1

Recognising polities in prehistoric Crete.T. Whitelaw.The Institute of ArchaeologyUniversity College London

To appear in: M. Relaki and Y. Papadatos (eds) From the Foundation to the Legacy of MinoanSociety. (Sheffield Studies in Aegean Archaeology.) Oxford: Oxbow Books.

I. Re-considering polities in prehistoric Crete.

For most of the first century of Minoanarchaeology, a reconstruction of the politicalstructure of palatial period Crete, structuredaround the three major palaces identified atKnossos, Phaistos and Malia, was universallyaccepted. This picture became naturalised,generating the expectation that the agrarianMinoan states were necessarily centred on themajor lowland basins, in what are today primeagricultural zones (e.g. Renfrew 1972: fig. 14.4).This led to the further expectation that additionalpalace centres might be discovered in only a fewcomparable locations, such as the major coastalplains near the medieval and modern centres ofRethymnon and Khania in the archaeologicallyunder-explored west of the island (e.g. Youngerand Rehak 2008a: 150; 2008b: 178). Anomaliesto these expectations were explained as sub-ordinate centres (Haghia Triadha, Gournia), orexceptional (Zakros) (e.g. Warren 1985: 74;Younger and Rehak 2008a: 150-2).

Palaces discovered in recent decades atPetras and Galatas, and elaborate structures atKommos, Archanes, Mochlos, Protoria,Makrygialos and Khania (and re-investigated atMonastiraki) have generally been subsumedwithin this accepted structure as eitherrepresenting subordinate centres, or side-lined asnon-canonical centres (e.g. Kommos). However,the chronologies of these newly recognisedcentres, and detailed re-assessments of thehistory of the three major palaces (e.g.Macdonald 2002; 2010; 2012; La Rosa 2002;2010a; Pelon 2005; Driessen 2010), have createdproblems for the temporal as well as spatialreconstruction of Minoan political organisation.As our chronological understanding of individualsites improves, it is increasingly difficult to alignthe history of any of the palaces with the basicPrepalatial, Protopalatial and Neopalatialtemporal scheme, or accept that this encapsulatesmajor island-wide organisational transformationsin Cretan political history.

This long-standing framework wasconstructed on the perceived paralleldevelopment of the three major palaces, based far

more on assumption than evidence. With limitedattention to the Protopalatial levels at the majorpalatial sites until recently, the final Neopalatialpicture (with ideas on the administrative structureretrodicted from the deciphered Linear B tabletsfrom the Final Palatial phase at Knossos) wasprojected back to the foundation of the palaces atthe start of the second millennium. Thisinevitably created a very static picture of thepalaces and palatial society (Bennet 1990: 198),and also required a fundamental transformation attheir inception. As the long accepted spatial andtemporal frameworks are challenged by newevidence and ideas, debates are increasinglytargeting the starting date for the process, thenature and scale of the states which developed,the pace of the transformations, and theuniversality of these characteristics and processesacross the island.

The overall interpretive problem is todocument and explain the transformation ofCretan societies from ubiquitous small,independent communities, to at least a limitednumber of integrated, bureaucratic states duringthe first half of the second millennium BC. Thestart and end points of this process are clearest, atleast for the polities centred on the three majorpalaces of central Crete. I have argued elsewherethat the communities at Knossos, Phaistos andMalia, large but not exceptional by Aegeanstandards in EMII, each witnessed rapidpopulation growth in the final Prepalatial phase(EMIII-MMIA) (Whitelaw 2004; 2012). Thiscorresponds to the evidence now widely acceptedfor the earliest monumental constructions at eachsite (Macdonald 2010: 532; La Rosa 2010a: 583-4; Driessen 2010: 559). These dramatictransformations mark, if not the start, thencertainly a fundamental step-change in theprocesses of state formation on Crete.

At the other end of the trajectory, ourclearest understanding of the nature of a state onprehistoric Crete comes from the LMIIIA2period, as documented through the decipheredLinear B tablets recovered from Knossos. Thesedemonstrate the integration of at least the centraland western two-thirds of the island into a single

2

polity, administered from the palace at Knossos(Bennet 1985). Debates remain about the datesof establishment and collapse of this polity, andits full extent, since it is anchored in space byonly a limited number of toponyms which can belinked convincingly to the names of laterClassical cities.

Did this extensive polity represent aninheritance by the LMIIIA administration atKnossos, of a pre-existing LMIB state centred onKnossos, or was this a new creation sometimeduring the phases LMII-IIIA2? Any answer iscomplicated by several recent arguments, first,for the preservation of an early Linear B archivein the Room of the Chariot Tablets at Knossos,possibly documenting a smaller, simpleradministration, which would then have expandedto the scale documented by the remainder of thearchive by the time the place was destroyed(Driessen 2000; 2001b). Support for such anexpansion has been built on the changingdistribution of ceramics in the Knossian LMII toLMIIIA1 styles (Popham 1980; Bennet 1985:242-5; Rehak and Younger 2001: 441-2) andperhaps changing burial practices (Preston 2004:333-7). Secondly, as ceramic assemblages fromrecent excavations are studied morecomprehensively, stylistic regionalism isincreasingly being proposed for LMIB (e.g. vander Moortel 2002; Floyd 1998; Barnard et al.2003; Platon 2002; Hatzaki 2007; MacGillivray2007; Knappett and Cunningham 2006; Broganand Hallager 2011), and now LMII (Arvanitakis2007). While caught in the problematic equationof pottery style with political affiliation (seebelow), such regionalism is thought by some toquestion the assumption of a major Knossos-centred state, and therefore any continuity ofpolitical structure from LMIB through LMII toLMIIIA1-2.

Pushing this uncertainty one phase back,we also have to recognise the 'Troubled Island'model which interprets the LMIB phase not asthe apogee of Minoan Crete, but as a period ofcrisis and possibly political fragmentation(Driessen and Macdonald 1997). Whilevigorously disputed, this has usefully questioneda wide range of assumptions which deservecritical attention.

An overall scenario often alluded to, andone that I expect most Minoan archaeologistswould broadly agree with, posits that states incentral Crete developed at the very start of theProtopalatial period at, and only at, the threecentres of Knossos, Phaistos and Malia, which

during all or most of the period, divided centralCrete amongst them. These were unified into asingle polity, controlled by Knossos, by somepoint within the Neopalatial period (Younger andRehak 2008a: 150). Debates then revolve aroundthe extent of such a polity, whether justencompassing central Crete, or how far to theeast or west it extended. To assess this scenarioor propose alternatives, we need to be able todefine the territories of polities, and track theirdevelopment in space and time.

While recognising the variety of statesocieties which have been defined byanthropologists, social and political theorists,within the restricted scope of prehistoric Crete, asan initial step, discussion can usefully focus onthe distinction drawn by Trigger between city-states and territorial states (Trigger 2003: 92-119). He explores a wide range of differencesbetween these; the most relevant points here arestructure, scale and integration. City-states areessentially single-city political entities, whichdominate a restricted hinterland necessary tosupport the inhabitants of that city, and providegoods and services to the population of the cityand its hinterland. Small city-states may havetheir population wholly resident within thecentral city, or also distributed across thehinterland in much smaller communities. Inlarger city-states, subsidiary communitiesdevelop to enable more effective exploitation ofthe larger territory, with resources to support theurban population channelled up through a moredeveloped settlement hierarchy (Steponaitis1981; Wright 2000). The cities themselves areconstrained in scale due to agriculturalproductivity and the effectiveness of bulktransport technology to provision the city’sresidents (Falconer 1987; Wilkinson 1994;Bintliff 1999; 2002).

Territorial states are more extensive, willinclude multiple urban centres, and encompassmore territory than the hinterland necessary tosupport the central city alone. It has been notedcross-culturally that city-states usually develop inthe competitive context of similar polities (Price1977; Renfrew 1975; 1986; Griffeth and Thomas1981; Yoffee 1991; Feinman 1994; 1998;Nichols and Charlton 1997; Hansen 2000a;Wright 2005), and that territorial states are oftencreated through the unification by alliance orconquest of neighbouring city-states (Trigger2003: 92-119; Marcus 1998). These largerpolities are usually unstable and short-lived,disaggregating into individual city-states which

3

are more effectively integrated social, economicand political entities.

What is far less often explored, is justhow difficult it is archaeologically to detectpolitical unification, and therefore to define theextent of territorial states, without writtenrecords. In large part, this is because the culturalunits which can be defined archaeologicallythrough variability in material culture are usuallyconsiderably more extensive than individual city-states. Material culture differences tend to markbroad cultural units (e.g. Yoffee 1991; Lightfootand Martinez 1995; Emberling 1997; 1999;Feinman 1998: 101; Brumfiel 2005; Trinkaus1987), but can represent identities at a wide rangeof social and spatial scales (Jones 1997; Stark1998; Lucy 2005). Compounding the difficultyof recognition, such political amalgamations areusually dynamic, expanding or contracting morerapidly than material culture distributions change,and crucially, tend to be shorter-lived than theresolution of most archaeological artefact-basedchronologies.

