ALLUVIAL TERRACES OF THE LOWER VAAL RIVER, SOUTH AFRICA: A REAPPRAISAL AND REINVESTIGATION1

KARL W. BUTZER, DAVID M. HELGREN, G. J. FOCK, AND ROBERT STUCKENRATH Departments of Anthropology and Geography, University of Chicago; Department of Geography, University of Chicago, 60637; McGregor Memorial Museum, P.O. Box 316, Kimberley, South Africa; and Smithsonian Radiation Biology Laboratory, Rockville, Maryland

ABSTRACT

The alluvial terraces of the Lower Vaal River, South Africa, are well known for their unique strati- graphic record of the Pleistocene period and as the focus of an apparent Acheulian sequence. Recent field examination indicates that this terrace suite is both more complex and more systematic than previously understood, extending downstream from the classical study area at Windsorton and Barkly West to the Orange-Vaal confluence. An expanded stratigraphic framework for the late Pliocene to early Pleistocene Older Gravels is described, while major subdivisions for the Younger Gravels (with Middle Pleistocene fauna and Acheulian occurrences) are proposed. A new late Pleistocene to Holocene lithostratigraphic en- tity, the Riverton Formation (with five members), is defined. Finally, problems of paleoenvironmental interpretation are discussed.

INTRODUCTION Butzer 1973a), no excavations of Acheulian In the 1940s southern Africa seemed to sites south of the Zambezi have been fully offer one of the best-known Pleistocene en- reported. The earlier work in southern vironmental and cultural successions of Africa was, for its time, excellent yet inten- the continent. However, in the last three sive reinvestigation of nearly all the early decades conceptual developments and dis- and mid-Pleistocene deposits seems to be coveries elsewhere have placed the South needed. African record in a backwater position. The One of the foremost study areas of early early and mid-Pleistocene time range is now and mid-Pleistocene environments and cul- relatively poorly understood from either the tures in South Africa has been the lower stratigraphic/environmental point of view Vaal River Basin. Here the Vaal and its (see, e.g., Flint 1959; Cooke 1967) or the tributaries have created complex terrace archeological one (see Cole 1961; Mason forms that provide an intermittent alluvial 1962, 1967). The study of this period in history spanning all of the Pleistocene. South Africa does not compare well with Since the Interior Plateau of southern Af- current information on the Rift Valley in rica has been subject to extensive erosion East Africa, where the discoveries at through most of the Cenozoic, the "Vaal Olduvai, along the eastern shores of Lake Gravels" comprise one of the few moderate- Rudolf, and in the lower Omo Basin have ly large areas of Pleistocene deposition. In proved so rewarding. The South African addition, the Vaal Gravels contain mam- australopithecine deposits and faunas stand malian fossils and Acheulian artifacts in in stratigraphic isolation, with the regional several of the younger alluvial units. implications of their paleoenvironmental in- The Vaal River gravels were "discov- ferences controversial (see Butzer 1971b; ered" in the early 1870s by diamond-dig- 1971c, p. 427-432; 1973d). With the ex- gers, whose relentless search for alluvial ception of Amanzi Springs (Deacon 1970; diamonds over the succeeding 100 years pitted most of the occurrences of the Vaal 1 Manuscript received September 11, 1972; re- Gravels into a cratered landscape. Such vised December 15, 1972. gravels can be readily traced at intervals [JOURNAL OF GEOLOGY, 1973, Vol. 81, p. 341-362] from the Vaal-Orange confluence upstream

© 1973 by the University of Chicago. All rights reserved. for 280 km to Windsorton, beyond which

341

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms 342 K. W. BUTZER, D. M. HELGREN, G. J. FOCK, AND R. STUCKENRATH exposures become few and sporadic. The sented here are based on sedimentological most productive gravels, between Barkly analysis of over 75 samples, involving hy- West and Windsorton, yielded fossils and drometer and wet-sieve determinations, artifacts in sufficient quantity to attract the Chittick calcium carbonate readings, pre- attention of Du Toit (1907, 1907-1908) at liminary quartz-grain micromorphology, pH the turn of the century, leading to closer values, and further, selective chemical tests. scientific investigation since the mid-1930s. This work is primarily associated with GENERAL GEOMORPHOLOGY OF THE Sdhnge et al. (1937), Cooke (1947, 1949), VAAL BASIN Breuil et al. (1948), and Van Riet Lowe The Vaal River rises on the High Veld (1952). Huzayyin (1941) drew attention Plateau of the southeastern Transvaal and to the importance of the Vaal Gravels, and eastern Orange Free State, an area of un- Zeuner (1958 [1945], 1959 [1946]) suc- dulating plains at 1,400-1,800-m elevation cessfully introduced the Vaal sequence into and formed primarily by shales and sand- the secondary literature of the northern stones of the Karroo System. Along the sub- hemisphere. After this first impetus for sequent section of its 1,350-km course to study passed, reservations, modifications, join the Orange River at 957-m elevation and even negative evaluations began to ap- near Douglas, the Vaal generally cuts below pear for the established terrace sequence Karroo shales, tillites, and diabase intru- (Maarleveld 1960; Partridge and Brink sions ("dolerite") into Precambrian quartz- 1967), the paleoclimatic interpretation ites, dolomites, andesitic lavas ("diabase"), (Flint 1959; Cooke 1967; Bond 1967; and granite. Beyond the grain of the com- Partridge and Brink 1967; Butzer 1971a, plex Precambrian structures of the Vaal's 1973d), and the archeological record itself superimposed course between Vereeniging (Cole 1961; Mason 1962, 1967). and the Vredefort Dome, the valley broad- As a result of these much-needed, pri- ens out onto the lower planation surface marily negative, reassessments, the Vaal ofse- the Middle Veld, an irregular plain stud- quence is now widely regarded with equivo- ded by hills and mountains of resistant cation. Meanwhile, ongoing geochronologi- igneous rocks. cal, paleoenvironmental, and archeological The beveled nature of the Veld surfaces, work elsewhere in eastern and southern gently cutting across the strata of the Kar- Africa continues to emphasize the potential roo System, was first emphasized by Davis importance of the Vaal Gravels for eluci- (1906), who interpreted it as a peneplained dating both environmental history and pre- landscape studded with monadnocks. Dixey historic subsistence patterns during the (1938) went further and interpreted the early and mid-Pleistocene time range. High Veld "peneplain" as part of a mid- Following brief reconnaissance in 1969 Tertiary erosional surface claimed to be al- and 1970, a reinvestigation was begun by most ubiquitous for the African continent. Butzer in 1971 with further detailed studies King (1963, p. 226-229; 1968, p. 271- (D.M.H.) currently underway. The pur- 274), on the other hand, considers the High pose of this paper is to review critically Veld to be part of an "extreme" planation the available evidence, both from the older surface, resulting from pedimentation pro- literature and particularly as gained in the cesses whereby subhorizontal plains are cut field since 1969. Whereas our interpretation when extremely broad, coalescent pedi- of the older erosional forms and deposition- ments remove upland interfluves. Like al features remains to be completed, study Dixey, King (1968, chap. 9) postulates a of the late Pleistocene and Holocene se- sequence of universal erosional cycles that quence now allows the definition of a new deserve consideration in discussing the evo- lithostratigraphic entity. The results pre- lution of the Vaal Valley:

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms ALLUVIAL TERRACES OF THE LOWER VAAL RIVER, SOUTH AFRICA 343

1. "Gondwana" Planation Surface, cut dur- interior watersheds were possibly uparched. ing the Late Triassic and Jurassic across the Deep dissection and headward cutting of the Stormberg basalts (Karroo System) and now coastal rivers into the escarpment has con- forming residuals along the crests of the Vaal tinued into the present, without significantly watershed. affecting the exceptionally low gradients of the 2. "Post-Gondwana" dissection, with cutting river courses (e.g., 6.5:10,000 in the case of the of incipient surfaces in valley heads below the Vaal) of the interior plateaus. "Gondwana" Surface. These Early Cretaceous readjustments adjacent to the upwarped Dra- The High Veld surface has been tenta- kensberg were followed by further tectonic tively dis- approximated in figure 1 by attempt- turbances along the Great Escarpment. ing to reconcile the 1:250,000 topographic 3. "African" Landsurface, cut across the maps with King's small-scale outline map High Veld during the time span of the Late (King 1968, fig. 119). Elevation of this Cretaceous to mid-Cenozoic. surface rises gradually from 1,200 m in the 4. Two cycles of pediplanation, in response Kimberley region to nearly 2,000 m at the to repeated epeirogenic uplift in Miocene and watershed of the Great Escarpment. In gen- again in Pliocene times, cutting the broad val- ley-side and tributary pediments of the Lower eral, the High Veld constitutes an impres- Vaal Middle Veld. sive planation surface. Nonetheless, it is 5. The Pleistocene stage was set by reeleva- disturbing that some sectors, such as the tion of the Great Escarpment while the old Gaap Plateau on Precambrian dolomites,

FIG. 1.-The Vaal and upper Orange Drainage, with High Veld planation surface and gravel occur- rences. In part modified after DuToit (1907), SShnge et al. (1937), Cooke (1947), and King (1968).

