WEATHERING SEQUENCE AND ALTERATION PRODUCTS IN …clays.org/journal/archive/volume...

7
Clays and Clay Minerals, Vol. 36, No. 5,448-454, 1988. WEATHERING SEQUENCE AND ALTERATION PRODUCTS IN THE GENESIS OF THE GRASKOP MANGANESE RESIDUA, REPUBLIC OF SOUTH AFRICA LORAINE C. HAWKERAND JAMES G. THOMPSON Soil and Irrigation Research Institute Private Bag X79, Pretoria, 0001, Republic of South Africa Abstract--Although numerous, small, manganese oxide deposits associated with dolomite in the Eastern Transvaal escarpment, Republic of South Africa, have been known for many years, their mineralogical make-up is somewhat controversial. Chemical, mineralogical, and morphological properties of the weath- ering products of dolomite and the coexisting manganese oxide material in the Graskop area were therefore determined. Mn and Fe occur only in minor accessory minerals in the original rock; however, in the weathering residue, these elements are concentrated and occur as separate mineral phases, chiefly bir- nessite, nsutite, and goethite. Thin veins of pure muscovite and quartz traverse the residua. Rare, pure calcite and maghemite nodules were noted throughout the residual manganese material. The properties of this weathering sequence suggest that the manganese wad deposits were formed in situ as a result of the congruent dissolution of dolomite, leaving a porous, sponge-like structure, highly enriched in Mn and Fe oxides. Key Words--Birnessite, Dolomite, Infrared spectroscopy, Manganese, Nsutite, Petrography, Wad, Weath- ering, X-ray powder diffraction. INTRODUCTION The existence of numerous, mainly small, man- ganese oxide deposits associated with dolomite (so- called "manganese wads") in the Eastern Transvaal escarpment has been known for many years. The larg- est and probably best-known of these occurs in the Graskop area Oat. 24~ long. 30~ which has a mean annual rainfall of 1800 ram. On the basis of X-ray powder diffraction (XRD) analysis alone, De Waal and Hiemstra (1966) stated that the main con- stituent of this deposit is cryptomelane. In the present work, preliminary selective dissolution studies showed that the potassium was linked to aluminum silicate minerals and not to the manganese and that the man- ganese was present almost entirely in a dioxide form (evolution of C12 with warm HC1; a near violent re- action with 30% H202). A small particle size and struc- tural disorder are characteristic of many manganese oxides and oxyhydroxides, making them difficult to identify by XRD alone. In the present paper, the mor- phology, chemistry, and mineralogy of the weathering and alteration products of the parent dolomite rock are described to explain more fully their very high Mn content in relation to the small amount of Mn in the original rock. An attempt is also made to clarify the Mn mineral identification of the Graskop manganese wads. The mineralogy of transecting veins and nodule accumulations is also reported. SITE AND SAMPLES The Graskop manganese deposit is located in a steep- sided valley; about 2 m of weathered material has been exposed in a quarry (Figure I). The heterogeneity of gravels and stones in the solum deafly reflects the col- Copyright 1988, The Clay Minerals Society luvial nature of the overlying soils, as is expected in a steep-sided valley. These colluvially transported soils do not contribute directly to the genesis of the under- lying deposits and were not examined in any detail. The manganese residua are black (N 1,5/0), light- weight, and friable, even in the moist state. Parts of the deposits are dense due to compression by over- burden, which is shown in some samples by the pres- ence of slickensides. Three samples of the manganese wads from positions progressively farther away from the weathering dolomite rock were analyzed. These positions, 1, 2, and 3, are shown diagrammatically in Figure 1. White veins, differing in hardness and mor- phology, transect the deposits. Sparse nodules of soft, white material are scattered throughout the deposit. Black, magnetic nodules were also found in the deposit. Samples of the gray, unweathered dolomite rock, the brown, weathering-front material (distinct transition between the rock and the manganese wad, ~ 1 cm thick), and the black, residual manganese wad were analyzed. Minerals in the veins and nodules, the residue of HC1- treated samples, and the magnetic products of an ig- nited wad sample were also identified. EXPERIMENTAL The samples were examined petrographically. Three samples of the residual manganese material were pre- pared in water for transmission electron microscopy (TEM). The materials were homogenized in a Sieb mill for 30 s. Carbonates were removed by treating the samples with 1.0 M acetic acid prior to the extractions. Samples were extracted with sodium citrate-bicarbon- ate-dithionite (CBD) (Mehra and Jackson, 1960) in preparation for infrared analysis. Organically bound 448

