 32.50675 515 Watchtower
Hill-top Butaweel
6‐7 AD TRS, building material.
TEL94 13.62791 32.50863 462 Gasr
Hill slope 4‐6 AD Hayes 181, building material.
TEL95 13.62677 32.50962 446 Small farm
2 Break of slope Two cisterns
2‐4 AD Trip.amph. II, III.
DOG69 13.70087 32.5604 284 Gasr
Plateau 4‐7 AD Building material.
TEL96 13.66168 32.50309 468 Small farm
2 Plateau A cistern 1‐4 AD Trip.amph. I, II, III. Hayes Form 45a.
TEL97 13.64874 32.5119 378 Small farm
1 Break of slope
2‐4 AD Building material.
TEL98 13.65053 32.51969 402 Gasr Break of
slope A cistern 4‐7 AD Building material.
TEL99 13.64784 32.5205 391 Small farm
2 Plateau 1‐3 AD Trip.amph. II, III.
TEL100 13.64935 32.51488 357 Bath Wadi
floor 2‐5 AD Building material.
TEL101 13.643 32.51032 353 Bath Wadi
floor 3‐6 AD Walters 79 Sigillata, Hayes 53 Rim, TRS
Hayes forms 4‐10.
TEL102 13.64342 32.50882 365 Pottery kiln Wadi
side 1‐3 AD Trip.amph. I, II.
DOG103 13.72604 32.59219 224 Small farm
2 Wadi side
Sida Masoud
2‐4 AD Trip.amph. II, III.
DOG104 13.7142 32.5988 210 Large farm (villa ?)
3
Wadi side Two cisterns.
1‐4 AD Dr. 35 Sigillata, Hayes 62, Trip.amph. II, III.
DOG105 13.72358 32.58633 221 Small farm (villa)
2 Break of slope Aulad Ali
Traces of bath building and colums. Two cisterns.
1‐4 AD Gualish Sigillata, ARS Hayes 94, Trip.amph. II, III.
DOG106 13.71959 32.597 217 Oilery
5 Break of slope
Sh'bet asc-Schood
1‐3 AD Trip.amph. I, II, III. CW Type 50.
DOG107 13.72423 32.60121 232 Oilery
5 Wadi side Three cisterns
1‐4 AD Trip.amph. I, II, III.
327
Site Code X Y Z Site type
Secondary type
No. Pres
s Location Local name Luxury element Other features
Period Evidence
TUT108 13.8958 32.53113 320 Pottery kiln Wadi
side Henscir Rmadia
Amphora stamps
1‐3 AD Trip.amph. I, II, III.
TUT109 13.88791 32.53415 353 Small Farm
2 Break of slope
Henscir ar-Rkkak
2‐5 AD Trip.amph. I, II, III. Hayes 58, Hayes Form 60.
GUM110 13.78214 32.48403 420 Pottery kiln Wadi
side 2‐4 AD Trip.amph. I, II, III.
DOG111 13.68581 32.61866 164 Large farm,
Dam, kiln,
quarry 3 Slope Wadi Mseel
1‐4 AD Trip.amph. I, II, III. CW Type 37.
TUT112
13.75801
32.59669
242 Oilery
Villa
8 Slope Two p.kilns, cistern and bath
1‐3 AD Trip.amph. I, II, III.
SRI113 13.50744 32.4071
335 Bath
Wadi side Wadi Seri
4‐5 AD Mosiac, 4th‐5th c. coins.
SRI114 13.50929 32.40657
338 Pottery kilns
Wadi side Wadi Seri 18 kilns
3‐5 AD Trip.amph. II, III.
SRI115 13.51289 32.40607
342 Cemetery
Plateau
2 BC‐3 AD Campana A finewares, Numidian coins.
SRI116 13.49779 32.40645
355 Gasr
4‐6 AD Building material.
Table 1: the archaeological sites recorded by the TAS 2007 in the Wadis Turgut and Doga.
ARS Form 58 TRS Form 1 ARS Form 181
328
ARS Form 62
(a)
(b)
a,b) Campanian pottery.
329
A b c
D e f
A) TRS Form 2, b) TRS Form 10, c) TRS Form 4c, d) ARS Form 27, e) ARS 185, f) TRS lamp.