A clear illustration of this recognitionproblem is the contrast in models for ClassicMaya territorial organisation proposed before thedecipherment of the Mayan glyphs. Someanalysts used broad ceramic and architecturalstyles and the scale of sites, architecturalcomplexes or monuments to infer a small numberof large-scale hierarchically-organised politiesacross the Mayan lowlands (e.g. Adams 1991:170-4; Adams and Jones 1981; Marcus 1983),whereas others proposed that the dozens of citieswere politically independent city-states (e.g.Mathews 1985; Sharer 1994: 494-512). Thesituation has been considerably clarified, thoughnot entirely resolved (e.g. Flannery 1998: 17-21)through the decipherment of the Mayan glyphs,allowing the reading of genealogies and histories,themselves of course subject to propagandistconstruction and interpretive ambiguity. Thesepaint a picture of complex and fluid relationsamong a wide range of small to largeindependent and quasi-independent city-states,with short-term amalgamations and alliancesconstructed through inter-dynastic marriages andconquests, almost entirely invisiblearchaeologically (Mathews 1991; Schele andMathews 1991; Grube 2000).

In a limited number of archaeologicalexamples, settlement pattern data documentinggaps between clusters of settlements have beenargued to represent buffer zones between

competing polities (Adams 1981: 63-7; Feinman1998; Whitelaw 1998). This usually requires alarger scale and finer resolution of survey datathan are available for most regions; in theseexamples, such interpretations were inspired byhistorical or ethnohistorical accounts of polityindependence.

The different nature and scale of city-states and territorial states may require verydifferent approaches for archaeologicalrecognition. A variety of approaches have beenused to define polities on prehistoric Crete,usually without explicit theoretical ormethodological justification. These will bereviewed, and a further approach explored,assessing its potential strengths and limitations.

II. Topography and territories.

The traditional model of the political structure ofMinoan Crete, focused on the three majorpalaces, assumes they 'make sense' in terms ofnatural divisions of the central Cretan landscape.Most of a century of investigation withoutrevealing additional palatial centres, encouragedthe view that these were the only centres, and lessexplicitly, that all of at least central Crete wasdivided amongst them (e.g. Warren 1985: 74;Cherry 1986: fig.2.2; Bennet 1990: 195-8).

In fact, the island has supported adiversity of political divisions (Fig. 1: Bennet1990; Faraklas et al. 1998; Perlman 2004;Chaniotis 1996; Sanders 1982; Harrison 1993;Detorakis 1994), with the four-fold north-coaststructured model significant only when the islandhas been politically dominated from outside, bynorth Mediterranean-based empires (Bennet1990). Figure 2 schematically represents thescale and number of independent polities onCrete through time. City-states of various sizeswere the norm, with territorial states relativelyinfrequent, and island unification extremely rare,except when the island was incorporated within amuch larger state or empire. The one prehistoricepisode of large-scale unification, during LMII-IIIA2, was extremely short-lived, probablyreflecting the difficulty of establishing andmaintaining an indigenous integration of theentire, attenuated island. This documentedhistorical diversity of political entities on Cretesuggests that there is no 'natural' structure to thepolitical organisation of the island; geographyand topography are relevant, but not determining,as is commonly assumed.

4

Fig. 1. The political organisation of Crete, prehistory to the present: A. Principal MM sites; B. LMadministrative centres; C. Linear B organisational structure; D. Archaic centres; E. Early Hellenistic cities; F.Late Hellenistic states; G. Roman cities; H. Saracen; I. Early Venetian; J. Late Venetian; K. Ottoman; L.Autonomous Crete; M. Axis occupation; N. Hellenic state.

Turning from environmentaldeterminism to historical contingency, if wesimply accept the locations of the known earlypalatial centres as given, can these be used topredict the extent of their associated polities?This is what Cherry's well-known map ofThiessen polygons imposed on the distribution ofknown and potential palaces represents (1986:fig. 2.2).

This has recently been re-examinedcritically, using GIS techniques to update theThiessen polygon approach with modelledwalking time, dividing central Crete among thethree known early and principal palaces (Bevan2010). But is this how people actually interactacross the Cretan landscape? Various categories

of social and biological data (e.g. dialects,marriage patterns, genetic data) should documentways in which recent populations have socialisedthe same landscape. These do not dictate howearlier populations would have behaved, butanalogically, may alert us to new possibilities, orchallenge our intuitive assumptions. A recentlypublished study of local differences invocabulary in 1960s-70s Crete (Kontosopoulos2006), documents variants of 172 words orphrases recorded for 163 communities across theisland. The degree of shared usage can arguablybe considered a rough index of the intensity ofinter-community interaction between residents atdifferent locations, and allows us to investigatethe assumed relevance of distance andtopography to inter-community interaction on

5

Fig. 2. Schematic representation of scale and duration of polities on Crete, through time.

Crete, before the widespread impact of tourismand personal motorised transport on mobility.

To explore the potential division of thecentral Cretan landscape among the three majorpalace sites, in Figure 3, the degree of similarityin word use is assessed for each community inthe study, in relation to the modern communitiesnearest to each of the three palatial centres. Thefourth map compares these datasets to definewhich communities are most similar invocabulary to each of these centres.

Overall, this indicates that vocabularydifferences increase systematically with distance,

providing support for the assumption that GISwalking-time calculations provide a realisticbase-line for analysing inter-communityinteractions. Not surprisingly, the major massifsof Lasithi and Ida represent strong barriers tointeraction. More surprisingly, there are onlyvery low levels of differentiation within centralCrete, with no clear boundaries reflecting thetopography commonly assumed to defineterritories for the three major palaces. In fact, theeastern Mesara is slightly closer in terms ofvocabulary, to both the Knossos and Malia areas,than it is to Phaistos, also predicted by thewalking-time models.

6

Fig. 3. Recent vocabulary variants: A. similarity with Knossos; B. similarity with Malia; C. similarity withPhaistos; D. partition among three palace sites, compared with territories estimated by Thiessen polygons andwalking times (Bevan 2010, fig.4).

From this preliminary exploration, itseems clear that the major topographic barriersshould be relevant to understanding how peopleinteract across the Cretan landscape. Butprevious studies have mapped additional

expectations onto the past landscape which seemunjustifiable, at least viewed against one proxymeasure of recent patterns of interaction on theground.

7

Of course, this approach to predictingpolities from centres only makes sense if we haveconfidence that we have recognised all therelevant polity centres, which the discoveries ofthe past few decades should call into question,particularly acknowledging how much of theisland has never received systematic or intensivearchaeological investigation. Additionally, as atop-down, partitive approach, it can only dividethe territory among known centres, it cannotestablish whether parts of the landscape layoutside the political control of those or any othercentres.

Before leaving geography andtopography, it is possible to work in a moreexploratory and systematic way, from the bottomup, to define potential landscape units on theisland (see also Bevan and Wilson 2013). Aspre-industrial societies, agriculture wasfundamental to supporting Minoan polities, ofwhatever scale. While many characteristics willaffect agricultural production, surface slopeseverely constrains the areas available foragriculture in the Mediterranean prior to thewidespread construction of agricultural terraces.Figure 4 maps areas of less than 10 slope,revealing numerous agriculturally productivelandscape blocks at a variety of different scales.As we shall see, the palatial centres could havebeen supported from fairly limited territories(Figs. 5-7). Simply taking the Malia plain as anexample, we can identify a considerable number

of low-slope basins on that sort of scale acrossthe island, far more than have ever beenconsidered as potential prehistoric palatialterritories. While this does not predict that all oreven most will have supported a palatial centre, itseems premature to assume that the additionalpalaces discovered in the past 25 years will notbe supplemented by others, when more of Cretereceives intensive archaeological investigation.

Over the first century of Minoanarchaeology, a model imposed top-down fromassumptions about the three major central Cretanpalace sites, has generated a remarkably resilientconceptualisation of the Minoan politicallandscape. This 'understanding' in turnencouraged assumptions about the deterministnature of Cretan geography which are notsupported by historical patterns, unbiasedanalyses of topography or patterns of recentsocial interaction, and have been challenged byrecent archaeological discoveries. Islandgeography and topography, and the mechanicalconstraints they impose on communication,interaction and transport, will be relevant to howpast communities constructed their social andpolitical landscapes, but in far less deterministicways than have been assumed.

III. Defining polities through administrativeevidence.

Administrative systems need to be considered ona sliding scale, and while the nature and content

Fig. 4. Island topography and centres: A. topographic relief and known palaces and other major centres; B. low-slope land.

8

of the existing records are accepted asdemonstrating a state level administrativestructure in Neopalatial Crete (Schoep 1999;2002), they only hint at its existence before theend of the Protopalatial period. The very limitedHieroglyphic and Linear A records of that dategive little idea of the nature or scale ofadministration. The control over access toseveral storerooms documented by the abundantsealings from Phaistos (Weingarten 1986), neednot represent a state-level administration, beinganticipated by centuries in the sealing system inuse at EHII Lerna. The scale of the Protopalatialpalace structures themselves (Macdonald 2010;2012; La Rosa 2010; Militello 2012), and thedifferentiated society they represent, support anassumption of state administration for theirconstruction, maintenance, and the specialisedactivities they housed, but tell us little about theirregional dominance or how they were organisedor functioned.

For the Protopalatial period, the use oftwo different script systems at Phaistos vsKnossos, Malia and Petras, suggests at least twoindependent administrations, distinguishingPhaistos, though not necessarily differentiatingany of the north coast sites.