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms 344 K. W. BUTZER, D. M. HELGREN, G. J. FOCK, AND R. STUCKENRATH are developed on highly resistant rocks and the true extent and differentiation of these represent exhumed pre-Karroo surfaces (see surfaces is a delicate matter, and will re- Wellington 1937). Whether this High Veld quire careful mapping. This is particularly surface is everywhere contemporaneous and so in the Kimberley region, where the de- attributable to a single set of geomorpho- limitation of the lower margins of the High logical forces in the Vaal drainage is, there- Veld surface is difficult, and where it is un- fore, uncertain. Even more debatable is certain whether one or more younger sur- whether it can be correlated beyond the faces can indeed be recognized. Finally, it Orange-Vaal watersheds with other, appar- remains to be considered whether several ently convergent, surfaces, and so estab- extensive erosional surfaces in the water- lished as a morpho-stratigraphic marker for sheds of the lower Vaal and Harts are not all of Africa. Any such universal surface after all lateral pediments penecontempo- could only be cut with reference to a single raneous with the highest gravels in these base level, the ocean surface. At this point areas. of the erosional "cycle," pediplanation be- DISTRIBUTION OF GRAVELS IN THE comes distinguishable from peneplanation only in terms of processual details and, as VAAL BASIN Wellington (1955, p. 32-34) has pointed Compared with the better-known terrace out, it is difficult to conceive of the High sequences of higher latitudes, the Vaal Veld peneplain (or pediplain) stretching gravels are thin, of limited areal develop- across the continent from the Atlantic to ment, and of small vertical differentiation the Indian Ocean. Instead, it is far more in relation to a shallow and unimpressive reasonable to postulate a continuous post- valley. Triassic erosional hemicycle, initially rather The uppermost 400 km of the river are rapid as the softer Karroo rocks were re- weakly incised into the High Veld Karroo duced (Wellington 1955, p. 33). Eventual- rocks. These shales and sandstones provide ly, the resistant Precambrian units were ex- no durable gravels, and there are no mid- posed in the lower Orange valley below the Pleistocene or older stream terraces. The Orange-Vaal confluence. At this point in oldest deposits consist of valley-bottom the erosional history, several stages of par- clays fragmentarily preserved under late tial planation may have occurred, reflecting Pleistocene silt terraces. Included in such both the history of cutting through rock erodible strata is the faunal site of Cornelia barriers downstream and possible, repeated (Butzer 1973c), with its Middle Pleisto- uparching of the Orange-Vaal watersheds. cene mammalian assemblage (Cooke 1973). Variable surface gradients of the upland The 100-km stretch of river near Veree- plains, as well as breaks of longitudinal niging traverses Precambrian rocks, and gradient in the Orange and Vaal river sys- gravel deposits to +30 m are indicated, pri- tems, can all be adequately explained in marily by well and foundation drillings this fashion. (A. B. A. Brink, personal communication, In other words, the identification of the 1969; Mason 1962, p. 35-39, 117), and High Veld with the postulated "African normally found under a complex mantle of land surface" of Dixey or King, either in late Pleistocene alluvial silts. Some of these terms of age or functional origin, is far from older gravels, consisting mainly of quartzite an established fact. Similarly, the universal and chert from the Precambrian basement, dicyclic nature of Late Tertiary erosion re- appear possibly to be reworked in the mains to be demonstrated. The immediate younger alluvial terraces well exposed by relevance of this is that the several erosion- quarries along several north-bank tribu- al surface of the Vaal Basin have not been taries. These include important mid-Acheu- "dated" and their origins can be determined lian assemblages excavated from +10 m only by detailed field study. Furthermore, gravels of the Klip River (Mason 1962, p.

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms ALLUVIAL TERRACES OF THE LOWER VAAL RIVER, SOUTH AFRICA 345

40-42). No gravels at all are exposed along bedded in a matrix of calcreted silty sand. the subsequent erosional stretch of the river Except for basal units with locally derived 300 km downstream to Commando Drif. rock, gravel rounding is good to excellent, Consequently, the first 900 km of the river while sorting according to grade is rather offer few possibilities of establishing a ter- poor. In most areas, Ventersdorp diabase race sequence and the Vereeniging gravels (andesite) is the dominant cobble and will remain difficult to relate to terraces of boulder component, with local significance the Lower Vaal. of indurated shale, dolerite, and silicified Between Commando Drif and a little up- wood of Karroo lithology, and a ubiquitous stream of Windsorton, a stretch of 175 kinm, residual component of quartzite, quartz, gravels are sporadically preserved at +15- chalcedony, chert, agate, and jasper pebbles 50 m above a shallow, low-gradient Vaal, derived from the Precambrian Basement along eroded, sloping spurs mantled by Complex. Gravel thickness is locally seldom sandy wash of eolian origin or partly ob- greater than 5 m, although a section of 18 m scured by late Pleistocene silt terraces. Near has been reported by Partridge and Brink Bloemhof these silts mantle crude colluvial (1967). detritus in which Mason (1969; see also Previous work in the Lower Vaal area led Cooke 1949) excavated a late Acheulian oc- to differentiation of two categories of Older cupation site and several as yet unpublished Gravels: primary and derived (see Sbhnge faunal sites. Older deposits are absent along et al. 1937; Cooke 1947; Breuil et al. the south-bank tributaries of this stretch, 1948; Van Riet Lowe 1952; Partridge and probably because of the eolian cover mantle Brink 1967). The primary gravels ("Basal and well-developed late Pleistocene to Older Gravels" of Cooke 1947) are in situ Holocene silt terraces. To the north, how- alluvial deposits, with appreciable quanti- ever, there are a number of tributary grav- ties of diabase and/or dolerite. By contrast, els (see Sihnge et al. 1937, p. 36-38; Cooke the derived gravels (synonyms include "Re- 1947), that merit further study. distributed Older Gravels," "Older Red Optimal terrace development is found Gravel," and "Potato Gravels") have been between Waterval/Windsorton and the reworked by colluvial agencies and weather- Harts River confluence at Delportshope, a ing has removed all but the resistant silica 130-km stretch marked by alternating steep rocks. As a result, the primary gravels con- and gentle longitudinal gradients, where sist mainly of poorly sorted, coarse-pebble there are multiple and systematic terrace to boulder-grade materials whereas derived levels, with Vaal as well as tributary facies gravels are well sorted in the coarse-pebble to a maximum relative elevation of almost grade. The majority of outcrops of the de- +100 m. This is the area of the classic rived gravels form unconsolidated surface studies of Sihnge et al. (1937) and Cooke veneers seldom more than 30 cm thick; (1947, 1949). these are found on flat surfaces as well as Finally, the remaining 155 km from the slopes. Local thicknesses may exceed 2 m, Harts to the Orange confluence has spo- however, and some occurrences are cal- radic, but nonetheless significant, gravel ex- creted-for example, in the Windsorton posures recognized in principle by Du Toit area, where derived gravels are found ce- (1907, 1907-1908) but ignored or forgot- mented in old soil pipes or in surface cal- ten by later authors. cretes over primary gravels (Partridge and Brink 1967; Mason 1967). "OLDER GRAVELS" OF THE LOWER VAAL Cooke (1947) pointed out that the pri- The oldest alluvial deposits of the Lower mary Older Gravels are found at progres- Vaal are found at relative elevations of 20- sively lower elevations nearer the river, 90 m above modern flood level. These are spread out laterally over as much as 10 km. cobble-grade conglomerates, generally em- Since they were presumed to lie in old hol-

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms 346 K. W. BUTZER, D. M. HELGREN, G. J. FOCK, AND R. STUCKENRATH lows of the original river bed, they could duced by splitting of the cycle across hardly be contemporaneous. Consequently, lithological barriers," (2) the local effect of he interpreted them as a discontinuous the barriers themselves and their subse- record of different stages in bedrock dissec- quent removal through breaching (3) in- tion as the river encountered ever more re- creased longitudinal gradients through axial sistant hard-rock barriers, requiring in- warping of the watershed, and (4) the de- creasing adjustments of channel profile, velopment of knickpoints due to rapid low- with increased local gradients. In view of ering and river capture by the Harts (Par- the fragmented nature and seemingly ran- tridge and Brink 1967, p. 35). dom vertical distribution of the Older The only way that the knickpoints can Gravels, it is understandable that Sdhnge et be understood and a terrace stratigraphy al. (1937) did not explicitly suggest a ter- elucidated is by detailed mapping of both race stratigraphy, even though their longi- gravels and fluvial platforms along the tudinal profile (Sbhnge et al. 1937, pl. 4) length of the Lower Vaal, combined with does show three stages with gradients paral- lithostratigraphic characterization of each leling those of the modern river bed. Cooke outcrop. Figure 2 shows our preliminary (1947) is even more cautious, recognizing map of gravel exposures and platforms of the difficulty of correlating these terrace the Barkly West-Windsorton area; detailed fragments across the two steep-gradient fieldwork is currently underway along the knickpoints upstream of Windsorton and entire Lower Vaal with this primary ob- downstream of Barkly West. To avoid these jective in mind. The tentative information problems of longitudinal correlation, Par- now available can be summed up as follows: tridge and Brink (1967) confined their a) The highest fluvial platform is cut across study to the Windsorton sector, where they diabase at 1,130-1,170-m elevation, south of identified discrete terrace bodies by air- the Vaal between Riverton and Schmidtsdrif. photo interpretation, field mapping, and es- At about +90 m, this surface may represent