330
a b
c d
a) LR TRS lamp, b) Eastern Sigillata A, c) a New‐Punic letter marks the eastern Sigillata dish, d) Campana A sherd.
331
a b
a) Campana A; b) religious status found in a sanctuary site at the Wadi Sri within the same level of the Campana A potsherds and the Numidian coins.
332
a b c
d e f
g h
i j
333
k l
m n
a) C ware Sabratha Type 37.481; b) Canpana A; c) C ware Sabratha Type 63.662; d) C ware Sabratha Type 50.538; e) C ware Sabratha Type 324.2342; f) C ware Sabratha Type 254.3205; g) C ware Sabratha Type 70.2511; h) C ware Sabratha Type 66.2499; i) C ware Sabratha Type 262.3205; j) C ware Sabratha Type 241.2153; k) C ware Sabratha Type 37.487; l) C ware Sabratha Type 147.2970; m) C ware Sabratha Type 133.1601; n) C ware Sabratha Type 147.2985 (Dore and Keay 1989).
Figure 1: Samples of the TAS ceramic evidence.
334
Tomb no.
Area
Location
Date of discovery
Structural characteristics
Material
Type of pottery Cronology Bibliography
19 Tarhuna El-Zagadna 1982 Hypogeal TRS (4); Common ceramic(2); Lamp (4)
4th – 5th centuries AD.
Inventory of Lepcis Magna no. 7872-7880; archivio Soprintendenza Leptis Magna.
20 Tarhuna Gasr ed-Daun
1978 Hypogeal with two rooms and vaulted complied barrel
Urns type 1,1(8) with New-Punic inscriptions. Italian sigillata (22); Eastern sigillata A (1); ARS (1); Common ceramic(13); amphorae (6); Lamp (11).
Eastern sigillata A: atlante II form35. ARS: Hayes 5a. Tripolitanian amphorae.
AD 50-150 Inventory of Lepcis Magna no. 5246-5347; archivio Soprintendenza Leptis Magna.
21 Tarhuna Gasr Doga 1980 Hypogeal shaft 2,60 X 1,60 m
Limestone urns(3); Amphorae (3); ARS(6); TRS(2); Common ceramic(4); Lamp (4); Terracotta (1).
ARS:Hayes forms: 15,25,42; TRS: 2,1
AD 200-300 Inventory of Lepcis Magna no. 6347-6420; archivio Soprintendenza Leptis Magna.
33 Tarhuna Sidi Asid 1974 Hypogeal TRS; Common ceramic 4th – 5th centuries AD.
Inventory of Lepcis Magna no. 1154-1165.
41 Tarhuna Gasr Doga 1977 Hypogeal ? ? ? Inventory of Lepcis Magna no. 4943-4953.
44 Tarhuna Gasr Doga 1978 Hypogeal ? ? ? Inventory of Lepcis Magna no. 6278-6289.
45 Tarhuna Gasr Doga 1978 Hypogeal Eastern sigillata(1); Lamp (3). Eastern sigillata: Atlante II form 12.
AD 0-50. Inventory of Lepcis Magna no. 6278-6289.
52 Tarhuna El-Zagadna 1984 Hypogeal with an entrance of an irregular shape.
TRS(7); Glass(1); Common ceramic(4); Lamp (5).
? 5th century AD.
Inventory of Lepcis Magna no. 7884-7907.
55 Tarhuna Sidi Asid 1978 Hypogeal Italian sigillata(1); Eastern sigillata A(1); Amphorae(6).
Italian sigillata: form IX; Eastern sigillata
AD 0-50. Inventory of Lepcis Magna no. 3869-
335
A: form Atalnte II 4; Amphora Trip. II; Van Der Werf 3; Benghazi ER3
3932; archivio Soprintendenza Leptis Magna.
56 Tarhuna Sidi Asid 1973 Hypogeal TRS(2); Amphorae(2). ? 5th century AD.
Inventory of Lepcis Magna no. 3155-3160; archivio Soprintendenza Leptis Magna.
59 Tarhuna Sidi Mohammar Algnega
1989 Hypogeal Four urns with Latin inscriptions; Amphorae(2); Lamp (1).
Trip. I AD 100-150. Inventory of Lepcis Magna no. 8123-8126; archivio Soprintendenza Leptis Magna.