Turning to the Neopalatial period, wherewe have considerably more abundant evidence,the political interpretation of the administrativeartefacts is far from clear. Differences inadministrative practices have been welldocumented, particularly between HaghiaTriadha and Zakros, the two sites with substantialsamples of administrative artefacts (Weingarten1986; Hallager 1996; Schoep 1999).Unfortunately, these assemblages largelyrepresent different types of specialisedadministrative activities, and are not directlycomparable for an assessment of differences inoverall administrative practices; they simplyemphasise how partial our evidence is foradministration at each site. There are minordifferences in administrative practices at differentsites (Weingarten 1986; Schoep 1999), but theirsignificance is difficult to assess withoutcomprehensive documentation of the overallvariability within and between each local system.Additionally, is it realistic to expect uniformadministrative practices at different centreswithin the same political system (Schoep 1999:203-13, 220-1)? If the system works locally, willa newly dominant centre necessarily change it,and even if so, how much standardisation will beimposed, and how long will such changes take topercolate down the levels of the administrative

hierarchy? We simply do not know, so whereasidentity of practices might be argued as evidencefor a common administration, the differencesdocumented to date need not indicate the inverse.

A specific category of administrativeartefact is represented by the 'look-alike' sealingsrecovered from a number of sites, and oftenconsidered to document the involvement of aKnossian administration across much of theisland (Betts 1967; Hallager 1996: 207-13;Wiener 2007: 236), now also documented bysealings at Akrotiri (Karnava 2008; Weingarten2010). While a range of arguments is regularlyassembled to support the idea of Knossianpolitical hegemony, their ambiguity is usuallyrecognised, and the look-alike sealings have beenproposed as the conclusive evidence for politicaldomination (e.g. Niemeier 2004: 394), so theytake on exceptional significance. Recentpetrographic analyses confirm the similarity ofthe clay used in the sealings at different sites, andits mineralogical compatibility with centralCretan geology, providing support for the earliericonographic interpretations (Goren andPanagiotopoulos 2010). While these artefactsdocument contact between (probably) Knossosand other communities, without being able toread the organic documents they originallysealed, we don't know what, if anything, theyindicate about political or administrativeorganisation (Cherry 1986: 26; Weingarten 1986:296, n. 26; 1991; 2010; Wiener 1987: 266, n. 46;Schoep 1999: 213-7; Krzyszkowska 2005: 167-8). Were these necessarily official palace-issuedadministrative documents, and even if so, werethey communications between individuals atindependent polities (cf. the Amarna lettersbetween Egypt and Hatti), instructions from adominant power (cf. the Amarna letters from thepharaoh to subordinate rulers in the southernLevant), or different types of communicationsbetween individuals at different sites; without thedocuments themselves, we have no idea.

While administrative artefacts shouldideally provide our most direct indications ofpolitical integration and polity extent, theirinterpretation is clear only if the documentsthemselves can be tied directly to identifiablesites, and provide information about the nature oftheir interactions. The Knossian Linear B tabletsdo this, recording administrative transactionslinking the centre to a large number of locationsthrough a variety of interactions involving largequantities of materials (documenting the scaleand nature of administration), and can beanchored to specific locations by a very small

9

number of toponyms in the centre and west of theisland (documenting its minimal extent).Equivalent information is not clearly recordedamong the very limited preserved Linear Adocuments (Bennet 1990: 198-99; Schoep 2001;2002), nor is it recognisable among the evenmore limited Hieroglyphic documents.

IV. Defining polities through material culturedistributions.

Various types and characteristics of Protopalatialand Neopalatial material culture have beeninvoked to support reconstructions of Minoanpolitical structure and history. Except for theLinear B texts, all other arguments areproblematic: theoretically, methodologically andempirically. For reasons of space, not all can bereviewed here, but the examples considered (eliteprestige artefacts, ceramic style) outline problemswhich apply to all attempts to define states inprehistoric Crete on the basis of material culturedistributions.

Three types of argument have beendeveloped:

1. specific types or traits document exchange ordiffusion from a single source, assumed toindicate political influence or dominance, usuallyargued or assumed to be Knossian;

2. contrasts are drawn between regional andisland-wide styles, usually in ceramics, and thelatter are assumed to imply political unification,the former fragmentation;

3. essentially a development from 2, specificceramic distributions are interpreted as mappingpolitical territories.

The first approach considers thedistribution of specific types or styles of materialculture as evidence of influence, regularlyassumed to represent political dominance,whether the materials are exchanged or locallyimitated (Wiener 1987: 266; 1990: 134-43, 150-1; 2007; Niemeier 2004: 393-4). Material cultureinterpreted this way includes ashlar masonry,mason's marks, lustral basins and Minoan halls(distributions documented by Driessen 1982;1989-90), figured frescoes (Rehak 1997), and thefinest ceramics (Betancourt 2004a; 2004b). Afundamental empirical and methodologicalproblem arises because Knossos is the mostintensively investigated palatial site, anddocuments most fully the range and sequence ofdevelopment of Minoan elite material culture. Ittherefore provides abundant comparanda formany categories of material evidence, but thisdoes not necessarily demonstrate Knossian

precedence, inspiration, production or control,which are usually assumed (however, see Bevan2010: 40-3). There is an almost inevitableKnossian sampling bias which needs to beacknowledged and assessed critically in eachcase (see also Knappett 2011:393-96).Illustrating the compound nature of theassumptions, pictorial fresco fragments recoveredat Galatas are identified as the earliest known onthe island, but are also assumed to have beenpainted by artisans trained at Knossos(Rethemiotakis 2002: 57). As other sites are morethoroughly investigated, they may revealexamples which antedate the earliest knownKnossian examples, such as the early forms oflustral basin and Minoan Hall at Quartier Mu atMalia (Driessen 1982: 54-5; Poursat 2007),though both are regularly suggested to beevidence for Knossian influence elsewhere on theisland (Wiener 1990: 140; 2007: 234).

More fundamentally, even if Knossianprecedence can be securely documented, how dowe establish that the adoption of such a traitelsewhere represents political imposition ordomination, rather than passive diffusion oractive emulation? That the spread of such traitscan be variously taken to represent Knossianpolitical dominance, cultural hegemony, peer-polity competition (Cherry 1986), or the de-centralisation of elite power (Driessen andMacdonald 1997: 71), indicates the absence ofclear theoretical justification for the Knossos-dominance model. While the 'Versailles effect'(competitive elite emulation) was coined to applyto Cretan influence in the wider Aegean (Wiener1984: 17; 1990), it has as much relevance as aprocess on Crete itself (Wiener 1987: 266;Warren 2002: 204). Aspiring individuals seek toacquire or copy status-enhancing artefacts,material styles and behavioural traits developedin the more competitive context of sociallydifferentiated urban centres. This indicates elitelevel communication, and illustrates processes ofcultural perception and valuation in activeidentity construction, but how might we actuallyestablish if, or in which cases, such adoptionsrepresent political domination?

A second set of interpretations usesceramic stylistic uniformity in contrast withregionalism as an indication of the degree ofisland-wide political unification. The equation ofceramic stylistic similarity with politicalaffiliation is dealt with in more detail below.Here, it is the contrast in relative similarity whichis taken to suggest differences in regionalpolitical integration. The focus on defining local

10

styles is relatively recent in Minoan ceramicstudies, and is highly constrained by recentexcavations which have retained andsystematically studied significant quantities ofmaterial, so the emerging evidence principallyconcerns a limited number of LMIB assemblages(e.g. van der Moortel 2002; Floyd 1998; Barnardet al. 2003; Platon 2002; MacGillivray 2007;Hatzaki 2007; Knappett and Cunningham 2006;Brogan and Hallager 2011). Not surprisingly, themore systematically and intensively one looks,the greater the distinctions and localcharacteristics which can be identified. Thecriteria used to define local styles at each site arevery mixed, and differ, not allowing systematiccomparisons or assessments of degrees ofdifference among assemblages. Equallyproblematic, there are no standards against whichto calibrate the significance of the differencesdetected. Interpretively, local styles mayrepresent different production traditions, scalesand modes of production, distribution andmarketing systems, as well as influence throughdifferent types of interactions with othercommunities. Without having criteria foridentifying and distinguishing these differentprocesses, we cannot interpret degrees ofregionalism in specifically political terms.

The third approach, using ceramic stylesto define political territories, essentially derivesfrom the previous, but has been developed moreexplicitly with Protopalatial ceramics, and is sodirectly involved with the question beingaddressed in this paper, that it requires moredetailed exploration.

Distributions defined by artefact style,predominantly pottery because of its abundance,are used world-wide to define archaeological'cultures', to establish our basic time-spaceframeworks for organising information about thepast. How such distributions should be defined,and what they actually represent in humanbehavioural and social terms, have been fiercelydebated for the past 60 years (e.g. Clarke 1968;Whallon and Brown 1982; Dunnell 1971; 1986;Hodder and Orton 1976; Hodder 1978; Adamsand Adams 1991; Lyman et al. 1997).Approaches to analysing and interpreting suchdistributions have become increasingly diverseand self-critical in recent decades, with debatesabout the meaning of artefact style (e.g. Wobst1977; Conkey and Hastorf 1989; Hegemon1992), material culture variation and transmission(e.g. Hodder 1978; 1981; Lemonnier 1993;Dobres and Hoffman 1994; Stark 1998; Stark etal. 2008; Gosselain 2000; Neiman 1995; Hurt

and Rakita 2001), and the creation andrepresentation, through material culture, ofindividual and group identities (e.g. Shennan1994; Stark 1998; Emberling 1997; Jones 1997;Robb 1999; De Atley and Findlow 1984; Greenand Perlman 1985; Lightfoot and Martinez 1995;Diaz-Andreu et al. 2005). These fundamentaland on-going debates about the identification,meaning and significance of material culturevariability have received almost no recognition inMinoan archaeology.

