ANTHROPOLOGICAL RECORDS Volume 28 MONTAGU CAVE IN PREHISTORY: A DESCRIPTIVE ANALYSIS BY r CHARLE M. KELR UNIVERSITY OF CALIFORNIA PRESS MONTAGU CAVE IN PREHISTORY: A DESCRIPTIVE ANALYSIS A' le a r ?i- .y ?( , _A i v p 4 lv? . J 0 i . ! ?, , p 0 O' Al I i #11 I : 10 i # v - 00, I 11 ;so 4 N 4 t 0 A ","t JOOZ 46 MONTAGU CAVE IN PREHISTORY: A DESCRIPTIVE ANALYSIS BY CHARLES M. KELLER ANTHROPOLOGICAL RECORDS Volume 28 UNIVERSITY OF CALIFORNIA PRESS BERKELEY * LOS ANGELES * LONDON 1973 UNIVERSITY OF CALIFORNIA PUBLICATIONS ANTHROPOLOGICAL RECORDS Advisory Editors: J. B. Birdsell, E. A. Hammel, R. F. Heizer, C.W. Meighan, H. P. Phillips, A. C. Spaulding Volume 28 Approved for publication April 23, 1971 Issued June 29, 1973 University of California Press Berkeley and Los Angeles California University of California Press, Ltd. London, England ISBN: 0-520-09401-8 Library of Congress Catalog Card No.: 79-635565 O 1973 by The Regents of the University of California Manufactured in the United States of America To Dr. Bertram McGarrity, who introduced me to an interest in form PREFACE No archaeological project can be conducted without the cooperation and assistance of many people, and this is particularly true of a project in a foreign country. For their help and interest I am indebted to Mr. Nico Kriel of Derdeheuval, South Africa; Mr. R. R. Inskeep, School of African Studies, University of Cape Town; Dr. A. 0. Fuller, University of Cape Town; Dr. Thomas Barry and the staff of the South African Museum, Cape Town; Mr. B. D. Malan, Secretary of the Commission for the Preservation of Natural and Historical Monuments, Relics, and Antiques; Dr. A. R. Hall, Bolus Herbarium, University of Cape Town; Mr. Michael Wells, Botanical Research Institute, Albany Museum, Grahamstown; Mr. and Mrs. Jalmar Rudner, South African Museum, Cape Town. For their sustained encouragement and advice I would like to express my gratitude to Dr. J. Desmond Clark and the late Dr. T. D. McCown. The help and patience of Miss Jay Adams and the Survey Research Center, Data Processing Services Facility, University of California at Berkeley, are also gratefully acknowledged. For their assistance in analyzing the sediment samples from the cave, I would like to thank Dr. Herbert Glass of the Illinois State Geological Survey, Urbana; Dr. Karl Butzer, Department of Anthropology, University of Chicago; Dr. Robert Jones, Department of Agronomy, University of Illinois; and Mr. Gary Horlick, Department of Chemistry, University of Illinois, Urbana. Dr. Charles Alexander and Dr. Alan Peshkin, Directors of the African Studies Committee of the University of Illinois were instrumental in providing funds for typing, and their assistance and that of the entire committee is gratefully acknowledged. Dr. Glynn Isaac kindly put at my disposal his data on the collection from the 1919 excavation. Financial support for the project was provided by the National Science Foundation and the Committee on Research, University of California, Berkeley. The type of excavation that was done, and the level of excellence at which it was performed, was possible only through the efforts of my crew. Their unflagging enthusiasm and energy over seven months is sincerely appreciated. I am indebted to Peter Saunders, Cedric Poggenpoel, and Andrew van Turha. For his help in describing the artifacts in Cape Town I would like to thank Peter Saunders. Finally, for her aid in cataloging, illustrating artifacts, editing, typing, and general encouragement, I thank my wife, Bonnie B. Keller. Charles M. Keller May 22, 1972 CONTENTS Preface . Introduction. The Environmental Setting ....... Stratigraphy and Method of Excavation Description and Discussion of Assembla Introduction. Definitions. Layer 1 Description Layer 1 Discussion Layer 2 De scription Layer 2 Discussion Layer 3 Description Layer 3 Discussion Layer 5 Description Layer 5 Discussion Comparisons. Layer 1 ....... Layer 2 ....... Comparisons of Laye Comparisons with Otl Conclusions. . . . . . . . . . . . . lges . . . . . . . . . . . . . . * . . . . . . * . . . . * . . . . . . * * . . * * . . . * @ * s s . . . . . * * * - s . . . . . . . . .* . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ,rs 3 and 5 . he r Sites s Appendix I: A Descriptive Classification for the East African Late Acheulian Assemblage . . Appendix II: A Provisional Interpretation of the Sedimentary Sequence from Montagu Cave (Cape Province), South Africa by Karl W. Butzer .................. Appendix III: Analysis of Botanical Specimens from Feature 3 .................. Bibliography. Abbreviations. Figures 1-53 Plate s I- LJI TABLES Table 1. Correlations of Raw Material with Selected Attributes of Layer 1 Scrapers. Table 2. Distribution of Tools, Cores, and Grindstones from Features in Layer 1 ...... Table 3. Correlations of Raw Materials with Selected Attributes of Layer 2 Scrapers . ... Table 4. Correlations of Raw Material with Selected Attributes of Layer 2 Trimmed Flakes Table 5. Tools and Cores from Layer 2. Table 6. Correlations of Size with Selected Attributes of Layer 3 Scrapers s......... Table 7. Tools and Cores from Layer 3 ............................. Table 8. Tools and Cores from Layer 5 ............................. Table 9. Hand-axes from 1919 Excavation ............................ Table 10. Hand-axes from the Transvaal Earlier Stone Age ................... vii 4 7 14 14 16 17 24 26 38 42 54 56 67 70 70 71 74 76 80 81 89 94 95 98 99 131 25 25 39 39 40 54 55 68 75 78 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . MONTAGU CAVE IN PREHISTORY: A DESCRIPTIVE ANALYSIS BY CHARLES M. KELLER INTRODUCTION recent years interest and research into the Acheu- n of Africa have progressed to the point that our owledge about this Industrial Complex is as complete precise as is that about any body of material on continent. Extensive work has been done in East East Central Africa at sites such as Isimila, orgesailie, Olduvai Gorge, and Kalambo Falls, and southern Africa important Acheulean sites at Doorn- te, Amanzi, and the Cave of Hearths have been vestigated. Although much more is known of the later final stages of the African Acheulean than of the ly and middle stages, some clear patterns have erged. Most notable of the results deriving from s research is the demonstration of considerable ation within the Acheulean Industrial Complex. ere the artifacts are present in a primary archaeo- ical context quite striking differences have been und in the inventory of the tool types present as eli as in the ratio of tools to debitage. These differ- ces are not explainable in terms of separation in e or space, and the basic resemblance of the occur- nces is ample evidence to support the interpretation the various occurrences represent parts of a igle whole. The situation was summarized by Howell d Clark (1963), and, following Kleindienst (1961), iey suggested that the differences within the Acheu- i1ean reflect the performance of different activities at Ndifferent sites. Variation at another level of analysis has been re- orted by Isaac (1969) from Olorgesailie. There he Feels that differences in the plan-forms of bifaces may e due to differences in stylistic preferences between ocial groups. And, finally, an increase in chronologi- pa control of the contexts that produce Acheulean Wgregates has shown that morphological differences 1 exist within the span of time represented by sites such as Olduvai Bed IV or Isimila. It is clear then that only one facet of the variation that exists within the Acheulean Industrial Complex is represented by the order of variation described by Kleindienst in 1961, but it will be our focus in the following discussion of the Acheulean artifacts from Montagu Cave because it is this sort of variation that appears when the Montagu and East African aggregates are compared. Much of the material from the East African sites has been described in terms of a system of categories defined by Kleindienst (1962) after her extensive analysis of East and Central and some South African Acheulean material, and this has facilitated comparisons between the sites in East Africa. However, it is only about 600 miles from Olorgesailie to Kalambo Falls, and, given the extremely wide area over which Acheulean material has been found, it is important to determine whether the kinds of phenomena reflected in the East African Late Acheulean were present in other parts of Africa, and also whether there were significant spatial differ- ences within the Late Acheulean. Unfortunately, the areas outside of East Africa from which the Acheulean is known in primary archaeological context are few, and in the one region where appropriate sites are known, South Africa, the published descriptions of the occur- rences are phrased in a way that precludes close com- parisons with the East African material. This report is an account of the excavation and the description and analysis of the artifacts from Montagu Cave situated in South Africa (200 10' E., 330 50' S.), some 2,000 miles south of the area in East Africa where the other well-known Acheulean sites are located (fig. 1). The primary objective of the work carried out I Anthropological Records at Montagu Cave was to recover a full lithic assem- blage from a sealed archaeological context and to ana- lyze this material using the type categories and termi- nology employed by Kleindienst (1962) and Howell and Clark (1963). Montagu Cave was suggested by Dr. J. Desmond Clark as a suitable site for this investigation because it had been partially excavated in 1919, and that exca- vation had shown that the cave contained Acheulean artifacts, as well as some later material (Goodwin, 1929). Work was begun at the cave on August 10, 1964, and continued until March 1, 1965. The crew consisted of myself and three Cape Coloured teenage boys, Peter Saunders, Cedric Poggenpoel, and Andrew van Turha. My wife supervised the cataloguing of the excavated material and the succession of small boys who washed artifacts after school. After the excavation was com- pleted, the artifacts, approximately 275,000 of them, were transported to the South African Museum in Cape Town where they were described by Peter Saunders and myself, and where my wife illustrated the artifacts shown in plates I to LIII. We left Cape Town on August 16, 1965, and returned to Berkeley, where the analysis was done between September 1965 and August 1966. An important aspect of the project was to describe as fully as possible a South African Acheulean assem- blage. Since the majority of the Acheulean artifacts in Africa, including those in South Africa, have come from geological rather than archaeological contexts, Montagu Cave is very important for providing the material for such a description. Sites of this period are much less common than those of later periods. It was known from the previous excavation at Montagu that there was more than one Acheulean stratum in the cave, and this would provide an opportunity to investigate chronological variation in the Acheulean at a single site in South Africa. Although the project had been designed to focus on the Acheulean material at Montagu, as the excavation progressed it became apparent that there was a con- siderable amount of post-Acheulean material present. In fact, nearly half of the artifacts recovered in the excavation were post-Acheulean. These latter assem- blages are described in the following pages, but be- cause of the inadequate state of knowledge about simi- lar assemblages the comparisons made between Mon- tagu and other sites are less elaborate than those for the Acheulean. Another objective of the project was to investigate the horizontal spread of artifacts on occupation sur- faces or horizons, if such could be found in the cave. Occupation horizons have been found, for the most part, on open sites, and it was hoped that if this kind of information could be recovered from a cave site it would contribute to our knowledge of the way pre- historic men used their habitation places. Acheulean material has not been found in many South African cave sites, with the possible exception of the Wonder- werk cave in the northern Cape Province and Olie- boompoort and Wonderboom in the Transvaal. The best-known site, the Cave of Hearths, has a brecciated deposit. It is obvious that the problems of excavating a heavily cemented deposit make it very difficult to recover information about the arrangement of artifacts on an old surface, and for this reason Montagu Cave, with its relatively unconsolidated deposit, was ideal. Occupation surfaces are of great interest to archae- ologists, not only because of the information-they can provide about horizontal distribution of artifacts, but also because they represent units of contemporaneity that are extremely useful for investigating variations within a single site. If, for instance, a site was occu- pied over a period of time by groups performing dif- ferent tasks, such as butchering, stone tool manufac- ture, and hide preparation, the artifacts associated with these activities would be left on the surface occu- pied by each group. If these surfaces were contained in a single stratum and that stratum excavated as a single unit, the different assemblages associated with each activity would be mixed. The characteristic assem blages could be kept separate only if the surfaces were excavated separately. At Montagu, in the strata in whicd we were able to isolate individual occupation surfaces, there was no evidence of this kind of variation within a stratum, but this could not have been demonstrated unless we had taken the time to isolate and excavate these surfaces. This is an interesting point, because at Isimila and Olorgesailie in East Africa, there was evidence of considerable assemblage variation within a fairly small area. As will become clear, Montagu was primarily a factory site and was visited for this purpose over a long period of time. The Acheulean material has been described in terms of Kleindienst's (1962) typology in order to facilitate comparisons with other Acheulean assemblages. In addition to listing the numbers of examples present of a particular type, the descriptions that follow in- clude other relevant information about the artifacts in question. This was done in order to point out the amount of subtypical or intratypical variation that was- present. A given type of artifact can be characterized in terms of a combination of attributes, but it is un- likely that all representatives of the type will be iden- tical, and it was these minor variations that we were interested in. It has been stated frequently that human 2 I vior is patterned, but that the patterns that are owed are not rigid and allow some variation within turally prescribed limits. With regard to the Acheu- Clark (1960:315) has made the distinction of ormal" and "informal" tools. These designations fer to fairly standardized forms such as hand axes, the one hand, and to more amorphous forms such scrapers, on the other. A distinction of this sort gests that there are variations in the mental pat- of the tool-makers and that a formal tool is the nifestation of a more complex and highly integrated ttern whereas an informal tool represents a simpler ttern. In the descriptions that follow, a measure of variation can be seen in the kinds of attributes esent and in the frequencies with which they occur. weakness of the typological method when applied ethnographic or archaeological data is that it em- sizes the large patterns of behavior but obscures e smaller variations that are likely to be present. The objectives of ethnography are to observe behavior d from these observations to infer the rules of bavior a people follow. The same kind of procedure nbe followed in archaeology. From the descriptions the artifacts recovered from a site, one should able to outline the rules that were followed with gard to the production of tools. There are obvious ficulties in this approach, since the archaeologist never be certain of the individual formal charac- ristics the makers of the tools felt were important, t even so, some important information of this sort a be gained. For instance, Kleindienst (1961:40) 3 refers to "flake scrapers" as a general tool category; yet of the scrapers that would seem to fall into this category, in Montagu Layer 3 only 33.8 percent of the scrapers are made on flakes and in Layer 5, only 21.4 percent; the remainder are made on other pri- mary forms. There are many possible interpretations of facts such as these, but the important point is that simply referring to a given number of "scrapers" gives no information such as the primary form on which they are made or where on the object the working edge is placed. As our knowledge increases, it is just this kind of information that will aid in making finer chron- ological, spatial, and activity distinctions between assem- blages and thus contribute to a much fuller understand- ing of the behavior of the human groups involved. Yet another objective of this study was to investigate the alleged association of fire with the Acheulean at Montagu. Goodwin (1929) is not clear about whether there was evidence of the use of fire, but Montagu has been cited by Oakley (1957:386) as one of the places in Africa where fire is associated with Acheulean material. The presence or absence of fire at Montagu is relevant to the oft-repeated idea that man could not occupy caves until he had fire. No further mention will be made of this facet of the project because no evidence of fire was found in either of the Acheulean layers at Montagu. There was no charcoal, nor were they any fireplaces like those found in the post-Acheulean layers. Subsequent chemical analysis may provide additional information, but at this point there is nothing to show that fire was used at Montagu Cave by the makers of Acheulean tools. Keller: Montagu Cave in Prehistory THE ENVIRONMENTAL SETTING Africa resembles a large shield, consisting of a broad flat central plateau fringed by coastal lowlands. The edge of the plateau is usually a scarp and the lowlands may be narrow or more than two hundred miles wide as in Mocambique. These two features, the plateau and the lowlands, have an important effect on the drainage, climatic, floral, and faunal patterns of the continent. The Republic of South Africa has been divided by Cole (1961) into four major physiographic regions, which are named from both their surface and subsur- face features. The two areas that lie on the central plateau are called the Karroo and Pre-Karroo pro- vinces because they are formed on bedrock of Karroo and Pre-Karroo age respectively. The other two re- gions lie south of and below the escarpment; one area is subdivided into the Eastern and Western Marginal provinces, and these are connected by the fourth, the Cape Folded province, with which we will be most con- cerned. The Montagu Cave, situated near the town of Montagu, lies in the western end of a long narrow valley called the Little Karroo, which makes up part of the Cape Folded province. Stretching from east to west across the southern part of the continent, this province is made up of parallel ridges and valleys. To the north lies the dry Karroo, and to the south lies the coastal foreland and the Indian Ocean. The northern boundary of the Little Karroo is the Swartberg range, part of the escarpment, and the southern boundary, the Langeberg range; the distance between them is about forty miles. These ranges, as well as the other ranges of the region, are made of folded Table Mountain Sandstone, and the valleys are eroded into Bokkeveld shales. In general, the topography of the area is very regular, with the mountains separated by smooth open valleys. The effect is a series of long corridors running from the Cape Flats on the west to the vicinity of Port Alfred, 500 miles to the east. The mountains of the Cape Folded province are joined in the west by the southern end of a series of ranges that extend to the northwest, and the combination makes the topography of the im- mediate area very complex. In the east the valleys are open to the coast. Most of South Africa receives its rainfall during the summer months (December-February), but the Cape Folded province is an exception, receiving most of its rain during the winter (June-August). In winter most of the country is dominated by a high-pressure system, but the southern coastal area is affected by stormy westerly winds, which at this time swing to the north and bring rain to the region (Fitzgerald, 1961:33). The situation is summarized by Schumann: When the winter anticyclone is established over the plateau there is an outflow of air over the coastal belt; along the west coast this is associated with berg winds and very high temperatures, which in turn cause the development of a low. . . . This lowi unites with the circumpolar low to form an inverted "V' shaped depression which moves eastwards alon4 the south coast . . . bringing the unsettled weather, strong north-westerly winds and rainfall associated with depression, to each part of the coast in turn. (Cole, 1951:43) The weather, moving from the southwest, drops most of its moisture on the coastal belt and on the seaward side of the mountains. Consequently, the Little Karroo.i is much drier than the coastal foreland only a few miles to the south on the other side of the mountains, Caledon, on the coastal strip, has a mean annual rain.m fall of 540 mm. whereas that of Montagu is 282 mm.. (Jackson, 1961:plate 8). The mean daily temperature in Montagu ranges from 5.0? C. in July to 17.50 C. in January (Jackson, 1961:plates 29-34). There are three characteristic floras of the south- ern part of South Africa, and their origins and relatiol have been discussed by Levyns (1964). The oldest flor which requires the most rain, is the Temperate Fore8 Flora found in the Knysna and Humansdorp areas, whe the mean annual rainfall is 450 to 650 mm., and also in other areas, where it is confined to ravines and other sheltered places. Large trunks of Podocarpus falcatus that were found buried in the sand of the Cap Flats have been dated by carbon 14 to from 30,000 to 40,000 years ago; this, coupled with the present disco tinuous distribution of forest, indicates that the tem- perate forest was once much more widespread than it is at present. Levyns feels that dry periods were un- favorable for forest and that the Cape Flora replacedi The Cape Flora is made up of shrubs and larger types such as Proteaceas and Leucadendron. It is suit to moderately high winter rainfall and dry summers and is found in the area of the Cape Peninsula and along the coastal forelands south of the mountains. North of the mountains, where the rainfall drops below 250 mm. per year and is seasonally erratic, youngest of the three floras, the Succulent Karroo Fl [ 4 ] I ,i I Al 4 ?j i1i .1 i Keller: Montagu Cave in Prehistory Wds optimum conditions. The 250 mm. per year wndary appears critical for the Cape and the Succu- at Karroo Flora, and rainfall in excess of this re allows the former to exist whereas less rain roduces the latter. As a result, the Cape Flora is ten found on the more moist summits of hills, sur- Funded by the Succulent Flora on the drier, lower [cpes. This discontinuous distribution is interpreted 1 indicating that the Succulent Flora occupies areas hce the territory of the Cape Flora. The vegetation of the Montagu area is the type scribed by Keay as "Cape Macchia," which is "com- Bed of