Transcript of WEATHERING SEQUENCE AND ALTERATION PRODUCTS IN …clays.org/journal/archive/volume...

Page 1: WEATHERING SEQUENCE AND ALTERATION PRODUCTS IN …clays.org/journal/archive/volume 36/36-5-448.pdfMicromorphology Figure 2b is a photomicrograph of a thin section of the dolomite rock

Clays and Clay Minerals, Vol. 36, No. 5,448-454, 1988.

WEATHERING SEQUENCE A N D ALTERATION PRODUCTS IN THE GENESIS OF THE GRASKOP MANGANESE RESIDUA,

REPUBLIC OF SOUTH AFRICA

LORAINE C. HAWKER AND JAMES G. THOMPSON

Soil and Irrigation Research Institute Private Bag X79, Pretoria, 0001, Republic of South Africa

Abstract--Although numerous, small, manganese oxide deposits associated with dolomite in the Eastern Transvaal escarpment, Republic of South Africa, have been known for many years, their mineralogical make-up is somewhat controversial. Chemical, mineralogical, and morphological properties of the weath- ering products of dolomite and the coexisting manganese oxide material in the Graskop area were therefore determined. Mn and Fe occur only in minor accessory minerals in the original rock; however, in the weathering residue, these elements are concentrated and occur as separate mineral phases, chiefly bir- nessite, nsutite, and goethite. Thin veins of pure muscovite and quartz traverse the residua. Rare, pure calcite and maghemite nodules were noted throughout the residual manganese material. The properties of this weathering sequence suggest that the manganese wad deposits were formed in situ as a result of the congruent dissolution of dolomite, leaving a porous, sponge-like structure, highly enriched in Mn and Fe oxides. Key Words--Birnessite, Dolomite, Infrared spectroscopy, Manganese, Nsutite, Petrography, Wad, Weath- ering, X-ray powder diffraction.

I N T R O D U C T I O N

The existence of numerous, mainly small, man- ganese oxide deposits associated with dolomite (so- called "manganese wads") in the Eastern Transvaal escarpment has been known for many years. The larg- est and probably best-known of these occurs in the Graskop area Oat. 24~ long. 30~ which has a mean annual rainfall of 1800 ram. On the basis of X-ray powder diffraction (XRD) analysis alone, De Waal and Hiemstra (1966) stated that the main con- stituent of this deposit is cryptomelane. In the present work, preliminary selective dissolution studies showed that the potassium was linked to a luminum silicate minerals and not to the manganese and that the man- ganese was present almost entirely in a dioxide form (evolution of C12 with warm HC1; a near violent re- action with 30% H202). A small particle size and struc- tural disorder are characteristic of many manganese oxides and oxyhydroxides, making them difficult to identify by XRD alone. In the present paper, the mor- phology, chemistry, and mineralogy of the weathering and alteration products of the parent dolomite rock are described to explain more fully their very high Mn content in relation to the small amount of Mn in the original rock. An attempt is also made to clarify the Mn mineral identification of the Graskop manganese wads. The mineralogy of transecting veins and nodule accumulations is also reported.