Providing evidence for these debates, awide range of ethnographic andethnoarchaeological research has established thatsome characteristics of material culture maymark cultural boundaries whereas others do not,representing economic distribution patterns,technological traditions or learning patterns,among other processes, and others simplyrepresent random variation (e.g. Hodder 1978;1981; Miller 1985; Stark 1998; Stark et al. 2008;Jones 1997; Emberling 1999; Gosselain 2000).Distinguishing among generating processes inethnographic contexts relies on detailed andsystematic analyses, and usually requires othercontextual information, observable in theethnographic present but not necessarilydeterminable archaeologically.

In Cretan prehistory, without explicitconsideration of the assumptions involved, somedistributions of artefacts are regularly interpreted,not as defining vague archaeological 'cultures',but far more specifically as mapping politicalentities (e.g. Cadogan 2011). The justificationfor this seems to be that the spatial units sodefined, correspond in location and scale to thepolitical entities we are expecting to find, but thisargument is obviously circular. We need toestablish why any specific material distributionsshould be considered to represent politicalaffiliation.

Stylistic distribution studies build from apassive information flow model, in which it isassumed that stylistic similarities betweenartefacts produced in different communities willdecline with the distance between them. Therationale is that producers and consumers incommunities which are closer together are likelyto interact more regularly with each other, andwill be more accepting of (and so, willing toacquire or copy) stylistic variants they arefamiliar with and accept as appropriate, throughregular exposure. So we should normally expecta fall-off in stylistic similarity in material culturewith distance (Plog 1980). Deviations from this

11

suggest that something a bit more interesting isgoing on, for example, factors inhibiting orenhancing interaction between members of thecommunities, or that specific meanings areattached to the stylistic characteristics which leadto them being differentially accepted or rejectedas imports, imitated closely, adapted, or ignored.

A significant analytical problem emergesif, like our vocabulary model, we choose two ormore base points and assess the similarity offinds from other points against them. This willinvariably define ‘territories’ of influence aroundthe base points, since the stylistic characteristicsat any other point will be more similar to one ofthe original base points than another. Because ofthe way we have structured the investigation, wewill create ‘territories’ centred on our initial basepoints, even if these are completely arbitrary andthere is, in fact, continuous variation across thearea.

A systematic and neutral approach wouldneed to give equal weight to the assemblages atall points, to try to define spatial clusters amongthem, which is the approach ideally used for thedefinition of archaeological cultures across aregion: any groupings in the data emerge throughanalysis, rather than being imposed from thestart. In contrast, Cretan studies have workedfrom what are presumed to be the mostinfluential centres outwards, with the result thatother communities are defined as responsive orsubsidiary to those initial points of reference (e.g.'provincial Middle Minoan pottery': Walberg1983). Unfortunately, because of the way Cretanarchaeology has developed, we rarely have eitherthe unbiased substantial samples, nor thequantified documentation of stylistic attributeswhich are essential for systematic exploratoryanalyses.

To illustrate the problem, forProtopalatial Crete, we have three palatial siteswhich have been the principal focus for majorinvestigations, and therefore provide the mostabundant samples of Protopalatial pottery. Wehave assumed they are the major and innovatingpottery-producing centres, and so usecomparisons with much smaller samples fromminor sites, to allocate the latter to the orbit ofone assumed territorial centre or another. If styleis purely passive, we should expect a regular fall-off in stylistic similarity with distance from acentre, resulting in roughly circular territoriesaround each centre (subject to travel/transportconstraints), which is approximately what hasbeen proposed, for example in the territory

defined around Knossos (Cadogan 1994).However, since no explicit analysis has beenundertaken with the assemblages, we do notknow if there is actually a decline in similarity;this is simply a presence/absence distribution ofsome material resembling that known from thecentre. Without knowing the quantities, andwhether the dispersed examples are exports fromthe palatial centres, local copies, or somecombination of both, or knowing the differentcontexts or modes of pottery production orexchange represented in different communitiesacross the region, or contexts of consumption, theprocesses involved may be economic, social,ideological or political, but we have no basis fordetermining which, individually or incombination, are responsible for producing thedistribution.

Beginning to address such questions wasthe focus of Knappett's research (1999), thoughthis subtlety is usually ignored, and his study issimply cited as support for the 'Malia-Lasithistate' model. It was initially assumed that thestylistic similarity in pottery between Malia andMyrtos Pyrgos was the result of intensiveinteraction between the communities, withabundant ceramic exchange representing strongeconomic links (Cadogan 1995; Poursat2010:263-64). Knappett's detailed petrographic,technological and stylistic study demonstratedthat very little material was actually movingbetween Malia and Pyrgos, but local fine wareswere extremely similar, and it was assumed thatthose produced at Pyrgos were closely modelledon those of the palatial centre. This wascontextualised by the significant contrast with thelocal and non-standardised styles of the coarseand cooking wares, so the close emulation in finetable wares was identified as a specific elitestrategy. This led to his suggestion that while theelites at Pyrgos were sub-ordinate members ofthe Malia state, the control of that state wasprincipally ideological, rather than economic, andhe proposed that the Malia-Lasithi state beconsidered a de-centralised or segmentary state(Knappett 1999). But with no demonstrablestrong economic links, was Pyrgos actuallysubordinate to the Malia state, or was itindependent, with the local elite simplyemulating elite behaviour at the closest majorcentre, either to facilitate their relations withthose elite, or to enhance their prestige at home,or both? We simply don't know: there is nomaterial evidence which clearly supports theview that Pyrgos was part of the Protopalatialstate centred at Malia.

12

Stepping back from the specific case, is itpossible to recognise archaeologically apolitically-determined material culturedistribution? Two sets of studies are informativeof the problems, but not particularly hopefulabout the prospects. Late Iron Age Celtic goldcoins in Britain are politically-identified,ideologically-charged artefacts, and are expectedto have circulated principally within theboundaries of the political units where they wereminted (Collis 1971; Cunliffe 1981; Sellwood1984). Such politically constrained distributionscould be expected to be defined by sharp edges atpolitical boundaries (Collis 1981; Hodder 1977;Kimes et al. 1982). While there are regionallydefined distributions, there are very considerableoverlaps, so they do not produce clear boundaries(Cunliffe 2005: figs 7.9, 7.13, 7.14, 8.3, 8.10,8.13). This makes the polities difficult to defineon the basis of coinage alone, but given that thereare few other bases for doing so (significantly,ceramic stylistic distributions are usually muchsmaller than the assumed political units: Cunliffe2005: Figs 4.10, 7.15, 17.15, 17.16, 17.17),coinage is used this way (Kimes et al. 1982;Cunliffe 2005: 130-79). Even moreproblematically, similar boundary effects will begenerated by economic competition betweenrival production centres (Hodder and Orton 1976:195-7), so even if there were clear boundaries,these would not, in themselves, indicate that thematerial distributions defined political entities.We might expect the economic distributionpatterns for different types of goods to vary tosome degree (though all may be affected bycommon constraints on transport, the distributionof consumers, etc.), so detecting the same sharpboundary in a range of types of material culturemight be suggestive of a political boundary. Thiswould only apply if that boundary wascompetitive or hostile, strongly policed, andcrucially, stable for a period longer than theresolution of the local archaeologicalperiodisation. But even political boundariessignificant and stable enough to be fortified andpatrolled, such as the Roman Germanic limes,can be remarkably permeable (Hedeager 1979;Wells 1992; 1999).

One of the most relevant studies ofceramic distributions, utilising a substantial,systematically collected sample (6,410 decoratedsherds from the survey of 130 sites, in 12contiguous city-states), considered the effect ofpolitical boundaries on both exchange andceramic styles in the south-east of the Valley ofMexico, in the period immediately before and

after Aztec unification, when political boundariescan be reconstructed through ethnohistoricsources (Hodge and Minc 1990; Hodge et al.1993; Minc et al. 1994; Minc 2006; 2009). Twolevels of political boundaries were considered,those of individual city-states and those definedby alliances among them. While some ceramicdistributions were largely concentrated withinregions formed by political alliances, noneclearly define individual city-states, and allcrossed the boundaries, with gradual fall-off inquantities with distance from source. Withoutpreviously knowing the polity or allianceboundaries, it is impossible to recognise whichdistributions are strongly affected by theboundaries and which are not. Following theAztec conquest of the area, ceramics generallycirculated more widely, so political relations didaffect exchange patterns, but not in the spatiallydefined ways which would enable polities to beidentified from the ceramic style distributions.

Neither cultural and political boundariesnor material culture distributions appear toconform in any straightforward sense to theassumptions necessary to support the directinterpretation of ceramic or other material culturedistributions as political maps (Lightfoot andMartinez 1995; Trinkaus 1984; 1987; De Atleyand Findlow 1984; Green and Perlman 1985;Emberling 1997; Jones 1997), undermining thepredominant approach to the definition of politiesemployed in Cretan prehistory. On the otherhand, spatial distributions can inform us about awealth of processes and behaviours, from theorganisation and control of exchange systems, todifferent types of identity construction. Theseprocesses may be affected by political structureand affiliations, but not necessarily in the director easily identifiable ways assumed.

V. Settlement pattern data and politicalstructure.

For Crete, an increasing number of intensivesurveys are making the island one of the mostthoroughly surveyed regions of theMediterranean, with island-wide coverage andsamples from a variety of topographic contexts.However, many projects are published only inpreliminary form, and the data are difficult toanalyse comparatively, having been collectedover several decades by projects with verydifferent approaches to fieldwork, documentationand publication. In addition, many surveys havebeen very small, often in areas peripheral to thepalace centres, and document only a limitedsegment of a local settlement system.