Table 2: The Hypogeal tombs discovered during 1970s and 1980s in the Gebel Tarhuna. All their finds transferred to stores of Lepcis Magna and have catalogued by Italian archaeologists.
336
Figure 2: Three dimensions of the Tarhuna plateau topographic map.
Tarhuna
Elevation850 - 900735 - 849625 - 734515 - 624400 - 514300 - 399180 - 29970 - 179 0 - 69 m
µ0 10 20 30 405Km
Tarhuna plateau
13.312570
13.312570
13.533408
13.533408
13.754246
13.754246
13.975084
13.975084
14.195922
14.195922
31.9
3225
2
31.9
3225
2
32.2
5429
4
32.2
5429
4
32.5
7633
6
32.5
7633
6
337
Site no. X Y Z Site type No. of Press
11 13.70646 32.49113 471 SF 117 13.73393 32.45511 455 SF 124 13.7195 32.39699 422 SF 126 13.70135 32.39952 441 SF 137 13.86652 32.44114 415 SF 138 13.86566 32.43217 402 SF 139 13.86873 32.42751 393 SF 140 13.89805 32.41667 362 SF 142 13.87394 32.4505 409 SF 144 13.92998 32.48258 280 SF 147 13.95416 32.49605 280 Villa 149 13.92642 32.46933 327 SF 168 13.86703 32.38327 315 SF 13 13.70622 32.60406 200 SF 27 13.72358 32.58633 221 Villa 2
13 13.73171 32.45731 440 SF 214 13.73535 32.45146 439 SF 216 13.74027 32.45639 458 SF 220 13.72961 32.42337 451 SF 221 13.7205 32.41577 416 SF 222 13.72316 32.42364 436 SF 223 13.72476 32.40891 406 SF 225 13.78326 32.42743 424 SF 228 13.68811 32.41776 470 SF 264 13.75263 32.46399 440 SF 267 13.84795 32.38352 345 SF 269 13.8859 32.37594 310 SF 2
338
4 13.70337 32.59869 215 LF 312 13.70083 32.47489 435 Village 319 13.72724 32.43228 443 LF 336 13.86021 32.4464 411 LF 341 13.91498 32.43428 382 LF 343 13.93182 32.43516 375 LF 363 13.78262 32.46054 427 LF 365 13.83684 32.42881 364 LF 366 13.83875 32.42366 363 LF 32 13.68954 32.6081 184 LF 4
32 13.76454 32.44357 412 LF 460 13.78007 32.55392 227 LF 4
61 13.77444 32.52088 275Large Farm 4
1 13.67798 32.6203 165 O&F 56 13.71959 32.597 217 Oilery 5
33 13.78001 32.44192 400 Oilery 556 13.75965 32.64525 159 Oilery 55 13.71008 32.59188 230 Oilery 6
58 13.79937 32.61133 150Oilery villa 7
59 13.78664 32.56657 219Oilery villa 7
45 13.96377 32.49023 270 Oilery 8
57 13.78341 32.64893 145Oilery villa 17
8 13.72856 32.57325 2459 13.71947 32.5548 284 G
10 13.69839 32.48787 420 Mausoleum
339
18 13.73806 32.46149 45729 13.69813 32.42985 46230 13.71065 32.44505 446 Mausoleum 31 13.75565 32.44667 426 Sancatury 35 13.85614 32.44946 410 G 48 13.96637 32.51036 45262 13.77655 32.49497 384 Dam 70 13.90647 32.39578 340
Table 3: The sites are located and recorded by Cowper during years 1895-6.