evergreen shrubs with hard leathery leaves, ,nerally small and often heath-like. Many of the bs contain oil or resin and have a brownish or eyish appearance. Trees are rare and grassland urs only spasmodically. It is a 'Mediterranean' erophyllous type of vegetation" (Keay, 1959:8). Cole 1) says that typical sclerophyllous bush occurs y where the mean annual rainfall is 20-30 inches 8-762 mm.). A typical area includes an upper story -bushes five to eight feet tall, mainly Proteaceae, t occasional stands of silver trees (Leucadendron enteum). Below this, there is a dense layer of the al shrubs described by Keay and a ground flora of all woody plants, herbs, and geophytes (Cole, 1961: "In the drier areas the vegetation is more open less distinctly layered. Small bushes with hard leaves dominate and there are generally few plants ii ericoid leaves. . . ." (Cole, 1961:69). This state- pnt describes well the vegetation around Montagu, fthough the larger shrubs and trees are found only bar water there. It seems doubtful that the sclero- llous bush represents a climax vegetation, because [rning has "undoubtedly retarded the normal succes- on and it may well have contributed to the abundance bulbous plants which would be least affected" (Cole, 161:69). At the present time, with the aid of irriga- on, the farmers of the Montagu area raise wine grapes, kricots, peaches, and pears. The history of climatic and floral changes in the ptreme southern part of Africa is not well known. Putzer's recent summary of climatic evidence illus- pates the complexity of the problem (Butzer, 1971). bie magnitude of these changes is still in question, It it is clear that they were sufficient to bring about ifts in the patterns of the floral mosaic found in lost areas, and there is no reason to suppose that he situation was different in the area with which we e concerned. It is therefore very likely that at times imewhat drier than the present Succulent Flora would ave been present around Montagu and that in wetter les patches of forest would have occurred along the water courses and in protected areas of the mountains. Drainage patterns are often important to archaeolo- gists because they suggest paths of movement for men and animals, and sources of raw material for stone- tool makers. Most of the rivers of the Cape Folded province rise on the escarpment, run through the moun- tains, and reach the sea directly. There are no large drainage systems in this area to compare with the Orange-Vaal system of the plateau. Rivers such as the Gouritz, Geelbeks, and Buffels are typical because they have their headwaters on the escarpment, run through passes in the northern ranges and into the Little Karroo, where they are joined by tributaries from that region, and then breach the southern moun- tains and flow across the coastal foreland to the sea. In considering the possible routes that prehistoric people may have followed in their movements, it is tempting to overemphasize the importance of the migra- tion routes of historic groups. Settlers started from Cape Town, moved to the north and to the east along the flat coastal strips, and then entered the more moun- tainous regions by passes through which the rivers flowed. The Little Karroo was entered by means of Robinson's Pass, which penetrates the Outeniquas Moun- tains north of Mossel Bay (see fig. 1). King (1942, 308) states that most of the passes through the mountains are incised meanders and consequently are often nar- row and steep-sided. Many passes of this kind, which would not be suitable for European settlers in ox-drawn wagons, would have been perfectly feasible routes for a group of hunter-gatherers moving on foot. We cannot, therefore, rely too heavily on historical evidence for comparison, since what appears to Europeans to be an area of difficult access may be relatively open for less encumbered groups. Movement into and out of the Little Karroo would have been easiest from the south, where the barrier is only one range of mountains, or from the east, where the mountain ranges stop near the coast. From the west, that is at the Cape Flats north of Cape Town, it is necessary to cross the DuToits Berge or one of the neighboring ridges, cross the valley of the Breede River, and then cross the western end of the Lange- berge to enter the Little Karroo. The Montagu area is drained by the Groot Appelkooskloof River, which flows west through Cogmans Kloof and becomes a tributary of the Breede River. Cogmans Kloof is used by the present-day road to enter the Little Karroo, and this is the most likely route from the west. To the north there are a series of mountain ranges, and the rivers rising on the escarpment flow through a maze of passes, which again would have been possible paths for pre- historic man to have followed. Anthropological Records Drainage patterns are also of interest in considering the raw materials used by the inhabitants of Montagu Cave for their stone tools. The tools found in the lower two artifact-bearing layers of the cave are made exclu- sively of Table Mountain Sandstone, which is the rock that forms the mountains and therefore the rock in which the cave is formed. However, pieces that came directly from the cliff or cave walls were rarely used; instead, rounded cobbles were collected from the stream that runs in front of the cave. In the upper two layers of the cave deposit chert and quartz were used in addi- tion to Table Mountain Sandstone. The quartz is avail- able locally, but the source of the chert is unknown. The most plausible explanation seems to be that the chert occurred as erratics and was collected from rivers that came from the north. Because most of the rainfall comes in the winter, soil formation in the Montagu area is slight. The mois- ture is present when the temperature is too low for chemical action, but not low enough for frost action to be important, and little organic matter is contributed by the sclerophyllous bush. The soils vary in depth from a few inches to about thirty inches and "usually consist of crumbly gravelly loams containing occasional ferruginous concretions, overlying compact and imper- vious gravelly or sandy clays" (King, 1942:89). In the area immediately in front of the gorge or small ravine, known as a kloof, in which the cave is located, there seemed to be little or no soil, and the farm road to the kloof actually runs on the steeply dipping beds of the Bokkeveld shales. Hot mineral springs form a tourist attraction a few miles from the town of Montagu. Although prehistoric man may not have appreciated the therapeutic effects of the waters, this constant source of water would prob- ably have drawn game and man to the region. Montagu cave is located in the west side of a kloof on the north side of the Langeberge, which separates the Little Karroo from the coastal foreland to the south. A small permanent stream runs in the bottom of the kloof and empties out into the valley. The present-day occupant of the farm uses the stream for drinking water and irrigation. Prehistoric inhabitants probably used it both for water and as a source of raw material. There are more trees along the stream and on the west side of the kloof than are common outside on the floor of the main valley. Direct sunlight enters the cave, which faces east-northeast, until about 10 A.M. in the winter and until about 11 A.M. in the summer. The wind in the area can be quite strong, but the currents in the kloof are complex, and as a result the wind changes frequently and gusts come from many directions. The cave consists of two chambers, a large open outer chamber, in which the excavation was carried out, and a long narrow tunnel-like chamber, which opens out at the junction of the roof and back wall of the outer chamber. This rear chamber extends several hundred feet back into the mountain. During the guano- mining operations carried out there in the late nine- teenth century, an iron ladder was fixed to the back wall of the cave, making it possible to climb from the floor of the outer chamber into the inner chamber. It probably would be possible to negotiate this climb with- out the ladder, but it is not likely that this was done regularly by the prehistoric inhabitants. Archaeological material occurs only in the outer chamber. The outer chamber is approximately 55 feet deep by 35 feet wide, and the roof is about 40 feet above the present floor level. Owing to the steep cliff face above the cave, the drip line extends some 17 feet into ihe cave, and rain blows to within about 25 feet of the back wall. Probably as a result, the area of most in- tensive occupation seems to have been near the rear of the outer chamber (see fig. 2). At the mouth of the cave the beds of Table Mountain Sandstone, which despite its name is really a quartzite, X are tilted about 710 from the horizontal. The long axs of the cave follows approximately the strike of the bedS Water drips slowly but regularly from the bedding plan in the roof of the cave, and there are stained areas around the places from which the water drips, indicating that this has gone on for some time. According to Arth Fuller, lecturer in Geology at the University of Cape Town, Table Mountain Sandstone is extremely variable with regard to hardness and mineral composition, and there are many felspathic areas that are much softer than the surrounding rock (pers. comm.). The cave was apparently formed by water seeping down along the bedding planes and dissolving a more soluble portion of the Table Mountain Sandstone. This solution, and the resultant undercutting and fall of piece of the roof, formed the cave. There are portions in the; upper chamber that appear water worn, and there are small funnellike holes, always centered on a bedding plane, which run horizontally into the rock. Very fresh- looking fallen rocks are present at the rear of the uppe chamber; the cave seems to be increasing in length. The cave today, as well as in the past, is a pleasant place. In the late nineteenth and early twentieth centuri it was a popular picnicking spot. There is a constant water supply nearby and around the stream there are a variety of plants and large shrubs and trees that be used as food and as a source of fuel. In addition, game is attracted by the water. Dassie, baboon, and grysbok were seen in the kloof, and the quantity and variety of animals would presumably have been greater in the past. From the front of the cave there is a cle view of the floor of the valley, which would make it possible to observe the movements of large bodies of game. The cave is cool in the summer as well as in the winter, and it affords shelter from the wind and rain. It is easy to see why the site was occupied re- peatedly over a long period of time. 6 STRATIGRAPHY AND METHOD OF EXCAVATION tagu Cave, known locally as the Guano Cave, was reported by a European in the 1880s. Mr. C. enscroft, the discoverer, reported the existence of cave to the owner of the land, Mr. A. Kriel, who turn discovered that there were deposits of bat o in the dark, inner portion of the cave. This o was removed in the 1890s and used as fertilizer Mr. Kriel's farm. In 1919 Mr. E. J. Jansen visited the site and dis- rered on the surface "an implement of true paleo- bc form" (Goodwin, 1929:3). This discovery was re- ed in the Cape Times of May 31, 1919. Mr. Jansen several small pits but did no extensive excavation. October of 1919 Drs. S. H. Haughton and K. H. Barnard, eontologists, and Mr. Tucker, all of the South ican Museum, Cape Town, visited the site, and in -eight days during October and November they Eioved approximately three-quarters of the deposit the cave. An account of their work, and a descrip- of the stratigraphy and artifacts, was published by J. H. Goodwin ten years later (Goodwin, 1929). Four act-bearing layers were reported, the uppermost ptaining "Later Stone Age" artifacts and the three er ones containing artifacts from the "Earlier Stone The stratigraphy of the cave, as it appeared to hton and Barnard, was summarized by Goodwin 29). The uppermost layer was described as "debris," ich was thought to have been thrown down from the r cave during guano-mining operations. Layer "A" s described as the "modern surface deposit," which a composed of brown earth about a foot thick in the nter and three-fourths of an inch of swallow guano tng he sides. Below this was an artifact-bearing yer "B," which "is formed on a basal black band; r this lies a white band, and above this is ordinary wn earth, with black fire-zones and occasional thin ite bands" (Goodwin, 1929:60). Layer "C" consisted brown earth capped with swallow guano and contained artifacts. Layer "D," which contained tools, was de up of "irregular black, grey, white, and brown nds."' Layer "EE" was sterile red sand. Below "EE" the artifact-bearing layer "'F" which was composed "black, grey, and brown bands quite irregularly mixed h white bands." Another sterile sand, layer "G," nderlay layer "F." The lowest occupation layer was ," which was composed of a series of black, grey, and white bands, and below this lay the decomposing bedrock of the bottom of the cave (see fig. 3). Goodwin's map (1929) shows the location of the pits and trenches dug by Jansen, Haughton, and Barnard. This map indicated that they left a fairly large area on the south side of the cave undisturbed except for one trench, which was stopped by large boulders. How- ever, the notebook kept by Haughton and Barnard, which is in the files of the South African Museum in Cape Town, contains two maps, one of which shows the trenches that they intended to dig, and another, at the end of the notebook, which shows the trenches actually dug. It is this first map that Goodwin has reproduced, but the second and more accurate map shows that the trench into the undisturbed area, trench 3 of Goodwin's map, was never dug. Consequently, an area, 20 feet by 25 feet, of undisturbed deposit remained, and it was this area that was excavated in 1964-1965. At the site the areas that had been previously exca- vated were easily recognizable since they were much lower and relatively free from rock. The unexcavated portion had a datum depth of about .2 feet and was covered with angular rubble, whereas the excavated portion had a datum depth of about 5.0 feet and was free of rubble. Along the edge of the undisturbed area lay a series of large boulders, which the notes of the previous excavators said could not be moved without the use of explosives. These were clearly fallen rocks and had not been moved by the previous archaeological or guano-mining activities. Consequently, it was a straight- forward matter to identify the undisturbed area, to see that the published map was in error, and to select the area for excavation. The initial phase of the excavation involved two ob- jectives: first, to move the rubble and boulders off the top of the undisturbed area, and second, to cut some preliminary sections that could serve as guides for the excavation. A grid system was laid out, and work was begun in squares 20 and 25 D and 40 E (see fig. 17), while part of the crew began moving the rubble out to the dump area in front of the cave. The rubble con- sisted almost entirely of angular pieces of rock although an occasional prehistoric artifact was found, and some corks and bits of green bottle glass were present, indi- cating that the rubble belonged to a recent period. Be- low the loose rubble, a loose brown sand was encoun- tered, and the work was halted. [7] iI i 1. 8 Anthropological Records This left only the large fallen rocks to be dealt with. These were moved, by means of a block and tackle and a crowbar, into the center of the area, away from the edge and the sections that were being cut, and there were broken up with sledgehammers and wedges and carried out to the front of the cave. A pit, 45 D, was excavated near what Goodwin called "the grotto," toward the back of the cave. The original report indicated that this was near the edge of the pre- vious excavation, and it was hoped that undisturbed fill might be found, but this was not the case. The pit was later filled with rocks broken up during the course of our excavation. After these steps were completed the main excava- tion was begun. The sections had been cut primarily to provide a vertical guide from which to work. One of the most striking features about the sections was that the strata dipped much more than the section drawings published by Goodwin had indicated, and it became rapidly clear that the use of arbitrary horizontal levels as excava- tion units would be unsuitable. One such six-inch level excavated in square 25 D contained microliths from one side of the square and hand-axes from the other, and as a result we decided to remove the deposit using the obvious strata as the units of excavation. However, this raised certain methodological and terminological problems. Generally speaking, an archaeologist deals with two kinds of deposits. The first is often referred to as being "sterile," meaning that it contains no artifacts; such deposits usually have been formed by noncultural agencies such as wind or water. Descriptions of the second, or "artifact-bearing," type of deposit are often confusing. Some of these deposits contain artifacts but man has had no influence on the formation of the de- posits; river gravel containing stone tools is an example. Other deposits, such as middens, contain artifacts and are artifacts themselves, since they owe their existence to human activity. The confusion arises in discussions of the latter kind of deposit. If the deposit is homogeneous in color, compactness, grain size, or whatever other criteria the excavator feels are impor- tant, then the deposit may be excavated as a single unit. Alternatively it may be divided into subunits by the excavator; these subunits are usually of uniform thickness, such as six inches or ten centimeters, and are called "arbitrary units" since they are imposed on the deposit by the excavator. If, however, the deposit is not homogeneous but is made up of sharply defined subunits varying in color, and the like, the archaeolo- gist may choose to excavate the deposit by these sub- units, which are often called "natural" units since they I are inherent in the deposit. In fact, as we have poin out above, these units are not "natural" but "cultural,' since they themselves are artifacts owing their comr sition in the main to cultural activities. This situation is particularly common in cave de- posits, and the confusion was thrown into focus at Montagu Cave, where the use of "arbitrary" units was not appropriate, and consequently we determined to excavate by so-called "natural" units. However, within these "natural" units we encountered concentrations of artifacts that appeared to represent material deposite during a single occupation and so were called occupa- tion horizons or surfaces. The occupation horizons were termed "cultural" units until it became apparent that the layer containing them was equally as cultural in its formation as the horizons themselves. Therefo we will speak of visual units and layers, indicating O that they were readily recognizable strata and car no connotation regarding processes of deposition. Wi the visual units in which occupation horizons were p sent the occupation horizons were excavated as entitie and the method of doing this is discussed below. In fa the only "natural" unit in the deposit was the sterile layer, Layer 4. The visual units or layers were the basic units in which the excavation was done and the vertical locations of the artifacts recorded. The artifact concentrations, referred to subsequentl as "surfaces" or "horizons," are not floors like those found at open sites such as Olorgesailie, Isimila, or Kalambo Falls. The Montagu surfaces were often more than one artifact thick but, although the thickness varie a surface was never more than about three inches thi An exception to the removal of the deposit by visuav units or layers was made in the case of the squares excavated to provide a guide section. These squares, 20 and 25 D, were excavated by six-inch levels, and the material recovered has not been included in the analysis. Parts of Layer 1 and Layer 2 were also ex- cavated by arbitrary levels, as discussed below. The upper layer, Layer 1, consisted of a light browE sand, which lay in a depression in the surface of the layer below it. A small niche in the south wall of the i cave about one foot above Layer 1 contained sand like., that which constituted Layer 1, and one of the fireplaci in Layer 1 had been partly dug away. These facts indicate that the top of this layer had been removed some time before the deposition of the rubble by the guano-miners and previous excavators. There was no visible stratifica- tion within the layer, and no horizontal concentrations of artifacts were found. Where the layer was more than six inches thick, it was removed in arbitrary levels of six inches each. The greatest thickness occurred at the sec- tion in square D, where it was nearly twelve inches. I Keller: Montagu Cave in Prehistory lSeveral fireplaces were found in Layer 1, which ire isolated and treated as features (following Heizer, F68:58) and the artifacts from them kept separate, kbough in the analysis all the material from Layer 1 i treated as a unit. There was some charcoal tered through the deposit, but most was concen- ted in the fireplaces. Layer 2 both underlay Layer 1 and extended later- r beyond it, since Layer 1 was confined to the low 1ea in Layer 2. Layer 2 consisted generally of a ttled dark gray sand. There were small (from a inches to one and one-half feet in diameter), ir- ular patches of lighter colors, but these were not ge enough to follow as strata. Some alternating k and light bands were present near the top of ner 2 in the 35 E and F section, but they lensed from twelve to fourteen inches from the section, the remainder of the layer was uniform. The er was about nine inches thick at the west end of '25 D section and about five feet thick at 20 G use the bottom of the layer sloped sharply toward south and east, that is, to one wall and the mouth hle cave. Charcoal was scattered throughout the er. !An attempt was made to follow several of the er poorly defined light stains in Layer 2, in the ef that they might represent some kind of feature, apparently this was not the case. These color tions were present throughout the layer and pro- d its faintly mottled appearance. Finally, square E was removed, and in the small section that ited a horizontal concentration of artifacts was This concentration was uncovered and followed, surface I thus identified. The procedure that we wed from that time on involved looking for a con- tion of artifacts in one of the sections and then vering this concentration. The artifacts were left ace, and by moving from one to the other follow - the dip of the artifacts it was possible for us to ver the whole surface. On occasion we would unter a bare spot, and since there was no visible raphy to follow, this created certain problems, solution of which usually lay in working around bare spot and determining from the arrangement he artifacts in the surrounding area whether the was in fact bare or if we simply had failed to deeply enough. This was necessarily very slow done with grapefruit knives and paint brushes, the rewards of following an artifact horizon and ng upon a fireplace or concentration of tools d to the interest of a rather tedious