13

To date, interpretation has been largelydescriptive or focused on the identification oflocal settlement hierarchies. The model for thelatter has been Processual settlement patternarchaeology (Flannery 1998: 16-21; Parsons1972; Wright and Johnson 1975; Cherry 1987;Balkansky 2006; Kowalewski 2008), aimedprimarily at the analysis and interpretation oflarge regional datasets, for example inMesopotamia and Mesoamerica (e.g. Adams1981; Johnson 1972; Wright and Johnson 1975;Hole 1987; Sanders et al. 1979; Blanton et al.1993; Blanton 2004). These drew interpretivelyon the central place models developed formodern industrial societies in Europe and NorthAmerica (Johnson 1972; 1977; Hodges 1987;Butzer 1982: 219-23), though these have beendemonstrated also to have relevance tounderstanding the regional organisation ofdeveloping economies (Smith 1974; 1976; 1980)and early modern Europe (de Vries 1990).

Because archaeological datasets areinvariably only partially preserved, accessible orrecorded, represent low resolution data, and existin a variable landscape, rather than thegeographers' idealised isotropic plain, it isaccepted that the subtle distinctions in spatialconfiguration necessary to distinguish amongdifferent central place models can rarely beconvincingly documented. Instead, the focus ison recognising an hierarchical relationshipamong sites in a region, and analysing their inter-relationships to understand the degree ofintegration within the system. The keycharacteristic for identifying a state level ofpolitical integration is usually argued to be afour-level settlement hierarchy, to differentiatesuch a system from the two to three levelsexpected for a regional chiefdom (Wright andJohnson 1975; Johnson 1977; Wright 1977; Earleand Johnson 2000), though this is only a 'rule ofthumb' (Flannery 1998: 16). In the mostconvincing studies, the settlement data arecorrelated with archaeological evidence foradministrative integration (Wright and Johnson1975; Johnson 1980a; 1987; Wright 1987; 1998;Marcus 1983), or other data supporting thedifferential administrative role of specificcommunities within the postulated settlementhierarchy (e.g. Sanders et al. 1979: 52-60; Smith1979; Blanton et al. 1982). More typically, thesettlement pattern data alone are relied on, withconsequent (usually unacknowledged)uncertainties.

Often ignored, but integrally linked tothe subject of this paper, is the difficulty in

distinguishing territorial states from multi-polityregions, unless the primary urban centre developsexceptionally in response to its regionaladministrative role. This depends on factors suchas how strongly centralised the system is; there isno single or unambiguous signature (Johnson1977; 1980b; 1981; Savage 1997; Drennan andPeterson 2004). In practice, many surveys workwithin naturally defined regions (islands,topographic basins, restricted sections of a rivervalley), and assume that the study region includesmost or all of one past settlement system, but notmultiple systems. Other studies simplyuncritically analyse a study region as if it was acoherent and integrated whole. Unfortunately,any analysis will be systematically distorted ifonly part, or parts of more than one system areincluded in the analysis.

A particular problem in the Aegean andmore widely in the Mediterranean, is thatrequiring a four-level settlement hierarchy toidentify a state, rules out most city-states. EvenAthens, one of the largest and most politicallycomplex Classical city-states, with its demecentres and dispersed hamlets and farms, hadonly two administrative levels, three levelsoverall; community sizes within the largest city-states may fit a rank-size model (Cavanagh2009), but not the political hierarchicalexpectation. Most city-states were much smaller(Hansen 2006b), and surveys such as aroundKoressos on Keos document only twohierarchical levels, the city and rural hamlets orfarms (Whitelaw 1998). Some analysts haveused the absence of a four-level hierarchy tosuggest that the Classical Greek city-state not beconsidered a state (Marcus 1998: 91). While thenature of Classical city-states is receiving criticaland comparative re-assessment (e.g. Berent 2000;Hansen 2006a; Vlassopoulos 2007; Anderson2009; Gehrke 2009), they should not be rejectedas states because they do not fit one very specificspatial model; after all, when it comes to definingthe nature of the state, they literally wrote thebook. The archaeological models of settlementhierarchies, often applied in a mechanisticfashion, appear to have been formulatedprincipally with reference to territorial states orvery large, developed city-states; they are notadequate for recognising or analysing small city-states.

Most diachronic settlement analyses facea similar problem, trying to define the point atwhich a state can be recognised in a scalarcontinuum. Settlement data from surveysthroughout the East Mediterranean and Near East

14

suggest that developing urban centres from 8-20ha were often surrounded by very smallvillages or hamlets, with the development ofsecondary centres as a later phenomenon, aspopulations expanded under relatively stableconditions (see also Falconer and Savage 1995).At what point in that process one defines theemergence of the state should depend principallyon the availability of relevant evidence for theadministrative structure of a state, not on whatcan be a fairly arbitrary exercise in defining sitesize modes.

The Cretan survey record is particularlyproblematic; different surveys engage, at best,variable segments of settlement systems, so it isnot surprising there is no clear aggregatepatterning; this may represent regional diversity(Driessen 2001a), or simply non-comparability ofvery partial datasets. From the surveys conductedto date on Crete, we have information on coreareas of two states, those centred on Phaistos andMalia.

Phaistos and the west Mesara.

Taking the 40km2 of the West Mesara (Watrouset al. 2004) and Kommos (Hope Simpson et al.1995) surveys together, these represent the onlypublished intensive survey of a segment of thecore of the territory of one of the major palatialcentres. The Ayiopharango (Blackman andBranigan 1977; Vasilakis 1990), South Coast(Blackman and Branigan 1975) and Odigitria(Branigan and Vasilakis 2010) surveys areassumed to provide smaller peripheral samples ofthe same polity (Fig. 5). Focusing on the coresample, these surveys provide useful, but farfrom straightforward data. Consistent with otherAegean surveys conducted in the late 1970s-80s,sherd collections were extremely limited andonly aggregate site sizes for all periods ofoccupation are usually documented, maskingperiod-specific changes at sites. Mitigating this,across all prehistoric periods, the only majoroccupations outside the palatial centre of Phaistosare at the previously known and extensivelyexcavated sites of Haghia Triadha and Kommos.Phaistos itself is presently being surveyed(Bredaki et al. 2009), but patchy excavationsbeyond the palace suggest a Neopalatial extenton the order of 55-60ha (Watrous et al. 2004:294) if occupation was continuous between allthe outlying soundings, and a minimum of 32hain the Protopalatial period, making a similarassumption (Whitelaw 2012; Militello 2012).

In analysing their data, the projectdirectors define site hierarchies of three (EMII-III) and four levels (MMIB-LMIB), though noclear modes in site size provide any basis for thisinterpretation. More significantly, despite theover-estimation inevitable with all-periodaggregate site size estimates, for the Prepalatialphases, only two sites outside of Phaistos mightreach or exceed 1ha (most being much smaller),and for all periods, only Phaistos itself and in theNeopalatial period, Haghia Triadha, exceed2.5ha, when the latter took over from Phaistos asthe administrative centre for the region (La Rosa2010b). The survey data provide no clearevidence for a developed regional settlementhierarchy; the divisions seem imposed on anundifferentiated distribution of very small sites,to meet the expectation of four hierarchical levelsfor states. However, the absence of clear second-order centres is not entirely surprising, sincegiven the scale of Phaistos itself from MMIA,these should only develop on the order of 4-8kmfrom the palatial centre, and it is only in thedirection of Kommos that the combined intensivesurvey area extends this far from Phaistos. So ifthere was a developed settlement hierarchy in thewestern Mesara, intensive survey has not yetbeen extensive enough to detect it. On the otherhand, we might anticipate that in the areasurrounding Phaistos and Gortyn, eachinvestigated fairly continuously for over acentury, extensive exploration should havelocated the major sites, as with Kommos andHaghia Triadha.

Taking a different approach to recognisethe development of an integrated settlementsystem, we can use site size, to the degree that itcan be calculated from such low-resolution data,to estimate probable community populations, andtherefore the notional cultivation areas aroundeach site necessary to support its population. Thesites outside Phaistos are so small and well-spaced that there is no overlap of suchcatchments, suggesting no necessary economicinteraction or integration in the Prepalatialperiods. The dramatic expansion of Phaistos inthe Protopalatial period leads to the completeoverlap of the catchment of Phaistos on those ofneighbouring hamlets, and indicates the necessityfor some sort of inter-site dependencyrelationships during the Protopalatial period, butnot earlier (Whitelaw 2012). On presentevidence, Phaistos would have been the centre of

15

Fig. 5. Phaistos and the west Mesara: main sites and Phaistos agricultural catchments by phase.

a very simple, two or just possibly three-tiersettlement system by the end of the Protopalatialperiod. Such simple, highly centralised, under-developed hierarchies are typical of many EastMediterranean and Near Eastern early urbanisingregional systems, and city-states, rather than thewell-developed, four-level hierarchically-structured systems characteristic of largerterritorial states.

Watrous et al. (2004: 286-7, 295),challenging previous assumptions that theMesara forms a 'natural' region (see also Relaki2004), suggest that Phaistos may never havedominated all of the Mesara plain, let alone all ofsouth-central Crete, based on the argument thatthe peak sanctuaries of Kophinas and Demati,overlooking the central and eastern Mesara, willhave served separate polities. There is nonecessary one-to-one relationship betweenpolities and major sanctuaries, and in the absenceof any systematic surveys east of Phaistos, this is

an interesting but completely speculativeproposal. The interpretation of Monastiraki inthe Amari valley as a sub-ordinate centre toPhaistos (Kanta 1999; Kanta and Tzigounaki2000; Watrous et al. 2004: 287) in theProtopalatial period, is based solely on stylisticsimilarities in the ceramics and seals.