SITENo X Y Z 1 13.39565 32.23533 4802 13.39674 32.24458 4453 13.39383 32.25177 4204 13.39816 32.25516 4545 13.38288 32.25429 4206 13.3879 32.28547 4407 13.40229 32.29391 4448 13.38869 32.28002 4439 13.3666 32.25546 390
10 13.358 32.2814 40811 13.3433 32.28415 39012 13.37531 32.29425 46513 13.37855 32.28963 46514 13.38324 32.30042 45015 13.42393 32.37233 36516 13.47229 32.36324 40017 13.4818 32.34126 39718 13.52013 32.40232 371
340
19 13.51497 32.37535 37820 13.51997 32.40229 37021 13.54128 32.43664 38422 13.59923 32.43141 49023 13.56681 32.45783 42524 13.6087 32.46528 37525 13.58673 32.46607 38226 13.56249 32.46981 43027 13.59012 32.47488 38628 13.58967 32.48138 36729 13.52934 32.5244 38530 13.5676 32.5144 35131 13.60061 32.49523 32632 13.60734 32.48902 31233 13.58955 32.48138 36834 13.64479 32.47365 40635 13.67028 32.48062 43236 13.65605 32.50828 45037 13.68033 32.46771 47538 13.69812 32.42981 45539 13.76384 32.44208 40240 13.75125 32.45901 44641 13.74343 32.48544 44642 13.74023 32.49428 45543 13.761 32.4855 447
Table 4: The gsur (ditched sites) are located on the Gebel Tarhuna and identified by the Google Earth.
341
SITE No X Y Z SITE TYPE 1 13.77401 32.49535 407 Fortified Gasr 2 13.77655 32.49497 384 Dam 3 13.77638 32.49361 397 Quarry 4 13.77941 32.49077 404 Pottery kiln 5 13.77681 32.49409 386 Bath 6 13.78418 32.48644 413 Pottery kiln 7 13.79228 32.49046 417 Pottery kiln 8 13.78214 32.48403 420 Pottery kiln 9 13.7936 32.49447 450 Small Farm
10 13.78748 32.49086 449 Small Farm 11 13.7857 32.49086 460 Small Farm 12 13.78437 32.49035 423 Small Farm 13 13.78232 32.48995 416 Opus siginum 14 13.78093 32.49036 410 Enclosure 15 13.78149 32.48921 419 Opus siginum 16 13.77966 32.48685 438 Small Farm 17 13.79299 32.48885 435 Small Farm 18 13.79167 32.49055 429 Dam 19 13.7804 32.49213 409 Enclosure 20 13.7802 32.49317 410 Terrace wall 21 13.7896 32.48298 430 Hoard
Table 5: Sites recorded by the intensive survey in the upper Wadi Guman.
342
SITE No. X Y Z
No. Press
3 13.87863 32.45702 380 1 4 13.87009 32.45591 388 15 13.86913 32.45022 411 17 13.87863 32.44533 409 1
13.85614 32.44946 410 113.87512 32.42846 393 113.86873 32.42751 393 113.91498 32.43428 382 113.93133 32.40709 360 114.01803 32.30004 240 113.97515 32.30328 248 113.86652 32.44114 415 113.86566 32.43217 402 113.86552 32.4285 395 113.99558 32.46621 235 113.92642 32.46933 327 113.92998 32.48258 280 113.9271 32.48679 290 1
13.95416 32.49605 280 113.96637 32.51036 452 1
12 13.95032 32.51603 331 113 13.9067 32.5178 310 1
13.92003 32.52158 296 113.86477 32.48152 446 113.85668 32.49682 358 1
6 13.86211 32.45023 420 213.87538 32.44249 418 2
343
13.92189 32.42049 360 213.95266 32.41151 355 213.99433 32.4162 338 213.95386 32.41708 354 213.88827 32.38554 326 213.87617 32.43171 385 213.98622 32.44853 272 213.92116 32.46653 342 213.92287 32.48839 305 213.92997 32.50511 266 213.88975 32.5293 318 213.85554 32.4836 450 2
2 13.8751 32.46083 395 313.86021 32.4464 411 313.86566 32.43217 402 313.89805 32.41667 362 313.98749 32.3793 295 3
10 13.96223 32.45625 265 313.90304 32.4558 331 313.88512 32.47379 430 313.85389 32.48881 456 313.86409 32.43187 391 413.93247 32.38261 318 413.87628 32.47522 430 413.88204 32.43122 382 513.96377 32.49023 270 813.97918 32.40416 311 913.86551 32.42853 40013.9243 32.4173 361
344
13.87552 32.42664 39213.90647 32.39578 34013.95629 32.51422 35413.90623 32.52312 322
Table 6: Sites recorded by Oates in the area around Gasr ed-Dauun 1953-4.