tablishing representative lithostratigraphic remnants of a river-cut platform or, perhaps, profiles for each. Three units were recog- an exhumed, Paleozoic, glacially eroded plain. nized: 200, 100, and 70 feet above low- However, there are appreciable gravel deposits water level or, about 57, 30, and 18 m on top of this feature near Nooitgedacht (see above flood stage. Nooitgedacht Platform, fig. 2), and derived The knickpoints found along the Lower gravels from this major outcrop were subse- quently reworked at lower levels by minor Vaal are crucial to any interpretation of the tributaries draining towards a Vaal River then terrace gravels since the former could either at or below +60 m. Similar colluvial "tribu- be a result of "fixed" bedrock barriers or tary" gravels are found southwest of Gong- of successive longitudinal stream adjust- Gong. ments migrating upstream. In the first case, b) A second platform can be traced con- ancient river gradients would have main- tinuously west of the river from above Water- tained some basic similarity to those of to- val to near Barkly West, at elevations of 1,150- day, even if barrier resistance increased 1,170 m cutting across both diabase and Dwyka with dissection; in the second instance, for- shale, with a gradient half that of the modern mer gradients would have had little similar- river (see Holpan Platform, fig. 2); relative elevation at Windsorton is +60 m, at Barkly ity to those of the modern river. Previous West +80 m. Here again, there is some con- authors have emphasized the notion of suc- formity with the exhumed, pre-Karroo topog- cessive stream adjustments and Partridge raphy. Primary deposits are restricted to the and Brink (1967), following up in part on Holpan area, where they have already been Cooke (1947), attribute the Older Gravel mapped by Partridge and Brink (1967) as terraces to: (1) headward advance of knick- "200-ft" gravels. Derived gravels form veneers points related to subphases of the post- over much of the platform. African "erosion cycle" of King and "pro- c) Another, subcontinuous fluvial platform

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms Platforms and Alluvial Formations Barkly West to Windsorton

Riverton Fm.

Younger Gravels

Older Gravels

Tributary' Gravels

FIG. 2.-Preliminary mapping of platforms, gravels, and Riverton beds in the Barkly West-Windsor- ton area.

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms 348 K. W. BUTZER, D. M. HELGREN, G. J. FOCK, AND R. STUCKENRATH can be identified at 1,105-1,130 m west of the or steeper valley margins include both par- Vaal between Windsorton and Gong-Gong, with ticularly coarse grades and high proportions a gradient only half that of the modern river of local rock, with some evidence of valley (see Proksch Koppie Platform, fig. 2). Thus margin dips and lateral intercalations of the relative elevation increases from +33 m rubble screes. Intervening stretches of lower upstream to +72 m downstream. This is the gradient or softer Karroo bedrock presum- "100-ft" gravel mapped by Partridge and Brink (1967) at Windsorton (Proksch Koppie). Vari- ably supplied few durable materials; local ous other gravel bodies are found on this plat- alluvium here was probably fine grained, form between Windsorton and Gong-Gong- so that sandy or silty facies must have once some primary, some weathered to residual been widely represented. This finds some products, most reworked, including derived support in recurrent sandy or loamy inter- "tributary" gravels north of Barkly West. Stud- beds, such as those of the "100-ft" terrace ies of gravels and matrices must be completed at Windsorton (see Partridge and Brink before a correlation with +45-50 m gravels in 1967, profile 4). Contrary to the opinion of the Orange-Vaal confluence area can be con- Van Riet Lowe (1952), these are fluvial firmed. rather than eolian facies. Nonetheless, d) Lower-lying gravels are preserved spo- radically and restricted to the immediate bed- eolian sands of Kalahari origin have been rock slopes or bed of the Vaal, where any plat- tapped by the Harts headwaters and there- forms are narrow. In addition to the Younger by introduced to the Vaal sediment complex Gravels discussed below, these younger de- on one or more occasions (Van Rooyen posits include Partridge and Brink's (1967) 1971, chap. 5); such materials were repeat- "70-ft" stage at Windsorton (Little Proksch edly reworked by fluvial and eolian agencies Koppie, Wedburg, Amandelhoogte)-a key al- on or downwind of the Vaal floodplain. luvial unit linked with gravel outcrops at Wil- The pedogenetic alteration of the Older lowbank (+26 m), and at +30-36 m in the Douglas area. Gravels remains to be studied systemati- cally. The two common patterns are (1) The distinction between primary and de- calcreted conglomerates with corroded but rived gravels can be a difficult one to make. intact silicate rocks, and (2) rubefied, un- In many instances, shallow gravels have consolidated gravels of exclusively resistant been deeply weathered in situ, without re- lithology. The latter would appear to indi- working on slopes of 1° or less; only the cate intensive weathering during formation more resistant silica rocks remain. In other of fersiallitic soil profiles. The former reflect instances, such residual pebbles are domi- massive calcification. In addition, the cal- nant or exclusive in intact, calcreted gravel crete facies is more often than not capped terraces. Examples such as this, including by younger calcretes developed in colluvial the Willowmore exposures, are character- rubble or even in superposed, derived istic wherever there are no local diabase or gravels (see Partridge and Brink 1967). dolerite bosses liable to erosion. Conse- The mutually exclusive occurrence of both quently, each gravel outcrop or veneer must pedogenetic forms remains to be explained, be evaluated separately. In fact, in the ma- as do their temporal relationships. jority of cases, apparently derived gravels do carry stratigraphic implications for a PALEONTOLOGY AND ARCHEOLOGY OF THE particular platform and (implicitly) stage "OLDER GRAVELS" in river evolution. The protracted period of bedrock dis-

It is improbable that the major terrace section and sporadic gravel aggradation re- stages were ever represented by uniform corded by the successive generations of gravel bodies over great longitudinal Older Gravels is poorly dated since there reaches of the Vaal. Preliminary studies are no materials suitable for radiometric show that gravel exposures adjacent to hills assessment, and faunas appear to be absent.

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms ALLUVIAL TERRACES OF THE LOWER VAAL RIVER, SOUTH AFRICA 349

However, problematic remains of several although their provenance is generally im- rather ancient forms of uncertain prove- possible to assess. However, J. D. Clark, nance have been described from the Lower K. P. Oakley, and Van Riet Lowe did re- Vaal. These include molars of Gompho- cover artifacts in the cemented, reworked therium, "Archidiskodon" subplanifrons, top of the "200-ft." gravels at Windsorton and "Archidiskodon" broomi (Cooke 1963, (Clark, personal communication, 1972). table 7; 1967, p. 180). Further in question Mason (1967) subsequently excavated a are the discordant remains of the archaic small collection from this same position, suine Notochoerus capensis and the giraffid dividing the materials into Later, Middle, Sivatherium (?) cingulatum (Wells 1964). and Earlier Stone Age on typological These fossils may originally relate to one or grounds. But, according to Clark (personal other of the Older Gravel units and in fact communication, 1972), "I saw nothing in may not have been found in the Younger the collection he recovered that could not Gravels. Unfortunately, the field records have been E.S.A. in the light of the Olduvai are wholly inadequate as a result of uncon- assemblages." Conceivably, therefore, in- trolled collecting over many decades, but formation of value may yet be forthcoming, Helgren's ongoing work has failed to find although any such artifacts are decidedly

traces of bone in any of the Older Gravel scarce.

units. However, the mastodont is a Pliocene "YOUNGER GRAVELS" OF THE LOWER VAAL

form, and Maglio's (1970) revision of the Prior to the next period of aggradation Elephantidae reclassifies the Vaal elephants along the Lower Vaal, Cooke (1947) postu- as Mammuthus subplanifrons and Loxo- lates epeirogenic deformation in the Orange- donta sp. nov., respectively, species that Vaal drainage, with uparching of the inter- date 4.0 m.y. B.P. in the East African po- fluves and intensified bedrock downcutting. tassium-argon chronology. This appears to Whether or not this was the case, accumu- lend some measure of support to the crude lation of a complex body of Younger estimate of 4.8 m.y. B.P. derived for the Gravels began along the length of the Vaal. "200-ft." gravels at Windsorton by geo- These Younger Gravels are as coarse as or morphologic criteria (Partridge and Brink coarser than the Older Gravels, and of simi- 1967). Therefore, it is possible that the lar lithology. Van Riet Lowe (1952), ampli- Older Gravels date from the terminal Plio- fying the interpretation of SShnge et al. cene as well as the early Pleistocene. (1937), identified four stages of deposition The primary Older Gravels may or may (I, IIa, IIb, and III), each supposedly rest- not contain man-made implements. Fock ing on freshly cut bedrock platforms. The (1960) and Mason (1967) have recovered scheme is roughly as follows, employing raw possible, but isolated, artifacts from well data from the scale sections of Sdhnge et al. within the matrix of "200-ft" conglomerates (1937, pls. 2-3) and revising the technical near Windsorton. Evaluation of older re- terminology: ports of possible artifacts from the same sediment body (see Breuil et al. 1948) is 1. Bedrock dissection followed by lateral rendered impossible by the multiple solu- planation at 16-18 m ("12 m") above modern river bed. tion pipes that penetrate the intact terrace 2. Aggradation of Younger Gravels I (7 m alluvia but which are filled with superficial- thick) on this rock-cut platform, to 12 m or ly similar materials (Partridge and Brink more above modern flood level. 1967; Mason 1967). 3. Renewed bedrock dissection, followed by Some of the derived Older Gravels con- lateral planation at 8-13 m ("6 m") above tain artifacts (Breuil et al. 1948; Van Riet modern river bed. Lowe 1952; Mason 1962, p. 45-46; 1967) 4. Aggradation of Younger Gravels IIa and