SITE AND SAMPLES

The Graskop manganese deposit is located in a steep- sided valley; about 2 m of weathered material has been exposed in a quarry (Figure I). The heterogeneity of gravels and stones in the solum deafly reflects the col-

Copyright �9 1988, The Clay Minerals Society

luvial nature of the overlying soils, as is expected in a steep-sided valley. These colluvially transported soils do not contribute directly to the genesis of the under- lying deposits and were not examined in any detail. The manganese residua are black (N 1,5/0), light- weight, and friable, even in the moist state. Parts of the deposits are dense due to compression by over- burden, which is shown in some samples by the pres- ence o f slickensides. Three samples of the manganese wads from positions progressively farther away from the weathering dolomite rock were analyzed. These positions, 1, 2, and 3, are shown diagrammatically in Figure 1. White veins, differing in hardness and mor- phology, transect the deposits. Sparse nodules of soft, white material are scattered throughout the deposit. Black, magnetic nodules were also found in the deposit.

Samples of the gray, unweathered dolomite rock, the brown, weathering-front material (distinct transition between the rock and the manganese wad, ~ 1 cm thick), and the black, residual manganese wad were analyzed. Minerals in the veins and nodules, the residue of HC1- treated samples, and the magnetic products of an ig- nited wad sample were also identified.

EXPERIMENTAL

The samples were examined petrographically. Three samples of the residual manganese material were pre- pared in water for transmission electron microscopy (TEM). The materials were homogenized in a Sieb mill for 30 s. Carbonates were removed by treating the samples with 1.0 M acetic acid prior to the extractions. Samples were extracted with sodium citrate-bicarbon- ate-dithionite (CBD) (Mehra and Jackson, 1960) in preparation for infrared analysis. Organically bound

448

Page 2: WEATHERING SEQUENCE AND ALTERATION PRODUCTS IN …clays.org/journal/archive/volume 36/36-5-448.pdfMicromorphology Figure 2b is a photomicrograph of a thin section of the dolomite rock

Vol. 36, No. 5, 1 9 8 8 Weathering sequence and alteration products of manganese residua 449

~ overlying soils

hard white vein (Quartz)

t - m a n g a n e s e residual material

. (~.~ " nodules

- brittle white vein (muscovite) =, Z-- �9 O ) �9

-- ~ distinct weathering frora of dolomite rock

-- I~ I ~ , ~ R dolomite rock

Figure 1. Schematic profile of manganese residua overlying dolomite (R) in quarry in Graskop area, Republic of South Africa. Sample locations are indicated by numbers in pa- retheses.

Mn was assumed to be negligible, inasmuch as the % C in the samples was minimal . X-ray fluorescence (XRF) techniques were used for the analysis of major elements in untreated samples (Norrish and Hutton, 1969).

X-ray powder diffraction (XRD) techniques were employed to identify crystalline materials, including, as far as possible, the poorly ordered manganese min- erals. XRD traces of randomly orientated (rear-loaded) powder samples were obtained on a Philips PW 1140/ PW1050 diffractometer fitted with a graphite mono- chromator using CoKa radiat ion generated at 40 kV and 30 mA (divergence slit = 1 ~ receiving slit = 0.1 mm, scatter slit = 1~ The finely particulate and dis- ordered nature of manganese oxides in these weathered samples made X R D identification difficult and some- times impossible. Orientation of the samples did not improve the clarity of the diffractogram, and X R D patterns generally showed only a few, broad lines, which commonly varied in position. Infrared (IR) spectros- copy was used to supplement X R D analysis because of its sensitivity to short-range order in manganese oxides (Potter and Rossman, 1979a). The IR spectra were obtained with a Bruker IFS45 IR spectrometer on 0.5 mg of powdered sample in 13-mm diameter, pressed, KBr (300 mg) pellets. IR spectra of the man- ganese oxides in samples were obtained by computer generation of difference spectra. Absorpt ion due to sil- icates was removed from a spectrum by subtracting from it that fraction of the extracted residue spectrum necessary to null the absorption near I000 cm- l, which is due to silicates.