technique of vation. All of the deposit removed was sieved gh one-eighth screens. After the surface had been uncovered, the distribu- tion of the artifacts was plotted and the type of artifact indicated on the plot. This was done in very general terms, that is, "flake or flake fragment," "chip or chunk," "natural stone," or "trimmed piece." The difficulty was that there was no way to wash the pieces as they were taken up, so in some cases a tool thought to be a core when dirty was clearly a scraper when washed. There- fore more specific type designations were not made during the plotting and are not indicated on the plots presented here. The concentrations of trimmed tools are, however, visible. Datum depth readings were taken on the surfaces, and the material that occurred between the surfaces was labeled as having come from, for example, "between surfaces II and III." There was always some artifactual material be ;ween the surfaces, but the surfaces stood out as distinct concentrations of artifacts, and the assump- tion is made that the artifacts on a surface were de- posited over a short period of time and were contempor- ary (see figs. 17-23). An inspection of the charts and plots will show quite clearly that the number of artifacts found on the sur- faces increased as we went down into Layer 2. A simi- lar situation was present in the lowest Acheulean layer, Layer 5. An explanation of a similar phenomenon can be found in Wells (1965:81). In analyzing the botanical remains from Scott's Cave, a "Later Stone Age" site in the Eastern Cape province, Wells found that the sample from the upper level contained a smaller per- centage of hard parts, such as grass stems, than the lower level. Wells interprets this as owing to differen- tial decomposition, the hard parts being more resistant than the soft. The lower sample was only about one- fourth the size of the upper sample, but Wells feels that they originally contained about the same amount of vegetation. The result is, then, that the lower and older sample was considerably reduced in bulk and the hard parts concentrated as decomposition and compac- tion took place. It is clear that considerable compaction has taken place in the lower layers of the Montagu Cave deposit as grass bedding and other perishable materials decomposed. However, this mechanism seems inadequate to explain the concentration of artifacts on surfaces or horizons. Work on the surfaces had indicated clearly that there was much less material toward the front of the cave than toward the back and that stones from the roof and sides of the cave were more common near the front. The time involved in excavating this front area was not commensurate with the amount of information gained, and so none of the squares in the 15 line were exca- vated below surface VII. I :E 9 Anthropological Records There were no identifiable concentrations of tools below surface VII in Layer 2, and the deposit below surface VII was removed as a unit. But the layer thickened considerably to the southeast, and we felt that it would be wise to have some sort of vertical control over the material being recovered. To achieve this, two adjacent squares, 20 and 25 G at the thickest part of the deposit, were excavated by six-inch levels. At the bottom of Layer 2 there was a thin, artifact- rich band, which was the same color as the layer above it. Artifacts and pieces of charcoal were the main constituents of this band, and it looked very much like some sort of deflation feature. Much of the band was made up of charcoal, but it does not seem likely that this would have remained if the deflating agency had been wind, nor was there any indication of water action. The artifacts from the band were analyzed separately and were similar in all respects to those from the rest of the layer. There is no evi- dence from the artifacts to indicate that this band represents an extensive period of time that would fill the apparent hiatus betwene Layers 2 and 3. Layer 3, which directly underlay Layer 2, was com- posed of fine sandy clay and was much more compact than Layer 2. As with Layer 2, the sections were somewhat misleading, since there appeared to be some very fine banding in this layer, but the bands lensed out within a few inches of the section faces. However, there were six of these thicker units (12, 13, 15, 17, 17a, and 17b) that ranged in color from reddish brown to white. No concentrations of artifacts were found within the bands, so no surfaces were uncovered, nor were any plots of artifact distributions made. The ex- tent of the bands varied, as can be seen in figure 4. Band 12 was extensive, covering most of the excavated area, and was about nine inches thick. Below lay bands 13 and 15, which covered a smaller area and were about three and four inches thick, respectively. Band 13 was confined to the center of the excavation and does not appear in the section. Band 17 covered a large area, whereas bands 17a and 17b were much smaller and lay above 17 with 17a being the highest. In the sections, 17 had appeared to be composed of a series of alternating dark and light bands, and it had been our plan to follow these. However, they lensed out, and everything between 17b and the sterile sand that lay below it was treated as 17. No charcoal was found. Layer 3 was underlain by a sterile red sand, Layer 4. Microscopic examination of the sand has shown that it consists of white quartz sand coated with red silt- and clay-sized material. Comparison of grains of this sand with samples taken from the wall of the cave indicates that the two are identical and that the sterile sand is derived from the disintegration of the bedroc_ The origin of the finer grained red material, an iron oxide, and its mode of deposition is unknown. There is a small amount of silt- and clay-sized material ia the bedrock so the finer-grained material might also be derived from the bedrock. Although the presence o red beds, and climatic inferences from their presence, has received considerable attention in the literature African prehistory, the mere presence of the red silts and clay seems an inadequate basis for asserting a climatic change at Montagu (Krynine, 1949 and 1950; Sherman, 1952; Van Houten, 1948). The sterile sand i! completely unsorted, and the extreme angularity of t grains rules out deposition by wind or water. A bedding plane in the bedrock runs through the roof of the cave roughly parallel to the D line of squares. It was along this line that most of the larg fallen rocks found on the top of the deposit occurred, and rocks were found in this area all through the ex- cavation, with the exception of Layer 3 and the steril4 red sand. Layer 3 was relatively thin, and one might argue that it represents only a fairly short period of time. However, the sterile sand ranged in thickness from two and one-half to three and one-half feet and must represent a considerable time, and no rock fall were present in it either. Apparently grain-by-grain disintegration of the bedrock took place for a long and periodically large rocks fell from the roof, parti- cularly along the bedding planes. Below the sterile sand was Layer 5, which was composed of alternating light and dark bands; its cot position was similar to Layer 3. The four subdivisiog of Layer 5 were called 20, 21, 22, and 23, and each contained an occupation horizon, numbered VIII, IX, and XI respectively. Preliminary comparisons of samples of the dark and light bands done by Dr. Arthur Fuller of the U versity of Cape Town indicated that one difference between the two is that the dark layers contain more organic matter. The darker bands always contained more artifacts than the lighter ones, and the averag thickness of the darker bands was greater than that the light, which frequently lensed out. The darker then, would seem to have been formed during times more intensive occupation. In the excavation of Layer 5, bands were followed and, where concentrations of tools were found within the band, the concentrations were treated as horizo and uncovered and plotted as with the horizons in 2. By following bands, and the tools within the band we found that this layer had been considerably com- pacted. Compaction features were apparent in the s A I i 10 I s, but not to the extent that they were later revealed the horizons in Layer 5; parts of horizon X, for ex- pie, showed evidence of having been pushed down t two feet around underlying rocks. Layer 5 was ubtedly considerably thicker than when it was ori- ly deposited, and as the softer material decom- d the weight of the overlying deposit compressed lowest layer. Four horizons were found in this er. No charcoal was present. In the examination of the sediment samples from yer 5 for pollen and other petrographic features, all pieces of carbonized insect carapaces were d. Samples were submitted to Mr. Fred Guess, an omologist at the South African Museum, who identi- d them as fragments of the shells of Col optra etles. There are a number of birds, loca. ly called ifts, that live in the roof of the cave; they have ding habits similar to swallows and catch insects the wing. The source of the beetles' carapaces was interesting question, and initially it seemed likely they were incorporated into the deposit by the sd, either by the birds' dropping bits of the shell in the birds' feces. However, an examination of the odern bird feces in the cave revealed no insect re- s. Mr. Guess felt that it was unlikely that insects the very small size found in the deposit would have en caught by the birds, and an ornithologist, Dr. J. , Winterbottom, concurred; very delicate parts of ects, such as legs and wings, were present, and ese probably would not have survived if the insects d been eaten by the birds. Mr. Guess suggested that e insects may have been present in grass that was ought into the cave, and then as the grass and in- cts decomposed only the hard parts of the insects mained. Insect remains were not present in the light- lored bands, a fact that would support this interpre- ion, since insect remains are associated only with idence of intensive occupation. Dr. Robert Jones, Department of Agronomy, Univer- y of Illinois, examined samples from this layer. He und quantities of plant opal, a silica formation pre- nt in epidermis of grasses. This is additional evi- nce that grass was carried into the cave from outside. Below Layer 5 lay a second sterile sand layer, ayer 6, seven inches thick and similar to the one tween Layers 3 and 5. Below this sand was a thin, ne-grained layer, Layer 7, similar in appearance to yers 3 and 5, but only about two inches thick. It Fontained weathered fragments of bedrock but no arti- cts. Below this was the decomposing bedrock of the ve. The stratigraphy described above coincides closely ith that outlined by Barnard and Haughton as reported by Goodwin (see fig. 3). The only significant difference 1 1 is the lack of a sterile layer, Goodwin's layer "'C," between our Layers 2 and 3, and the previous excavator's apparent failure to differentiate between our Layers 1 and 2. It is, of course, entirely possible that Layer 1, which was partially removed on the south side of the cave, had been removed entirely in the central and north parts. One aspect of the deposit in the cave that has not yet been discussed is the presence of holes and dis- turbed areas. Caves and rock shelters commonly have their deposits disturbed near the walls, and Montagu is no exception. This was shown most clearly in the 35 E-F-G section where all the layers are disturbed next to the wall. This disturbed area was from about one and one-half to two feet wide and ran immediately next to the wall. It was composed of brown sand that was less compact than the undisturbed deposit and con- tained a mixture of artifact types. In addition, there were six holes, approximately from one to one and one-half feet in diameter, which ran through the deposit in different directions. Four of these were filled with artifacts and other material from Layer 2, mixed with material from lower layers, and two were empty. One hole in 35 F went nearly vertically through Layers 3, 4, and into 5, and then twisted toward the back of the cave. This hole was filled with a mixture of material from Layers 2, 3, and 4. Another hole ran across squares 35 and 30 E into 25 D in Layer 5 and con- tained only material from that layer. Yet another hole in Layer 4 ran parallel to the one in Layer 5 and was filled with a mixture of material from Layers 2 and 3. A fourth hole followed a similar path within Layer 4, but unlike the others it was empty. All four of these holes sloped from the back of the cave toward the mouth and ended near a group of fall rocks in square 25 D. Under these rocks there was a hole that went down into the bedrock. In addition, there was a hole in 25 E running almost straight down to these same rocks in 25 D and filled with artifacts and material from Layers 3, 4, and 5. As the surfaces in Layer 5 were uncovered, it be- came clear that an area in 25 E, adjacent to the rocks in 25 D, was disturbed. The section at this point showed what looked like a miniature fault. The bands were broken and slipped downward, apparently as the result of removal of material immediately below. The weight of the overlying deposit then forced this part of the layer into the void below. The last hole was encountered on surface XI, in square 30 G, and ran straight down, disappearing among some large rocks. The origin of these holes is unknown. All those that were empty looked as if water had played an important part in their enlargement, but it is not clear whether or not water action was responsible for their origin. One Keller: Montagu Cave in Prehistory I Anthropological Records suggestion has been that they were the result of animal burrows and subsequently were enlarged by water. Al- though there are no burrowing animals living in the area at this time, some may have been present in the past under different environmental conditions. Water drips from the ceiling and could have dissolved the holes through the deposit and gone out through the bedding planes in the bedrock at the bottom of the cave. These holes provide some suggestions regarding the formation of the deposit. One explanation suggested for the alternating black and white bands of Layer 5 was that they were caused by a small pond standing in the cave. This explanation seems very unlikely, however, since it is hard to imagine that water would stand in the cave with two large holes running down into the bedrock. If there had been sufficient water in the cave to form a pool, it either would have run out or would have washed the holes full. To summarize, then, the cave is formed in a meta- morphosed sandstone that breaks into large blocks as well as disintegrating grain-by-grain. The coarse sand produced by this grain-by-grain disintegration makes up Layers 4 and 6, which contain no archaeological material. Layers 1, 2, 3, and 5 contain archaeological material and are made up of a mixture of the coarse sand derived from the bedrock and larger amounts of fine-grained material. However, the mechanism or mechanisms specifically involved in the deposition of Layers 1, 2, 3, and 5 are not known. Certainly two factors are involved, the first the decomposition of the bedrock, the second the addition of organic and perhaps some inorganic material by the human and other occupants of the cave. Evidence from more re- cent South African sites, such as Scott's Cave or Melkhoutboom, indicates that quantities of grass, brush, and twigs are carried into caves by the people living in them. This material introduces large amounts of organic matter that would not otherwise be present. Analysis of samples from Layers 3 and 5 have demon- strated the presence of sand grains, which are not found in the sample from the bedrock nor in Layer 4. The sand grains that make up Layer 4 are coated with an iron oxide. From the size of the grains it seems possible that this material was blown into the cave, but its specific source is unknown. It may be derived from the shales that occur as the bedrock of the valley the cave overlooks. The literature on the accumulation of cave deposits in general is very meager, and that dealing with the formation of archaeological cave de- posits smaller yet, so many of these questions remain unanswered. Analysis of solidified guano found adhering to the wall of the upper cave proves that guano has not been a signifi- cant factor in the formation of the sediments of the lower cave. There is no similarity between the chemi- ^ cal constitution of the guano and that of the sediments. A second problem that must be considered is the relation of the upper and lower caves to each other and to the formation of the deposit in the lower cave. Some parts of the walls of the upper cave appear A smooth and rounded, but it is not clear whether this is from weathering or from water action. The question 1 of the mechanism of water action is raised here be- cause the suggestion has been made that at some point in the past a small stream could have been present in the upper cave that ran down into the lower cave and out its mouth. I am certain that this was not the case during the time represented by the deposits in the lower cave. Although no survey has been made that is sufficiently accurate to permit a precise estimation of the volume of material that has been removed in the formation of the upper cave, it is clear that much mo material has been removed than is represented by the sediments in the lower cave. Since most caves are essentially funnel shaped, individual sand grains, by the action of the wind, could, without actually being blown out, work their way down along the slope of the cave floor to the mouth, and then out of the cave. Alternatively, a stream, either perennial or periodic, may have been responsible. Whatever the explanation, the phenomenon has not functioned in the recent past since it would have inhibited the accumulation of the deposits with which we are concerned in the lower ca In addition, there has been an appreciable accumulation of wall-derived sand in the upper cave. The analysis of the archaeological material that follows will indicate that there is a considerable differ- ence between the artifacts from Layer 3 and those fro Layer 2, and evidence from other sites suggests that these two assemblages may be separated in time by as much as 40,000 years. Yet there is no stratigraphice feature between Layers 2 and 3 in the Montagu Cave to account for a time gap of that length. There is no reason to assume that the disintegration of the bedroe went on at a constant rate, and it may have stopped altogether at some periods. The sand grains are too large to have been blown out of the cave, and there is no evidence of water action that could have removed the particles. Either deposition of the sand did stop, or the two archaeological assemblages are not sepa- rated by so great a time as had been commonly held and the evidence from other sites reflects a situation different from the one at Montagu. 12 I harcoal samples have produced dates of 23,200 + B.P. (GRN. 4726) for the top and 45,900 + 210 B.P. 4728) and > 38,000 (GXO 947) for the bottom of r 2. The > 38,000 date reflects the limits of the ting ability of the laboratory. The laboratory re- suggests that this confirms the 45, 900 date. A ple, GRN. 5123, from between horizon 5 and 6 has uced a date of 19,100 + 110, and another sample, * 5124, from between 6 and 7, a date of 50,800. impossible to explain either the inconsistency of e dates or the extreme age of some of them. The ous reason for a sample to produce a date older expected is that the sample has been contaminated old charcoal. This is not possible in this situation, e the lower layers do not contain any charcoal. A ple from the Howieson's Poort shelter, near amstown, which contains material similar to that Layer 2, has given a date of 18,740 + 320. This ests that the 19,100 and 23,200 dates may be about t. An alternative explanation is that the samples have produced the younger dates are the contami- d ones and that the older dates are a more accur- reflection of the age of this material. The date for the bottom is surprisingly old but, if rect, could indicate that there is no great time gap een Layers 2 and 3 since a date of from 50,000 to 000 years for the Acheulean of Layer 3 is not impos- e. However, there is a date of 7,100 + 45 B.P. iN 4725) for Layer 1, which would mean that the lerence in age between Layer 1 and the upper part Layer 2 is about 16,000 years. In the absence of 13 evidence of any deposition during that time, one cannot assume that the disintegration of the bedrock went on at a constant rate, since there was no deposition during the 16,000 years that elapsed between the deposition of Layers 1 and 2. Archaeologists frequently state that the most impor- tant data they recover are not the artifacts themselves, but information about the relationship of the artifacts to each other in the deposit. I have tried to suggest that an understanding of the deposit in which these rela- tionships obtain is equally important, since the deposit may itself be an artifact. Linguists and ethnologists are aware that the context in which behavior occurs is important for a full understanding of that behavior, and the context in which archaeological artifacts are found may be equally important to their interpretation. Later, the material from Layers 3 and 5 will be shown to be the result of tool-making or workshop activities. Workshop sites are often not considered to be "habita- tion" sites, but in Montagu Cave knowledge of the deposit in which the material occurs demonstrates that, in fact, the site was occupied repeatedly, apparently for the purpose of exploiting the source of raw material in the stream below. An additional reason for this extensive discussion is that the literature on the sedimentology of cave de- posits in general, and archaeological cave deposits in particular, is very small, and it is hoped that this dis- cussion will point out some of the problems involved and perhaps stimulate further research in this area (see Appendix II). Keller: Montagu Cave in Prehistory DESCRIPTION AND DISCUSSION OF ASSEMBLAGES INTRODUCTION Archaeological descriptions involve a substantial amount of subjectivity, and if they are to be useful it is impor- tant that they be as explicit as possible. In the description and analysis of the Montagu collec- tions we used the Acheulean terminology published by Kleindienst (1962) to facilitate comparisons with the East African material. For the post-Acheulean material a terminology was developed with the cooperation of Miss Barbara Anthony. Definitions of the post-Acheu- lean tool categories are included in the body of the descriptions, and the substance of Kleindienst's termi- nology is included in Appendix I. Definitions of more general terms are presented below. Miss Anthony was consulted on the assumption that it would be useful to have basically similar termino- logies used for describing material from Montagu Cave and Peer's Cave. At the time this description was done (1965) no other Middle Stone Age sites in southern South Africa had been excavated for some years, and eventual comparison with the Peer's Cave material seemed appropriate. A second objective of these descriptions, beyond the purely comparative, is to illustrate variability within the descriptive categories. Clark (1950) has