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms 350 K. W. BUTZER, D. M. HELGREN, G. J. FOCK, AND R. STUCKENRATH

IIb, separated by lenses of current-bedded clusions of S6hnge et al. (1937), we find sands, with a total thickness of 12-14 m, on that the field evidence also negates some of this platform, to 12 m or more above modern these specific criticisms of Cooke (1947) flood level. and Partridge and Brink (1967). Our views 5. Aggradation of silt and sand, with diato- maceous lenses and Achatina shells, resting on and tentative results are as follows: current-bedded sands that fill the base of swales a) The bedrock base to the Younger Gravel in an undulating topography formed by gravels complex (as, e.g., illustrated by the excellent I-II. Local thickness of these "Calcified Sands" scale sections of Sdhnge et al. [1937, pls. 2-3 is almost 12 m, cumulative thickness probably 17 m or more. and figs. 5-6]) shows no demonstrable fluvial 6. Calcification of the previous fine alluvium. platforms of stratigraphic significance. In fact, the discrepancies between the raw observation- 7. Renewed bedrock dissection to modern channel floor, seemingly preceded or followed al data and the textual generalizations of Sihnge et al. (1937) are too great to be overlooked. by lateral fill-cutting of terrace spurs at 7.5 m above modern flood level. The field evidence shows clearly that the Vaal's bedrock channel had already been cut approxi- 8. Aggradation of Younger Gravels III (5 m thick) on floor of modern channel, to 3 m mately to its modern depth and morphology below modern flood level. prior to aggradationr of the Younger Gravels at 9. Sheet erosion ("peneplanation") of previ- Windsorton. b) The Younger Gravels can indeed be sub- ous deposits at 12 m above modern flood level, divided into three successive alluvial bodies contemporary with aggradation of so-called Youngest Gravels in minor, local thattributaries. are distinct both on facies and stratigraphic 10. Wind-blown sands. criteria. The basal unit ("A") consists of mas- sive, traction bed-load deposits of cobble-grade In evaluating this sequence, Cooke (1947) gravel, with a rubefied matrix (a red cambic argued that the Younger Gravels I are soil horizon), and with subsequent formation found at a single locality, that the distinc- of a calcrete or silcrete duricrust on top. The intermediate unit ("B") is more typically rep- tion between units IIa and IIb is based resented by uncemented coarse sands and peb- solely on the presence of some discontinu- bles showing complex current-bedding, trough ous sandy units, and that gravels II and III cross-bedding, and gravel "dunes" moving are conformable at several sites. He conse- downstream. The top unit ("C") is a thick quently regarded gravels I, II, and III as accumulation of coarse to cobble-grade gravel, conformable deposits recording "a single deposited as gravel bars splaying across the and practically continuous cycle of down- former channel axis in pool and riffle sequences; cutting" (Cooke 1947, p. 256). However, a major postdepositional duricrust cements the younger unit of "Calcified Sands" (5) most of the sediment body. This tripartite sequence can be traced downstream from Wind- was considered to be distinct. Partridge and sorton to Delportshope, despite minor facies Brink (1967) lump all the gravels and variations within each unit and repeated pinch- alluvial sands of phases (2-5) under the ing out of "B." Whereas "B" and "C" may be misleading designation "Current Gravels," conformable, "A" is appreciably older, with while not implying that vertical and lateral a long interval of polygenetic weathering inter- changes of facies are absent. The total is vening. Altogether, the Younger Gravels repre- ascribed to "a single cycle of channel de- sent at least two discrete sedimentary intervals, velopment involving entrenchment over a with bed-load sediments deposited on an ir- considerable period of time and culminating regular bedrock floor and derived primarily in the development of the present flood- from "head'ward" erosion along successive knickpoints. This is difficult to reconcile with plain" (Partridge and Brink 1967, p. 25). the concept of a "single" cycle of "downcut- The "Youngest Gravels" are labeled "Gully ting," since there were at least two hemicycles Wash Gravels" and given little stratigraphic of net aggradation along most of the lower significance within this overall time range. Vaal. Although we do not accept all of the con- Our own tripartite subdivision of the Young-

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms ALLUVIAL TERRACES OF THE LOWER VAAL RIVER, SOUTH AFRICA 351 er Gravels appears to match the IIa and IIb g) Although there are wind-blown sands in units, with intermediate sandy beds, of Sihnge the Windsorton area, many exposures of "red et al. (1937); their unit I has no apparent sands" prove to be colluvial loams that thicken validity, and their unit III is either truncated in footslope areas. Such colluvia commonly unit "A" or reworked gravel. include a component of reworked eolian sand. c) The Younger Gravel complex rises from The stratigraphy of these reddish loams and at least +12 m (above modern flood level) at sands generally requires a good deal of local Windsorton to +16 m at Canteen Koppie fieldwork due to regional facies peculiarities. (under a thick mantle of colluvial loam), and Some units are equivalent to the Riverton For- even higher elevations of +22-23 m at Gong- mation, others are younger, but all distinctly Gong, dropping off again toward Delportshope, postdate the Younger Gravels complex. where in situ unit "C" gravels extend from below the modern river bed to +19 m, but unit PALEONTOLOGY AND ARCHEOLOGY OF "A" gravels are at +27 m. Further down- THE "YOUNGER GRAVELS" stream, comparable gravels to +15-18 m are found near the Riet confluence and at Douglas. The age of the Younger Gravel complex It appears that the major reason for these vari- is thought to be late Middle Pleistocene on ations of gradient lies in the flow adjustments the basis of disparate and sporadic faunal of a greatly augmented river discharge to local collections made over the years (Cooke constrictions of the valley. 1949, 1963). These come primarily from d) The "Calcified Sands" are stratigraphi- Windsorton, from diggings into the gravel cally distinct from and disconformably above body, although some materials appear to the Younger Gravels, rendering Partridge and Brink's (1967) characterization of the "Cur- come from superposed sands and loams, or rent Gravels" as a one-phase, subcontemporary even from the base of the "Calcified Sands" feature totally invalid. Geochemically too, these (Cooke 1949). The most recent faunal in- units are distinct. The various outcrops of ventory has been provided by Wells (1964). Younger Gravels at Windsorton have 4%-45% Excluding the five Plio-Pleistocene elements calcium carbonate and 5%-15% amorphous discussed above, it includes 36 species or silica in their matrix; by contrast, the local genera, of which at least 17 (or 48% ) are "Calcified Sands" have only 1.0%-1.5% amor- extinct; Wells (1964) considers it probable phous silica with similar carbonate ranges as the that further specific determinations will Younger Gravels. These "Calcified Sands" are augment the proportion of extinct species. coeval with one of the late Pleistocene units (Member II) of the Riverton Formation, as Since at least some of the Vaal faunal col- defined further below. lections substantially postdate the Younger e) The morphological terrace formed by the Gravels "C" and another part is of appar- Younger Gravels (plus Calcified Sands) was not ent Plio-Pleistocene age, a simple "late sculptured by sheet erosion (see Cole 1961, Middle Pleistocene" age assignment is be- p. 112-113). Instead, it is a bonafide but con- coming increasingly difficult to uphold. vergent +12-m floodplain surface, reflecting Even within the undisputed "younger" as- the final aggradation stages of both the Younger semblage there are at least four species of Gravels and the Calcified Sands. Erosional steps zebrine and caballine horses, four genera subsequently cut on the flanks of this sediment and at least seven species of suines, and complex presumably reflect on later fluvial epi-

sodes. three forms of elephant (see Cooke 1963; f) The "Youngest Gravels" are of interest Wells 1964). This is difficult to reconcile in that they contain large boulders moved by with one, brief faunal span. In addition to gravity processes. Their stratigraphy is difficult the "derived" Plio-Pleistocene forms, there to determine at Windsorton; elsewhere, crude probably are three distinct assemblages re- local detritus may underlie or be intercalated lating to Younger Gravels "A," to Younger with the younger Riverton Formation, whereas Gravels "B/C," and to the Calcified Sands similar deposits above Bloemhof contain good Acheulian occurrences and mammalian fauna and other superficial deposits of late Pleisto- (see Mason 1969). cene age. There is then reason to believe