RESULTS A N D DISCUSSION

Micromorphology

Figure 2b is a photomicrograph of a thin section of the dolomite rock and the morphologically and tex- turally different material above it; the transit ion is in-

dicated by the arrows. The residual manganese mate- rial is shown in Figure 2a. Figures 2c, 2d, and 2e are higher magnification micrographs of the residual man- ganese material, the transit ional material (weathering front), and the dolomite, respectively. The weathering front (Figures 2b and 2d), which constitute a distinct transition between the dolomite and residual man- ganese material, has a skeletal framework of dolomite which contains dark material inclusions. The high bi- refringence of the dolomite can be seen under cross- polars, particularly in weathered zones adjacent to areas of dark Fe-Mn oxides. In Figures 2a and 2c, the do- lomite rock appears to be completely altered, and only a porous, spongy microstructure of fine-grained, op- tically isotropic, manganiferous material is left. The original dolomite has apparently been "replaced" by a skeletal framework with a large amount of pore space. No downward migration or reprecipitation of migrat- ing manganese was observed. A few quartz grains are scattered throughout the fabric.

Some of the most important factors contributing to the weathering patterns of minerals are their crystal- lographic properties (e.g., cleavage) and chemical com- position. The dolomite rock (Figures 2b and 2e) con- tains fine, intergranular, contact microcracks, fractures, and cleavage planes, which probably were the first con- duits for the weathering solution into the rock. During weathering, water presumably flowed into these mi- crocracks and reacted with the dolomite. Weathering proceeded along these planes of weakness by dissolu- t ion and removal of dolomite, creating new porosity. The end result seems to be a porous mass o f Fe and Mn oxides (Figures 2a and 2c), hence, the very low bulk density of the residual material. No clay minerals were apparently formed from the dolomite.

Chemical composition and mineralogical identification

Table 1 gives the bulk oxide composi t ion of the weathering sequence samples. Mn and Fe occur only in minor amounts in the original rock. The relative decrease in bulk density, established in prel iminary analyses, from unweathered rock (2.97 g/cm 3) to the manganese oxide residua (average 0.13 g/cm 3) is 22.8: 1 and is similar to the relative increase in total Mn content of 1:21.5. The removal by leaching of the Ca and Mg carbonates (totaling 80% in the dolomite rock) would give a relative Mn increase of 1:21.5, which is also in close agreement. Figure 3 shows the bulk com- posit ion of this sequence plotted on the composi t ional diagrams CaO-MgO-MnO and CaO-MnO-Fe203. The solution of dolomite during weathering leads to the almost complete removal of Mg and Ca and the relative accumulation of Mn and Fe (Table 1, Figure 3). The manganese residua do not appear to be chemically ho- mogeneous, and the Mn (and Fe) concentration varies from one sample to another. The trends, however, are

Page 3: WEATHERING SEQUENCE AND ALTERATION PRODUCTS IN …clays.org/journal/archive/volume 36/36-5-448.pdfMicromorphology Figure 2b is a photomicrograph of a thin section of the dolomite rock

450 Hawker and Thompson Clays and Clay Minerals

Figure 2. Photomicrographs of(a) residual manganese material above (b) the weathering front and parent dolomite (boundary shown by arrows). Under higher magnification the (c) manganese sample clearly exhibits a "spongy" microstructure of fine, optically isotropic material above the (d) weathering front which illustrates the transition from the parent dolomite (e) to the residua material.

very much the same. Interestingly, apart from residue 3, the MnO/F%O3 ratio is of similar order (~2.6) (Ta- ble 1; Figure 3, straight line), which suggests that the original Mn and Fe in dolomite rock were responsible for the skeletal framework of the manganese wads.

The A1 and K contents of the manganese wads sug- gest the presence of variable amounts of clay minerals. The XRD data of the residue after HC1 treatment in- dicate that the dominant clay mineral is muscovite.