made the distinction between "formal" and "informal" tools with reference to the Acheulean. This distinction involves the notion that tools that have been produced by considerable modification of the primary form-for instance a hand-axe, which requires a large number of operations in manufacture-exhibit less variation and are more standardized than those that require fewer operations, for instance a scraper. It is not a simple matter to compare categories like "hand-axe" and "scraper" since the attributes that define the two are usually mutually exclusive. However, in the follow- ing descriptions univariate frequencies are included in an effort to indicate the kinds of variation present within the categories. It must be emphasized that the attributes for which frequencies are included are not necessarily the most important ones for determining membership in any classificatory group. Other attri- butes may have been as important, or even more important for classificatory purposes. Many of the attributes listed are not criterial (Bruner, et al., 1956, 31) or essential (Clarke, 1968:137), but are of interest to archaeologists for purposes other than classification, Thus characteristics such as raw material or type of striking platform are included, even though they had no bearing on classification. Since most of the criteria attributes do not vary but form a categorical frame- work within which other attributes vary, it is often possible to see evidence of the choice of techniques and raw material to produce certain tool forms, where. as other techniques and raw material were preferred for other forms. Another possibility that bears on the "formal- informal" distinction is that the "informal" tools repre- sent a residual category; that is, when all possible pieces have been assigned to some category the rema ing pieces constitute the "informal" tools. To clarify this question I will summarize the way the various attributes were used to categorize the artifacts. When classifying Acheulean material, I first sepa- rated artifacts into trimmed and untrimmed categories The untrimmed pieces included flakes, flake fragments chips, and chunks. Any piece that showed secondary modification of its primary form was included with the trimmed pieces. Pieces showing only hammerstone modification were placed in a special category. These categories were not the traditional "waste" and "tool" units, since some pieces, for example cores, were treated as tools and only later included with the un- trimmed pieces for the purpose of figuring percentage Trimmed pieces were then segregated into various types There were minor fluctuations in procedure depen on the particular lot of artifacts. No codified check list was used, although a mental one was. Pieces with a large number of flake scars were usually picked out first and described as belonging to one or another category (hand-axe, cleaver, pick) on the basis of plan and cross section. Obviously the relative weighting of a given attribute is impossible to assess when a pro- cedure of this kind is used, since a number of attri- butes are perceived at the same time. One picks out the hand-axes (or whatever) and describes them and goes on to something else. However, it is important to note that some basic attributes are very important in influencing categorization. Amount of modification is an initial one. No matter what the plan-form, if secondary modification is absent, then the piece is 4 I I Keller: Montagu Cave in Prehistory ed in what ultimately becomes the waste category. i-form is of secondary importance with cross sec- third; that is, presence or absence of a cleaver r pointed end is then modified by cross section to both relative thickness and degree of symmetry) ategorize a piece as a hand-axe, pick, cleaver, or plane. ifter these more readily recognized pieces for which and section are important are described, whatever ft must be categorized. Some pieces fell outside sits of Kleindienst's terminology for hand-axe, so forth and these were placed in special categories, as "hand-axe chopper," which are dealt with below. ,special categories contain pieces that have differ- rxpressions of the same attributes we have been ssing. r other categories other attributes are more im- nt. The plans and cross sections of choppers are rent from those of hand-axes and cleavers, but the nction between cores and choppers is made on the of type of flake scars rather than on any of the kously mentioned attributes. Pieces with identical 3 and sections but different kinds of flake scars d be placed in separate categories. That is, if two imens were identical in all attributes but one had e complete scars and the other had small step s and a crushed edge, the former would be classed core, the latter as a chopper. When cores, pieces large complete scars, showed battering, crushing, small step scars along part or all of their margin, were described as cores with chopper use. he residuum now consists of pieces with less modi- ion than those already classified. The remaining kes in the Montagu Acheulean collection consisted fragments of tools, large pieces with a few large scars, and pieces with minor modification of their tgins. The large pieces appeared to represent an kly stage in the manufacturing process and so were into a separate category. The pieces with marginal jdification were placed in the "scraper" category. pse items with characteristic asymmetrical sections steep edges were placed in Kleindienst's "core raper" category. Everything else was lumped as a ,aper. In classifying the post-Acheulean material, the pro- lure was essentially the same, within the limits posed by the nature of the modification. For instance, plan was a much less important attribute and the na- ture of margin modification was more important than for the Acheulean material. In general, however, those pieces with more modification (of whatever kind) were categorized first, and those that were modified less were categorized last. For scrapers then, especially in the Acheulean, the "informal" tools do represent a residual category. This is not to say that this category is defined in purely negative terms. There are attributes that define the category, but they are fewer and for cultural reasons they are more difficult to describe than are other attri- butes. The characteristics used to define the more formal tools are those that are important in categori- zation of western European hand tools: amount of modi- fication (or number of attributes), plan-form, and cross section. In the following description the frequency of various attributes is set out in tabular form. Some attributes for some types have been used for correlation analysis and these are discussed individually. In general, how- ever, one can readily perceive the pattern of combina- tions present. For most attributes there is one expres- sion that accounts for a majority of the pieces. A com- bination of the major expressions of each attribute produces a composite of the category, and wholesale correlations seem unwarranted. For instance, among the hollow scrapers from Layer 2 the most common raw material (chert, 90 percent) occurs most commonly in association with the most frequent primary form (end-struck flakes, 41.6 percent), which occurs most commonly with the most frequent plan (irregular, 71.9 percent) and the most frequent location of trimming (one side, 73.2 percent). The classification procedure has been described in order to clarify the way Kleindienst's terminology was used and also to give some indication of the classifi- catory use of the attributes listed. Some were of con- siderable importance whereas others, such as raw material, were not used at all. Therefore it is not possible to test these categories by carrying out a correlation study of the attributes listed since many of the necessary characteristics are not included. Many of the attributes that are not listed are criterial, in that they determine assignment to a category, but are "inessential" (in Clarke's terms) since they are present on each piece in the category. Dv 1 5 Anthropological Records DEFINITIONS Tool-Those pieces that appear to have been altered in a purposeful way, excluding cores. The alteration may have been intended to modify the shape of the pri- mary form or to produce a particular kind of edge, or both. Tools are indicated by the term "shaped" in the figures. Waste-Unaltered chips, chunks, flakes, flake frag- ments, and cores. All waste except cores is referred to as flake waste. Certainly some pieces classed as waste could have been used as tools, but if no altera- tion is visible, these pieces have been classed as waste. Utilized-Pieces that have been altered only by use. Primary form-The kind of object on which a tool or core is made. For instance, the primary form on which a hand-axe is made might be a flake, a chip, or a chunk. For some it is impossible to tell what the primary form was; for others it is clear that the primary form was a flake, but the flake has been altered to the point where it is impossible to say whether it is end-struck or side-struck. The place- ment of the trimming is described in terms of the primary form, so where possible the trimming is said to be on the dorsal or ventral face. But if such a designation is impossible, the piece is simply described as "unifacial" or by some other appropriate term. The same is true of the placement of the edge. A chip could be described as "trimmed on one side," but a side- struck flake could be described as "trimmed on the distal side" or "proximal side," and so on. Flake-A piece that retains the complete bulb of percussion and tapers to thin edges at its margins. Split flake fragment-A piece that retains all or part of the bulb of percussion, but which has broken parallel to the direction in which the flake was struck. Ideally the split face and platform would intersect at a 900 angle. Snapped flake fragment-A flake fragment in which the split face is parallel to the platform; that is, it runs at 900 to the direction in which the flake was struck. Chip-An artifact that obviously has been detached from some larger piece but lacks a bulb of percussion or part thereof is called a chip. The bottom or distal part of a step-flake fragment would be classified as a chip. Chunk-An artifact from which other pieces have been detached. The surface of a chunk is usually made up of partial flake scars or break faces. The scars, however, are not regular or numerous enough to justify calling the piece a core. Plain platform-A smooth platform on a flake or I core. On a flake the plain platform could be cortex, a break face, or part of a large flake scar. Negative-scar platform-The platform of a core where a single negative scar has been used as a striking platform. Simple-facetted platform-A platform consisting of two or in the case of cores three flake scars. The term is comparable to "pseudo-facetted," used by some writers. Facetted platform-A platform with three or more flake scars. A facetted platform on a core is one thai would have produced flakes with facetted platforms. "Bulb only" platform-The kind of platform found on some flakes (in Layers 1 and 2) that have practi- cally no platform, but only a very thin face above the bulb. Thin platform-The kind of platform found on some cores; two faces intersect at a sharp angle and force has been applied to one face to remove flakes from the other face. End-struck-Flakes that are longer than they are wide. Length is measured along a line perpendicular to the striking platform; width is measured at the widest point on a flake along a line parallel to the striking platform. Side-struck-Flakes that are wider than they are long, or that are equal in these two dimensions-that is, flakes as wide as they are long are counted as side- struck flakes. Dorsal face-The face of a flake opposite the face with the bulb of percussion. On other artifacts-such as a chunk-with a plano-convex cross section, the convex face is the dorsal face. Ventral face-The face having the bulb of percussio The term is equivalent to the "main flake surface" of some other writers. End-The narrow part of the margin of a flake. OD an end-struck flake the end is opposite the platform and on a side-struck flake it is adjacent to the plat- - form. Side-The broad part of the margin of a flake. (See "fend" above.) Proximal end or side-The platform end of an end- struck flake or the platform side of a side-struck flal Distal end or side-The end or side opposite the striking platform of a flake. Edge-The working edge of a tool cr the part of the margin that has been trimmed; an "edge" may be foun on either the sides or ends. 16 Keller: Montagu Cave in Prehistory Shallow-Describes an edge with an angle that falls etween 0 - 250 Blunt-Describes an edge with an angle that falls etween 260 - 550. Steep-Describes an edge with an angle of more ban 550. Guillotine left and right-Kleindienst (1962) uses the Irm "guillotine" to refer to cleaver bits that are not perpendicular to the long axis of the tool. Where it is Ss5ible to determine the ventral and dorsal faces of cleaver, the tool is oriented with the ventral face bwn and the bit away from the recorder, and then ,e bit is described as to whether it slopes to left or pght. Where it is impossible to identify the dorsal or entral face, the bit is simply described as "guillotine." In the following pages the artifacts from the four tifact-bearing layers are described. Range, means, id standard deviations are given for the lengths, idths, and thicknesses of the tools, cores, and util- ed pieces, and other characteristics are described well. Ranges, means, standard deviations, and modes also given for width/length and thickness/width. hese ratios are often useful in expressing something the shapes of the tools in question. The modal fre- ency for these ratios is also included, since it may flect more precisely than the mean a pattern that e makers of the tools were attempting to produce. re several values are represented an equal number times the values are listed. Where there was no o1 value it has been omitted. Broken specimens re not used in determining the w/l and t/w values. name of each type designation is followed by two nbers: the first is the number of examples of that and the second is the percentage the type rep- resents of the larger general category, such as tools or cores, to which it belongs. For example, in Layer 1 there are 9 crescents that comprise 4.5 percent of the tools, and there are 18 single-platform cores that comprise 77.4 percent of the cores. Similarly, the per- centages given for utilized pieces are in terms only of the utilized pieces. In the descriptions the most precise terms have been used wherever possible. For instance, in some categories reference is made to trimming located on the "distal end" and in the same category trimming is described as located simply on "one end." This has been done because it is sometimes impossible to dis- tinguish distal from proximal ends. If the primary form is a flake, one can make the distal/proximal distinction in terms of the platform, but on other pri- mary forms where no platform exists this distinction is not possible, and the more general term "one end" is applied to those trimmed on the "proximal" or the "distal" end. The same is true of the face that was trimmed; if possible the term "dorsal" or "ventral" was used, but if it was not possible to make this dis- tinction then the more general "unifacial" was used. Nongeometric terms have been used to describe the plan-form of some tools. Among the scrapers from Layer 1, for example, terms such as "pebble" or "crys- tal" are used. These terms reflect the primary form on which the tool was made; thus a piece made on a split quartz crystal is described as having a "crystal"- shaped plan-form. The same is true of "pebble" when it is used to describe plan-form. A brief discussion of the assemblage from each layer follows the description, but comparisons will be discussed in the comparative section. LAYER 1 DESCRIPTION Tools (201) Crescents (9, 4.5 percent; plates I:12-13; II:12-13) All these are made of quartz and are trimmed on the dorsal face. All are made on flakes, but the platforms have been removed by the backing. Standard Dimensions Range Mean deviation Mode Length 10-20 mm. 14.4 mm. 3.1 mm. Width 5-9 mm. 6.4 mm. 1. 3 mm. Thickness 1-4 mm. 2.5 mm. 1.0 mm. W/L T/W 33 -90 20-67 47.0 40.0 15.9 15.1 40 33 I I 1 7 Anthropological Records Backed blades (2, .9%) These are flakes that are trimmed, apparently in order to blunt one side. These specimens are quadrilateral in shape and are made on end-struck quartz flakes with plain platforms. They are trimmed on one side on the dorsal face to straight edges. Standard Dimensions Range Mean deviation Mode Length Width Thickness W/L T/W 14-15 mm. 4-5 mm. 2-3 mm. 29-33 50-60 14.5 mm. 4.5 mm. 2.5 mm. 31.0 55.0 0.7 mm. 0.7 mm. 0.7 mm. 2.3 5.0 Obliquely truncated blades (3, 1.5%; plate II:10-11) These are pieces that are trimmed obliquely on one end and sometimes in other places as well. The trimming is steep and resembles backing. One is made of quartz and two of chert. All are trimmed on the dorsal face and distal end; one is trimmed on two sides as well. Two have "bulb only" platforms and one has a plain platform. Standard Dimensions Range Mean deviation Mode Length Width Thickness W/L T/W 22-36 mm. 6-10 mm. 2-4 mm. 22 -45 30-50 27.0 mm. 8.0 mm. 3.0 mm. 31.0 38.0 7.8 mm. 2.0 mm. 1.0 mm. 10.1 8.7 Thumbnail scrapers (31, 15.4%; plates 1:5-6, 8; II:1-2, 4, 7-8) These tools are trimmed only on one edge and are usually roughly trapezoidal in plan with a convex working edge Dimensions Length Width Thickness W/L T/W Range 6-36 mm. 7-27 mm. 1-9 mm. 58-167 10-61 Mean 14.0 mm. 13.1 mm. 4.4 mm. 99.0 34.0 Standard deviation Mode 5.1 mm. 3.5 mm. 1.8 mm. 24.8 11.9 100 27 No. Percent Trimmed face 14 45.2 17 54.8 Dorsal Unifacial No. Percent 27 87.1 4 12.9 Primary form Chips Chunks End-struck flakes Side-struck flakes Split flake Edge plan Convex Straight Hollow scrapers (4, 2%) 7 2 8 1 3 1 22.6 6.5 25.8 41.9 3.2 Striking platform (21) Bulb-only Plain Simple-facetted Removed 30 96.7 1 3.2 These tools are characterized by concave, semicircular scraping edges. Standard Dimension Range Mean deviation Length Width Thickness W/L T/W 17-38 mm. 6-23 mm. 3-7 mm. 19-135 25-60 27.0 mm. 14.8 mm. 5.2 mm. 68.0 41.0 9.3 mm. 8.0 mm. 1.7 mm. 47.8 14.2 No. Percent Trimmed face No. Percent Chert Primary form Chips Chunk Snapped flake 4 100 2 1 1 50 25 25 Dorsal 1 Opposite sides and faces 2 Unknown 1 Material Quartz Chert 4 11 2 4 19.1 52.4 9.5 19.1 Material Mode 25 50 25 . . _ . . . . 1 8 Keller: Montagu Cave in Prehistory Scrapers (108, 53.7%; plates I:7,9; II:3, 5, 14, 16) All tools in this large and variable category have steep unifacial edges. Three of the 108 scrapers are fragmentary Dimensions Length Width Thickness W/L T/W Range 5-123 mm. 5-98 mm. 2-66 mm. 23 -264 11-100 Mean 28.4 mm. 22.6 mm. 9.7 mm. 87.0 43.0 Standard deviation 23.5 mm. 17.0 mm 11.1 mm. 38.2 19.0 Mode 60 40 No. Percent Location of trimming No. Percent Table Mountain Sandstone Quartz Chert Primary form Chips Crystal fragments End - struck flakes Side-struck flakes Split flakes Snapped flakes Chunks Plan Round Pebbles Slabs Crystal- shaped Short quadrilateral Sub-quadrilateral Ovoid Semicircular Long quadrilateral Irregular 15 32 61 19 16 14 16 6 1 36 3 4 2 6 4 2 1 1 1 84 13.9 29.6 56.4 17.6 14.8 12.9 14.8 5.6 0.9 33.3 2.8 3.7 1.9 5.6 3.7 1.9 0.9 0.9 0.9 77.7 Two sides Two ends End and side Distal end One end All the way around On distal end and side One side and both ends One side Trimmed face Bifacial Unifacial Ventral Dorsal Platform (31 pieces) "Bulb only" Plain Facetted Simple-facetted Outils ecailles (23, 11.4%; plate I:10) These tools have been described by van Riet Lowe (1946) and Clark (1958a and 1958b). Tools with characteristic bifacial outil edge and the characteristic battering are included in this category. Those with the outil type working edge but lacking the battering are included in the chisel category (see below). Standard Dimensions Range Mean deviation Mode Length Width Thickness W/L T/W 9-30 mm. 5-21 mm. 3-16 mm. 38-175 24-100 18.9 mm. 11.7 mm. 7.6 mm. 65 63 5.6 mm. 3.9 mm. 3.7 mm. 27.5 2.0 50, 63, 67, 77 60, 100 Material Quartz Chert Primary form Chip Chunks Crystals Crystal fragments End-struck flake Side-struck flake No. Percent 22 95.7 1 4.3 1 4 3 13 1 1 4.3 17.4 13.1 56.5 4.3 4.3 Location of trimming Worked on two ends Worked on one end Piece made on the end-struck, flake trimmed on distal end No. Percent 6 26.1 16 69.5 1 4.3 19 Material 4 3 4 10 1 9 1 1 1 52 1 38 3 66 2 25 3 1 3.7 2.8 3.7 9.3 17.6 0.9 0.9 0.9 48.2 0.9 35.1 2.8 61.0 6.5 80.7 9.7 3.2 - - - . Anthropological Records Chisels (2, 0.9%) Standard Dimensions Range Mean deviation Mode Length 20-21 mm. 2 0.5 mm. 0.7 mm. Width 16 -17 mm. 16.5 mm . 0.7 mm - Thickness 4-14 mm. 9.0 mm. 0.7 mm. W/L 76-85 80 4.4 T/W 24-88 55 31.9 Material No. Percent Location of trimming No. Percent Quartz crystal 2 100 One end 1 50 Two ends 1 50 Burins (3, 1.5%) Standard Dimensions Range Mean deviation Mode Length Width Thickness W/L T/W 19-25 mm. 7-12 mm. 2-7 mm. 28-63 25-100 22.3 mm. 9.0 mm. 4.0 mm. 42 50 3.0 mm. 2.6 mm. 2.6 mm. 15.2 35.3 No. Percent Primary form 3 100 Chip Crystal Crystal fragment Discoid (1, 0.5%) This is a round, flat chert tool, the primary form of which is is trimmed bifacially and is not a core. Dimensions Length 26 mm.; Width 21 mm.; Thickness 11 mm. No. Percent 1 1 1 33.3 33.3 33.3 unidentifiable. It Pointed tool (1, 0.5%) This is an irregularly shaped chunk of quartz with a unifacial point trimmed on one end. Dimensions Length 15 mm.; Width 6 mm.; Thickness 2 mm. Trimmed flakes (6, 3%; plate I:3) Most of the tools included in this category, as well as those called "trimmed chips" and "trimmed chunks," have shallow or blunt edges whereas most scrapers have steep edges. The distinction of cutting edges and scraping edges made in the Acheulean material is based on two features: one, the shallowness or sharpness of the edge; and two, the way in which the edge is worked, that is, a cutting edge, in the Acheulean, is usually bifacial, whereas a scraping edge is usually unifacial. Un- fortunately, such variation does not exist in this later material since virtually every edge is unifacial. The trimmed flakes, however, seem to possess a characteristic combination of attributes in the placement of trimming, the object on which they are made, and the angle of the edge, and as a result, they are placed together in a single category. The other similar tools, that is, "trimmed chips" and "trimmed chunks," possess similar edges but the primary form is different. It is tempting to think of these tools as cutting tools, but no real evidence exists for this presumption. Dimensions Length Width Thickness W/L T/W Range 15-98 mm. 15-27 mm. 3-21 mm. 38-112 21 -37 Mean 48.3 mm. 33.5 mm. 9.8 mm. 78 28 Standard deviation Mode 30.3 mm. 16.0 mm. 6.3 mm. 27.5 6.8 Material Quartz ___ _ . 