The known administrative documentsprovide minimal information relevant to polityscale. No toponyms can be recognisedunambiguously in the Linear A documents fromHaghia Triadha, giving no idea of its dependentterritory (Bennet 1990: 198-9; Schoep 2001: 98-9; 2002: 192). While the quantities ofagricultural products listed in the Linear A tabletsdo not require an extensive dependent territory(Palaima 1994: 318-21; Schoep 2001: 97-9;2002: 176-92), the small number of recoveredtablets is unlikely to document production fromthe entire territory administered by the site (seealso Palaima 1994: 316-7). To date, there are

16

insufficient data with which to define the extentof the territories administered from Phaistos orHaghia Triadha in the Protopalatial orNeopalatial periods.

Malia and the Malia-Lasithi state.

Intensive survey was initiated at the palatialcentre of Malia and expanded to include thecoastal plain and neighbouring inland valleys, theonly Cretan survey with both urban andhinterland data (Fig. 6: Müller 1996; 1997; 1998;2003; Müller-Celka 2007). To date, preliminarysummaries have been published, while analysisof the recovered material continues (Müller-Celka 2007; Puglisi 2007). At the palatial centre,overall sherd density has been recorded, butperiod-specific material is only reported on apresence/absence basis by survey unit (Müller-Celka 2007: fig. 5; pers comm.). The figure of50-60ha for the maximum extent of the site in theProtopalatial period (Müller 1997: 52; Müller-Celka 2007: 856; Driessen 2001a: 61), includesas well as the city, the extensive cemeteries nearthe shore and the outlying port at Haghia Varvara(Müller-Celka 2007: fig. 5). The urban sitesurrounding the palace appears to cover 40ha,with up to another 17ha of low densityoccupation in two locations outside thefortification wall on the east. Evidencesupporting a contraction of the city in theNeopalatial period (Driessen 2001a: 63), has yetto be published; all the excavated residentialareas except Quartier Mu document Neopalatialas well as Protopalatial occupation throughoutthe core of the site. However, since thedistribution of dated Protopalatial ceramicsmatches well the overall extent of dense surfacematerial (Müller-Celka 2007: fig. 5), thecommunity does not appear to have expanded inthe Neopalatial period.

In addition to the coastal plain, surveyextended to the inland basins of Mochos andKrasi, in the foothills of Lasithi, in all coveringsome 40km2. The largest sites outside the cityappear to be a handful of sites of limited extent,distributed across the coastal plain (Müller 1996;1998; Puglisi 2007), with no significantsubsidiary centres except possibly Sissi (ca. 3ha)reported within the surveyed territory. Thevalleys of Mochos and Krasi might be expectedto have encouraged the development of at leastone sub-ordinate centre in each, as perhaps seenin the Sissi valley.

The Malia survey and other studies in theregion provide information not available for thewest Mesara, which may help to define the

boundaries of the Protopalatial polity. On thesouthern fringe of the investigated area, well upin the foothills of Lasithi, several small fortifiedsites originally identified by Evans, have been re-studied and are now dated to the MM period(Müller 2003; Müller-Celka 2007: 859; Nowicki1995; 1996; 2000). These should indicate somesort of boundary, whether to protect outlyingcommunities of the Malia polity, protectcommunities outside it from aggression fromMalia, or simply reflect instability or limitedcentral control on the periphery of the polity.Whatever specific interpretation, they suggest aneffective limit to palatial control on the northernslopes of Lasithi during at least some phases ofthe MM period.

This obviously raises serious doubtsagainst the argument that the upland plain ofLasithi or areas further south or east wereincorporated into a polity centred on Malia.Fortified or defensible sites within and aroundthe Lasithi basin also suggest the area may nothave been integrated into any polity during theProtopalatial period (Nowicki 1996). Potteryimports (Betancourt 2007), as well as ceramicdedications from the Malia lowlands at thePsychro cave (Watrous 2004), could havecrossed the polity boundary, either episodicallyor continuously, and do not require politicalintegration.

On the basis of the preliminary publishedinformation, the Malia survey appears to presenta more comprehensive but comparable picture tothe Phaistos region, of a highly centralisedsettlement system in both the Protopalatial andNeopalatial periods, with a two or just possiblythree-level settlement hierarchy. The evidencesuggesting a southern polity limit, during at leastsome part of the MM period, falls approximatelywhere the boundary of the catchment necessaryto support the estimated population of the centreshould fall (Fig. 6), suggesting that Malia in theProtopalatial period was a small city-state, ratherthan a territorial state.

VI. Knossos: from site to territory.

To date, there has been no intensive survey in thewider region around Knossos. On the other hand,over a century of investigations at the palatialcentre (Hood and Smyth 1981), supplemented bypreliminary observations from an intensivesurvey of the city (Bredaki et al. 2010; Whitelawet al. in press), provide our most detailedunderstanding of the development of a Cretanpalatial centre. The changing occupation area ofthe site through time can be used to estimate the

17

Fig. 6. Malia and east-central Crete: main sites and Malia agricultural catchments by phase.

agricultural catchment necessary to support thecentre’s resident population.

While this exercise does not define theactual extent of the polity centred on Knossos, itdoes indicate the approximate scale of theminimum region which must have beencontrolled by Knossos economically andpolitically, simply to guarantee subsistencesupport for its population in each phase (note 1).The importance of this subsistence-basedperspective is that it addresses one of the majoruncertainties expressed in alternativeexplorations: whether the territories defined (bywhatever means) were actually dominatedeconomically or politically from the centre (e.g.Knappett 1999; 2007; Knappett and Schoep2000; Poursat 2008: 195; Warren 2004; Niemeier2004: 393-4; Müller-Celka 2007).

By the end of the late Prepalatial period(MMIA), occupation at Knossos extended over aminimum of 20 and more likely 40ha (Whitelaw2012). Its support catchment will have extendedsouth nearly to Archanes (Fig. 7), and I havepreviously suggested that this may explain the

cessation of competitive new construction ofburial monuments in the Phourni cemetery afterMMIA, as the developing centre at Archanes wassubsumed under Knossian control (Whitelaw2004: 244-5).

A second test point is provided by thefoundation of the palace (Rethemiotakis 2002),and re-establishment or very significantexpansion of the community (Evely 2008: 104;Whitelaw and Morgan 2009: 94-7; Watrous perscomm.) at Galatas in the Pediada, some 16kmsouth-east of Knossos. The sparse Protopalatialsites in the region are noted as usually fortified orin defensible locations (Panagiotakis 2003; 2004;Whitley et al. 2007: 107; Evely 2008: 105),suggesting that the area was outside anyintegrated palatial territory at that time.Established in MMIIIA, the palace went throughseveral transformations before abandonment inLMIA (Rethemiotakis 2002). Three alternativescenarios may be considered:

1. the foundation of the palace represents theimposition of Knossian political control in theregion, and its abandonment, some change in the

18

Fig. 7. Knossos and central Crete: main sites and Knossos agricultural catchments by phase.

exercise of that control (Rethemiotakis 2002);

2. the foundation of the palace represents theemergence of a local elite, and its abandonmenteither a local collapse, or the suppression of localindependence through Knossian expansion(Watrous et al. 2004: 287);

3. the foundation of the palace represents theestablishment of an independent polity by a cadetor dispossessed line of the Knossian (or Malliote)elite, following the model of Mycenaeanperipheral polity formation suggested by Wright(1984).

Held to support the first interpretation, isa shift from stylistically local ceramics, to styleswhich closely follow Knossian models

(Rethemiotakis 2002; Rethemiotakis andChristakis 2004). The nature andcomprehensiveness of this stylistic shift has yetto be documented in detail (e.g. some pithoicontinue to be produced in the local tradition:Christakis 2006: 125), and the stylistic argumentwill be subject to all the ambiguities outlinedabove. In this case, the argument based onceramics is considered to be strengthened by thecontemporary introduction of the palace layout,ashlar masonry (with masons' marks), andfrescoes. All of these are elements of Minoanpalatial elite culture, so need not document aspecifically Knossian origin, and they need notrepresent an imposition; they could have beenadopted by an emerging local elite, asserting their

19

power and seeking legitimation throughemulating palatial fashions at pre-existingcentres.

An additional perspective is contributedby preliminary reports on the intensive survey ofthe immediate hinterland of Galatas, whichappears to document the intensive colonisation ofa previously under-populated landscape,contemporary with the establishment of thepalace (Whitley et al. 2006: 107; Whitley et al.2007: 107). This need not rule out the second orthird options, but the preliminary evidence seemsconsistent with a rapid and organised intrusion,and the established Knossian power provides thenearest and most likely source.

Implicating Knossian political expansionin the foundation of the palace at Galatas inMMIIIA, provides a second test point for thecity-state model, since the expanding populationof Knossos would push its minimum supportcatchment well into the Galatas region by the endof MMIII, and as far as the site itself in LMI(Fig. 8).