The brittle, white veins consist of almost pure mus- covite (10% K20 ~ 100% muscovite) (Table 1, XRD). The hard, white veins consist of almost pure quartz (99.9% SiO2 ~ 100% quartz) (Table 1, XRD). The white nodules consist of calcite (46.1% Ca ~ 100% calcite) (Table 1, XRD), and the magnetic nodules con- sist of maghemite (80-90%), together with some he- matite (XRD). The maghemite and hematite may have formed by slow oxidation and simultaneous dehydra-

Page 4: WEATHERING SEQUENCE AND ALTERATION PRODUCTS IN …clays.org/journal/archive/volume 36/36-5-448.pdfMicromorphology Figure 2b is a photomicrograph of a thin section of the dolomite rock

Vol. 36, No. 5, 1988 451 Weathering sequence and alteration products of manganese residua

Table 1. Total chemical analysis (%) expressed on an air-dried basis.

Sample Fe203 MnO CaO K20 SiO2 A1203 MgO LOI TotaP

Dolomite rock 0.65 1.72 29.2 <0.01 0.90 0.14 20.9 46.1 99.6 Weathering front 1.14 2.75 28.8 <0.01 0.83 0.34 20.5 45.2 99.7 Manganese residue (1) 10.2 26.5 10.94 1.80 10.1 5.62 7.1 26.3 98.9 Manganese residue (2) 10.96 31.58 0.64 1.66 15.6 5.13 0.59 31.4 97.7 Manganese residue (3) 19.36 37.01 0.06 3.40 15.3 8.43 0.53 12.2 97.0 Muscovite vein 1.71 0.19 0.04 9.89 48.8 30.0 2.2 5.0 99.0 Quartz vein 0.09 0.06 0.03 0.02 99.9 0.12 <0.01 0.1 100.3 Calcite nodules 0.53 0.35 46.1 0.01 4.0 0.45 0.54 43.0 95.3

J Including minor amounts of PzOs, TiO2, Na20, SO3.

tion of goethite (Taylor and Schwertmann, 1974). On ignition of the manganese wad material at 1000~ over- night, some separate, Fe- and Mn-rich, magnetic min- erals were formed, namely magnetite and partridgeite (Figure 4d). Goethite and maghemite normally convert to hematite on heating > 600"--800~ the conversion

%Ca

50 50

/ ,

%Mg %Mn

%Ca

%Mn %Fe Figure 3. Composition diagrams of (a) parent dolomite, (b) weathering front, and (c) manganese material immediately above the weathering front, and (d-h) various residual man- ganese samples, illustrating trend of element composition with dolomite weathering.

temperature being related to the Fe 2+ content, i.e., higher for more ferrous compositions (Brown, 1980). This may explain the formation of magnetite on ig- nition instead of hematite; however, the Fe z+ content of these samples was not determined. Moreover, the influence of possible Mn substitution in Fe minerals or vice versa can also not be excluded.

XRD lines of the residual manganese minerals in samples 1-3 are broad and relatively indistinct (Figure 4). Samples 1 (Figure 4a) and 2 (Figure 4b) contain different amounts of birnessite (~7.2, 2.46, 2.33 A) and todorokite (9.69, 4.8 A), whereas sample 3 (Figure 4c) is essentially nsutite (2.44, 2.15, 1.66 A). Some poorly crystalline goethite is also present (4.18 ~).

In Figure 5, numbers A, B, and C represent IR data for samples 1, 2, and 3, respectively, whereas spectrum a in each case is the difference spectrum of the unex- tracted sample (spectrum b) and that of the CBD ex- tracted sample (spectrum c, corrected intensity). Potter and Rossman (1979a, 1979b) were the chief source references for the IR band allocations. [The IR spectra were compared with calibrated IR spectra of well-crys- tallized materials, published by Potter and Rossman (1979a).] The IR bands of the wads are broad and not clearly resolved, indicating disorder. The two intense, poorly resolved bands near 500 cm- 1 are characteristic ofbirnessite and are said to distinguish it from all other manganese oxides, although other bands of birnessite may easily be confused with those of highly disordered todorokite (Potter and Rossman, 1979a). A broad band near 580 cm-~ and weak bands at wavenumbers slight- ly less than 400 cm-1 indicate nsutite (Figure 5C). The spectrum of sample 1 is similar to that of birnessite, but the greater intensity of the major absorption bands, together with XRD evidence, suggests that sample 1 probably consists of a highly disordered todorokite (Figures 4a and 5A). The positions of the weak shoul- ders at higher wavenumbers (673 cm -~, Figure 5A; 700 cm -1, Figure 5B) also suggest the presence of some birnessite in samples 1 and 2. The distinction between the two minerals by IR analysis is more difficult if the bands are broad and not clearly resolved. The present observations could support Potter and Rossman's (1979a) suggestion that a cont inuum exists between the