20 Keller: Montagu Cave in Prehistory Material Table Mountain Sandstone Chert Primary form End-struck flakes Side-struck flakes Snapped flakes Location of trimming One side Two sides Two ends No. Percent 3 50 3 50 2 1 3 3 2 1 33.3 16.7 50.0 Trimmed face Dorsal Parti-bifacial Platform Plain Facette d "Bulb only" 50.0 33.3 16.7 Trimmed chips (3, 1.5%) Dimensions Length Width Thickness W/L T/W Range 20-59 mm. 11-24 mm. 4-7 mm. 28-55 29-45 Mean 39.3 mm. 15.3 mm. 5.3 mm. 41 37 Standard deviation 19.5 mm. 7.5 mm. 1.5 mm 10.9 6.6 Material Chert Location of trimming Two sides One side No. Percent Trimmed face 3 100 Dorsal Bifacial No. Percent 2 66.6 1 33.3 1 33.3 2 66.6 Point (1, 0.5%; plate I:4) This point is made on an end-struck, chert flake with a plain platform. It is trimmed only on the dorsal face (and is probably intrusive). Dimensions Length 58mm.; Width 46 mm.; Thickness 14 mm. Choppers (2, 0.9%) Dimensions Length Width Thickness W/L T/W Range 76-100 mm. 56-72 mm. 28-55 mm. 72 -74 50-76 Mean 88.0 mm. 64.0 mm. 41.5 mm. 73 63 Standard deviation 16.9 mm. 11.3 mm. 19.0 mm. 0.8 13.1 Mode Material Table Mountain Sandstone Shale Primary form Chunk No. Percent Location of trimming No. Percent 1 50 One end 1 50 1 50 One side 1 50 2 100 Core scrapers (2, 0.9o) Tools included in this category are usually high-backed and made on chunky objects. The characteristic type of edge is like that on the Acheulean core scraper, that is, steep and often undercutting the side of the object on which it is made. Standard Dimensions Range Mean deviation Mode Length Width Thickness W/L T/W 95-107 mm. 56-84 mm. 56-65 mm. 59-79 67-116 101.0 mm. 70.0 mm. 60.5 mm. 68 41 8.4 mm. 19.7 mm. 6.4 mm. 9.7 24.7 No. 5 1 2 2 2 Percent 83.4 16.7 33.3 33.3 33.3 Mode - 21 Anthropological Records Material No. Percent Table Mountain Sandstone 2 100 Primary form Chunk Location of trimming One end Complete circumference No. Percent 1 50 1 50 2 100 Cores (23) Single-Platform Cores (18, 77.4%; plates I:14-15; II:9) These cores are trimmed with more or less parallel flakes from a single plat- form. Dimensions Length Width Thickness W/L T/W Range 10-36 mm. 8-28 mm. 4-17 mm. 38-167 39-120 Mean 22.6 mm. 15.1 mm. 9.9 mm. 73 68 Standard deviation Mode 7.7 mm. 5.5 mm. 3.8 mm. 33.7 21.9 50, 69 73 Material Quartz Chert Primary form Chunks Crystal fragments Crystals Location of trimming on chert cores (4) Trimmed one side Trimmed one end No. 14 4 5 9 4 Percent 77.9 22.1 27.8 50.0 22.2 Platforms Plain Simple-facetted Facetted Negative scars Crystal faces Location of trimming on quartz cores (14) 3 75.0 Trimmed one end 1 25.0 Trimmed one side No. 6 1 4 4 3 1 3 93.8 1 6.2 Double -Platform Cores (4, 17.4%) Cores in this category are trimmed with more or less parallel flakes from two platforms. Dimensions Length Width Thickness W/L T/W Range 18-25 mm. 12-26 mm. 3-15 mm. 67-104 25-83 Mean 20.8 mm. 18.7 mm. 9.5 mm. 80 50 Standard deviation 3.0 mm. 5.7 mm. 4.9 mm. 13.7 21.6 Mode Material Quartz chunks Chert chunks Location of trimming Both ends worked of same face Both ends worked on opposite faces Trimmed from one end and one side but on opposite faces No. Percent Platform 2 50.0 Thin 2 50.0 Plain Trimmed faces from natural 2 50.0 1 25.0 1 25.0 Disc Core (1, 4.4%) Cores placed in this category are usually round, trimmed radially, and thin. This core is sub-quadrilateral in shape, made of Table Mountain Sandstone, and the pri- mary form is unidentifiable. It has been trimmed bifacially with a flat cross section. Dimensions Length 58 mm.; Width 48 mm.; Thickness 20 mm. Percent 33.3 5.6 22.2 22.2 16.7 No. 2 1 1 Percent 50.0 25.0 25.0 - _ 22 Keller: Montagu Cave in Prehistory Utilized Pieces (67) Grindstones (4, 6%) These are slabs of Table Mountain Sandstone that appear to have been utilized for grinding. One is stained red. Standard Dimensions Range Mean deviation Mode Length 144-214 mm. 173.5 mm. 34.7 mm. Width 130-194 mm. 152.2 mm. 29.7 mm. Thickness 45-75 mm. 61.7mm. 14.2 mm. W/L 68-106 89 13.4 T/W 29-57 41 10.0 Utilization No. Percent Unifacial 3 75.0 Bifacial 1 25.0 Hammerstone (1, 1.5%) This is an irregular chunk of Table Mountain Sandstone that has been used as a hammerstone. Dimensions Length 97 mm.; Width 91 mm.; Thickness 70 mm. Flakes (29, 43.3%) There are twenty (69%) quartz flakes and nine (31%) chert flakes that have been utilized. Flake Fragments (2, 3%) One of these is made of quartz and one of chert. Chips (15, 22.4%) Nine (60%) of the utilized chips are made of quartz and six (40%) of chert. Chunks (8, 11.9%o) Five (62.5%) of these are made of quartz and three (37.5%) of chert. Cores (1, 1.5%) This consists of a single platform core made of chert that has been utilized on one end. Pigment (7, 10.4%) These are pieces of abraded hematite. 23 Anthropological Records LAYER 1 DISCUSSION The material from Layer 1 is the kind that has often been called "microlithic," but an examination of the size ranges on the preceding pages will make it quite clear that whatever "microlithic" means it would prob- ably not cover such a range. An examination of the lengths of flakes made from the three different mater- ials (see fig. 5) indicates that larger pieces are of Table Mountain Sandstone and that smaller ones are of the finer grained materials. The same is true of the tools, the crescents and backed blades being made of quartz and chert and the larger tools, choppers or large scrapers for instance, are made of Table Moun- tain Sandstone. By far the most commonly occurring type of tool is the scraper, which accounts for 72.0 percent of all the tools. The correlation tables show that for all three raw materials a piece of angular waste was most commonly chosen for modification to produce a scraper. Modification was usually on the side of what- ever piece was chosen. (The quartz angular waste pieces are an exception; no reason for this is appar- ent. The same is true for Layer 2 scrapers.) The number of scrapers found made from pieces of angular waste parallels the amount of angular waste found in the assemblage as a whole (fig. 10). We can assume therefore that scrapers were usually made by choosing one of the most commonly occurring artifact forms and trimming all or part of its longest available sec- tion of margin, "side" in my terminology, to a steep unifacial edge. Scrapers are followed in frequency by outils ecailles, 11.4%o, and crescents, 4.5 percent. These three types alone account for 87.9 percent of all the tools. The point and the trimmed flakes are included with the Layer 1 material for the sake of completeness, but these tools are characteristic of the Layer 2 assemblage and may well have been picked up by the later people and then incorporated in Layer 1. In the core category, single-platform cores pre- dominate, and these plus double-platform cores account for 94.8 percent of the cores present. Figure 6 shows the incidence of various platform types found on the flakes. The number of "bulb only" platforms, coupled with the predominate types of cores, strongly suggests that parallel-sided flakes were being produced by the punch technique. But it must be emphasized that, although punched, parallel-sided flakes are characteristic and common in this assemblage, they are by no means the predominant type numerically. To describe this assem- blage as a completely "microlithic" or "blade" industry would convey an erroneous impression. Seven features were discovered in Layer 1, and the material that these features contained is outlined in table 2. With the exception of feature 3, all were fire- places or hearths. All these were more or less oval and the long axis ranged from about 50 to about 97 centimeters, and in depth from 2.5 to 25 centimeters. It is impossible to say if these fireplaces were actu- ally hollowed out before use, but it seems unlikely. They were discernible by a concentration of charcoal surrounded by a light-colored ring of what appeared to be ash. In excavation, the charcoal and then the light-colored material were removed until the unaltere` deposit was reached. A similar excavation of the fire that we had used for making our lunch during a period of several weeks produced a pit some 20 inches deep. We had not initially dug a fire pit, and the light-colorn material below the fire appeared to be the result of combustion of the organic matter in the fill and not ash in the sense of residue left by the combustion of wood. This has interesting implications for the interpreta- tion of the artifacts found in these fireplaces. If the fireplaces were not excavated prior to building the fir and the "ash" is, in fact, only the result of combustioa of the deposit already present, then the artifacts found in the fireplace are not associated with the fireplace any more than those outside but immediately adjacent to it. None of the artifacts in the fireplaces were actu ally found among the charcoal but were located below the wood in the "ash." Also none of these artifacts showed any signs of having been subjected to intense heat. There is no clear evidence, then, that the arti- facts found in the fireplaces were either thrown into the fire or fell out of material placed in or on the f The single feature that was not a fireplace, feature 3, was a small grinding or pounding pit. It consisted of an oval pit, 86 by 56 centimeters and about 13 ce meters deep at the deepest point. Near the center of the pit was an irregular slab of Table Mountain Sand- stone, 214 mm. by 194 mm. by 72 mm., which was both smoothed and battered on one face. The slab w set so that it sloped forward to the deepest part of pit. On the bottom of the pit toward its shallow end there was a considerable amount of carbonized vege- table matter, which ceased abruptly at a point about three-quarters of the way toward the deep end. At point there was a shoulder where the bottom of the dropped 1.5 to 2 centimeters. The absence of carbo material and the shoulder suggest that a container w placed at the lower end of the stone. 24 I I. o Keller: Montagu Cave in Prehistory TABLE 1 Correlations of Raw Materials with Selected Attributes of Layer 1 Scrapers Location of Trimming 'Primary Form End Side Two sides End and side Other Total No. Percent No. Percent No. Percent No. Percent No. Percent No. Percent Material: Table Mountain Sandstone lar waste 0 0.0 7 50.0 2 14.3 1 7.1 1 7.1 11 78.5 kes 2 14.3 1 7.1 0 0.0 0 0.0 0 0.0 3 21.4 efragments| 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 Total 2 14.3 8 57.1 2 14.3 1 7. 1 1 7.1 14 99.9 Material: Quartz lar waste 11 35.5 10 32.3 2 6.5 0 0.0 0 0.0 23 74.3 es 1 3.2 6 19.4 0 0.0 0 0.0 0 0.0 7 22.6 efragment 1 3.2 0 0.0 0 0.0 0 0.0 0 0.0 1 3.2 Total 13 41.9 16 51.6 2 6.5 0 0.0 0 0.0 31 100.1 Material: Chert lar waste 6 10.0 23 38.3 1 1.7 1 1.7 3 5.0 34 56.7 es 5 8.3 10 16.7 1 1.7 1 1.7 2 3.0 19 31.7 e fragments 3 5.0 3 5.0 0 0.0 0 0.0 1 1.7 7 11.7 otal 14 23.3 36 60.0 2 3.4 2 3.4 6 10.0 60 100.1 TABLE 2 Distribution of Tools, Cores, and Grindstones from Features in Layer 1 Feature 1 Feature 1A Feature 2 Feature 3 Feature 4 Feature 5 Feature 6 scents 1 - _ _ _ 1 1 uely truncated blade - _ _ _ _ 1 bnail scrapers 1 - 12 1 - 4 - pers 1 1 12 - 3 6 1 6caill6s | 3 - 2 1 1 2 - ted tool _ _ _ _ _ 1 ed flakes | _ - _ I 2 | le-platform cores 1 _ 1 1 _ 1 stones - - 3 1 - - - Totals 7 1 31 5 4 19 1 The vegetable matter was identified by M. J. Wells le Botanical Research Institute, Grahamstown. The hits of his analysis are summarized in Appendix III. 4t of the material was from the tubers of plants of eria and Morea, members of the Iris family, many [Which are eaten by people in the area today. Feature then, seems to have been used for the preparation, grinding and/or pounding, of plant foods collected the inhabitants of the cave. It seems probable that Ine kind of receptacle-perhaps a piece of skin or ket-was placed in the deep end of the pit toward which the grinding/pounding stone sloped, to receive the pounded pulp as it fell from the stone. The locations of the features are indicated in figure 7 and suggest that the area of most intensive occupation was toward the rear of the cave. The present drip line, shown in figure 2, tends to support this, indicating that the usable portion of the cave, at least so far as keep- ing dry was concerned, was a relatively small part of the total area. This interpretation assumes that the ancient drip and rain-strike lines were approximately what they are today. i I V? 25 Anthropological Records LAYER 2 DESCRIPTION The definitions used for Layer 1 apply here where the types are the same. Tools (992) Crescents (37, Dimensions Length Width Thickness W/L T/W Material Quartz Chert Trimmed face Unifacial 3.7%; plates XI:11-12; XIII:12, XV:11; XVI:4) Standard Range Mean deviation 15-42 mm. 29.9 mm. 7.0 mm. 3-19 mm. 12.5 mm. 3.3 mm. 1-8 mm. 3.3 mm. 1.2 mm. 17-65 44 10.7 13-57 27 8.6 No. Percent 6 16.2 31 83.8 Edge plan Crescent Trapezoidal Irregular 37 100.0 Backed Blades (12, 1.2%) Dimensions Length Width Thickness W/L T/W Range 17-36 mm. 6-19 mm. 2-4 mm. 35 -62 16 -50 Mean 26.3 mm. 12.8 mm. 3.1 mm. 51 27 Standard deviation 6.6 mm. 4.1 mm. 0.6 mm. 8.5 10.0 . Material Quartz Chert Primary form Unidentifiable Chips End-struck flakes Snapped flakes Condition Broken Whole Platform (6 pieces) " Bulb only" Plain Removed Vo. Percent Location of trimming 3 25.0 End and side 9 75.0 One side Trimmed face 2 16.7 Dorsal 3 25.0 Ventral 6 50.0 Undeterminable 1 8.3 Edge form Straight 3 25.0 Convex 9 75.0 Pln 3 1 2 Long quadrilateral 50.0 Short quadrilateral 16.7 Crescent 33.3 Irregular No. Percent 2 16.7 10 83.2 9 2 1 75.0 16.7 8.3 9 75.0 3 25.0 5 1 1 5 41.6 8.3 8.3 41.6 Backed Flakes (17, 1.7%; plate XVI:8, 14) Backed flakes are similar to backed blades except that they are not quadrilateral in shape and hence are not "blades" in the usually accepted sense. A few other pieces are included in this category that are not flakes, but to assign them to a separate category would serve no purpose. Dimensions Length Width Thickness W/L T/W Mean Range 22-65 mm. 10-24 mm. 2-8 mm. 32 -86 16-38 33.7 mm. 17.3 mm. 4.3 mm. 53 25 Standard deviation Mode 10.6 mm. 4.1 mm. 1.7 mm. 13.2 5.7 42, 43, 47 27 Mode 32 21 No. 26 8 3 Percent 70.4 21.6 8.1 Mode . 26 Keller: Montagu Cave in Prehistory Material Quartz Chert Primary form End-struck flakes Chips Split flake Plan Long quadrilateral Triangular Irregular No. P 2 15 13 3 1 1 3 13 'ercent Location of trimming 11.7 Both sides 88.3 One side Trimmed face 76.5 Ventral 17.6 Dorsal 5.8 Undeterminable Platform (13 pieces) 5.8 "Bulb only" 17.6 Plain 76.5 Facetted Simple facetted Obliquely Truncated Dimensions Length Width Thickness W/L T/W Blades (42, 4.2%; plates V:8; X:8; XV:14; XVI:5, 13) Range 16-47 mm. 6-18 mm. 2-6 mm. 25 -67 13 -67 Mean 26.8 mm. 12.6 mm. 3.0 mm. 44 27 Standard deviation 6.3 mm. 2.9 mm. 0.9 mm. 11.6 Material Quartz Chert Primary form Chip Unidentifiable Snapped flakes End-struck flakes Plan Long quadrilateral Crescent- shaped Sub-quadrilateral Short quadrilateral Irregular Trimmed face Dorsal Ventral Undeterminable No. I 3 39 1 6 11 24 13 3 3 2 21 38 1 3 Percent Location of trimming 7.1 Distal end and one side 92.9 One end Distal end Proximal end 2.3 Distal end and two sides 14.3 Proximal end and two 26.2 sides 57.1 One end and one side Platform (24 pieces) 30.9 Plain 7.1 "Bulb only" 7.1 Facetted 4.7 Simple-facetted 50.0 Removed Condition 90.5 Fragmentary 2.3 Whole 7.1 No. Percent 4 9.5 8 19.1 19 45.2 2 4.7 1 2.3 6 14.3 2 4.7 1 1 45.8 7 29.2 1 4.2 1 4.2 4 16.6 12 71.4 30 28.6 Thumbnail scrapers (4, 0.4%) Dimensions Length Width Thickness W/L T/W Range 14-44 mm. 16-19 mm. 3-5 mm. 36 -1 36 19-31 Material Chert Primary form Chip Side-struck flakes Chunk No. Percent Location of trimming 4 100 Distal side-flake Chip, one side Chunk, two ends 1 25 Plan 2 50 1 25 Irregular Crescent- shaped No. 1 16 Percent 5.8 94.1 1 5.8 15 88.3 1 5.8 2 8 2 1 15.4 61.5 15.4 7.6 Mode 41 27 Mean 21.5 mm. 16.7 mm. 3.7 mm. 100 22.4 Standard deviation 15.0 mm. 1.5 mm. 0.9 mm. 37.8 5.1 Mode No. 2 1 1 Percent 50 25 25 3 75 1 25 . 27 - Anthropological Records Hollow Scrapers (71, 7.1 %; plate XV:6) Dimensions Length Width Thickness W/L T/W Range 16-61 mm. 8-53 mm. 2-18 mm. 25-212 mm. 11-93 mm. Mean 32.4 mm. 17.3 mm. 4.8 mm. 56 28 No. Percent Location of trimming No. Percent Table Mountain Sandstone Quartz Chert Primary form Chips End-struck flakes Side-struck flakes Snapped flakes Plan Sub-quadrilateral Long quadrilateral Short quadrilateral Irregular Platform (34 pieces) " Bulb only" Plain Simple -facetted Facetted 5 2 64 15 30 2 24 5 12 3 51 7 24 2 1 7.0 Distal end and two sides 2.8 Proximal side 90.0 Split face of a split flake Two sides Two sides and one end 21.1 Distal end and side 42.2 One end 2.8 Distal side 33.8 One side 7.0 16.9 4.2 71.9 Trimmed face Ventral Dorsal Undeterminable 20.6 70.6 5.8 2.9 Multiple Hollow Scrapers (21, 2.1%; plate XV:4-5, 15) These are similar to hollow scrapers but they have more than teristic semicircular concave scraping edges. Dimensions Length Width Thickness W/L T/W Rangae 26-67 mm. 11-37 mm. 2-14 mm. 22-89 16-93 Mean 37.2 mm. 16.0 mm. 4.8 mm. 44 30 Standard deviation 10.6 mm. 5.9 mm. 2.5 mm. 14.6 15.9 two of the charac Mode 39 31, 33 No. Percent Location of trimming No. Percent Chert Primary form Chips End-struck flakes Snapped flakes Plan Irregular Long quadrilateral Short quadrilateral Triangular 21 100 Two sides One side 3 10 8 16 3 1 1 14.3 Trimmed face 47.6 Unifacial 38.1 Ventral Adjacently one side on opposite face 76.2 Platform (11 pieces) 14.3 4.7 Plain 4.7 "Bulb only" Facetted Standard deviation Material Mode 9.7 mm. 7.5 mm. 3.1 mm. 27.6 13.0 50 18 1 1 1 12 1 1 1 1 52 3 66 2 1.4 1.4 1.4 16.9 1.4 1.4 1.4 1.4 73.2 4.2 92.9 2.8 Material 10 47.6 1 1 52.4 18 85.7 1 4.7 2 9.5 8 1 2 72.7 9.1 18.2 -.-P, - - - - - - - - - 28 Keller: Montagu Cave in Prehistory Strangulated Scrapers (16, 1.6%; plates XV:12, XVI:7) These tools owe their name to the fact that they exhibit at least two irregular concave edges which are opposite to each other and produce the characteristic constricted plan. Dimensions Length Width Thickness W/L T/W Range 25-54 mm. 11-38 mm. 2-9 mm. 28-94 8 -36 Mean 39.8 mm. 19.3 mm. 4.0 mm. 50 22 Standard deviation 7.3 mm. 7.0 mm. 1.7 mm. 18.8 7.4 Mode 41 14, 19 Material No. Percent Chert Primary form Chips End-struck flakes Snapped flakes Plan Irregular Long quadrilateral Short quadrilateral Location of trimming 16 100.0 Two sides Two ends 4 10 2 12 3 1 25.0 Trimmed face 62.5 Dorsal 12.5 Undeterminable Platform (11 pieces) 75.0 " Bulb only" 18.7 Plain 6.2 Facetted No. Percent 15 93.7 1 6.2 15 93.7 1 6.2 4 5 2 36.4 45.5 18.2 Scrapers (189, 18.9%) plates IV:2; V:l1; VI:2-3; IX;4; XI:10; XII:1; XIV:6) Dimensions Range Mean Standard deviation Length Width Thickness W/L T/W Material 14-108 mm. 5-128 mm. 2-59 mm. 32-191 11-133 37.5 mm. 28.5 mm. 12.7 mm. 76 46 19.5 mm. 18.8 mm. 10.1 mm. 25.6 21.3 No. Percent Location of trimming Table Mountain Sandstone Quartz Shale Chert Primary form Unidentifiable Chips End-struck flakes Side-struck flakes Crystal fragments Crystal Split flakes Snapped flakes Flake fragments broken in two directions Chunks Plan Ovoid Round Sub-quadrilateral Sub-triangular Triangular Short quadrilateral Crystal- shaped Long-quadrilateral Quadrilateral Pointed Pebble - shaped Irregular Semicircular 60 25 2 102 14 30 27 10 11 1 6 15 2 73 4 6 4 3 5 6 7 6 1 1 1 144 1 31.8 End and side 13.1 All around 1.1 Two sides 54.0 One end Distal end Distal end and two sides 7.4 Distal end and one side 15.9 Both ends and both sides 14.3 Distal side and one end 5.2 Both ends and distal side 5.8 Two ends and one side 0.5 Distal side 3.2 Two sides and one end 7.9 One side 1.1 Trimmed face 38.6 Unifacial 2.1 3.2 2.1 1.5 2.6 3.2 3.7 3.2 0.5 0.5 0.5 76.2 0.5 Ventral Opposite sides and opposite faces Dorsal Bifacial Platform (40 pieces) "Bulb only" Plain Simple-facetted Facetted Removed 64 39, 50 No. Percent 14 7.4 4 2.1 27 14.3 25 13.2 10 5.2 3 1.5 3 1.5 1 0.5 1 0.5 1 0.5 1 0.5 5 2.6 1 0.5 93 49.2 90 6 1 90 2 3 23 3 8 3 47.6 3.2 0.5 47.6 1.1 7.5 57.5 7.5 20.0 7.5 Mode -----CD - : - - - --- - - - 29 Anthropological Records Outils Ecaill6s (10, 1.0%) There are two pieces that are double-edged and these are combined with burins. Dimensions Length Width Thickness W/L T/W Range 10-51 mm. 8-35 mm. 3-16 mm. 41-85 33 -80 Mean 24.4 mm. 16.0 mm. 8.3 mm. 66 51 Standard deviation 10.9 mm. 7.9 mm. 4.6 mm. 13.0 15.4 Material Table Mountain Sandstone Chert Quartz Primary form Chips Crystal fragments Chunks No. Percent 1 10.0 2 20.0 7 70.0 2 6 2 Location of trimming Two sides Two ends One end 70.0 60.0 20.0 Chisels (7, 0.7%) Dimensions Length Width Thickness W/L T/W Material Table Mountain Sandstone Quartz Chert Primary form Chunks Chips Crystal fragments No. 1 3 3 2 2 3 Percent 14.3 42.8 42.8 Location of trimming One end Two ends No. Percent 6 85.7 1 14.3 28.6 28.6 42.8 Burins (10, 1.0%; plates XI:7, 9; XV:10, 13) Dimensions Length Width Thickness W/L T/W Range 9-50 mm. 6-32 mm. 2-14 mm. 29-95 33 -75 Mean 27.9 mm. 14.8 mm. 7.7 mm. 55 51 Material Table Mountain Sandstone Quartz Chert Primary form Chips Chunks Crystal fragment End-struck flake No. 1 3 6 4 4 1 1 Percent Location of trimming 10.0 One end 30.0 Distal end 60.0 One side 40.0 40.0 10.0 10.0 30 Mode 69 No. 1 1 8 Percent 10.0 10.0 80.0 Range 13-36 mm. 9-25 mm. 3-11 mm. 44-86 12-79 Mean 26.6 mm. 16.4 mm. 7.6 mm. 64 50 Standard deviation 8.9 mm. 5.6 mm. 3.1 mm. 14.7 20.1 Mode Standard deviation 13.9 mm. 8.2 mm. 4.5 mm. 17.5 13.7 Mode No. 6 1 3 Percent 60.0 10.0 30.0 . - - - Keller: Montagu Cave in Prehistory Discoid (1, 0.1%) This is an artifact that is trimmed unifacially. The trimming is smaller than that on the disc cores and for that reason it is classed as a tool rather than a core. It is made of chert and trimmed on one side unifacially. Dimensions Length 19 mm.; Width 12 mm.; Thickness 5 mm. Nosed Tool (1, 0.1%) This is a chunk of chert trimmed on one end. Dimensions Length 3 1mm.; Width 16 mm.; Thickness 6 mm. Pointed Tools (3, 0.3%) These are small, amorphous Dimensions Length Width Thicknes s W/L T/W Range 18-69 mm. 11-41 mm. 6-22 mm. 46-61 54-91 pointed tools. Mean 37.0 mm. 21.0 mm. 12.0 mm. 55 66 Material Table Mountain Sandstone Quartz Primary form Chunk Crystals No. Percent 1 33.3 2 66.6 1 33.3 2 66.6 Trimmed Flakes (300, 30%/; plates III:1, 5-6, 10; IV:1, 4; V:3; VI:4, 9; VII:2, 6-7; VIII:6; X:6-7, 15; XI:15; XII:4-5, 8, 17; XIII:8, 11; XIV:4; XVI:1, 9) Range 9-147 mm. 3-116 mm. 1-52 mm. 17-147 10-550 Mean 44.5 mm. 26.1 mm. 8.1 mm. 58 34 Standard deviation 26.5 mm. 16.6 mm. 3.9 mm. 22.2 37.8 No. Percent Location of trimming No. Table Mountain Sandstone Quartz Chert Primary form Side-struck flakes Split flakes Snapped flakes Flake fragments End-struck flakes Plan Sub-quadrilateral Long quadrilateral Short quadrilateral Triangular Ovoid Pointe d Irregular 98 14 188 9 13 95 6 177 1 64 39 24 1 2 169 32.6 Distal end and one side 8 4.7 Distal end and two sides 5 62.7 Proximal end and two sides 2 Distal side 3 One e nd 4 3.0 One end and one side 7 4.3 Distal end 10 31.6 Two sides 86 2.0 Split face 2 59.0 Two ends 1 One side 172 0.3 21.3 13.0 8.0 0.3 0.6 56.4 Trimmed face Ventral Bifacial Unifacial Adjacently on the same side but on opposite faces Parti-bifacial Dorsal 25 8 5 4 1 257 Mode 46 38 Percent 2.7 1.6 0.6 1.0 1.3 2.3 3.3 27.6 0.6 0.3 57.6 8.3 2.7 1.6 1.3 0.3 85.7 Standard deviation Mode 12.9 mm. 17.3 mm. 8.3 mm. 6.8 17.3 Dimensions Length Width Thickness W/L T/W Material - - - - - - - - 3 1 Anthropological Records Platform (197 pieces) No. Percent Plain Simple -facetted Facetted "Bulb only" Removed 109 26 41 18 3 55.4 13.2 20.8 9.1 1.5 Trimmed Chips (139, 14.0%; plates VIHI:7; X:10, 17) Dimensions Length Width Thickness W/L T/W Range 12-160 mm. 6-83 mm. 2-29, mm. 22-14 11-100 Mean 34.5 mm. 22.6 mm. 6.8 mm. 69 30 Standard deviation 20.8 mm. 12.6 mm. 4.8 mm. 22.8 14 Material Table Mountain Sandstone Quartz Chert Trimmed face Ventral Bifacial Unifacial Parti-facial Dorsal No. Percent Location of trimming 34 24.4 Two sides 8 5.7 One end 97 69.9 All around End and one side Two ends 7 5.0 One side 14 31 2 85 10.1 22.3 1.4 61.1 No. Percent 27 5 1 10 1 95 19.4 3.6 0.7 7.2 0.7 68.3 Trimmed Chunks (45, 4.5%; plate VII:8) Dimensions Length Width Thickness W/L T/W Range 10-174 mm. 8-130 mm. 2-92 mm. 40- 133 5-121 Mean 54.4 mm. 38.9 mm. 20.0 mm. 73 49 No. Percent Location of trimming No. Percent Table Mountain Sandstone Quartz Che rt Trimmed face Unifacial Ventral Dorsal Bifacial 24 53.4 One end 6 13.3 Two sides 15 33.4 One end and one side One side 19 5 12 9 5 9 4 27 11.1 20.0 8.9 60.0 42.2 11.1 26.7 20.0 Points (45, 4.5%; plates VII:10; X:12, 16; XI:2, 13; XUI:15, 18; XV:7-9) The tools in this category are sharp. Pieces that are pointed in the scraper category. Untrimmed be found in the waste categories. trimmed to a sharp point. Dimensions Length Width Thickness W/L T/W Range 19-109 mm. 12-56 mm. 3-18 mm. 42-103 19-46 characterized by a tip that, in plan at least, is plan but not trimmed to the tip are included in triangular flakes are not included here but will All "points" appear to have been intentionally Mean 44.8 mm. 28.8 mm. 9.0 mm. 68 32 Standard deviation Mode 20.4 mm. 10.4 mm. 3.4 mm. 16.5 6.8 58 29 Mode 19 Standard deviation Material Mode 36.8 mm. 27.3 mm. 18.5 mm. 18.1 21.9 73 42 _ . . . .. . . 