345
Figure 3: TUT 38.
346
Figure 4: DOG 66.
347
Figure 5: DOG 60.
348
Figure 6: TUT 5.
349
Figure 7: TUT 1.
Figure 8: TUT 3.
350
Figure 9: TUT4.
351
Figure 10: TUT5.
352
TUT 6 ................................................................... TUT 7
Figure 11
Figure 12: TUT 8.
353
Figure 13: TUT 9.
Figure 14: TUT 10.
354
Figure 15: TUT 11.
Figure 16: TUT 12.
355
Figure 17: TUT 13.
Figure 18: TUT 14.
356
Figure 19: TUT 15.
Figure 20: TUT 16.
357
Figure 21: TUT 17.
358
Figure 22: TUT 18.
Figure 23: TUT 19.
359
Figure 24: TUT 20.
TUT 21 TUT 22 TUT 23
Figure 25: Dams in the Wadi Turgut.
360
Figure 26: TUT 26.
Figure 27: TUT 27.
361
Figure 28: TUT 28.
362
Figure 29: TUT 29.
Figure 30: TUT 30.
363
TUT 33 ......................................................................... TUT 34 ........................................................................ TUT 35
Figure 31.
Figure 32: TUT 38.
364
TUT 41 ............................................................... TUT 42
Figure 33.
Figure 34: TUT 43.
365
TUT 44 ...................................................................... TUT 46
Figure 35.
TUT 48 ....................................................................................... TUT 50
Figure 36.
366
Figure 37: TUT 51.
367
Figure 38: TUT 52.
Figure 39: TUT 53.
368
Figure 40: A comparative standard deviation between some press elements of Methana (Greece) and Tarhuna (Done by Lin Foxhall).
0.84
0.86
0.88
0.9
0.92
0.94
0.96
0.98
1
1.02
Methana trapeta
MS75
MS218
MS114
MS109
A9.1
A21.2
MS123
St Dev 0.113411738
0
0.5
1
1.5
2
Methana press beds
0
0.5
1
1.5
2
2.5
1 2 3 4 5 6 7 8 9 10 11 12
Tarhuna mills St Dev 0.22394534
1.25
1.3
1.35
1.4
1.45
1.5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Tarhuna press beds
369
Methana site trapetum D Methana site press bed D Tarhuna mill D orthostat H
p.beds int
C20.2 0.69 MS123 0.55 1.5 2.2 3.1 1.33 C27.1 0.87 C29.1 0.55 1.5 2.2 3.1 1.33 MS19/20 0.9 D28.3 0.61 1.6 2.3 3.15 1.35 MS75 0.9 MS123 0.62 1.65 2.3 3.15 1.35 MS218 0.9 MS218 0.66 1.7 2.4 3.2 1.37 MS114 0.96 MS22 0.7 1.8 2.4 3.2 1.38 MS109 0.98 MS109 0.77 1.9 2.4 3.3 1.4 A9.1 0.99 C29.1 1.06 1.9 2.45 3.35 1.4 A21.2 1 MS70 1.1 2 2.55 3.35 1.4 MS101 1.13 MS122 1.12 2.05 2.65 3.4 1.42 MS22 MS210 1.22 2.1 2.65 3.5 1.42 MS70 MS123 1.45 2.1 2.8 3.6 1.44 C29.1 MS19/20 St Dev 2.85 3.8 1.44 MS122 MS75 0.223945 2.85 St Dev 1.45 MS123 MS101 2.9 0.388942 1.45 MS123 C20.2 2.9 1.48 MS123 C27.1 2.9 St Dev MS210 MS114 2.9 0.045821 C29.1 A9.1 2.9 D28.3 A21.2 2.95
3
St Dev 0.113411738 0.305736815 3
3
Methana data from: Foxhall, L. (1997) Ancient farmsteads, other agricultural sites and 3
equipment, in Mee, C.B. and Forbes, H.A. (eds) 3.05
A Rough and Rocky Place: the landscape and settlement history of the Methana Peninsula, 3.1 Greece, 257-68. Liverpool: Liverpool University Press. 3.1
Table 7: The standard deviation of some press elements recorded in Methana (Greece) and Tarhuna (Libya).