Page 5: WEATHERING SEQUENCE AND ALTERATION PRODUCTS IN …clays.org/journal/archive/volume 36/36-5-448.pdfMicromorphology Figure 2b is a photomicrograph of a thin section of the dolomite rock

452 Hawker and Thompson

80 70 60 50 40 30

Clays and Clay Minerals

20 10

Range = 400cps Dt 298;

; 8,87

f.~ 1 Td 2.45

~ ~ ~ 1 6 6 ' !S 2"15d NSn G : R n T

iu

Ns 3"1~1 5.0 Mu Mu 9.97 I / , ~ Mu N8 244 IJ

NS .oo 2.15 Mu Bn ~ ~ ~ , 1 , 4 1 Mu G 7.o C

Pdg 2.71 Pdg 1.66

1:' IJ 'r/ Pdg 3.83

80 70 60 50 40 30 20 10

Degrees 29 (Co K~ radiation)

Figure 4. X-ray powder diffraction patterns: (a) manganese wad sample 1, (b) manganese wad sample 2, (c) manganese wad sample 3, (d) magnetic products from ignited manganese wad sample 3. Td = todorokite; Bn = birnessite; Ns = nsutite; Go = goethite; Pdg = partridgeite; Ma = magnetite.

two minerals. Recent studies of todorokite samples using high-resolution transmission electron micros- copy (HRTEM) by Turner and Buseck (1981) and Turner et aL (1982) suggest families o f manganese ox- ide minerals with different end-members. The IR data also preclude cryptomelane identification, for which absorption bands occur at 300 cm -] . In all samples, Fe was found to be a major constituent; thus, Fe could have prevented the formation of pure, crystalline Mn structures. Fe-substi tution in the structures could have caused the shifted and broad bands o f the IR spectra.

Sample 1 (collected immedia te ly above the weath- ering front) also contains a large amount of dolomite which dilutes the manganese minerals and thus hinders

their posit ive identification. Sample 2 (taken higher in the profile) contains birnessite and a lesser amount o f todorokite, whereas sample 3 (taken still further from the weathering front) is essentially nsutite.

Crystal morphology using transmission electron microscopy

Manganese sample 2 consists of chains of small (670 ~), rounded structures which with the help o f a hand lens, seems to be made up of concentric tings of par- ticles (arrowed) (Figures 6a and 6b). TEMs of sample 3 (Figures 6c and 6d) show an abundance of muscovite among groups of irregularly shaped particles (nsutite),

Page 6: WEATHERING SEQUENCE AND ALTERATION PRODUCTS IN …clays.org/journal/archive/volume 36/36-5-448.pdfMicromorphology Figure 2b is a photomicrograph of a thin section of the dolomite rock

. ~

- - mica ~ j / /

I I t I

1200 1000 800 600 400 waven~nber (crn l)

1200 1000 800 600 400 wavenu~mber (cm-i) wavenumber (cm -1)

Figure 5. Infrared spectra of: (A) manganese oxide sample 1, (B) manganese oxide sample 2, (C) manganese oxide sample 3. a = difference spectrum ofb and c (corrected intensity); b = unextracted sample; c = citrate-bicarbonate-dithionite-extracted sample.

Figure 6. Transmission electron micrographs of (a, b) sample 2 (birnessite and todorokite), showing the rounded structures exhibiting a concentric arrangement (arrowed) of minute particles; (c, d) sample 3 (nsutite). Muscovite platelets are indicated by arrows.