32 . Keller: Montagu Cave in Prehistory Material Table Mountain Sandstone Quartz Chert Amount of trimming by face Bi-facial Parti-bifacial Unifacial Trimmed face Dorsal Ventral No. Percent Primary form 13 28.8 Unidentifiable 2 4.3 Chips 30 66.6 End-struck flakes Side- struck flakes Platform (25 pieces) 7 15.5 Facetted 7 5. Plain 3 6.6 Pli 35 77.9 Simple-facetted Reduced Condition 44 97.0 1 3.0 Fragmentary 1 3.0 ~Whole No. Percent 12 26.7 6 13.3 26 57.7 1 2.2 10 40.0 11 44.0 3 12.0 1 4.0 9 20.0 36 80.0 Point Tips (3, 0.3%) All are made of Table Mountain Sandstone. One is bifacial and two unifacial. Core Scrapers (5, 0.5%) Dimensions Length Width Thickness W/L T/W Range 31-104 mm. 26-92 mm. 13-71 mm. 78 -88 50-91 Mean 72.4 mm. 60.6 mm. 46.6 mm. 83 75 Standard deviation 35.1 mm. 29.9 mm. 24.7 mm. 3.5 15.5 Mode Material Table Mountain Sandstone Chert Location of trimming One side One end Two sides All the way around No. 4 1 2 1 1 1 Percent 80.0 20.0 40.0 20.0 20.0 20.0 Choppers (5, 0.5%) Dimensions Length Width Thickness W/L T/W Range 13-89 mm. 15-65 mm. 5-56 mm. 63-115 33-100 Mean 66.0 rr 49.6 rr 49.0 rr 83 73 Primary form Chunks- I- -- Chunks Pebbles Trimmed face Dorsal Opposite sides on opposite faces Standard deviation im. 30.6 mm. im. 19.8 mm. nm . 17.4 mm. 18.8 27.7 Material No. Percent Table Mountain Sandstone 4 80.0 Quartz 1 20.0 Location of trimming One side 4 80.0 Two sides 1 20.0 Primary form Chunks Crystal fragment No. Percent 4 80.0 1 20.0 No. 3 2 4 1 Percent 60.0 40.0 80.0 20.0 Mode - - - 33 - Anthropological Records Unidentified (9, 0.9%) These are amorphous trimmed pieces. Dimensions Range Mean 15-64 mm. 11-31 mm. 3-17 mm. Length Width Thickness W/L T/W Material Table Mountain Sandstone Chert Trimmed face Unifacial Bifacial Plan Irregularly shaped Quadrilateral 31.7 mm. 19.2 mm. 7.2 mm. 14.9 mm. 7.2 mm. 4.6 mm. No. Percent Primary form 2 22.2 7 77.7 Unidentifiable Chip Chunk 6 66.6 Location of trimming 3 33.3 One side Two sides One end 8 88.8 One end and side 1 11.1 Cores (614) Single -Platform Cores (237, 38.8%; plates V:5-6; VII:4, 13; IX:2; X:3, 9, 18; XI:5, 7; XIII:10, 18; XIV:1; XVI:6) Dimensions Length Width Thickness W/L T/W Range 12-151 mm. 6-80 mm. 2-61 mm. 33 -193 7-124 Mean 32.8 mm. 26.6 mm. 14.4 mm. 82 61 Standard deviation Mode 17.9 mm. 14.7 mm. 10.2 mm. 18.1 21.3 100 64 Material Table Mountain Sandstone Quartz Chert Trimmed face Dorsal Ventral Bifacial Multiple Unifacial Platform Cortex Facetted Plain Negative scar Simple-facetted Thin Crystal faces Reduced No. 42 60 1;35 14 2 3 1 217 83 49 18 57 10 12 7 1 Percent 17.7 25.3 57.0 5.9 0.8 1.2 0.4 91.6 35.1 20.7 7.6 24.1 4.2 5.0 2.9 0.4 Primary form Unidentifiable Chips Chunks End-struck flakes Side-struck flakes Flake fragment which is broken in two directions Snapped flakes Split flake Crystal fragments Crystals Location of trimming One end One side All around One end and side Distal end From a split face and on one end Undeterminable Multiple-Platform Cores (82, 13.3%; plates VIII:12; XI:8, 16; XIII:17; XVI:10) Range 14-140 mm. 2-103 mm. 4-60 mm. Mean 32.5 mm. 26.6 mm. 13.9 mm. 4-116 83 19-100 61 Standard deviation Mode 17.7 mm. 14.7 mm. 8.6 mm. 15.9 86, 96, 100 73.1 50 Standard deviation Mode No. 7 1 1 Percent 77.7 11.1 11.1 66.6 11.1 11.1 11.1 6 1 1 1 No. 76 7 109 5 3 1 4 1 29 2 124 66 3 9 8 1 26 Percent 32.0 2.9 46.0 2.2 1.2 0.4 1.8 0.4 12.2 0.8 52.3 27.8 1.2 3.8 3.6 0.4 10.9 Dimensions Length Width Thickness W/L T/W - - - -- - - - - - - - - - I 34 Keller: Montagu Cave in Prehistory No. Percent Primary form No. Percent Table Mountain Sandstone Quartz Chert Worked faces Bifacial Unifacial Platform Facetted Crystal face Simple -facetted Negative scar Plain Cortex Thin Disk Cores (158, 25.8%; p XVI:2-3, 11) Dimensions Length Width Thickness W/L T/W Range 18-112 mm. 5-94 mm. 2-75 mm. 55-119 11-90 17 13 52 _37 20.7 Chunks 15.8 Crystal fragments 63.4 Crystal End-struck flakes Snapped flakes 45.1 Unidentifiable 45 54.9 29 2 6 43 14 44 4 20.2 1.4 4.2 30.0 9.8 31.0 2.8 Location of trimming Two sides One end and one side One side and both ends Two ends and one side Both sides and both ends Both ends plates V:4; VII:5, 9; VIII:2, 5; IX:1; X:4-5; XI:14; Mean 43.3 mm. 37.8 mm. 19.5 mm. 89 50 Standard deviation 19.5 mm. 17.0 mm. 12.1 mm. 11.1 15.5 No. Percent Primary form No. Percent Table Mountain Sandstone Quartz Chert Plan Triangular Irregular Semi- circular Sub-quadrilateral Quadrilateral Sub-triangular Elongated Ovoid Circular 77 20 61 2 17 7 10 3 1 1 26 91 Biconical Cores (5, 0.8%) These are similar in shape in size. Dimensions Length Width Thickness W/L T/W Range 19-64 mm. 29-69 mm. 21-54 mm. 74-100 65-90 48.6 Chip 12.6 Chunks 38.8 Crystal fragments Side- struck flakes Split flake 1.2 Snapped flakes 10.7 Unidentifiable 4.4 6.3 1.9 0.6 0.6 16.4 57.6 Trimmed face Unifacial Parti-bifacial Bifacial to the Acheulean biconical cores but are smaller Mean 51.4 mm. 43.4 mm. 33.6 mm. 85 77 Standard deviation 21.2 mm. 18.1 mm. 15.0 mm. 8.9 8.3 No. Percent Primary form No. Percent Chunk Unidentifiable Plan Round Ovoid Material 35 4 1 1 1 40 42.6 4.8 1.2 1.2 1.2 48.7 12.1 28.0 1.2 3.7 1.2 53.7 10 23 1 3 1 44 Material Mode 90 50 1 2 1 1 1 4 148 51 17 90 0.6 1.2 0.6 0.6 0.6 2.5 93.6 32.3 10.7 57.0 Material Mode Table Mountain Sandstone 3 Quartz 1 Chert 1 60.0 20.0 20.0 1 20.0 4 80.0 3 60.0 2 40.0 - - = - K - - - - - - - - - - - - - - - - -- 35 Anthropological Records Radial Cores (12, 1.9%; plates XIII:20; XIV:2) These cores are trimmed radially like disc cores but have less regular cross sections than the latter. Dimensions Length Width Thickness W/L T/W Range 23-74 mm. 22-74 mm. 6-34 mm. 68-1000-- 20-77 Mean 47.4 mm. 41.0 mm. 18.5 mm. 87 45 Standard deviation Mode 17.1 mm. 16.0 mm. 9.0 mm. 11 16 43 Material Table Mountain Sandstone Chert Primary form Chunks Unidentifiable No. Percent Trimmed fa 7 58.3 Unifacially 5 41.7 Bifacially Parti-facial 4 33.3 8 66.6 Irregular Cores (77, 12.5%; plates VI:1; XI:14; XII:3; XIII:16, 19; XV:1-2) The cores in this category are similar to disc cores in plan and cross section, but they lack the regular trimming found on disc cores and on single-platform cores. The trimming is generally radial but appears random. Individual flakes produced from such cores could have been either irregular or parallel-sided. Although no charac- teristic patterns of edge damage are present, it is possible that these are tools rather than cores. Dimensions Length Width Thickness W/L T/W Range 2-139 mm. 13-82 mm. 3-48 mm. 53 -123 15 -93 Mean 38.1 mm. 32.1 mm. 16.8 mm. 84 51 Standard deviation Mode 19.2 mm. 14.7 mm. 10.2 mm. 13.5 16.2 94 40 Material Table Mountain Sandstone Quartz Chert Cross section Lenticular Asymmetrical Bevelled-base High-backed Flake Pebble Crystal Irregular No. 21 7 49 16 11 1 1 1 1 2 44 Percent Primary form 27.2 Chunks 9.1 Crystals 63.6 End-struck flakes Unidentifiable 20.8 14.3 1.3 1.3 1.3 1.3 2.6 57.1 Worked face Multiple faces Bifacial Parti-bifacial Unifacial No. Percent 30 3 1 43 7 33 1 36 38.9 3.8 1.3 55.9 9.1 42.9 1.3 46.7 Struck Cores (19, 3.1%; plate VUI:1) Cores in this category are characterized by a single large flake scar on one face that truncates radial scars on the same face. They are similar to "levallois" cores. Dimensions Length Width Thickness W/L T/W Range 19-64 mm. 17-56 mm. 4-23 mm. 66-106 21-71 Mean 35.1 mm. 29.6 mm. 12.7 mm. 84 44 Standard deviation Mode 12.8 mm. 11.4 mm. 5.4 mm. 10.3 11.1 81 46 No. Percent Primary form No. Percent Table Mountain Sandstone 3 15.7 Chunksi2n10.5 Chert 16 84.3 Unidentifiable17 8. No. 9 2 1 Percent 75.0 16.7 8.3 Material _ - - -1 - - - - - -,-- - -- 36 2 10.5 17 89.5 Keller: Montagu Cave in Prehistory Trimmed face Parti-bifacial Unifacial Bifacial No. Percent 2 10.5 13 68.5 4 21.0 Platform Facetted Thin Negative scar Plain No. Percent 9 47.4 6 31.6 1 5.2 3 15.7 Formless Cores (24, 3.9%/) These cores are similar to the Acheulean type trimmed in three or more directions on three or Dimensions Length Width Thickness W/L T/W Range 18-84 mm. 13-72 mm. 2-62 mm. 52-112 8-104 Mean 37.4 mm. 32.7 mm. 22.3 mm. 86 66 of the same more faces. Standard deviation 17.9 mm. 17.4 mm. 15.4 mm. 14.1 19.5 name; they are Mode 92 63 Material Table Mountain Sandstone Quartz Chert Primary form Chunks Unidentifiable No. 6 4 14 Percent 25.0 16.6 58.4 Plan Sub-quadrilateral Irregular Cross section 18 75.0 Trihedral 6 25.0 Irregular cross section No. 1 23 2 8.3 22 91.7 Utilized Pieces (338) Grindstones (2, 0.5%) Both specimens are made of Table Mountain Sandstone and are of irregular shape. One is unifacial and the other bifacial. Standard Dimensions Range Mean deviation Mode Length 97-171 mm. 134.0 mm. 52.3 mm. Width 94-120 mm. 107.0 mm. 18.4 mm. Thickness 78-91 mm. 84.5 mm. 9.2 mm. W/L - T/W Hammerstones (7, 2.7%o) All are made of Table Mountain Sandstone. Five are pebbles and two chunks. Standard Dimensions Range Mean deviation Mode Length Width Thickness W/L T/W 13-89 mm. 25-94 mm. 15-70 mm. 68-98 5-81 90.0 mm. 69.7 mm. 48.0 mm. 79 68 33.4 mm. 23.3 mm. 18.6 mm. 9.4 9.2 Pestle (1, 0.3%; plate IX:3) This is a pebble of Table Mountain Sandstone that as a pestle. Length 140 mm.; Width 60 mm; Thickness 48 mm. Dimpled Pounder (1, 0.3%) This is a Table Mountain Sandstone pebble. has been used on one end Length 110 mm.; Width 77 mm.; Thickness 64 mm. Percent 4.1 95.8 - = 37 Anthropological Records Anvil (1, 0.3%%) This is made of a chunk of Table Mountain Sandstone and is battered on one end. Length 133 mm.; Width 122 mm.; Thickness 50 mm. Flakes (113, 33.3%) Material No. Percent Table Mountain Sandstone 14 12.4 Chert 81 71.7 Quartz 18 15.9 Chips (72, 21.3 %) Material No. Percent Table Mountain Sandstone Che rt Quartz 7 50 15 9.7 69.5 20.8 Flake Fragments (56, 16.5%) Material No. Percent Table Mountain Sandstone 5 8.9 Chert 44 78.5 Quartz 7 12.5 Chunks (31, 9.1%) Material No. Percent Table Mountain Sandstone Chert Quartz 7 13 11 22.6 42.0 35.4 Pigment (54, 15.9%) These are abraded pieces of hematite. LAYER 2 DISCUSSION The most striking feature of the artifacts from this level is their variety in size, form, and raw material. Large flakes of Table Mountain Sandstone up to 15 cm. long are associated with flakes less than 1 cm. long made of quartz or chert; triangular flakes with con- vergent dorsal scars and facetted platforms are asso- ciated with small, parallel-sided flakes with "bulb only" platforms; and large trimmed flakes, disc cores, and "levallois type" cores are associated with backed blades, crescents, and small single-platform cores. Undoubtedly, if this assemblage had been collected from the surface of an open site, it would have been interpreted as a mixture and separated into a "Middle Stone Age" and a "Later Stone Age" industry, but the association of all those elements in Layer 2 is incontrovertible in a context where mixing is impossible. It will be recalled from the discussion of the strati- graphy that, although there was no visible stratigraphy within Layer 2, there were seven concentrations of artifacts that were excavated as occupation surfaces. The deposit below surface VII was removed as a single unit with the exception of two adjacent squares in the corner of the excavation where the deposit was thickest; these were excavated in six-inch levels. Also, at the very bottom of Layer 2 there was a thin band that b a higher density of charcoal and artifacts than the r of the layer. In addition, there were a few squares which the upper surfaces were dug through before tb were discovered. The material, then, has been colle from five kinds of excavation units in this layer: the material from above surface I and from those squar where the upper surfaces were dug through; the Bur faces and the material from between them; material from below surface VII; material from the arbitrary levels below surface VII; and the band with high art fact and charcoal density at the bottom of the layer, The differential concentration of artifacts between the top and bottom of the layer was also mentioned the description of the stratigraphy. This presents problems in the determination of what differences be present in the assemblages contained within Lay 2 because there are more artifacts on the lower faces than on the upper ones. An inspection of table will show quite clearly these differences in the num bers of tools present at the top of Layer 2 and at bottom. The thin band at the bottom contained forty.. i 1 1 II 44 J i 38 Keller: Montagu Cave in Prehistory TABLE 3 Correlations of Raw Materials with Selected Attributes of Layer 2 Scrapers 39 i ~~~~~~~~~~Location of Trimming rPrimary Form End Side Two sides End and side Other Total r i ~~~No. Percent No. Percent No. Percent No. Percent No. Percent |No. Percent Material: Table Mountain Sandstone 'Angular waste 5 8.5 18 30.5 5 8.5 3 5.1 1 1.7 32 54.3 Flakes 3 5.1 5 8.5 1 1.7 4 6.8 2 3.4 15 25.4 Plake fragments 4 6.7 4 6.7 1 1.7 0 0.0 0 0.0 9 15.3 Other 0 0.0 0 0.0 1 1.7 1 1.7 1 1.7 3 5.1 Total 12 20.3 | 27 45.7 | 8 13.6 8 13.6 4 6.8 59 Material: Quartz Angular waste 8 32.0 12 48.0 1 4.0 0 0.0 1 4.0 22 88.0 Flakes 1 4.0 1 4.0 0 0.0 0 0.0 0 0.0 2 8.0 lF1ake fragment 0 0.0 0 0.0 0 0.0 1 4.0 0 0.0 1 4.0 Other 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 Total 9 36.0 13 52.0 1 4.0 1 4.0 1 4.0 25 Material: Chert Angular waste 9 8.8 37 36.2 7 6.8 5 4.9 1 0.9 59 57.8 Flakes 2 1.9 12 11.7 2 1.9 2 1.9 2 1.9 20 19.6 Flakefragments 1 0.9 5 4.9 6 5.8 0 0.0 1 0.9 13 12.7 Other 1 0.9 1 0.9 3 2.9 3 2.9 2 1.9 10 9.8 Total 13 12.7 55 53.9 18 17.6 10 9.8 6 5.8 102 Material: Shale Angular waste 0 0.0 1 50.0 0 0.0 0 00.0 0 0.0 1 50.0 Other 1 50.0 0 0.0 0 0.0 0 0.0 0 0.0 1 50.0 Total 1 50.0 1 50.0 0 0.0 0 0.0 0 0.0 2 TABLE 4 Correlations of Raw Material with Selected Attributes of Layer 2 Trimmed Flakes Table Mountain Sandstone Quartz Chert Total Platform No. Percent No. Percent No. Percent No. Percent Bulb only 3 1.5 3 1.5 11 5.6 17 8.5 S?lain 35 17.8 5 2.5 68 34.5 108 54.8 imple-facetted 11 5.6 0 0.0 16 8.1 27 13.7 Facetted 13 6.6 2 1.0 27 13.7 42 21.3 %emoved 1 0.5 0 0.0 2 1.0 3 1.5 Total 63 32.0 10 5.0 124 62.9 197 Anthropological Records TABLE 5 Tools and Cores from Layer 2 .- ca C' d q LO CD Q) a) Q) ) a) Q > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~> > > a) > > 0 Q Q Q Qb >x?a,a>> a0: o >0 .0 Cd >0 14t044 . Cd Cd - o- a- - - -a -a a a a > >> > > > > g ad a Cd C a C . 3 a Cd a Cd Cd a~~~~~~~~~~~~~~~~~~~~~~~~~~~~~C) C) C.) a.) C.) a) Cd cd -- Cd 4 JC C Cd > ~~~~~ ~ ~~~~~~~~~~~~~~~~~~~~Cd Cd Cd d c . ) a) a) a) a) a) a) a) a) a) ) a) a) a m ) ( / 1 m ( m C) C) C) C) a) C) a) C) a C. ) C ) C. ) C.) Cd C.) Cd Cd Cd Cd Cd Cd ~-Cd cd 4- 4-4 ~ 4 + .4- 4- ~ - 4 4, -4 4'4 0 0 o o 0 0 a a a (aLa~ a a a ) a) a1) a) aL) (12 (/2 (1 (/2 c (/ En P (/2 p (/2 m cr2 PQ (/2 u( 2 m Crescents Backed blades Backed flakes Obliquely trun- cated blades Thumbnail scrapers Hollow scrapers Multiple hol- low scrapers Strangulated scrapers Scrapers Outils ecailI-s Chisels Burins Discoid Points Trimmed flakes Trimmed chips Trimmed chunks Pointed tools Choppers Core scrapers Nosed tool Point fragments Unidentifiable Single plat- form cores Double plat- form cores Disc cores Formless cores Irregular cores Biconical cores Struck cores Radial cores 2 1 1 - - 1 2 1 16 2 2 - 5 1 1 - 11 - - - - - 1 - 1 -29 1 3 6 1 1 1 1 1 3 1 2 4 1 2 - 2 -1 1 1 - - 1 2 4 - 10 - 28 7 1 12 - 3 1 1 - 3 _ - 1 - 5 1 4 2 2 - - - 1 - - 1 - - 1 2 5 4 17 2 2 1 - 1 _ - ~~~~~~~~1 1 3 2 1 1 1 - 1 - 7 2 1 1 - 1 _ - 1 2 3 1 11 1 7 2 3 _ - - 13 - 3 4 24 10 47 1 1 1 1 - 2 1 3 -_ - - 1 1 1 1 2 2 12 2 39 2 7 - 7 1 2 3 8 6 3 7 1 31 27 23 2 58 24 13 5 17 3 1 1 - - 1 - - 1 - - 9 7 1 - 3 3 1 1 1 - 3 7 7 1 2 2 14 6 2 - - - 6 2 1 - - 1 - 2 1 1 4 3 5 3 8 7 3 19 7 87 18 18 19 1 3 - _ _ - 1 - - 1 - - 1 1 - 1 _ - 4 1 9 3 3 7 40 7 8 1 4 30 8 50 1 5 8 5 - - - 1 - - - - 2 1 14 1 _ - 2 1 1 1 2 8 - 28 10 12 - 1 - - 1 2 _ - 3 1 1 - 1 - 6 1 3 1 2 1 1 3 4 9 - 1 - 1 1 1 1 1 3 1 3 8 - 1 -2 1 12 26B 1 - - 3 - 2 - I1 1 1 - _ - -L 14 3 11 6 4 2 3 - 1 % 1 1 - 1 2 8 4 - 1 2 3 2 - - 2 - - 1 I 40 Totals 11 12 2 2 6 6 5 18 12 33 16 26 46 26 172 55 620 139 114 92 75 37 21 65 i 14 i N I I I I 1 Keller: Montagu Cave in Prehistory pee tools, and in order to have a comparable number in the top, it is necessary to combine the tools from ive surface I through surface III, a total of forty- en tools. If we then compare these two units, we ,that the two most numerous types of tools in both the trimmed flakes and scrapers, 31.9 percent and percent at the top, and 28 percent and 21 percent hbe bottom. This is also true of Layer 2 as a whole, pre trimmed flakes make up 30 percent and scrapers percent of the tools. ghe only types that are not present in the top part hbe layer but are found in the bottom are crescents, ked blades, and backed flakes. This may not be a "ficant difference, however, since the samples are Small-less than fifty-and since in the total assem- We these types constitute only 3.7 percent, 1.2 per- , and 1.7 percent respectively. The mean length of end-struck flakes in the total ier 2 assemblage is 4.06 cm.; in the bottom band s 3.87 cm.; and in the top units 4.04 cm. Of the 33 pieces of flake waste in the bottom band, 65.9 ~cent are Table Mountain Sandstone, 18.9 percent rt, and 15.2 percent quartz. Of the 3,925 pieces of ke waste on surface II and above, 63.4 percent are de Mountain Sandstone, 23.9 percent chert, and 12.7 ~cent quartz. Since the material from the top and :bottom of Layer 2 is so very similar, the assem- ge will be discussed as a whole. Tools with more or less shallow, unifacial edges Lt is, trimmed flakes, chips, and chunks) make up percent of the tools and are followed by all types scrapers, which total 30.6 percent. The three next it numerous types are points, obliquely truncated des, and crescents, which account for 4.5 percent, percent, and 3.7 percent of the tools respectively. variety of flake and platform types present was nijoned above, and figure 12 shows that flakes with tcular kinds of platforms were used for making Wcular kinds of tools. The correlation table of platform types and raw berial for trimmed flakes shows that a flake pro- tion technique involving a plain platform was used It frequently, followed by a technique involving a tted platform. This was true for all raw materials. ong he scrapers the preferences were the same as e from Layer 1. A piece of angular waste, the it commonly occurring artifact form, was trimmed ig one side, that is along part of the longest part Phe margin. The same relationships among raw brials are present in the Layer 2 material as in rer 1. More quartz pieces were trimmed on the than specimens of other raw material. Surface I (see fig. 17) was very near the top of the Layer 2, and parts of it were removed before we were able to isolate it as a horizon. The excavation pro- cedure was outlined earlier. There is very little that can be said about the distribution of the artifacts on surface I because they are so few. Although feature 3 belonged to Layer 1, it cut through the upper surface of Layer 2, and so is indicated on the appropriate plots. Much the same is true of surface II (see fig. 18), although there seem to be more artifacts in squares 30 F and G than in the others. There were also three fireplaces on this surface, two in 30 F and G and one in 20 E and F. On surface III (see fig. 19) the material is evenly distributed with the exception of two concentrations, one against the wall of the cave in square 30 G and another at the edge of 30 F. There also seems to be a less obvious concentration of material in 20 E near the large fall rock. The area plotted on surface IV (see fig. 20) is greater than the previous three because we were able, in excavating this surface, to move out into the 15 line of squares for the first time. These had been occupied by large fallen rocks, which were not moved until the excavation reached the level of the surface on which they were resting. Again, as with the other surfaces, there are concentrations of artifacts in squares 30 F and G. An even larger area was plotted on surface V (see fig. 21) than on IV as it was possible to remove more of the fallen rocks toward the front of the cave. There is a concentration of artifacts next to a small fireplace in 30 F and also a dense strip of material more or less along the boundary between the F and the G squares. On surface VI (see fig. 22) there is a concentration of artifacts in the southwest half of square 30 F and an interesting concentration of natural stones on the other half of the square, but unfortunately it is impos- sible to interpret these concentrations in terms of functional units. In 20 G and extending into 15 G there is a long narrow area which is devoid of stone, either natural or artifactual; this again is interesting but functionally uninterpretable. By far the greatest density of material on any of the surfaces was found on surface VII (see fig. 23). In squares 20 F and extending into 20 G and 25 F and G there is an interesting oval ring of material with a heavy concentration along its west side. There is also another small concentration in 25 F. There are two justifications for excavating surfaces or horizons of artifacts in the way we have done. First, uncovering and plotting the distribution of artifacts I 41 Anthropological Records makes it possible to speak more precisely about the concentrations that occur than if the artifacts were simply labelled by their grid unit. This latter tech- nique has been used by Cooke in his report on Pomongwe (1963), and he was able to show that more cores occurred near the front of the cave than toward the rear and so was able to interpret something about areas in which different activities were practiced. But if the Pomongwe material had been plotted as it was done at Montagu, a more detailed interpretation might have been possible. Second, even if there is no interest in the distribution of the artifacts on a surface, the excavation of such a surface in this way provides a unit of contemporaneity; that is, it can be assumed that the material was deposited over a short span of time and so represents more accurately the tools and waste being produced