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms 352 K. W. BUTZER, D. M. HELGREN, G. J. FOCK, AND R. STUCKENRATH that the Younger Gravel complex spans a river. At the rate at which primary gravels much broader range-and possibly all--of are being worked over, it is consequently the Middle Pleistocene. urgent that pit sections be examined where- Of potential stratigraphic value are a ever the related tailings contain little-worn number of faunal concentrations in situ artifacts, and that new in situ sites be lo- excavated by Mason (1969) near Bloem- cated in finer-grained facies that offer some hof, in 1968. An analysis of this fauna is likelihood of at least semiprimary contexts. awaited. The Younger Gravels, as well as the over- SEDIMENTARY UNITS POSTDATING THE lying, loamy red colluvia, contain what are "YOUNGER GRAVELS": THE RIVERTON best described as Acheulian and "Faure- FORMATION smith" artifacts (Cole 1961). The key col- The late Pleistocene and much of the lections come from six sites: Homestead, Holocene time range are spanned by a series Larsen, Newman's Pont, and Riverview VI of deposits that postdate the Younger Grav- at Windsorton, and Canteen Koppie and els. These are well exposed in erosional Pniel at Barkly West. In part as a result of gullies (Afrikaans: dongas) near Riverton- dispersal to museums of many countries, the on-Vaal (28°'S, 20046'E), where we have total number of artifacts preserved in the established a complex stratigraphic se- Witwatersrand- University is a mere 614. quence. Five units (members) have been Each collection was made almost exclusively recognized and they can be summarized as from mine dumps from several pits dispersed follows, with reference to figures 3-4: over an area of several acres. Each collection I. Over 6 m, base not exposed. Intensively was made selectively by several individuals gleyed, white to light gray or pale yellow (Mun- over a span of several decades, and subse- sell hues 5Y), generally massive, clayey silt quent "sortings" must be assumed for most with slickensides. Interdigited with lenses of or all of the collections. Finally, each collec- sandy or gravely detritus, in part well stratified tion represents a spectrum from rolled to un- to laminated. A Vaal floodplain deposit, 8.5 m worn artifacts. As a result, the collections above modern flood water, including local tribu- are inhomogeneous, from neither primary tary interbeds. Postdepositionally calcified, and nor semiprimary contexts, as well as heavily followed by erosional disconformity. biased in favor of bifaces, cleavers, and II. 7 m. Weakly to moderately gleyed, white large flakes. Nonetheless, Cole's typological (SY), massive to well-stratified, silty sands to sandy silts, with dispersed basal pebbles. Fol- and metric analyses show significant differ- lowed by development of a 60-cm, light gray, ences among the collections, although their loam vertisol with prismatic structure, local meaning, as with all other gross measure- slickensides, and secondary calification. A Vaal ment typologies, is unclear. flood silt to + 10.5 m. Followed by at least 5 m Archeological research in the Younger dissection by tributary streams. Gravel complex has been exploratory only, III. 9 m. Local alluvia grading into +13-m with the volume of discursive and second- Vaal terrace. The various facies, both alluvial ary literature out of all proportion to the and colluvial, include basal gravels up to more significance of the limited and inadequate than 3 m thick, followed by weakly gleyed, materials in hand. The only systematic ex- white (5Y), calcareous, silty sands with gravelly cavation was devoted to the (at best) quasi- lenses and' horizons of reworked concretions. At the type site, the tributary terrace has a sterile Older Gravels (Mason 1967), rather relative elevation of 8.5 m above the donga bed, than to a careful survey and testing of suit- downstream, falling off to +5 m midstream, able deposits within the Younger Gravel ultimately converging with the valley floor up- complex. Artifacts can still be recovered stream (fig. 3). Locally, the tributary facies in quantity at Canteen Koppie and from may preserve a truncated calcareous paleosol, the tailings of diggings next to the modern with a profile in excess of 1 m, or a calcrete

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms ALLUVIAL TERRACES OF THE LOWER VAAL RIVER, SOUTH AFRICA 353

FIG. 3.-Longitudinal scale section of Riverton Boat-House Donga (west branch), exposing type sec- tion of Riverton Formation. Vertical exaggeration: 8X. duricrust with abundant amorphous silica. Fol- tributary terraces show remnants of a deep, lowed by at least 15 m of fill dissection by the relict vertisol with a dark gray (5Y) 180-cm Vaal River. A + Ca horizon of loamy texture and strong IV. 9 m. Vaal terrace at +8.5 m, embanked prismatic structure. Followed by at least 6 m against unit III, and consisting of grayish brown of fill dissection in the major valleys. (10YR) to light brownish gray (2.5Y), strati- fied, clayey to sandy silts. These flood deposits Younger upland sediments are restricted extend up the tributary streams, but no local to low dunes of yellow red medium sand and deposits are preserved, although older deposits light brown, laminated colluvial sediments appear to have been regraded (e.g., erosional of similar or finer texture. The donga floors shoulders) to conform to the new gradients. carry pink to very pale brown sands and Followed by at least 11 m of fill dissection silty by sands, well stratified, with grit lenses the Vaal River. and rare clayey lenticles. Commonly, too, V. 6 m. Tributary alluvia interdigited with there may be remnants of a +1-1.5 m wash +4.5 m Vaal alluvial terrace. The tributary terrace, consisting primarily of well-strati- facies is a massive-bedded, grayish brown fied, derived reddish sands of eolian origin, (2.5Y), silty sand, forming a terrace that, at the type site, rises from +3 m downstream to together with lenses of unit III or V mate- +6 m midstream before gradually converging rial. Finally, the subrecent Vaal flood silts with the donga floor upstream (see figs. 3-4). consist mainly of light brown (10YR), The Vaal facies is a well-stratified, dark gray laminated clayey to sandy silts. brown (10YR), clayey silt embanked against The Riverton Formation provides a dis- unit IV and spilling into tributary valleys. The tinctive and useful lithostratigraphic datum

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms 354 K. W. BUTZER, D. M. HELGREN, G. J. FOCK, AND R. STUCKENRATH

FIc. 4.-Selected scale cross-sections of Riverton Donga (west branch). Locations and patterns as in fig. 3. Vertical exaggeration: 4.5X. with which to identify and correlate Pleisto- Riverton Formation in the Barkly West- cene deposits between Windsorton and Del- Windsorton area is shown by figure 2. portshope and, to a lesser extent, down- stream to the Orange confluence, where PALEONTOLOGY, ARCHEOLOGY, AND facies change progressively. However, de- RADIOCARBON DATING OF THE tailed correlations of the Riverton Forma- RIVERTON FORMATION tion with the "Calcified Sands" and other Fossiliferous exposures in the Riverton post-Younger Gravels units at Windsorton Formation are restricted to unit III, the remain to be completed. Distribution of the upper part of which includes proliferations

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms ALLUVIAL TERRACES OF THE LOWER VAAL RIVER, SOUTH AFRICA 355 of Achatina zebra obesa (Pfeiffer), identi- ples. This was borne out by the radiocarbon fied by J. P. E. Morrison. These are gener- determinations. ally shells of uniform size (5.0-6.0 cm in Samples were collected from two locali- length), referrable to subadult individuals. ties (see fig. 3) in the eastern and western A fragment of ostrich eggshell was also branches of the Riverton donga. Based on found in the basal detritus of unit III. the old half-life, the Riverton dates are as Archeological vestiges are also confined follows: to unit III, where we found derived Acheu- SI-1110.-Member III, base; 8,640 ± 260 lian and Middle Stone Age materials, as B.P. Fragments of ostrich eggshell within well as several interesting Middle Stone gravel matrix. Very small sample, no pretreat- Age surface collections that appear to have ment given, and sample diluted. Since sample been in semiprimary context before being is subject to recrystallization, age must be con- exposed by recent erosion. The homoge- sidered as minimum due to contamination by neous Middle Stone age collections include, younger CaCO3. apart from pieces of quartzite, andesite, SI-1111.-Member III, middle; 9,610 ± 95 B.P. Massive calcrete matrix of conglomeratic and chert, mostly artifacts of indurated lense related to former land surface or ground- shale (lydianite). Dominant tool types are water level. Washed in 0.5N HC1 before CO2 Levallois points (convergently patterned) evolution. Must be considered as minimum age. and true blades, some of Levallois type, SI-1112.-Member III, top 30 cm; 17,240 ± generally with minimal retouch. Other com- 485 B.P. Shell of Achatina zebra obesa in termi- ponents include large side-scrapers, a lance- nal, calcrete duricrust. Small sample, diluted. olate point, cores (among which is a large, Sample soaked in distilled H20 to soften matrix, high-backed, prepared example), as well as matrix then washed off, and shells washed in a number of rather poorly standardized 1N HC1 before CO2 in H3P04. small tools. The overall impression is that SI-1113.-Member III, top 120 cm; 17,180 ± 215 B.P. Shell of Achatina zebra obesa in of a later, as yet undiagnostic variety of marly sand. Same pretreatment as SI-1112. Middle Stone Age. At Windsorton, the fine- Samples agree closely and provide basis for grained alluvia that we correlate with the estimating age of top of Riverton Member III. Riverton Formation also contain Middle Shell from this same collection was dated 11,400 Stone Age artifacts (Sihnge et al. 1937; ± 175 B.P. (UW-167A), with the porous, cal- Breuil et al. 1948; Van Riet Lowe 1952). careous matrix giving an apparent age of only Later Stone Age artifacts are definitely 7,360 ± 115 B.P. (UW-167B); these divergen- absent from units IV and V, although mi- cies indicate high levels of contamination by crolithic bladelets of Smithfield type occur younger carbonates both in the matrix and in the external shell walls, so that adequate pre- both in and under the post-Riverton sands. treatment is essential to realistic dating. Materials suitable for radiocarbon assays are limited to inorganic carbonates and Altogether samples SI-1112/1113 would shell. Except for a few outcrops of Pleisto- appear to indicate that alluviation of Mem- cene calcretes, there is no calcareous bed- ber III terminated no later than 17,000 rock anywhere in the Vaal Valley. Instead, B.P. Samples SI-1110 and 1111 are useless carbonates are derived from weathering of as dates for Member III, but their con- soda-lime feldspars, primarily at local dia- vergence near 9,000 B.P. may possibly re- base and dolerite outcrops, or in association flect on a younger carbonate cycle active with mineral springs related to doleritic perhaps 6,000-8,000 years ago. Implicitly, intrusions. Dead carbonates are consequent- therefore, they may be related to the ag- ly no problem, but carbonate cycles are gradation of Members IV or V. In any actively underway and the recrystallization event, this sample suite shows that C14 of calcite in even very recent sediments dates on calcretes (e.g., Netterberg 1969a) promises that contamination by younger and inorganic carbonates in the Lower Vaal carbonates is a serious liability in all sam- region are of limited value. As a final cross-