Significant changes in the local powerstructures at Archanes and Galatas correspondbroadly to the periods at which the minimumexpansion of the Knossian city-state would haveimpinged on each community's local settlementsystem. Poros and Amnisos are close enough tohave been absorbed within the expanding orbit ofKnossos before the end of the later Prepalatialperiod; the evidence for off-island connections atPoros makes most sense if it was already

integrally linked with the significant consumingpopulation at Knossos in EMII (Wilson et al.2004; 2008; Dimopoulou-Rethemiotaki et al.2007). Tylissos and Vitsila will have beenintegrated into Knossos' political territory withinthe Neopalatial period, if not earlier. These, withArchanes and Amnisos, are substantialcommunities which can be suggested assubordinate secondary centres within aNeopalatial four-level settlement hierarchy(Whitelaw 2001: 27-9). This would probablyinvolve three levels of administrative hierarchy(e.g. Knossos-Archanes-Vathypetro andKnossos-Tylissos-Sklavokambos), documentedthrough Linear A tablets and sealings. Such anhierarchy cannot, so far, be recognised forProtopalatial city-states, either on the basis of thesettlement or preserved administrative evidence.

For Malia, the modelled catchment forthe Protopalatial city-state (Fig. 6) extends intothe foothills of Lasithi to the general area of thefortified sites, consistent with the interpretationthat these marked the southern boundary of theMalia polity. These provide a test point for thereconstruction of Malia as a small city-state(Poursat 2008; 2010), rather than territorial state.

VII. Knossos: from city-state to territorialstate?

The catchment-based, minimalist definition ofterritories developed here only applies to city-states, though we cannot yet document anynecessary departure from this bottom-up modelfor any central Cretan palace-centred polity, at

Fig. 8. Central Crete and the potential expansion of Knossos: principal sites, Neopalatial catchments andwalking time boundaries.

20

least into the early Neopalatial period. This is asfar as this model can take us, and arguments forthe development of larger territorial states, basedat Knossos or any other palatial centre, will haveto be established on other, far more ambiguousgrounds.

The recent and on-going re-assessmentsof the construction histories at all palatial sites inthe Neopalatial period present a challenginglyunsynchronised picture. Whether all or some ofthese reflect independent and locally contingenthistories, or should be choreographed into adrama of Knossian expansion or conquest (e.g.MMIIIA: Galatas; MMIIIB: Phaistos; LMIA:Malia, and LMIB: Lasithi, Zakros), remainsunclear.

If Knossos expanded into a territorialstate, this will arguably have post-dated itsproposed expansion into the under-populated andunintegrated area around Galatas. Intriguingly,preliminary reports on the Galatas survey suggestthat settlements in the region south of the palaceretained the nucleated defensible charactertypical of the Protopalatial period, through theNeopalatial period (Evely 2008: 105), perhapssuggesting that this region remained outside ornear the periphery of the expanded polity.

If Galatas was a Knossian imposition inMMIIIA, why was it no longer important tomaintain a palatial control centre in the area afterLMIA? Did the community at Galatas and itslocal settlement system also decline, or wasadministration of the region maintained but re-organised? Did Knossos pull out of the region,or did it relocate its administrative sub-centre inthe Pediada elsewhere, perhaps further east toKastelli (Warren 2004: 163; though the knownsubstantial building is said to decline in parallelwith the palace at Galatas: Rethemiotakis 2002:65; Rethemiotakis and Christakis 2011:226),possibly reflecting further eastward extension ofKnossian control.

Looking east, the evidence forNeopalatial contraction at the urban centre atMalia has yet to be presented, but for theexploration here, a static situation is assumed, sothe support catchment estimated for theProtopalatial period is maintained for theNeopalatial period; there is no evidence tosuggest expansion into a territorial state in theNeopalatial period (Fig. 6). With the expansionof Knossos, its hinterland is likely to havebumped up against that of Malia during theNeopalatial period, at least along the northcoastal strip (Fig. 8). This convergence may

have set the scene for a new type of predatoryexpansion, involving not just encroachment on,but incorporation of the entire neighbouring city-state at Malia into an expanding Knossian polity.

Does the early LMIA rebuilding of thepalace at Malia on a layout resembling that atKnossos, represent its reconstruction as a newsecond-order centre under Knossian control(Poursat 2008; 2010), or increasing convergencethrough direct competition with the neighbouringpower, and what does the destruction andabandonment of the palace at the end of LMIA orearly in LMIB represent in terms of regionalpolitical structure (for arguments for a laterLMIB destruction, see particularly van deMoortel 2011:542-45)?

To the south-east, the situation in Lasithiis even less clear. Fortified or defensible sitesseem characteristic of the Prepalatial andProtopalatial periods, with the development ofsites in non-defensible locations around the plainparticularly in the Neopalatial period (Nowicki1996; Watrous 1982). But whether thesecommunities were ever integrated, or linkedpolitically to a lowland palatial centre, remainsunknown. The influx of Knossian ceramicdedications at the Psychro cave in LMIB(Watrous 2004) is intriguing, but need notrepresent Knossian political dominance.

Looking south, does the shift ofadministration from Phaistos to Haghia Triadhaafter MMIIIA represent a re-structuring of localpower, or (along with the non-reconstruction ofthe ceremonial palace) a decisive move to cutlinks with the previous independent elites at theimposition of Knossian dominance (La Rosa2010a: 590; 2010b: 499)? Does the rebuilding ofthe palace at Phaistos in LMIB represent somedegree of local resurgence in the face of post-Theran eruption Knossian weakness (La Rosa2010a: 591), or a confident re-inscription ofKnossian control (Warren 2004: 163)?

The limited expansion suggested here forMalia, and the gradual expansion of ProtopalatialKnossos, raise the prospect that Phaistos, beforeits eclipse in MMIIIA, may never have expandedto dominate the entire Mesara, and other, as yetundocumented polities may have existed in theProtopalatial and Neopalatial periods in thecentral and/or eastern Mesara. There may alsohave been areas, even within central Crete, whichwere never incorporated into state-level politiesbased at one of the three main palatial centres.

The suggestion of a southern boundaryfor Malia in the northern slopes of Lasithi in the

21

MM period, seriously questions whether Maliaever extended its control (as opposed toinfluence: Müller-Celka 2007; Knappett 2007;Poursat 2008; 2010) as far as the Lasithi plain orbeyond to Myrtos Pyrgos. This should also openup a reconsideration of developments around theGulf of Mirabello, for which the unparalleledsettlement data from the combined Vrokastro,Gournia and Kavousi surveys, as well as theextensive recent excavation data from the centresat Gournia, Mochlos, Pseira and PriniatikosPyrgos, provide a unique resource for studyingthe development of a region.

In the far east, Petras provides a strongargument for independent local development(Tsipopoulou 1997; 1999; 2002; Tsipopoulouand Papacostopoulou 1997). The history of thepalace at Zakros, and any substantial but non-canonical antecedent structure, is under revision(Platon 1999; 2002; 2004; 2010). Theinterpretation that it served as an eastern port forKnossos (Bennet 1990: 196, n.20; Warren 2004:164; Platon 2004; Wiener 2007: 234-5) seems torest on the expectation of Knossian dominance,the assumption that it could not have been locallyself-sufficient, and Knossian influence in someceramics, all deserving documentation andcritical appraisal.

To the west, the patterns of recentinteraction (Fig. 3), as well as the evidence forHellenistic conquests and alliances (Fig. 1),question whether central Cretan polities shouldbe expected to have had any significant politicalimpact in west or west-central Crete. Theintegration documented by the Knossian Linear Barchive is clearly an exception, perhapsaccounting for its short duration and instability.Even this control may have been more restrictedand strategic than is usually assumed (Driessen2001b), fitting better a network (Smith 2005)rather than territorial state model.

VIII. Recognising polities: problems andprospects.

The questions outlined above all deserveexploration, but none of the speculativesuggestions warrant the certainty with whichparticular political interpretations are regularlyespoused in the literature. The latter have almostall been framed with an expectation of Knossiandominance. It is refreshing when the politicalstatus of a site or local region is considered in itsown terms (e.g. Andreadaki-Vlasaki 2002; 2010;Tsipopoulou 1997; 2002; Shaw 2006;Cunningham and Driessen 2004), and then in thecontext of a range of wider regional possibilities.

Many of the assumptions which structurepresent approaches to interpretation go back tothe origins of Minoan archaeology, and are sofundamentally ingrained that they are difficult torecognise, let alone unpack and examinecritically. Underlying nearly all conceptualis-ations of Minoan political landscapes has beenthe naturalisation of a pattern of palatial centresin central Crete, originally constructed onextremely limited information. This hasdiscouraged investigations which might test theassumed understanding, and tended to focus ourattention on spatial scales and landscape unitswhich are not necessarily appropriate for theentities we are trying to recognise. A longer-term historical perspective makes it clear thatCretan geography is far less deterministic ofpolitical structures than has been assumed byprehistorians. Looking at the relatively smallterritory required to support a major palatialcentre like Protopalatial Phaistos, we canrealistically expect there to have been more,whether like Petras, developing locally, orGalatas, perhaps representing an intrusivecolonisation of relatively unpopulated territory.There are also numerous agriculturally suitableareas, not hitherto considered, which could havesupported independent polities (not necessarilyall states) of various scales in the Protopalatial orNeopalatial periods. While the island has beenrelatively well investigated when compared withmost Mediterranean landscapes, it is easy toforget how little of Crete has been systematicallyor intensively investigated archaeologically.