by the site's occupants at a given time than would the material from any arbitrary level. Therefore, even though one is presently unable to interpret the concentrations of artifacts described above in terms of functional units, from the plots of these seven surfaces it is possible to make a general statement about the use of the cave. It seems clear that the area of heaviest occupation, if the density of the artifacts is a true reflection of this, was the rear of the cave. Unfortunately, most of this area was re- moved by the 1919 excavation, and the portion of the cave that we excavated covered only part of it. In Layer 1 the features were concentrated toward the rear of the cave, and the plots of artifact distributiox on the four surfaces in Layer 5 give the same impr sion. The rear of the cave seems, therefore, to have been the area most intensively occupied throughout tb time that the cave was inhabited. It had been hoped that plotting the distribution of the tools would produce some information about the functional units (the tools that were used together) which make up the assemblage. But this was not the case. The greatest number of tools found on any sing living surface in Layer 2 was thirty on surface VII, which covered more than two hundred square feet; clearly this is not a high enough density to be infor tive, since the tools do not occur close enough toge to show any association of types. The results are somewhat disappointing, then, at least in terms of functional interpretations, but they are very important for understanding the constitution, of the assemblage since they show quite clearly the intimate association of the typologically varied mate that was found. In the comparative section it will be indicated that an assemblage of this kind has never been adequately described, and, further, that the de stration of the association of these diverse elements on a series of occupation horizons is of great value. LAYER 3 DESCRIPTION Where Kleindienst's typology does not apply, the tool types are defined below. Incomplete specimens were not used in determining the W/L and T/W figures and so those categories consisting entirely of broken pieces, such as "Biface Butts," have no W/L or T/W ratios as part of their description. All artifacts from Layer 3 are Table Mountain Sandstone. Tools (56 3) Hand-axes (76, 13.5%; plates XVIJI:2,8; XIX:2-4; XX:2-5; XXI:2; XXII:1,8; XXVI:4; XXVII:3 -4; XXVIII:1) Dimensions Length Width Thickness W/L T/W Range 9-27 cm. 5-17 cm. 2-6 cm. 38-13 29 -75 Mean 14.7 cm. 8.8 cm. 4.0 cm. 60 47 Standard deviation 3.4 cm. 1.8 cm. 0.9 cm. 13.5 9.5 Mode 53 50 - - - i II .1 I N I I 42 Keller: Montagu Cave in Prehistory Primary form Side-struck flakes Chunks Chip Unidentifiable Trimmed face Parti-bifacial Unifacial Bifacial Butt trimming Untrimmed Trimmed unifacially Partially trimmed Rolled Bifacially trimmed Cortex Finish Rough Fine Very fine Coarse No. Percent Plan 9 2 1 64 10 2 64 16 9 16 1 33 1 18 19 6 33 11.8 Triangular 2.6 Sub-triangular 1.3 Ovate 84.2 Irregular ovate Asymmetrical ovate Long ovate 13.2 Asymmetrical long ovate 2.6 Irregular long ovate 84.1 Ovate-acuminate Lanceolate Elongated lanceolate 21.0 Irregular lanceolate 11.9 Asymmetrical lanceolate 21.0 Pointed 1.3 Cordiform 43.4 Elongate 1.3 Limande 23.7 25.0 7.9 43.4 Condition Broken Whole No. Percent 2 2.6 6 7.9 10 13.2 3 3.9 5 6.6 13 17.2 6 7.9 3 3.9 4 5.3 6 7.9 1 1.3 2 2.6 3 3.9 2 2.6 2 2.6 1 1.3 7 9.2 4 5.3 72 94.7 Hand-axe flakes (6, 1.1%) Dimensions Length Width Thickness W/L T/W Range 12-15 cm. 7-10 cm. 2-5 cm. 47 -83 22-56 Primary form Unidentifiable End-struck flakes Trimmed face Dorsal Ventral Butt trimming Bifacially Unifacially Untrimmed No. Percent Plan 4 66.6 Lanceolate 2 33.3 Cordiform Irregular ovate Asymmetrical broad ovate 5 83.4 Ovate 1 16.6 16.6 50.0 33.3 1 3 2 Finish Rough Coarse Fine No. Percent 1 16.6 1 16.6 2 33.3 1 16.6 1 16.6 2 1 3 33.3 16.6 50.0 Cleavers (64, 11.4%; plates XVIII:4; XIX:1; XXIII:4; XXIV:1-4; XXVI:2; XXVII:2) Dimensions Length Width Thickness W/L T/W Range 9-25 cm. 5-13 cm. 2-8 cm. 45-78 33 -89 Mean 15.7 cm. 9.2 cm. 4.5 cm. 65 53 Standard deviation 3.2 cm. 1.7 cm. 0.9 cm. 9.9 8.9 Mode 56, 62, 63 50 Primary form No. Percent Trimmed face No. Percent Unidentifiable Chip Chunk End-struck flakes Side-struck flakes 27 42.2 Unifacial 3 1 1.5 Opposite side/opposite face 1 1 1.5 Parti-bifacial 14 4 6.3 Bifacial 46 31 48.5 Mean 13.3 cm. 8.7 cm. 3.7 cm. 66 42 Standard deviation 1.4 cm. 1.0 cm. 1.2 cm. 11.0 10.5 Mode 69 4.7 1.5 21.9 71.9 - - - - ------c3 - - - 1 -- 43 Anthropological Records No. Percent Butt trimming No. Percent Irregular Convergent Asymmetrically convergent Ultra- convergent Parallel Asymmetrically parallel Irregular parallel Divergent Asymmetrically ultra- convergent Irregular convergent Bit plan Straight Guillotine to the right Guillotine to the left Guillotine Platform (35 pieces) Reduced Removed Plain 5 12 1 4 31 5 2 2 1 1 7.8 18.8 1.5 6.3 48.5 7.8 3.1 3.1 1.5 1.5 14 20 11 19 26 8 1 21.9 31.3 17.2 29.7 74.2 22.8 2.9 Untrimmed Unifacially Partially Partially trimmed ventral face Partially trimmed dorsal face Bifacially Butt plan V-butts Square butts U -butts Finish Rough Fine Very fine Coarse Condition Whole Broken Cleaver Flakes (11, 1.9%; plates XXIII:3; XXIV:5) Dimensions Length Width Thickness W/L T/W Range 9-18 cm. 8-17 cm. 2-6 cm. 57-142 25 -56 Mean 13.6 cm. 10.2 cm. 4.3 cm. 78 43 Standard deviation 2.2 cm. 2.6 cm. 0.9 cm. 22.7 9.2 Primary form Unidentifiable End-struck flakes Side-struck flakes Plan Irregular Parallel Divergent Irregular convergent Short quadrilateral Trimmed face Unifacial Dorsal Parti-facial Butt trimming Bifacial Untrimmed Partially trimmed on ventral face No. Percent 3 27.3 3 27.3 5 45.4 2 6 1 1 1 6 4 1 7 3 1 18.2 54.5 9.1 9.1 9.1 54.5 36.4 Butt plan Square butts U-butts V-butts Bit plan Straight Guillotine to the right Guillotine to the left Platform Plain Reduced Removed ., v.- Finish 9.1 Fine Coarse 2736 Condition 1 Whole 9.1 Fragmentary No. Percent 7 63.6 2 18.2 2 18.2 8 72.7 2 18.2 1 9.1 9 81.8 1 9.1 1 9.1 3 27.3 8 72.7 10 90.9 1 9.1 Knives (14, 2.5%; plates Dimensions Length Width Thickness W/L T/W XXII: 7; XXVI: 3) Range 8-19 cm. 4-12 cm. 3-5 cm. 40-75 36-100 Mean 14.0 cm. 8.6 cm. 4.1 cm. 62 50 Plan 13 4 3 2 1 41 13 13 38 20.3 6.3 4.7 3.1 1.5 64.0 20.3 20.3 59.5 21.9 17.2 4.7 56.2 14 11 3 36 61 95.3 3 4.7 Mode 64, 67 44, 50 Standard deviation 3.4 cm. 2.4 cm. 0.7 cm. 8.2 15.7 Mode 58 - ---- -- -C? - - - - - -- . 44 - Keller: Montagu Cave in Prehistory Primary form Unidentifiable Chip End-struck flake Side-struck flake Trimmed face Unifacial Trimmed ventral Bifacial No. Percent 8 57.1 1 7.2 4 28.6 1 7.2 1 1 1 2 7.2 7.2 85.6 Plan Round Sub-quadrilateral Elongate Semicircular Quadrilateral End and side Pointed Discoid (3, 0.5%; plate XX:1) Dimensions Length Width Thickness W/L T/W Range 8-16 cm. 7-13 cm. 3-5 cm. 63-88 30-43 Plan Round Ovoid Trimmed face Bifacial No. Percent 2 66.6 1 33.3 Finish Fine No. Percent 3 100 3 100.0 Scrapers (174, 30.9%; plates XXII:2; XXV:1-4, 7-11; XXVI:5; XXVII:1; XXVIII:2) For comparative purposes in figure 25 the scrapers are divided into large and small categories on the basis of their maximum dimension, but all are combined for descriptive purposes here. description. Dimensions Length Width Thickness W/L T/W Range 3-28 cm. 2-22 cm. 1-14 cm. 33 -2 25 11-100 Three broken specimens were omitted from the Mean 9.0 8.5 3.3 93 44 cm. cm. cm. Standard deviation 3.6 cm. 3.3 cm. 1.7 cm. 33.2 18.4 Mode 100 50 Primary form Unidentifiable Chips End-struck flakes Side-struck flakes Split flakes Snapped flakes Flake fragments broken in two directions Chunks Pebble Plan Short quadrilateral Triangular Irregular Ovoid Round Semicircular Sub-quadrilateral Sub-triangular A pebble Elongate No. Percent 4 2.3 9 5.2 21 12.1 39 22.5 13 7.5 4 2.3 3 1.7 80 46.0 1 0.6 8 2 154 1 1 1 1 1 1 4 4.6 1.2 88.5 0.6 0.6 0.6 0.6 0.6 0.6 2.3 Location of trimming Distal side Distal side and one end Two sides Two ends Two sides and one end One end A corner One end and one side Distal end Distal end and one side Proximal end Proximal side Split face One side Platforms (59 pieces) Simple-facetted Plain Reduced Removed No. 1 1 1 1 1 5 4 Percent 7.2 7.2 7.2 7.2 7.2 35.7 28.6 Standard deviation Mean 13.0 cm. 10.0 cm. 3.7 cm. 78 42 Mode 4.4 cm. 3.0 cm. 1.1 cm. 11.0 5.3 No. 23 7 14 3 1 21 4 15 4 5 1 1 1 74 12 42 4 1 Percent 13.3 4.0 8.0 1.7 0.6 12.0 2.3 8.5 2.3 2.9 0.6 0.6 0.6 42.5 20.3 71.2 6.7 1.7 _ _ _ _ _ . 45 Anthropological Records Core Scrapers (8, 1.4%; plate XXV:9) Dimensions Length Width Thickness W/L T/W Range 7-21 cm. 7-15 cm. 4-10 cm. 67-118 40-80 Mean 12.9 cm. 10.8 cm. 6.5 cm. 88 59 Primary form Chunks Snapped flakes No. Percent 7 87.5 1 12.5 Location of trimming Two sides One end and one side One side The margin (periphery of equidimensional pieces) Choppers (5, 0.9%; plate XVIII:1) Dimensions Length Width Thickness W/L T/W Range 7-11 cm. 6-8 cm. 4-6 cm. 60-100 50-86 Primary form Unidentifiable Chunks Side-struck flakes No. 2 2 1 Core Choppers (3, 0.5%) These are cores used on one Dimensions Length Width Thickness W/L T/W Range 8-11 cm. 7-9 cm. 6 cm. 82 -89 67-86 Percent 40.0 40.0 20.0 Location of trimming One side Two sides and one end Two ends Distal side No. 2 1 1 1 side as choppers. All are bifacial. Standard Mean deviation 9.3 cm. 8.0 cm. 6.0 cm. 86 75 1.5 cm. 1.0 cm. 0.0 cm. 3.0 7.7 Pointed Tool (1, 0.2%) This has an irregular shape and is trimmed bifacially. The primary form on which it is made is unidentifiable. Dimensions: Length 9 cm.; Width 7 cm.; Thickness 4 cm. Burins (2, 0.3%1; plate XXV:2, 5) Both are made on chunks, trimmed on one end, and are unifacial. Dimensions Length Width Thickness W/L T/W Range 5-10 cm. 5 cm. 2-3 cm. 50-100 40-6 0 Mean 7.5 cm. 5.0 cm. 2.5 cm. 75 50 Standard deviation 3.5 cm. 0.0 cm. 0.7 cm. 20.4 21.7 Mode Standard deviation 4.7 cm. 2.5 cm. 2.1 cm. 17.3 10.9 Mode 67, 100 67 No. 2 2 2 2 Percent 25.0 25.0 25.0 25.0 Standard deviation Mean 9.6 cm. 7.0 cm. 4.8 cm. 74 69 1.5 cm. 0.7 cm. 0.8 cm. 13.4 14.0 Mode 70 Percent 40.0 20.0 20.0 20.0 Mode - - - 46 Keller: Montagu Cave in Prehistory Nosed Tools (5, 0.9%) Dimensions R Length 8-: Width 4-_ Thickness 3 - ' W/L 50-; T/W 40-. Lange 12 cm. 10 cm. 5 cm. 100 125 Primary form Chunks End-struck flakes Side-struck flakes No. 3 1 1 Percent 60.0 20.0 20.0 Location of trimming One side Two sides Trimmed face Dorsal Unidentifiable No. Percent 4 80.0 1 20.0 3 60.0 2 40.0 Point (1, 0.2%; plate XXV:6) This is a flake that has been trimmed been partially reduced. bifacially. The bulb and platform have Dimensions: Length 5 cm.; Width 5 cm.; Thickness 2 cm. Triangular Tool (1, 0.2%) The primary form on which this tool is made is unidentifiable. It has a tri- angular plan and has been trimmed on two sides on the dorsal face. Dimensions: Length 7 cm.; Width 6 cm.; Thickness 3 cm. Minimally Trimmed Flakes (41, 7.3%; plates XXIII:5; XXVIII:3) These are flakes that have had a few flakes detached but lack sufficient trimming to be placed in any of the more specific categories. Most show some attempt to remove or reduce the bulb and platform. The trimming was done with large flakes, and it appears that the intent was to alter the shape of the object rather than to produce a specific kind of edge. Dimensions Length Width Thickness W/L T/W Range 5-45 cm. 5-23 cm. 2-8 cm. 34-169 17 -60 Mean 15.0 cm. 12.4 cm. 4.2 cm. 95 35 Standard deviation 9.2 cm. 4.4 cm. 1.4 cm. 32.3 9.3 Mode 100 40 Primary form Unidentifiable End-struck flakes Split flakes Snapped flakes Side-struck flakes Trimmed face Bifacial Parti-bifacial Dorsal Ventral Unifacial No. 12 7 1 2 19 21 4 8 6 2 Percent 29.2 17.0 2.4 4.9 46.4 Platform "Bulb only" Plain Simple-facetted Removed Reduced Unidentifiable 51.2 9.8 19.5 14.6 4.8 Mean 9.2 cm. 7.0 cm. 4.0 cm. 75 65 Standard deviation 1.8 cm. 2.5 cm. 0.7 cm. 15.8 31.2 Mode 75 No. 1 8 4 16 9 3 Percent 2.4 19.5 9.8 39.0 21.9 7.3 - - 47 Anthropological Records Minimally Trimmed Chips (5, 0.9%) These are similar to the trimmed flakes but are made on chips. All are trimmed on one side. Standard Dimensions Range Mean deviation Length 3-54 cm. 23.1 cm. 16.4 cm. Width 5-22 cm. 13.8 cm. 5.3 cm. Thickness 2-6 cm. 4.0 cm. 1.4 cm. W/L 30-167 81 38.9 T/W 19-40 31 6.8 Trimmed face No. Percent Dorsal 3 60.0 Unifacial 1 20.0 Bifacial 1 20.0 Minimally Trimmed Chunks (14, 2.5%; plate XXI:1) These are similar to the above two categories, but the primary form is a chunk. Dimensions Length Width Thickness W/L T/W Range 7-19 cm. 5-12 cm. 2-7 cm. 53 -86 33-100 Mean 10.2 cm. 7.4 cm. 3.9 cm. 74 54 Standard deviation 3.1 cm. 1.8 cm. 1.4 cm. 8.8 16.7 Mode 78 43, 50 Location of trimming One side One e nd One end and one side Unifacial but location unrecorded No. Percent 4 28.6 2 14.3 1 7.2 7 50.0 Biface Butts (32, 5.7%) All are bifacial, and the primary forms on fiable. Dimensions Length Width Thickness Range 5-17 cm. 6-11 cm. 3-6 cm. Finish Coarse Rough Fine No. 17 9 6 Mean 10.4 cm. 8.8 cm. 4.1 cm. which they are made are unidenti- Standard deviation Mode 5.6 cm. 3.9 cm. 1.9 cm. Percent 53.1 28.1 18.7 Biface Tips (3 3, 5.8%) These are the broken tips of bifaces. Dimensions Length Width Thickness Range 2-28 cm. 3-26 cm. 1-14 cm. Mean 5.9 cm. 5.8 cm. 2.7 cm. Standard deviation 1.2 cm. 4.4 cm. 3.8 cm. Mode 48 Mode 100 38 - . _ - Keller: Montagu Cave in Prehistory Trimmed face Unifacial Dorsal Parti-bifacial Bifacial No. Percent 2 6.1 1 3.0 2 6.1 28 84.9 Finish Rough Coarse Fine Very fine Unrecorded No. Percent 5 15.1 7 21.2 9 27.2 11 33.3 1 3.1 Amorphous Bifaces (6, 1.1%) These are irregular, bifacially trimmed pieces which have had no particular attention paid to any feature such as a point or working edge. They look like roughouts or unfinished tools. Dimensions Length Width Thickness W/L T/W Range 13-21 cm. 8-11 cm. 4-7 cm. 52-6 2 44-70 Mean 17.5 cm. 9.7 cm. 5.2 cm. 56 53 Standard deviation 2.9 cm. 1.0 cm. 1.2 cm. 3.7 8.0 Mode 53 50 Primary form Unidentifiable Chunk Side-struck flake No. 4 1 1 Percent 66.6 16.6 16.6 Trimmed face Bifacial Unifacial No. 5 1 Percent 83.4 16.6 Chisel-ended Bifaces (12, 2.1%o; plates XXIII:1, 2; XXVI:1) On all the primary form is unidentifiable. Dimensions Length Width Thickness W/L T/W Range 8-15 cm. 6-10 cm. 3-5 cm. 50-88 43-71 Mean 12.2 cm. 7.9 cm. 4.1 cm. 66 53 Standard deviation 2.0 cm. 1.5 cm. 0.8 cm. 9.9 8.9 Mode 64 50 Plan Ovate Parallel- sided Ovoid Sub-quadrilateral Pointed Elongate Irregular ovate Trimmed face Bifacial Dorsal No. Percent 1 8.3 3 25.0 1 8.3 1 8.3 2 16.6 3 25.0 1 8.3 Butt trimming Untrimmed butts Bifacially trimmed butts Trimmed butt Finish Coarse Rough 1 1 91.6 1 8.3 Gouge-ended Bifaces (3, 0.5%; Dimensions Length Width Thickness W/L T/W Range 9-14 cm. 6-13 cm. 3-4 cm. 67-93 31 -50 plate XXI:6) Mean 11.3 cm. 9.0 cm. 3.6 cm. 77 44 Primary form Side-struck flake Unidentifiable Finish Coarse No. Percent 1 33.3 2 66.6 Trimmed face Bifacial Unifacial (dorsal face) No. Percent 2 66.6 1 33.3 3 100.0 No. 5 6 1 Percent 41.6 50.0 8.3 9 75.0 3 25.0 Standard deviation 2.5 cm. 3.6 cm. 0.6 cm. 11.1 9.0 Mode 50 - - - - 49 Anthropological Records Beaked Bifaces (8, 1.4%; plate XXII:3,6) These tools are similar in many ways to Kleindienst's "picks," but they are far from being identical and so a more descriptive name has been used here. They look as if they might have been the functional equivalents of picks, but they are smaller than most of the East African examples. On all the primary form is unidentifiable. Dimensions Length Width Thickness W/L T/W Range 8-13 cm. 6-8 cm. 3-5 cm. 46 -88 50-83 Mean 10.6 cm. 6.8 cm. 4.5 cm. 65 67 Standard deviation 1.8 cm. 0.7 cm. 0.8 cm. 14.1 9.5 Mode 46, 70 71 Plan Ovate Pointed Sub-triangular Elongate Irregular ovate Finish Rough Coarse No. 2 3 1 1 1 Percent 25.0 37.5 12.5 12.5 12.5 Trimmed face Bifacial Parti-bifacial Butt trimming Trimmed butts Untrimmed 3 37.5 5 62.5 Twisted-Bit Bifaces (2, 0.4%; plate XXII:4-5) Dimensions Length Width Thickness W/L T/W Range 9-10 cm. 6 cm. 3-5 cm. 60-67 50-83 Mean 9.5 cm. 6.0 cm. 4.0 cm. 63 67 Standard deviation 0.7 cm. 0.0 cm. 1.4 cm. 3.3 16.6 Plan Irregular elongate Ovate No. Percent 1 50.0 1 50.0 Push Planes (6, 1.1%; plate XXI:7) On all the primary form is unidentifiable. Dimensions Length Width Thickness W/L T/W Range 10-18 cm. 7-11 cm. 3-6 cm. 60-70 33-71 Mean 12.8 cm. 8.2 cm. 4.3 cm. 64 54 Plan Parallel- sided Elongate Asymmetrically parallel Asymmetrically convergent Finish Coarse No. Percent 2 33.3 2 33.3 1 1in 7 Trimmed face Bifacial Parti-bifacial 1 16:7 Butt trimming Bifacial Untrimmed 6 100.0 Partially trimmed 50 No. Percent 7 87.5 1 12.5 6 75.0 2 25.0 Mode Standard deviation 3.0 cm. 1.6 cm. 1.0 cm. 3.5 11.3 Mode 64 57 No. 4 2 Percent 66.6 33.3 50.0 33.3 16.7 3 2 1 = - - . - - Keller: Montagu Cave in Prehistory Bifaces (34, 6%; plate XXVII:6) These are bifacially trimmed tools that are more or less amorphous; they exhibit a range of features, but cannot be put into any of the biface categories. On all the primary form is unidentifiable. Mean 11.3 cm. 8.4 cm. 3.9 cm. 72 49 Standard deviation 6.1 cm. 5.3 cm. 1.0 cm. 12.2 8.9 Mode 65, 67, 70, 78 50 Plan Short quadrilateral Irregular Ovate Long ovate Parallel Ovoid Sub-quadrilateral Truncated triangular Pointed Sub-triangular Elongate Trapezoid Condition Broken Complete No. 1 12 1 1 1 1 2 1 3 2 8 1 Percent 2.9 35.3 2.9 2.9 2.9 2.9 5.9 2.9 8.8 5.9 23.5 2.9 15 44'1 19 55.9 Trimmed face Unifacial Parti-bifacial Bifacial Finish (15, broken) Rough Coarse Fine Finish (19, complete) Rough Coarse Fine Unrecorded Unidentifiable (1, 0.2%) The primary form on which this tool is made is unidentifiable. It is irregular in shape and trimmed unifacially on one end to an irregular blunt edge. It appears to be fragmentary. Dimensions: Length 5 cm.; Width 7 cm.; Thickness 2 cm. Cores (110) Single-Platform Cores (1, 0.9%) The primary form on which this core is made is unidentifiable. It has a pyramidal cross section and is trimmed from the base of the triangle. Dimensions: Length 6 cm.; Width 5 cm.; Thickness 3 cm. Disc Cores (68, 62.7%; plates XVIII:5,7; XXI:3,4) One has been trimmed as a scraper. Dimensions Length Width Thickness W/L T/W Range 5-48 cm. 5-40 cm. 2-15 cm. 65-100 27-75 Mean 10.8 cm. 9.1 cm. 4.5 cm. 86 50 Standard deviation 6.0 cm. 4.9 cm. 1.9 cm. 9.4 10.3 Mode 83 50 Dimensions Length Width Thickness W/L T/W Range 7-30 cm. 5-15 cm. 3-7 cm. 50-100 29-6 7 No. 1 1 32 4 2 3 6 8 2 3 Percent 2.9 2.9 94.1 26.6 13.3 20.0 31.6 42.0 10.5 15.8 - - - - 51 Anthropological Records Primary form Chunks End-struck flake Unidentifiable Trimmed face Unifacial Dorsal Parti-bifacial Bifacial No. 6 1 62 8 2 9 50 Percent Plan 8.7 Irregular 1.5 Round 89.9 Semicircular Sub-quadrilateral Sub- triangular 11.6 Ovoid 2.9 Condition 13.1 72.5 Fragmentary Complete Formless Cores (9, 8.2%) Dimensions Length Width Thickness W/L T/W Range 5-17 cm. 6-14 cm. 4-12 cm. 2-120 57-109 Primary form Chunks Unidentifiable No. 7 2 Mean 10.2 cm. 8.2 cm. 6.1 cm. 77 74 Standard deviation 3.2 cm. 2.6 cm. 2.5 cm. 30.6 18.0 Percent 77.7 22.2 Irregular Cores (6, 5.4%) Dimensions Length Width Thickness W/L T/W Trimmed face Bifacial Parti-bifacial Unifacial Range 6-26 cm. 7-22 cm. 3-15 cm. 69-117 38-83 No.- 3 1 2 Mean 14.7 cm. 11.8 cm. 6.7 cm. 85 54 Standard deviation 7.9 cm. 5.8 cm. 4.4 cm. 16.4 15.0 Percent 50.0 16.7 33.3 Biconical Cores (22, 20%) Dimensions Length Width Thickness W/L T/W Range 5-18 cm. 3-12 cm. 3-7 cm. 33-100 42-100 Plan Irregular Round Ovoid Semicircular Sub-quadrilateral Elongate No. 3 7 8 1 1 2 Percent 13.6 31.8 36.4 4.6 4.6 9.1 Primary form Chip Chunk Unidentifiable Trimmed face Unifacial Parti-bifacial Bifacial Spindle Core (1, 0.9%o) This has an elongate plan and has been trimmed bifacially. The on which it is made is unidentifiable. Dimensions: Length 20 cm.; Width 11 cm.; Thickness 7 cm. primary form No. 8 26 3 2 2 28 Percent 11.6 37.7 4.4 2.9 2.9 40.5 2 2.9 67 97.1 Mode 73, 88 57, 67 Mode 69 Mean 9.0 cm. 6.9 cm. 4.7 cm. 79 71 Standard deviation 3.4 cm. 1.9 cm. 1.0 cm. 15.3 16.0 Mode 83 63, 71, 100 No. 1 1 20 2 4 16 Percent 4.6 4.6 90.9 9.1 18.2 72.7 - . - - 52 Keller: Montagu Cave in Prehistory Struck Core (1, 0.9%; plate XXI:5) This has an oval plan; the primary form on which it is made is unidentifiable. It is bifacial with one large flake detached from one face. Dimensions: Length 5 cm.; Width 5 cm.; Thickness 3 cm. Split Cobble Core (1, 0.9%) This is an irregular split cobble. Dimensions: Length 11 cm.; Width 9 cm.; Thickness 6 cm. Pebble Core (1, 0.9%o) This is a pebble trimmed on one side bifacially as a core.. Dimensions: Length 17 cm.; Width 15 cm.; Thickness 9 cm. Utilized Pieces (73) Hammerstones (6, Dimensions Length Width Thickness W/L T/W Primary form Flat slabs Pebbles 8.2%) Range 9-12 c] 7-9 ci 5-7 c] 58 -100 56 -88 Mean m. 10.1 cr m. 8.0 cr m. 5.7 cr 80 72 No. Percent 2 33.3 4 66.6 Standard deviation In. 1.2 cm. M. 0.9 cm. m. 0.8 cm. 14.1 12.7 Location of Use Two sides Ends Mode No. Percent 1 16.6 5 83.3 Anvils (4, 5.5%) All are made Dimensions Length Width Thickness W/L T/W on pebble s. Range 14-16 cm. 9-17 cm. 6-11 cm. 64-106 47-89 Flakes (17, 23.2%) Flake Fragments (13, 17.8%) Chips (6, 8.2%) Chunks (25, 34.2%) Cores (2, 2.7%) 53 Mean 15.0 cm. 12.5 cm. 8.3 cm. 82 68 Standard deviation 1.2 cm. 3.7 cm. 2.0 cm. 16.1 16.1 Mode - - Anthropological Records LAYER 3 DISCUSSION It will become apparent in the following pages that the assemblages from Layers 3 and 5 are considerably different from other recently excavated Acheulean assemblages in Central and East Africa. The differ- ences lie not only in the percentages of the artifacts in the assemblages but also in the presence of some characteristic tool types at Montagu. Therefore, a re- view of the salient features of the Layer 3 assemblage is appropriate. An examination of table 7 will clearly indicate that there is no significant variation within the occurrences that make up Layer 3, and consequently the layer will be discussed as a whole. It will be remembered from the discussion of the stratigraphy that this layer was composed of different-colored bands. However, the density of artifact distribution was uniform throughout these bands, and no occupation horizons were present. In figure 25 a certain amount of lumping has been done to facilitate comparisons. The category labelled "heavy duty" combines core scrapers and choppers. The "other small tool" category includes the unidenti- fied tools, burins, the pointed tool, triangular tool, and nosed tools. Minimally trimmed flakes, chips, and chunks are combined in the "minimally trimmed" category, and biface butts and tips are included in the "fragment" category. The "other biface" category com- bines the tools called simply "bifaces," as well as the chisel-ended, amorphous, twisted-bit, gouge-ended, and beaked bifaces and push planes. These categories do not duplicate exactly those used by others (Kleindie 1961 and 1962; Clark: 1964a; Howell and Clark: 196 but the attempt has been made to present the Mon material in a format that will allow comparisons to made with these other publications yet still describe the assemblage adequately. In any case, the raw f are present in the descriptions, and they can be or ized differently if the reader desires. Scrapers are the most frequently occurring type tool, and when they are divided into large and small the small scrapers are still the most frequent-16.8j percent. The correlation tables for scrapers show th for both the large and small categories the type of scraper most often found was