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms 356 K. W. BUTZER, D. M. HELGREN, G. J. FOCK, AND R. STUCKENRATH check we attempted to date the buried Ca- since river gradients were less and drainage horizon of a paleosol within the Riverton areas similar to those of today, the coarse Formation at the supposed "Fauresmith" Vaal gravels would require a very large type-site (28033'S, 21020'E), on Groen- river with greater stream competence, and a vallei Farm, Orange Free State: far more abundant or less seasonally con- SI-1114.-2,664 ± 65 B.P. Pedogenetic car- trasted rainfall. Unlike Sihnge et al. (1937), bonate horizon with root drip, disconformably Cooke (1947) favored nonclimatic, "tempo- under 1.5 m vertisolic alluvium and modern rary increases in erosive activity" to ex- soil. This probably correlates with Riverton plain the successive stages of bedrock dis- Member III as locally developed, forming top section that separate the several discrete of unit 1 of the Older Fill of Butzer (1971a). bodies of gravel alluvium. The "Calcified Sample subject to contamination by younger Sands" (Riverton Formation) are attributed carbonates and age must be considered mini- to a "sluggish, much diminished stream mum for formation. 8C13 = -1.79%o. flowing over a sandy floodplain" at a time In fact, the carbonates of the Fauresmith/ of drier climate, and pedogenetic calcifica- Groenvallei Ca-horizon appear to be pri- tion to upward capillary movement during marily related to accumulation of the over- the low-water season. lying dark cracking clays and the attendant Van Riet Lowe (1952), commenting on mobilization of carbonates. Since this verti- the "Youngest Gravels," suggests marked sol is equivalent to Riverton Member V, sheet erosion and seasonality of rainfall. the final period of aggradation may have This point is developed by Maarleveld terminated less than 3,000 years ago. (1960), who proposes maximum runoff at times when vegetation was sparse, but in- STRATIGRAPHIC INTERPRETATION creasing in response to increasing rainfall; The preceding information on the geo- at the height of a wet phase, on the other morphic evolution of the Lower Vaal is hand, the vegetative mat would have been summarized in figure 5. The chart is pro- too effective for rapid runoff. Flint (1959) visional, since ongoing fieldwork will re- expressed strong reservations regarding quire corrections and amplification for both oversimplified interpretation of phenomena the Older and Younger Gravel complexes. or processes, while Partridge and Brink However, this attempt does reflect a sub- (1967) and especially Partridge (1969) are stantial improvement in our comprehension reluctant to recognize climatic factors in the of the Vaal terraces, and the scheme pro- vicissitudes of the Vaal River's evolution. vides a working hypothesis that empha- The suggestions put forward below must sizes the true complexity of river evolution be considered as preliminary and incon- through time. clusive, although we do wish to emphasize that the Vaal sequence is potentially far PROBLEMS OF PALEOENVIRONMENTAL from uninformative in terms of paleoenvi- INTERPRETATION ronmental inferences. The evidence mar- Sahnge et al. (1937) interpreted each shaled here allows the following commen- hemicycle of bedrock cutting, gravel depo- tary: sition, and sand alluviation as the result of Ecological zonation of the Vaal Basin.- a wetter climate that began to wane as ag- Any paleoclimatic interpretation of the his- gradation began. Pedogenetic calcification tory of the Vaal River must consider the of the Riverton beds was ascribed to a semi- interplay of local and distant geomorphic arid climate, and eolian activity and "pene- balances. The highest parts of the eastern planation" related to an arid climate, drier watershed are moist subhumid (C2B'2r by than today. the Thornthwaite classification), the broad Cooke (1941, 1947) emphasized that central parts of the basin dry subhumid

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms Recent Vaal downcutting and gullying Alluviation of Vaal flood terrace and 1.5m sandy tributary terrace. Traces of Later Stone Age (Smithfield). Dissection of fill (vertical amplitude >6m). Eolian activity.

Mb V Alluviation: 6m of Vaal flood silts and tributary loams, climaxed by vertisol development. Floodplain +4.5m. Dissection (>11m). Mb IV Alluviation: 9m of Vaal flood silts. Floodplain +8.5m.

Dissection (>15m). Mb III Alluviation: 9m of tributary sands and gravels. Floodplain +13m. Terminated by 17,000 B.P. Partly coeval with Middle Stone Age settlement. Dissection (>5m). Mb II Alluviation: 7m of Vaal flood silts. Floodplain +10.5m. Dissection (amplitude ?). Mb I Alluviation: 6m of gleyed floodplain silts. Floodplain +8.5m. Dissection of fill (vertical amplitude >12m).

Calcareous paleosol YOUNGER Unit C (gravels) Middle to Late Achelia settlement. GRAVELS Unit B (sands and gravels) (+12 to 22m) Cambic and calcareous Mddle lestocenepaleosols fauna. Unit A (gravels) Bedrock dissection (vertical amplitude >18m).

(iv) Alluviation: '70ft' gravels. Bedrock dissection (>12m). (iii) Planation of Proksch Koppie Platform (+33/72m) with terminal aggradation of '100ft' gravels. Related tributary gravels (Barkly West). Bedrock dissection (>27m). (ii) Planation of Holpan Platform (+60/80m) with terminal aggradation of '200ft' gravels. Earliest artifacts ? ? Source of derived Pliocene fauna? Related tributary gravels (Nooitgedacht, Gong-Gong). Bedrock dissection. (i) Formation of Nooitgedacht Gravels (+90m) and Platform. Protracted, alternating bedrock dissection and planation.

FIG. 5.-Tentative late Cenozoic evolution of the Vaal Valley between Barkly West and Windsorton

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms 358 K. W. BUTZER, D. M. HELGREN, G. J. FOCK, AND R. STUCKENRATH

(C1B'2d), and the lower basin, below Bloem- and shallow veneer of river gravels. Lateral hof, semiarid (DB'3d) (Schulze 1958). Cor- planation by river channels is still a difficult respondingly, rainfall decreases from over subject, and peripheral pedimentation (see 800 mm in the upper basin to 300 mm near discussion of planation by Rust 1970) can the Orange confluence. The spontaneous hardly explain the sharply restricted ero- vegetation of most of the central and upper sional surfaces between Barkly West and watershed is grassveld, but the crucial lower Windsorton. There are as yet no specific basin has thorn savanna (thornveld) with explanations for these broad surfaces and patches of semidesert Karroo shrub (Acocks the long time spans they represent. Only 1953). In fact almost all of the gravel oc- careful fieldwork can provide new insights currences below Commando Drif are found on the complexity of processes favoring within these more xeric vegetation zones. longitudinal versus lateral change. Consequently, any study of the depositional Gravel aggradation.-The former trans- sequence must emphasize the distinction port of cobble- to boulder-size gravel by a between local and distant pebble compo- stream now as incompetent as the modern nents, and between alluvial and colluvial Vaal can hardly be brushed aside by in- facies. voking the 1-in-100-year flood catastrophe Bedrock versus fill cutting.-Unlike stream (e.g., Partridge 1969). As Wolman and dissection of solid bedrock, downcutting Miller (1960, p. 72) have shown, deposi- into alluvial fills can be a rapid process, tional features owe their origin primarily to easily initiated and often easily reversed. "events of moderate magnitude which recur On the other hand, incision into the shale relatively frequently rather than [to] rare bedrock and, particularly, diabase or doler- events of unusual magnitude." Prior to the ite, requires long periods of suitable me- construction of regulatory dams and irriga- chanical abrasion, preferably in combina- tion devices, monthly discharge at Riverton tion with chemical weathering. Partridge for January (the month with greatest aver- and Brink (1967, p. 34) chided Sohnge et age flow, or 21.6%) varied from as much al. (1937) for invoking climatic changes as 3,974 X 106 m3 (1917-1918) to as little to explain (bedrock) incision by the Vaal; as 13 X 106 m3 (1932-1933) over a 26- however, their counterexamples of rapid year period (see Wellington 1955, p. 374). (fill) cutting by American streams in recent Nonetheless, the Younger Gravels that decades are spurious . Sihnge et al. (1937) commonly clog the river bed are seldom did raise a difficult problem, noting that disturbed, and all the late Pleistocene and there has been little appreciable bedrock Holocene bed-load deposits of the Vaal dissection since the Younger Gravels epi- channel studied so far contain nothing sode. It is not immediately obvious why the coarser than granule-size gravel. Thus, the Vaal should have cut down 60 m or more Vaal gravels do require a stream compe- into resistant rock between late Pliocene tence that is quite unusual if not outside and mid-Pleistocene times, with minimal the range of recent variability. change thereafter. Surely the extreme Relevant to simple tectonoerosional inter- stepped gradients of the Delportshope-War- pretation of the Vaal terraces, there is a renton sector, if indeed applicable as the gathering body of evidence that a system- basic explanation, should have continued to atic terrace sequence extends from the provide ample nonclimatic stimulus to Orange-Vaal confluence to upstream of "grading" of the longitudinal profile. Windsorton. This sequence transcends a Platform cutting and planation.-A prob- variety of discontinuities, both in lithology lem that had been given little attention is and gradient. Thus, the terrace system im- the extensive development of fluvial plat- plies long periods of general geomorphic forms and marginal pedimentation surfaces, stability alternating with phases of dyna- compared with a remarkably discontinuous mism (in the sense of Erhart 1956), ruling