If we cannot rely on geographical'givens', we need to use period-specificarchaeological evidence to define and trackchanges in contingent and dynamic politicalformations. A first step has to be to accept howdifficult it is to recognise the territories ofindividual states archaeologically. Afundamental problem arises because states aredynamic: they expand and contract throughconquest and alliance, often combiningindividual small-scale city-states or incorporatingnon-state territories beyond their borders, intolarger, less stable, regional territorial states.These processes will often take place on time-scales shorter than can be monitored using ourmaterial culture chronologies, and constantchanges will blur any boundaries we might hopeto detect in material distributions.

Material culture analyses havetremendous potential for informing us aboutinteractions between the residents of differentcommunities, and their motivations for doing so,

22

but interpreting these directly in terms of politicsis rarely attempted elsewhere in worldarchaeology, with good reason. We areconsiderably more experienced at interpretingindividual processes, such as exchange;comparisons of multiple patterns may then besuggestive of the political contexts within whichdifferent types of exchanges took place, pointingtowards the actors, their potential motivations,and constraints. If we wish to use materialculture distributions for such analyses, we needto consider a wide range of material culture, anddocument large and reliable samplessystematically using standardised criteria, toallow detailed comparisons and robust spatialanalyses.

In a similar way, settlement data need tobe approached more subtly, as there are nodiagnostic patterns that will allow us to define thelimits of states on the ground with confidence.Rather, we can compare site types and theirdistributions (e.g. site sizes, spacing, clustering,relative population nucleation), as these patternagainst landscape characteristics (e.g. resources,agricultural and other productive potential,defensible locations, routes), and track changes inrelationships and configurations through time.These patterns can then be exploredinterpretively, in terms of what they may indicateabout the nature and organisation of relationsbetween communities, and the locations andconfigurations of sites may suggest the limits ofintegrated settlement systems, or the re-structuring of local systems when communitiesbecome incorporated into (or drop out of) largerregional systems. Such studies have begun (e.g.Haggis 2005: 59-81; Hayden 2004: 35-137;Nowicki 2000; Cunningham and Driessen 2004),though interpretations and larger-scale patternrecognition are inhibited by the small scale ofindividual surveys and the limited and non-comparable recording and presentation of data bydifferent projects.

The catchment-based approach to thedefinition of polities explored here attemptsseveral things. It tries to define what sort of datacan be used most effectively to address veryspecific questions, in this case, to define theagricultural hinterlands of city-state polities. Italso recognises that, logical as the approach mayseem, it cannot simply be asserted, but needs tobe tested against archaeologically recoverabledata. The patterns generated can also helpsupport specific interpretations of other evidence,so that while the material culture arguments forthe palace at Galatas as a Knossian foundation

are not conclusive, that the hinterland necessaryto support the expanding population of Knossosis likely to have imposed on the western Pediadaat about the same time as these characteristicsappear in the area, provides a different line ofsupport for that political interpretation of thematerial culture changes.

Comparatively, the catchment approachalso raises interesting questions about thedifferential constraints on expansion faced byeach of the central Cretan polities. In Figure 9,settlement size and catchment estimates aremapped, by phase, for the three palatial centres atthe same scale, facilitating comparisons.Phaistos, at the western end of the Mesara plain,will have had direct access to abundant primeagricultural land, and its necessary catchmentwill only have reached the foothills north andsouth by the end of the Protopalatial period.Further expansion will have been easiest andmost productive east, down the plain, but only ifthere were no competing polities in the central oreastern Mesara. Any attenuated expansion of thepolity will have required significant second-ordercentres, perhaps incorporating previouslyindependent local centres.

In contrast, the hinterland ofProtopalatial Malia would have requiredexpansion into the small inland valleys, and bythe time of the maximum documented extent ofthe city, in MMII, would have been stronglycircumscribed by relatively unproductiveuplands. Intriguingly, its estimated MMIIcatchment borders hit against the hills separatingit from the Pediada plain, the Lasithi basin, andthe Mirabello coastal lowlands. To incorporatethese more distant areas would probably haverequired the development of a more decentralisedform of administration, probably also resulting inweaker control.

Between these two extremes, expansionfrom Knossos was through fairly dissected butlargely productive terrain, all fairly comparableboth in terms of agricultural productivity andtransport constraints. A pattern of controldeveloped for the immediate territory might beextended, with no natural impediments orboundaries. Despite no intensive regional surveywithin its territory, Knossos is the only region ofCrete for which we know of significant likelysecond-order centres, at Archanes, Tylissos,Poros, Amnisos and potentially Vitsila. Theexpansion of Knossos to approximately doublethe size of the other central Cretan palatialcentres, suggests that it did indeed develop a

23

Fig. 9. The principal palatial sites and their development through time: Knossos: A. EMII; B. EMIII-MMIA; C.MMIB-III; D. LMI; E. catchments; Phaistos: F. EMII; G. EMIII-MMIA; H. MMIB-II; I. MMIII-LMI; J.catchments; Malia: K. EMII; L. EMIII-MMIA; M. MMIB-II; N. MMIII-LMI; O. catchments.

significant regional administrative role, to adegree that the other central Cretan palatialcentres never did. While I have arguedelsewhere for multiple pathways to complexityamong Cretan Prepalatial communities(Whitelaw 2004), these need to be pursuedbeyond the common elements of small-scale stateorigins (Whitelaw 2012), to document thedivergent individual histories of each politywhich we are increasingly confronted by, throughthe on-going re-assessments at each palatialcentre.

The approach developed here worksfrom the known to the unknown, exploringregional structures from the bottom up. But thisis most relevant to the scale of city-state polities.To move beyond this and confirm whether therewere any territorial states in Crete before thatdocumented by the Knossos Linear B tablets, anddevelop an understanding of their nature, requiresthat we figure out how to work more effectivelywith a wider range of material data, with moreexplicit recognition and critical assessment of theassumptions we are making, and engagementwith the full range of interpretive possibilities.

We will also need to work out how to usematerial distributions to monitor different typesof inter-community and inter-individualinteractions, which themselves may be subject topolitical influence, organisation or control. Todo this, we need to identify appropriate,theoretically-justified approaches to interpreting

the material record, and to developmethodologically-sound, empirically-supportedinterpretations. What is clear from the evidenceof new centres and re-assessments at those longknown, is that political developments on theisland are far more locally varied, fluid anddynamic than our traditional approaches allowedus to recognise or interpret.

IX. Notes.

1. For calculating catchments in Figures 5-9, onlyland of 10 slope or less is included, and distanceaway from the site is assessed in terms ofwalking time, so the catchments include all lowslope land within an equivalent walking timefrom the site, sufficient to support the estimatedpopulation in each phase, allowing 0.5ha tosupport each individual. This allowance is fairlyconservative, thereby defining minimumcatchments. The mapped catchments alsoincorporate the areas necessary to support thepopulations estimated for subsidiary sites withinthe catchment, calculated from the regionalpopulation density estimated from the MM andLM rural sites in the West Mesara and Kommossurveys (100 persons/km2). For NeopalatialKnossos, the estimated catchments for the majorsites at Tylissos, Archanes, Poros and Amnisosare incorporated, as well as a purely notionalfigure for the known but unstudied centre atVitsila. For calculations, Prepalatial andProtopalatial site areas and reconstructedpopulations follow Whitelaw 2012, with the area

24

for Protopalatial Phaistos estimated at 40ha,given the patchy distribution of excavationsacross the site and the suggestions of wideroccupation from the on-going survey, and forMalia 50ha, incorporating also the evidence fromthe survey. For the Neopalatial period, Phaistosis estimated at 55ha (Watrous et al. 2004: 294),and Malia kept stable at 50ha.

X. Acknowledgements.

I am very happy to offer this paper to Keith, inrecognition of 30 years of encouragement,support and friendship. For decades he wasalmost a lone voice asking interpretive questionsabout Prepalatial and particularly Protopalatialsocieties. I am grateful to the editors for theinvitation to participate in the Round Table. I amalso grateful to the other participants for theircomments and their own stimulatingcontributions, to Glynis Jones for her hospitalityin Sheffield, and to Andrew Bevan for commentson a draft. I owe a significant debt to AndrewBevan for running a range of GIS analyses forme; they were more time consuming then eitherof us anticipated when I asked and he agreed. Iwould also like to thank Andonis Vasilakis andMaria Bredaki, co-directors of the KnossosUrban Landscape Project, for allowing me todraw on some preliminary data from that project.The project is a collaboration between the BritishSchool at Athens and the 23rd Ephorate ofPrehistoric and Classical Antiquities of theHellenic Ministry of Culture. Principal fundinghas been provided by the Institute for AegeanPrehistory and the British Academy, withadditional funds and support from the BSA, 23rdEphorate and the Institute of Archaeology, UCL.This paper was written while on sabbatical, and Iwould also like to thank my colleagues at UCLwho covered my teaching and administrativeduties during that time.

XI. References.Adams, R.E. 1991 Prehistoric Mesoamerica (2nd ed.).

Norman: University of Oklahoma Press.Adams, R.E. and R. Jones 1981 Spatial patterns and

regional growth among Classic Maya cities.American Antiquity 46: 301-22.

Adams, R.M. 1981 Heartland of Cities: Surveys ofAncient Settlement and Land Use on the CentralFloodplain of the Euphrates. London: Universityof Chicago Press.

Adams, W. and E. Adams 1991 ArchaeologicalTypology and Practical Reality: a DialecticalApproach to Artifact Classification and Sorting.Cambridge: Cambridge University Press.

Anderson, G. 2009 The personality of the Greek state.Journal of Hellenic Studies 129:1-22.

Andreadaki-Vlasaki, M. 2002 Are we approaching theMinoan palace of Khania? In J. Driessen, I.Schoep and R. La