453

Page 7: WEATHERING SEQUENCE AND ALTERATION PRODUCTS IN …clays.org/journal/archive/volume 36/36-5-448.pdfMicromorphology Figure 2b is a photomicrograph of a thin section of the dolomite rock

454 Hawker and Thompson Clays and Clay Minerals

which seem more condensed than the spherical par- ticles of birnessite or todorokite (Figures 6c and 6d).

SUMMARY AND CONCLUSIONS

The main factors controlling the distribution and reorganization of Mn in the residual oxide material appear to have been properties of the parent material and the amount of moisture in the soil. Processes of weathering have tended toward an in situ accumulation of soft and porous material. The dolomite seems to have altered by congruent dissolution, thereby dimin- ishing the cores of the original material. Eventually only voids were left, leaving a porous, "sponge-like" structure highly enriched in Mn and Fe.

Under these well-drained, oxidizing conditions, Mn and Fe appear to have been relatively immobile dur- ing the stages of weathering. MnO2 concentrations and oxide structures vary with depth and from site to site. From the XRD and IR data, and the fact that only traces of HCl-extractable cations other than Mn were found, these manganese wads appear to be a contin- uous series of compositions of manganese oxides. Mn and Fe essentially crystallized into separate mineral phases--Mn into a series of manganese oxide com- positions within which there were several stable and metastable mineral forms (i.e., nsutite, birnessite, and poorly ordered todorokite) and Fe essentially into goe- thite. Natural manganese oxides, however concentrat- ed, are difficult to distinguish from each other, even by the most sophisticated techniques. The explanation of the natural accumulation and concentration of Fe in the form of magnetic nodules (maghemite and he- matite) has not been attempted.

The accessory minerals, i.e., veins of quartz and muscovite, do not seem to have any direct bearing on the manganese residua development. They were prob- ably formed during another cycle of rock weathering, sediment transportation, and deposition.

ACKNOWLEDGMENTS

We thank H. J. van Tonder of the Plant Protection Research Institute for the electron micrographs and A. Loock for the preliminary analytical dissolution data. We are indebted to H. C. H. Hahne and F. A. Mumpton for their valuable comments and critical reviews of this manuscript.

REFERENCES

Brown, G. (1980) Associated minerals: in Crystal Structures of Clay Minerals and their X-ray identification, G. W. Brindley and G. Brown, eds., Mineralogical Society, Lon- don, 362-372.

De Waal, S. A. and Hiemstra, S.A. (1966) X-ray analysis of six samples of manganese ore from Graskop, Eastern Transvaal: NatL Inst. Metallurgy Rept. 97, 3 pp.

Mehra, O. P. and Jackson, M.L. (1960) Iron oxide removal from soils and clays by a dithionite-citrate system buffered with sodium bicarbonate: in Clays and Clay Minerals, Proc. 7th NatL Conf., Washington, D.C., 1958, Ada Swineford, ed., Pergamon Press, New York, 317-327.

Norrish, K. and Hutton, J. T. (1969) An accurate X-ray spectrographic method for the analysis of a wide range of geological samples: Geochim. Cosmochim. Acta 33, 431- 453.

Potter, R. M. and Rossman, G. R. (1979a) The tetravalent manganese oxides: Identification, hydration and structural relationships by infrared spectroscopy: Amer. Mineral 64, 1199-1218.

Potter, R. M. and Rossman, G. R. (1979b) Mineralogy of manganese dendrites and coatings: Amer. Mineral 64, 1219- 1226.

Taylor, R. M. and Schwertmann, U. (1974) Maghemite in soils and its origin. I. Properties and observations of soil maghemites: Clay. Miner. 10, 289-298.

Turner, S. and Buseck, P. R. (1981) Todorokites: A new family of naturally occurring manganese oxides. Science 212, 1024-1027.

Turner, S., Siegel, M. D., and Buseck, P.R. (1982) Structural features of todorokite intergrowths in manganese nodules. Nature 296, 841-842.

(Received 7 0ctober1987 ; accepted 3 0 May1988 ; Ms. 1716.)