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms ALLUVIAL TERRACES OF THE LOWER VAAL RIVER, SOUTH AFRICA 359 out the knickpoint and piracy theories of erating effects of successive dams upstream terrace development. Furthermore, the same since the 1930s, the Riverton flood-silt ter- terrace sequences now appear to maintain races nonetheless seem to fall within the comparable mean gradients along a river modern range of variation. The facies and stretch in excess of 250 km, making epeiro- distribution of the silty basal units (Mem- genic deformation increasingly unlikely as bers I and II), in particular, imply a much a primary factor. Finally, the interdigita- broader floodplain with well-developed tion of "tributary gravels," lateral colluvial backswamps or sloughs in the past, subject rubbles, and tributary-channel boulders to widespread seasonal inundation and ac- with the basic Vaal gravels all indicate un- cretion of fine, suspended sediments. It usually vigorous slope denudation and local would appear that erosion/weathering ra- runoff during at least some of the phases of tios and runoff on the watershed, as well as aggradation. Since comparable slope rub- discharge and sediment relationships within bles nowhere overlie the Riverton Forma- the river, were quantitatively different from tion, a change of environmental parameters those of today, and qualitatively distinct must be posited. from those applying to times of gravel The field evidence in the Douglas-Wind- aggradation. sorton area further suggests that the Lower Alluvial and eolian sands.-Massive al- Vaal and its tributaries responded to local luvial sands, such as the "Calcified Sands" changes in geomorphic equilibrium and run- of Windsorton and those of Member III off, not to changing discharge of the distant of the Riverton Formation are difficult to Vaal High Veld drainage. If this is fully explain. Although detailed facies relation- verified by the systematic study of facies shipsin at Windsorton remain to be estab- the different gravel bodies, then it should lished, Riverton exposures indicate that be possible to develop reasonable models of sands was common to Vaal and tributary (1) vegetation mat, (2) rainfall intensity, channels as well .as to local colluvia. The duration, and seasonality, (3) runoff fac- last include the terrestrial snails washed tors, and (4) balances between rates of into sedimentary contexts after mass deaths erosion and weathering that could have of fa- single-age populations, presumably in vored such gravel transport and deposition response to severe (seasonal?) desiccation. (see parallel discussion in Butzer and Grain-size spectra of these moderately Helgren 1972). sorted sands peak between 50 and 200 4, Alluvial silts.-Suspended sediments are with a trace of coarse sands and granules characteristic of Vaal alluviation during the in most samples. Quartz sands above 100 4 present century in particular, and the Holo- are generally subrounded to rounded in cene (e.g., Members IV and V of the River- shape and, except for the absence of ferric ton Formation) in general. The most com- skins, are indistinguishable from so-called mon facies is a clayey silt deposited within Kalahari-type eolian sands (see Van Roo- low-velocity channel sectors, embanked yen 1971, chap. 5; Piaget 1963). Frosting against channel berms, and, very locally, properties, unduly emphasized by Bond flooding into low-lying areas such as tribu- (1954) in an earlier study of eolian com- tary mouths. At the Vaal-Orange conflu- ponents in Zambezi River sands, are incon- ence, there is a prominent contrast of facies; clusive since surface micropitting varies in the Orange is rapidly depositing a sandy siltpart according to the chemical environment (with a strong maximum in the 20-100 E of the depositional medium (Kuenen and grade) that is building up into a prominent Perdok 1962; Butzer and Gladfelter 1968). spit, behind which the Vaal slowly accumu- Polygenetic, reddish sands of Kalahari-type lates clayey silts (almost exclusively under are prominent both at Riverton and Wind- 50 f). Although this trend to fine sedimen- sorton, where they constitute several gen- tation has been accentuated by the mod- erations of dune sands or colluvial wash.

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms 360 K. W. BUTZER, D. M. HELGREN, G. J. FOCK, AND R. STUCKENRATH

Such sands are known to lose their oxide known so far are either secondarily cal- coatings soon after being blown into rivers creted or involve relict soils under no more or seasonal lakes. Altogether we believe that than a thin colluvial veneer. Interpretations an appreciable part of the sand component of these soils must therefore await comple- in the Riverton Formation was derived ei- tion of the final fieldwork. ther from existing eolian deposits by col- Faunal evidence.-The mixed Vaal fauna luvial action or possibly by direct eolian includes black rhino, hipparion, several sedimentation over permanent streams and zebrine and caballine horses, hippo, several standing waters. This implies that eolian suines (including warthogs), a giraffid, a sands were particularly abundant during variety or of hartebeests and wildebeests, an just prior to accumulation of Riverton extinct eland, a duiker, impala, springbok, Member III, although a large, expanded gazelle, a hippotragine antelope, greater non-outlet lake in the Kimberley area in- kudu, oryx, two genera of buffalo, several dicates a very moist climate at about the elephants, and spotted hyena (see Cooke same time (Butzer et al. 1973). 1963; Wells 1964). This basic faunal as- Paleosols.-Buried and relict soil phe- semblage, no matter how archaic and mixed, nomena in the Lower Vaal region include corresponds closely with the regional fauna (1) massive calcretes cementing primary or of the nineteenth century as well as with reworked gravels, (2) silcrete duricrusts the spectrum of animals depicted by late in some exposures of the Younger Gravels, prehistoric rock engravings in the general (3) deep, reddish cambic horizons in pri- area (Fock 1966, 1972). In the context of mary or reworked eolian deposits as well the Ethiopian faunal province it has little as many gravel exposures, and (4) localized ecological import. vertisols, primarily in Riverton sediments. Holocene vertisols are commonly encoun- CONCLUSION tered in Riverton floodplain deposits of the The recent field studies reported here western Orange Free State and they pre- provide a new stratigraphic framework sumably indicate slow geomorphic change, (fig. 5) for the Older and Younger Gravels, reduced proportions of runoff, and more ef- and for deposits of the subsequent time fective moisture (Butzer 1971a). Calcrete range. Furthermore, basic interpretation of chemistry has been summarily discussed by the several alluvial formations requires con- Goudie (1972); possible modes of origin of siderable revision. Although specific paleo- the Windsorton calcretes have been out- environmental interpretations are prema- lined by Netterberg (1969b). In particular, ture at this time, the present study does the calcification of the primary gravels may underscore that ongoing field and labora- well be related to the periodic oscillations tory studies will continue to provide new of a water-table, falling subsequent to ag- gradation. The calcreted colluvial mantles, and substantive data on a stratigraphic se- on the other hand, appear to be more strict- quence truly rich in implications for the ly pedogenic phenomena. In general, cal- Pleistocene record of southern Africa. cretes are found in a wide variety of arid to semiarid South African environments ACKNOWLEDGMENTS.-The first-named au- and their presence in the Lower Vaal region thor (K. W. B.) is grateful to J. D. Clark has no immediate paleoclimatic implica- (Berkeley), G. H. Cole (Field Museum), H. B. S. Cooke (Dalhousie University), G. L1. tions. The same would appear to apply for Isaac (Berkeley), and R. G. Klein (University the more difficult to interpret silcretes (see of Washington) for information, suggestions, Butzer 1973b). Finally, the red paleosols or discussion; A. B. A. Brink (Witwatersrand) are difficult to evaluate since few intact and T. C. Partridge (Loxton, Hunting Techni- profiles are preserved and all examples cal Surveys) kindly introduced K. W. B. to

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms ALLUVIAL TERRACES OF THE LOWER VAAL RIVER, SOUTH AFRICA 361 their study area at Windsorton in 1969. The gren is being aided by a Ford Foundation fieldwork and laboratory studies (in the Paleo- stipend in archeology awarded by the Univer- ecology Laboratory, University of Chicago) sity of Chicago. The maps and diagrams were were supported by the Wenner-Gren Founda- drawn by C. Mueller-Wille. The C14 dating tion (grant-in-aid 2344, to Butzer and Klein); was done at Radiation Biology Laboratory of the Anthropology Department of the Univer- the Smithsonian Institution. This paper is pub- sity of Chicago; and by NSF grant GS-3013, lished with the approval of the secretary of the to Klein and Butzer. Current fieldwork of Hel- Smithsonian Institution.

REFERENCES CITED

AcocKS, J. P. H., 1953, Veld types of the Union of Cape St. Francis, South Africa: Quaternary Re- South Africa: Bot. Survey South Africa Mem. search, v. 2, p. 143-169. 28, 192 p. COLE, G. H., 1961, Culture change in the Middle- BOND, GEOFFREY, 1954, Surface textures of sand Upper Pleistocene transition in Africa south of grains from the Victoria Falls area: Jour. Sed. the Sahara: Unpub. Ph.D. dissert., Univ. Chi- Petrology, v. 24, p. 191-195. cago. --- 1967, River valley morphology, stratigra- COOKE, H. B. S., 1941, A preliminary survey of the phy and palaeoclimatology in southern Africa, in Quaternary period in southern Africa: Bur. Ar- BISHOP, W. W., and CLARK, J. D., eds., Back- chaeology, Pretoria, Archeol. Ser., v. 4, p. 1-60. ground to evolution in Africa: Chicago, Univ. --- 1947, The development of the Vaal River Chicago Press, p. 303-312. and its deposits: Geol. Soc. South Africa Trans., BREUIL, H.; VAN RIET LOWE, C.; and DU TOIT, v. 46, p. 243-260. A. L., 1948, Early man in the Vaal River Basin: --- 1949, Fossil mammals of the Vaal river de- Archeol. Survey South Africa, Pretoria, Archeol. posits: Geol. Survey South Africa Mem. 35, pt. 3, Ser., v. 4. 117 p. BUTZER, K. W., 1971a, Fine alluvial fills in the --- 1963, Pleistocene mammal faunas of Africa, Orange and Vaal Basins of South Africa: Assoc. with particular reference to southern Africa, in Am. Geographers, Proc., v. 3, p. 41-48. HOWELL, F. C., and BOURLIERE, F., eds., African --- 1971b, Another look at the australopithe- ecology and human evolution: Chicago, Aldine- cine cave breccias of the Transvaal: Am. Anthro- Atherton, p. 65-116. pologist, v. 73, p. 1197-1201. - 1967, The Pleistocene sequence in southern --- 1971c, Environment and archeology: an Africa and problems of correlation, in BISHOP, ecological approach to prehistory: Chicago, Al- W. W., and CLARK, J. D., eds., Background to dine-Atherton, 703 p. evolution in Africa: Chicago, Univ. Chicago --- 1973a, Spring sediments from the Acheulian Press, p. 175-184. site of Amanzi, Uitenhage District, South Africa: --- 1973, The fossil mammals of Cornelia, Quaternaria, v. 16 (in press). O.F.S., South Africa: Natl. Mus. Bloemfontein --- 1973b, Geomorphology and geo-archeology Mem. (in press). of two Acheulian 'Pan' sites in the Kimberley DAVIS, W. M., 1906, Observations in South Africa: District: Doornlaagte and Rooidam: Cape Pro- Geol. Soc. America Bull., v. 17, p. 377-450. vincial Mus. (Nat. Hist.) Annals (Grahams- DEACON, H. J., 1970, The Acheulian occupation at town) (in press). Amanzi Springs, Uitenhage district, Cape Prov- --- 1973c, Geology of the Cornelia Beds, north- ince: Ann. Cape Prov. Mus. (Nat. Hist.) Annals, eastern Orange Free State: Natl. Mus. Bloem- v. 8, no. 11, p. 89-189. fontein Mem. (in press). DIXEY, FRANK, 1938, Some observations on the --- 1973d, Paleo-ecology of South African physiographical development of central and australopithecines: Taung revisited: Current southern Africa: Geol. Soc. South Africa Trans., Anthropology (in press). v. 41, p. 113-170. ---; Focx, G. J.; STUCKENRATH, A.; and Du ToIT, A. L., 1907, Geological survey of the ZILCH, A., 1973, Paleo-hydrology of late Pleisto- eastern portion of Griqualand West: Geol. cene lakes in the Alexandersfontein Pan, Kim- Comm. Cape of Good Hope (Cape Town) llth berley, South Africa: Nature (in press). Ann. Rept. (1906), p. 87-176. -, and GLADFELTER, B. G., 1968, Quartz-grain --- 1907-1908, Geological map of the Colony micromorphology, in BUTZER, K. W., and HAN- of the Cape of Good Hope: Cape Town, Govern- SEN, C. L., Desert and river in Nubia: Madison, ment Printer Sheets 46 and 42, at 1:238,000. Univ. Wisconsin Press, p. 473-481. ERHART, M., 1967, La genese des sols en tant que -, and HELGREN, D. M., 1972, Late Cenozoic phenomene geologique (2d ed.): Paris, Masson evolution of the Cape Coast from Knysna to [1st ed. 1956].

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms 362 K. W. BUTZER, D. M. HELGREN, G. J. FOCK, AND R. STUCKENRATH

FLINT, R. F., 1959, Pleistocene climates in eastern calcrete types: South African Archeol. Bull., v. and southern Africa: Geol. Soc. America Bull., 24, p. 117-122. v. 70, p. 343-374. PARTRIDGE, T. C., 1969, Fluvial features and cli- FOCK, G. J., 1960, Die Pri-Chelles-Kultur am Vaal, matic change during the Quaternary in South in Steinzeitfragen der alten und neuen Welt, Africa: South African Archeol. Bull., v. 24, p. Festschrift fiir L. Zotz: Bonn, Rohrscheid, p. 106-116. 151-160. --, and BRINK, A. B. A., 1967, Gravels and ---- 1966, Distribution of animals on rock-en- terraces of the lower Vaal Basin: South African gravings in the Northern Cape: Cape Provincial Geog. Jour., v. 49, p. 21-38. Mus. (Nat. Hist.) Annals, v. 5, p. 85-90. PIAGET, J. E. H., 1963, Coarse fraction mineralogy --- 1972, Wild animals in rock engravings: and granulometry of selected soils of the West- South African Mus. Assoc. Bull., v. 10, p. 2-11. ern Orange Free State: Unpub. D.Sc. thesis, GOUDIE, A. S., 1972, The chemistry of world cal- Univ. Orange Free State. crete: Jour. Geology, v. 80, p. 449-463. RUST, UWE, 1970, Beitrige zum Problem der In- HUZAYYIN, S. A., 1941, The place of Egypt in pre- selberglandschaften aus dem mittleren Siidwest- history: Inst. d'Egypte Mem., v. 43, 474 p. afrika: Hamburger Geog. Studien, v. 23, p. 1- KING, L. C., 1963, South African scenery: a text- 280. book of geomorphology: Edinburgh, Oliver & SCHULZE, B. R., 1958, The climate of South Africa Boyd. according to Thornthwaite's rational classifica- --- 1968, The morphology of the earth, a study tion: South African Geog. Jour., v. 40, p. 31-53. and synthesis of world scenery (2d ed.): New SOHNGE, P. G.; VISSER, D. J. L.; and VAN RIET York, Hafner. LOWE, C., 1937, The geology and archeology of KUENEN, PH. H., and PERDOK, W. G., 1962, Ex- the Vaal River Basin: Geol. Survey South Afri- perimental abrasion 5. Frosting and defrosting ca Mem. 35, pts. 1 and 2, 184 p. of quartz grains: Jour. Geology, v. 70, p. 648- VAN RIET LOWE, C., 1952, The Vaal River chro- 658. nology-an up-to-date summary: South African MAARLEVELD, G. C., 1960, Uber die pleistozinen Archeol. Bull., v. 7, p. 135-149. Ablagerungen im siidlichen Afrika: Erdkunde, VAN ROOYEN, T. H., 1971, Soils of the Central v. 14, p. 35-46. Orange River Basin: Unpub. D.Sc. thesis, Univ. MAGLIO, V. J., 1970, Early Elephantidae of Africa Orange Free State. and a tentative correlation of African Plio-Pleis- WELLINGTON, J. H., 1937, The Pre-Karroo pene- tocene deposits: Nature, v. 225, p. 328-332. plain in the south-central Transvaal: South MASON, R. J., 1962, Prehistory of the Transvaal, African Jour. Sci., v. 33, p. 281-295. a record of human activity: Johannesburg, Wit- --- 1955, Southern Africa: a geographical watersrand Univ. Press. study, v. 1, Physical geography: Cambridge, - 1967, The archaeology of the earliest sur- Cambridge Univ. Press. ficial deposits in the lower Vaal Basin near Hol- WELLS, L. H., 1964, The Vaal River "Younger pan, Windsorton District: South African Geog. Gravels" faunal assemblage, a revised list: South Jour., v. 49, p. 39-56. African Jour. Sci., v. 60, p. 91-93. --- 1969, The Oppermansdrif Dam archaeo- WOLMAN, M. G., and MILLER, J. P., 1960, Magni- logical project: Vaal Basin: South African Ar- tude and frequency of forces in geomorphic cheol. Bull., v. 24, p. 182-192. processes: Jour. Geology, v. 68, p. 54-74. NETTERBERG, F., 1969a, Ages of calcretes in south- ZEUNER, F. E., 1959, The Pleistocene period (2d ern Africa: South African Archeol. Bull., v. 24, ed.): London, Hutchinson [1st ed. 1945]. p. 88-92. --- 1958, Dating the past (4th ed.): London, _--- 1969b, The interpretation of some basic Methuen.

This content downloaded from 128.83.56.173 on Tue, 18 Jul 2017 14:18:25 UTC All use subject to http://about.jstor.org/terms