a piece of angular wag trimmed on one side. These correlation tables for large and small scrapers from Layer 3 indicate the same set of relationships as among the scrapers froE Layers 1 and 2. Most commonly a piece of angular waste was trimmed along one side to a steep unifacil edge. These tools seem to represent a very simple solution to a problem in that the most common arti- fact form was modified in the most obvious place. Among the artifacts from Layer 3, more than half were classified as angular waste (fig. 26). Hand-axes and large scrapers occur almost as often as scraperi 14.5 percent and 14 percent respectively. These are followed closely in frequency by the cleavers, other TABLE 6 Correlations of Size with Selected Attributes of Layer 3 Scrapers Location of Trimming Primary Form_l | One End One Side Two Sides. End and Side Other Total No. Percent No. Percent No. Percent No. Percent No. Percent No. Percent Small Scrapers Angular waste 0 0.0 58 61.1 0 0.0 0 0.0 0 0.0 58 61.1 Flakes 0 0.0 23 24.2 1 1.1 0 0.0 1 1.1 25 26.4 Flake fragments 0 0.0 9 9.5 0 0.0 0 0.0 0 0.0 9 9.5 Other 0 0.0 3 3.1 0 0.0 0 0.0 0 0.0 3 3.1 Total 0 0.0 93 97.9 1 1.1 0 0.0 1 1.1 95 Large Scrapers Angular waste 0 0.0 33 40.2 0 0.0 0 0.0 0 0.0 33 40.2 Flakes 2 2.4 30 36.6 2 2.4 0 0.0 1 1.2 35 42.6 Flake fragments 0 0.0 11 13.4 1 1.2 0 0.0 0 0.0 12 14.6 Other 0 0.0 2 2.4 0 0.0 0 0.0 0 0.0 2 2.4 Total 2 2.4 76 92.6 3 3.6 0 0.0 1 1.2 82 I 54 ., Keller: Montagu Cave in Prehistory TABLE 7 Tools and Cores from Layer 3 12 13 15 16 17a 17b 17 Hand-axes and hand-axe flakes 8 2 6 11 4 - 51 Cleavers and cleaver flakes 9 2 6 9 2 2 45 Knives 1 - - 4 1 - - Discoids 1 - - 2 - - - Large scrapers 7 1 11 12 4 2 46 Core scrapers 3 - 2 1 - 1 1 Choppers 3 - - 2 - - - Core choppers 3 - - - - - - Small scrapers 3 2 15 25 1 2 46 Nosed tools - - 2 1 - - 2 Pointed tool - - - 1 - Burins - - - 1 - - 1 Point - - - - - - 1 Triangular tool - - - - - - 1 Unidentified 1 - - - - - - Minimally trimmed flakes 7 - 3 6 7 - 18 Minimally trimmed chips 2 - - 1 - - 2 Minimally trimmed chunks 2 - 3 1 - - 8 Biface tips 3 - 2 7 1 - 20 Biface butts 3 - - 7 1 - 21 Bifaces 3 - - 4 2 1 24 Chisel-ended bifaces 1 - 2 - - 1 8 Push planes 1 - 1 - 1 - 3 Amorphous bifaces 1 - 2 1 - - 2 Beaked bifaces 1 - - - - - 7 Twisted-bit bifaces - - - - - - 2 Gouge-ended bifaces 2 - - - - - 1 Single-platform core - - - - - - 1 Disc cores 13 2 6 5 2 1 40 Formless cores 1 1 4 2 - - 3 Irregular cores - - - 1 - - 9 Biconical cores 5 - 3 4 2 - 8 Struck core - - - - 1 Spindle core - - 1 - - - Split-cobble core - - - - 1 Pebble core - 1 - - - - - Totals 84 11 69 108 28 10 381 bifaces, biface fragments, and minimally trimmed pieces. Knives, discoids, heavy-duty tools, and other small tools, each make up less than 3 percent of the total tools (see fig. 25). As pointed out in the descriptions, the biface butt -category contains those objects that are broken bifaces and are too fragmentary to permit their placement in -one of the more specific categories, such as hand-axe or cleaver. It is interesting to note that 60.5 percent of the hand-axe tips are of "fine" or "very fine" work- manship, whereas only 36.7 percent of the complete band axes are "fine" or "very fine." Undoubtedly the most striking feature of the assem- blage from this layer is the tremendous amount of waste. Of the 66,816 artifacts recovered, 99.1 percent are waste, and of this 50.5 percent are chips. Cores account for 0.1 percent of the waste and are subdivided by shape in figure 27. Disc and biconical cores are the prevalent types, with the former occurring more fre- quently. Forty-eight percent of the flakes are end-struck and fifty-two percent side-struck. The objection may be raised that the minimally trimmed pieces should not have been included with the tools but placed in a special "modified" category. This was not done, however, because the resulting histogram would make it very difficult to compare the minimally trimmed pieces with the rest of the assemblage because of the large quantity of waste. When placed in a separate category, the minimally trimmed pieces account for only 0.08 percent of the total assemblage, and this com- pletely obscures the fact that there are nearly as many minimally trimmed pieces as hand-axes or cleavers. There is very little utilized material, the primary form utilized most frequently being a chunk. Hammer- stones and anvils combined make up 13.7 percent of the utilized pieces. 55 Anthropological Records LAYER 5 DESCRIPTION The definitions used for the Layer 3 material apply here, and where new categories are used, they are defined below. Tools (1166) Hand-axes (165, 14.2%; plates XXIX:1; XXXII:2-3; XXXIII:2-4; XXXIV:1; XXXVII:1-2; XXXVIII:3; XXXIX:1,4; XLIII:1; XLIV:1; XLV:2; XLVI:1; XLVII:1, 6; XLVIII:5; XLIX:4; L:1; LIII: 3) Dimensions Length Width Thickness W/L T/W Range 9-31 cm. 3-17 cm. 2-12 cm. 23-91 29-133 Mean 16.6 cm. 9.3 cm. 5.2 cm. 55 56 Standard deviation Mode 4.4 cm. 2.0 cm. 1.5 cm. 8.3 10.5 50 50 Primary form Chunks End-struck flakes Side-struck flakes Unidentifiable Trimmed face Parti-bifacial Unifacial Bifacial Butt trimming Untrimmed Unifacial Cortex Partially trimmed Partially trimmed on dorsal face Partially trimmed on ventral face Bifacial Condition Broken Whole No. Percent Plan 1 3 8 153 31 2 132 31 14 4 8 3 3 102 0.6 Irregular 1.8 Ovate 4.9 Long ovate 92.6 Irregular long ovate Asymmetrical long ovate Irregular ovate 18.8 Broad ovate 1.2 Asymmetrical ovate 80.0 Lanceolate Asymmetrical lanceolate Irregular lanceolate 18.8 Narrow lanceolate 8.5 Block-shaped 2.4 Double-pointed 4.8 Asymmetrical ovate- acuminate 1.8 Ovate-acuminate Pointed 1.8 Sub-triangular 61.8 Finish Rough 15 9.1 Fine 150 90.9 Very fine Coarse Hand-axe Flakes (10, 0.9%0) Dimensions Length Width Thickness W/L T/W Range 14-21 cm. 7-12 cm. 3-6 cm. 53-80 44-60 Mean 17.0 cm. 10.0 cm. 5.1 cm. 66 52 Standard deviation 2.6 cm. 1.5 cm. 1.1 cm. 9.6 5.8 Primary form Unidentifiable End-struck flakes Side-struck flakes Trimmed face Unifacial Trimmed on ventral face Trimmed on dorsal face No. Percent Plan 2 20.0 Ovate 3 30.0 Asymmetrical long ovate 5 50.0 Triangular Long ovate Lanceolate 1 10.0 Irregular long ovate 6 60.0 Irregular lanceolate 3 30.0 No. 1 11 39 10 5 4 3 3 40 11 4 1 1 2 3 8 3 16 Percent 0.6 6.6 23.8 6.1 3.0 2.4 1.8 1.8 24.2 6.6 2.4 0.6 0.6 1.2 1.8 4.8 1.8 9.8 30.3 14.5 4.2 50.9 50 24 7 84 Mode No. 1 1 1 2 2 2 1 Percent 10.0 10.0 10.0 20.0 20.0 20.0 10.0 - - - - Mean Mode 56 Keller: Montagu Cave in Prehistory Butt trimming No. Percent Finish No. Percent Bifacial Untrimmed Trimmed on ventral face Cortex 3 30.0 4 40.0 2 20.0 1 10.0 Coarse Fine Rough 3 5 2 30.0 50.0 20.0 Cleavers (160, 13.7%; plates XXIX:5; XXX:1; XXXII:1, 4; XXXIII:1; XXXVI:1; XXXVIII:2; XXXIX:3,5; XLII:l; XLIII:3-5; XLV:1, 4-5; XLVIII:1-2; L:2; LI:1) Dimensions Length Width Thicknes s W/L T/W Range 7-28 cm. 5-16 cm. 2-11 cm. 42-100 27-110 Mean 17.5 cm. 10.3 cm. 5.1 cm. 60 51 Standard deviation 4.1 cm. 2.2 cm. 1.2 cm. 9.2 10.9 Mode 50 50 Primary form Chunks End-struck flakes Side-struck flakes Unidentifiable Butt trimming Untrimmed Partially trimmed Partially trimmed on dorsal face Partially trimmed on ventral face Trimmed unifacially Cortex Bifacially trimmed Butt shape Square butts V-butts Irregular butt Truncated Vs U-butts Bit Straight Guillotine to the right Guillotine to the left Quillotine Concave Convex No. Percent Plan 2 8 41 109 40 8 4 6 11 1 90 35 25 1 2 97 58 32 33 31 1 5 1.2 Irregular 5.0 Asymmetrical convergent 25.8 Convergent 68.1 Ultra-convergent Irregular convergent Asymmetrically ultra- 25.0 convergent 5.0 Irregular parallel Parallel 2.5 Asymmetrical parallel Divergent 3.8 Asymmetrically divergent 6.8 0.6 56.2 21.9 15.6 0.6 1.2 60.6 36.3 20.0 20.6 19.3 0.6 3.1 Trimmed face Trimmed on dorsal face Parti-bifacial Urnifacial Bifacial Platforms (64 pieces) Reduced Removed Plain Finish Rough Fine Very fine Coarse No. Percent 1 0.6 10 6.2 31 19.3 19 11.9 4 2.5 8 5.0 6 3.8 54 33.8 20 12.5 6 3.7 1 0.6 1 35 1 123 18 29 17 38 30 1 91 0.6 21.9 0.6 76.8 28.2 45.4 26.6 23.8 18.7 0.6 56.8 Cleaver Flakes (24, 2.1%; Dimensions Length Width Thickness W/L T/W Primary form Range 7 -26 7 -15 3 -7 47 -87 31 -75 plate s XXXVIII: 1; XLVII: 3) Mean cm. cm. cm. 17.1 cm. 10.7 cm. 4.9 cm. 62 46 No. Percent Standard deviation 3.7 cm. 1.9 cm. 1.1 cm. 9.7 11.2 Plan End-struck flakes Side-struck flakes Trimmed face Ventral Dorsal Butt shape V-butts U-butts Square butts 5 20.8 Short quadrilateral 19 79.1 Irregular Convergent Parallel-sided 4 16.7 Divergent 20 83.3 Asymmetrically parallel Ultra- convergent Asymmetrically divergent 10 9 5 41.6 37.5 20.8 Finish Coarse Rough Fine Mode 60 50 No. Percent 1 1 1 14 2 3 1 1 1 8 3 3 4.2 4.2 4.2 58.3 8.3 12.5 4.2 4.2 75.0 12.5 12.5 - - - ~ ~ ~~~~~ ~~ ~~~~~~~- - - - - - - - - - - - - - _ j-.,P jL.L I I q-- L I -J jL %J L . 10 --- L %,- ILS - v - - _ 5 7 Anthropological Records Butt trimming Bifacially trimmed Untrimmed Trimmed on dorsal face Cortex Unifacially trimmed N 1 To. Percent 3 12.5 .1 45.7 8 33.3 1 4.2 1 4.2 Bits Straight Guillotine to the right Guillotine to the left Guillotine Splayed Convex Knives (26, 2.2%; plates XXXV:3; XLIV:5; XLIX:2) Dimensions Length Width Thickness W/L T/W Range 9-23 cm. 5-11 cm. 3-6 cm. 42-83 33 -80 Mean 14.9 cm. 8.1 cm. 4.4 cm. 55 56 Standard deviation 3.7 cm. 1.7 cm. 0.8 cm. 8.3 10.5 Primary form Chunk End-struck flakes Side-struck flakes Split flake s Unidentifiable Trimmed face Parti-bifacial Bifacial No. Percent Plan 1 3.8 Irregular 1 3.8 Long ovate 1 3.8 Lanceolate 3 11.5 Round 20 76.9 Sub-quadrilateral Pointed Asymmetrical lanceolate 2 7.7 Elongate 24 92.2 End and side knives No. 1 1 1 1 1 14 1 3 3 Percent 3.8 3.8 3.8 3.8 3.8 53.8 3.8 11.5 11.5 Discoids (2, 0.2%; plate XXXV:1) Both are round and bifacial, one parti-bifacial. Both are coarse. Dimensions Length Width Thickness W/L T/W Range 11-14 cm. 10-12 cm. 5 cm. 86 -9 1 42-50 Mean 12.5 cm. 11.0 cm. 5.0 cm. 88 46 Standard deviation Mode 2.1 cm. 1.4 cm. 0.0 cm. 2.5 4.1 Scrapers (215, 18.4%; plates XLV:3, 6; XLVII:2; Dimensions Length Width Thickness W/L T/W Range 4-25 cm. 2-24 cm. 1-13 cm. 41-220 20-100 XXIX:3; XXXI:2-4; XXXIV:2; XXXIX:2; XLVIII:4; XLIX:5; LII:2; LIII:2) Standard Mean deviation 10.1 cm. 8.4 cm. 3.9 cm. 83 50 XLI:3; XLII:3-4,6; Mode 3.3 cm. 3.3 cm. 1.5 cm. 30.1 18.8 100 50 Primary form Unidentifiable Chips End-struck flakes Side-struck flakes Split flakes Snapped flakes Split and snapped flakes Chunks Trimmed face Dorsal Ventral Bifacial Unifacial No. 13 21 25 27 16 4 1 108 75 34 3 103 PE Ercent Location of trimming 6.1 Trimmed two sides 9.8 Proximal 11.6 One end 12.5 Two sides and one end 7.4 Corner 1.9 End and side 0.5 Two ends 50.2 Distal end Distal end and two sides Distal side 34.9 Distal side and one end 15.8 Proximal side 1.4 One side 47.8 Condition Complete Broken No. 9 5 7 1 1 Percent 37.5 20.8 29.1 4.2 4.2 4.2 Mode 53 50 No. 28 2 21 2 8 10 1 7 2 18 1 3 112 Percent 13.0 0.9 9.8 0.9 3.7 4.6 0.5 3.3 0.9 8.4 0.5 1.4 52.1 214 99.5 1 0.5 . . 58 Keller: Montagu Cave in Prehistory Core Scrapers (27, 2.3%; plates XXIX:4; XLIV:3; LIII:4) Mode 100 63, 83, 86, 100 Primary form Unidentifiable Chunks Location of trimming One side Two sides No. Percent 1 3.7 26 96.3 15 55.5 3 11.5 Location of trimming (continued) No. Percent One end and one side 4 14.7 One end 4 14.7 Corner 1 3.7 Hand-axe Choppers (4, 0.3%; plates XLI:1, 4; XLVIII:5) This term has been used by Clark (see comparative section) to describe large, rough, pointed tools usually with little or no trimming at the butt. The tools from Montagu in this category are large, usually have linguate tips and untrimmed butts, lack the keeled points of beaked bifaces, and lack the fine trimming around the tip found on core-axes. All are bifacial and rough. Dimensions Length Width Thickness No W/L or Range 13-22 cm. 10-15 cm. 5-7 cm. T/W figures are Mean 15.7 cm. 12.0 cm. 6.0 cm. available for these Primary form Chunks Unidentifiable No. Percent 2 50.0 2 50.0 Plan Sub-triangular Convergent Pointed Choppers (51, 4.4%; plates XXX:2; XXXVII:4; LII:4) Dimensions Length Width Thickness W/L T/W Range 7-20 cm. 6-14 cm. 4-12 cm. 55-100 40-92 Mean 12.2 cm. 9.2 cm. 5.8 cm. 77 63 Primary form Chunks Side-struck flake Unidentifiable No. 24 1 26 Percent 47.1 2.0 51.0 Location of trimming Two sides One end One end and one side One side Pointed Tools (2, 0.2%) One is made on a chip and One is less than 10 cm. long. Dimensions Length Width Thickness No W/ L or Range 9-23 cm. 7-10 cm. 2-6 cm. T/W figures are the other on a side-struck flake. Both are bifacial. Standard Mean deviation Mode 16.0 cm. 9.9 cm. 8.5 cm. 2.1 cm. 4.0 cm. 2.8 cm. L available for these tools. Dimensions Length Width Thickness W/L T/W Range 5-17 cm. 2-11 cm. 4-19 cm. 17-100 50-300 Mean 10.5 cm. 7.6 cm. 6.6 cm. 77 93 Standard deviation 2.9 cm. 2.0 cm. 2.8 cm. 21.3 49.2 Standard deviation 4.2 cm. 2.2 cm. 1.2 cm. tools. Mode No. 2 1 1 Percent 50.0 25.0 25.0 Standard deviation 2.8 cm. 1.9 cm. 1.5 cm. 10.9 12.3 Mode 67, 90 56 No. 4 3 2 42 Percent 7.9 5.9 3.9 82.4 - - - w . e 59 Anthropological Records Burin (1, 0.1%; plate XXIX:6) This is a triangular chunk trimmed on two ends with burin flakes. Dimensions: Length 10 cm.; Width 8 cm.; Thickness 4 cm. Nosed Tools (3, 0.3%) All are made on chunks and are unifacial. Two are trimmed on one side and one on a corner. One is less than 10 cm. long. Standard Dimensions Range Mean deviation Mode Length 5-15 cm. 11.3 cm. 5.5 cm. Width 4 -1 0 cm. 7.3 cm. 3.0 cm. Thickness 3-8 cm. 5.3 cm. 2.5 cm. No W/L or T/W figures are available for these tools. Minimally Trimmed Flakes (94, 8.1%; for illustration see Layer 3) Standard Dimensions Range Mean deviation Mode Length 5-25 cm. 12.2 cm. 4.5 cm. Width 4-24 cm. 11.5 cm. 3.9 cm. Thickness 2 -1 0 cm. 4.3 cm. 1.0 cm. W/L 50-200 102 37.8 100 T/W 15-92 38 10.4 38,40 Primary form No. Percent Trimmed face No. Percent Unidentifiable flakes 20 21.3 Ventral 31 32.9 End-struck flakes 18 19.1 Dorsal 7 7.5 Split flakes 8 8.5 Parti-bifacial 6 6.4 Snapped flakes 5 5.2 Bifacial 50 53.2 Split and snapped flakes 1 1.1 Platforms Side-struck flakes 42 44.7 Pltom Removed 63 67.0 Reduced 17 18.1 Plain 14 14.9 Minimally Trimmed Chips (4, 0.3%) Standard Dimensions Range Mean deviation Mode Length Width Thickness W/L T/W 10-14 cm. 7-12 cm. 2-4 cm. 50-92 29-50 12.3 cm. 8.5 cm. 3.2 cm. 70 39 1.7 cm. 2.4 cm. 0.9 cm. 16.8 8.3 Trimmed face Ventral face Unifacial Bifacial No. Percent 1 25.0 1 25.0 2 50.0 60 - Keller: Montagu Cave in Prehistory Minimally Trimmed Chunks (77, 6.6%; plate XL:1) Dimensions Range Mean Length 6-27 cm. 13.1 cm. Width 4-18 cm. 9.0 cm. Thickness 2-13 cm. 4.9 cm. W/L 26-110 71 T/W 30-129 56 Trimmed face No. Percent Dorsal Ventral Unifacial Parti-bifacial Bifacial 2 2 14 2 57 Standard deviation 4.3 cm. 2.6 cm. 1.7 cm. 17.0 16.6 Mode 100 50 2.6 2.6 18.2 2.6 74.0 Biface Butts (67, 5.7%) Standard Dimensions Range Mean deviation Mode Length 5-20 cm. 11.7 cm. 4.9 cm. Width 6 -14 cm. 9.4 cm. 3.7 cm. Thickness 3-7 cm. 5.0 cm. 1.7 cm. Finish No. Percent Trimmed face No. Percent Rough 18 26.9 Unifacial 3 4.5 Fine 2 3.0 Parti-bifacial 3 4.5 Coarse 47 70.0 Bifacial 61 91.0 Biface Tips (6 3, 5.4%) Standard Dimensions Range Mean deviation Mode Length 4-20 cm. 7.5 cm. 2.8 cm. Width 3-10 cm. 5.9 cm. 1.8 cm. Thickness 2-7 cm. 3.2 cm. 1.1 cm. Trimmed face No. Percent Finish No. Percent Dorsal 3 4.8 Rough 19 30.2 Unifacial 2 3.2 Fine 10 15.9 Parti-bifacial 2 3.2 Very fine 2 3.2 Bifacial 56 88.9 Coarse 28 44.4 Unrecorded 4 6.4 Cleaver Bits (2, 0.2%) These are fragments of broken cleavers. Standard Dimensions Range Mean deviation Mode Length Width Thickness 7-10 cm. 10 cm. 4-5 cm. 8.5 cm. 10.0 cm. 4.5 cm. 2.1 cm. 0.0 cm. 0.7 cm. - 6 1 Anthropological Records Amorphous Bifaces (5, 0.4%) On all the primary form is unidentifiable. Dimensions Length Width Thickness W/L T/W Trimmed face Dorsal face Bifacial Range 18-25 cm. 8-12 cm. 5-7 cm. 44-6 7 47-60 No. 1 4 Mean 20.8 cm. 10.6 cm. 6.0 cm. 54 54 Standard deviation 2.7 cm. 1.7 cm. 0.7 cm. 8.3 7.1 Percent 20.0 80.0 Chisel-ended Bifaces (21, 1.8%; plates XXXV:2; XLIII: 2) On all the primary form is unidentifiable. Dimensions Length Width Thickness W/L T/W Plan Range 9-29 cm. 5-14 cm. 3-8 cm. 44-80 44-86 Irregular Ovate Lanceolate Convergent Parallel- sided Sub-quadrilateral Elongate Irregular convergent Asymmetrical convergent Irregular parallel Trimmed face Bifacial Parti-bifacial No. 1 1 1 8 3 1 3 1 1 1 Mean 14.0 c 8.8 c 5.1 c 59 61 Percent 4.8 4.8 4.8 38.1 14.3 4.8 14.3 4.8 4.8 4.8 18 85.7 3 14.3 Standard deviation m. 4.5 cm. m. 2.1 cm. m. 1.4 cm. 10.6 12.8 Butt trimming Bifacial Untrimmed Partially, on the dorsal face Partially, on the ventral face Partially, on both faces Finish Rough Coarse Fine Gouge -ended Bifaces (9, 0.8%) On all the primary form is unidentifiable. Dimensions Length Width Thickness W/L T/W Plan Irregular Ovate Lanceolate Convergent Parallel- sided Elongate Range 10-15 cm. 6-11 cm. 4-6 cm. 46-110 36 -83 No. 1 1 1 3 1 2 Mean 11.4 cm. 8.1 cm. 5.0 cm. 76 61 Percent T: 11.1 Bi 11.1 PE 11.1 33.3 Fi 11.1 R( 22.2 C( rimmed Standard deviation 1.7 cm. 1.3 cm. 0.7 cm. 19.0 15.4 face ifacial arti-bifacial inish ough oarse Mode 80 63 No. Percent 8 88.8 1 11.1 5 55.5 4 44.4 Mode Mode 47, 63, 67 50, 60 Percent 61.9 23.9 4.8 4.8 4.8 42.8 52.4 4.8 No. 13 5 1 1 1 9 1 1 - - - - | - - 62 L Keller: Montagu Cave in Prehistory Beaked Bifaces (38, 3.3%; plates XXXVI:3; XLVI:3; LII:1) Standard Dimensions Range Mean deviation Mode Length 9-20 cm. 14.4 cm. 3.0 cm. - Width 4 -14 cm. 9.5 cm. 2.2 cm. - Thickness 3-11 cm. 6.0 cm. 1.7 cm. - W/L T/W Primary form Chunks End-struck flake Side-struck flake Unidentifiable Plan Triangular Ovate Long ovate Pointed Sub-triangular Elongate Asymmetrical ovate Broad ovate Block- shaped 36 -100 38-150 No. 7 1 1 29 1 1 1 21 8 3 1 1 1 71 65 Percent 18.4 2.6 2.6 76.4 2.6 2.6 2.6 55.4 20.5 7.9 2.6 2.6 2.6 14.1 22.1 Trimmed face Bifacial Parti-bifacial Dorsal face No. 25 7 6 Butt trimming Bifacial Untrimmed Partial, on dorsal face Partial, on ventral face Unifacial Cortex Partial Finish Rough Coarse 13 13 3 1 1 3 4 56, 80 50, 56, 60 Percent 65.8 18.4 15.7 34.2 34.2 7.9 2.6 2.6 7.9 10.5 24 63.2 14 36.8 Twisted-Bit Bifaces (6, 0.5%; plate XXX:6) On all the primary form is unidentifiable. Dimensions Length Width Thickness W/L T/W Rarge 10-18 cm. 7-11 cm. 3-9 cm. 47-110 50-82 Mean 12.7 cm. 8.7 cm. 5.5 cm. 71 69 Standard deviation 3.8 cm. 1.5 cm. 2.2 cm. 24.1 11.8 Mode Plan Convergent Parallel- sided Sub-triangular Elongate Irregular ovate Broad ovate No. 1 1 1 1 1 1 Percent 16.7 16.7 16.7 16.7 16.7 16.7 Butt trimming Trimmed Untrimmed Finish Rough Coarse No. Percent 5 83.3 1 16.7 2 33.3 4 66.6 Double-Pointed Bifaces (3, 0.3%) On all the primary form is unidentifiable. All are bifacial Standard Dimensions Range Mean deviation Mode Length 7 -1 0 cm. 9.0 cm. 1.7 cm. - Width 5-11 cm. 7.0 cm. 3.5 cm. - Thickness 3-5 cm. 3.6 cm. 1.6 cm. - W/L 50-110 77 24.8 T/W 27-100 62 29.7 No. 1 1 1 Percent 33.3 33.3 33.3 Finish Fine Rough Coarse No. Percent 1 33.3 1 33.3 1 33.3 Plan Irregular Pointed Elongate - . . - - - - - 63 Anthropological Records Push Planes (16, 1.4%; plates XXXV:4; XLI:2) Dimensions Length Width Thickness W/L T/W Range 9-19 cm. 5-13 cm. 4-10 cm. 50-80 38-111 Mean 14.4 cm. 9.3 cm. 6.2 cm. 64 70 Primary form Unidentifiable End-struck flake Side-struck flakes Trimmed face Dorsal Bifacial Parti-bifacial Finish Rough Coarse Fine Bifaces (70, 6.0%; Dimensions Length Width Thickness W/L T/W No. 13 1 2 6 9 1 4 11 1 ercent Plan 81.2 Convergent 6.3 Parallel-sided 12.5 Elongate Irregular convergent Asymmetrical ovate 5672 Butt trimming 6.3 Bifacial Untrimmed Partial, on the dorsal 25.0 Unifacial 68.7 6.3 plates XXXI:1; XLVU:4; XLIX:1, 3) Range 7-22 cm. 3-12 cm. 2-8 cm. 45-100 38 -88 Mean 11.3 cm. 7.7 cm. 4.3 cm. 69 59 Standard deviation 4.2 cm. 2.1 cm. 1.5 cm. 13.7 14.6 Primary form Chunks Side-struck flakes Unidentifiable Plan Irregular Ovate Long ovate Lanceolate Parallel- sided Semicircular Pointed Sub-triangular Elongate Asymmetrical long ovate Irregular ovate Asymmetrical ovate No. 2 2 66 10 7 2 2 7 2 10 5 15 2 7 1 Percent 2.9 2.9 94.2 14.3 10.0 2.9 2.9 10.0 2.9 14.3 7.2 21.4 2.9 10.0 1.4 Trimmed face No. Percent Parti-bifacial 6 8.6 Unifacial 2 2.9 Bifacial 62 88.6 Condition Broken 15 21.4 Whole 55 78.6 Finish (of 15 broken pieces) Rough 4 Coarse 3 Very fine 1 Unrecorded 7 Finish (of 55 complete pieces) Coarse 26 Rough 21 Fine 2 Unrecorded 6 26.6 20.0 6.6 46.6 47.3 38.2 3.6 10.9 Cores (291) Single platform Cores (4, 1.4%) All are unifacial Dimensions Length Width Thickness W/L T/W Range 6-12 cm. 4-7 cm. 2-5 cm. 33 -88 50-100 Mean 8.2 cm. 5.5 cm. 3.7 cm. 73 68 Standard deviation 2.6 cm. 1.2 cm. 1.5 cm. 22.6 20.5 Mode Standard deviation Mode 3.4 cm. 2.3 cm. 1.6 cm. 7.6 16.9 59 60 No. 6 5 2 2 1 8 5 face 1 2 Percent 37.5 31.2 12.5 12.5 6.3 50.0 31.2 6.3 12.5 Mode 60, 67 63, 67 ----ZI2 - -.---- - - - | = 64 PE Keller: Montagu Cave in Prehistory Primary form Chunk End-struck flake Unidentifiable No. Percent 2 50.0 1 25.0 1 25.0 Location of trimming One end One side No. Percent 1 25.0 3 75.0 Disc Cores (170, 58.4%; plates XLVII:5; LI:2-3) Dimensions Length Width Thickness W/L T/W Primary form Chunks End-struck flakes Side-struck flakes Unidentifiable Trimmed face Unifacial Trimmed dorsal Parti-bifacial Bifacial Range 5-22 cm. 4-16 cm. 2-13 cm. 24-130 33-175 No. 5 1 1 163 XXIX:2; XXXI:5; XXXIV:3; XXXVII:3; XLIV:4; Mean 9.8 cm. 8.2 cm. 4.7 cm. 85 59 Percent 2.9 0.6 0.6 95.9 24 14.1 6 3.5 39 22.9 101 59.4 Standard deviation 2.6 cm. 2.0 cm. 1.4 cm. 12.9 15.0 Plan Triangular Irregular Ovoid Semicircular Sub-quadrilateral Sub- triangular Elongate Round Formless Cores (14, 4.8%; plate XXX:3) Standard Dimensions Range Mean deviation Mode Length 7-16 cm. 10.9 cm. 2.8 cm. Width 5-16 cm. 8.5 cm. 3.1 cm. Thickness 4-11 cm. 6.7 cm. 2.0 cm. - W/L 50-108 78.0 15.8 67 T/W 63-117 81.1 15.4 - Primary form No. Percent Chunks 8 56.1 Unidentifiable 6 43.9 Irregular Cores (12, 4.1%) Standard Dimensions Range Mean deviation Mode Length 7 -14 cm. 9.8 cm. 2.2 cm. Width 5-11 cm. 7.4 cm. 1.6 cm. Thickness 2-7 cm. 4.1 cm. 1.3 cm. - W/L 54-90 77 11.2 71, 78, 89 T/ W 33;-100 58.3 26.9 71 Primary form No. Percent Trimmed face No. Percent Unidentifiable 9 75.0 Bifacial 7 58.3 Chunks 3 25.0 Unifacial 3 25.0 Parti-bifacial 2 16.7 Biconical Cores (80, 27.5%; plates XXX:5; XXXV:5; XLII:5; XLVI:2; LII:5) Standard Dimensions Rarge Mean deviation Mode Length Width Thickness W/L T/W 5-29 cm. 5-18 cm. 2-13 cm. 29-120 25-120 9.9 cm. 7.9 cm. 5.7 cm. 82 74 3.3 cm. 2.0 cm. 1.4 cm. 13.8 14.2 100 83 Mode 100 50 No. 3 20 42 5 15 2 10 73 Percent 1.8 11.8 24.7 2.9 8.7 1.1 5.9 42.9 - - 65. Anthropological Records Primary form Chunk Unidentifiable Trimmed face Dorsal Ventral Unifacial Parti-bifacial Bifacial No. Percent Plan 1 1.3 Ovoid 7 9 98.8 Semicircular Sub-quadrilateral Irregular 1 1.3 Elongate 1 1.3 Round 5 10 63 6.3 12.5 78.8 Plano-Convex Cores (9, 3.2%; plates XL:2; XLII:2; LIII:1) The distinctive traits of these cores are their plano-convex cross section and radial trimming on the plano-, or flatter, face. In addition to the radial trimming on the flat face, they are often trimmed on one side on the convex face. These cores closely resemble unstruck examples of Victoria West cores. On all the primary form is unidentifiable. Dimensions Length Width Thickness W/L T/W Range 6-22 cm. 6-17 cm. 3-10 cm. 56 -100 42 -78 Mean 13.1 cm. 9.9 cm. 5.8 cm. 79 59 Standard deviation 5.1 cm. 3.4 cm. 2.0 cm. 15.7 10.7 Mode 67 50 Plan Irregular Ovoid Round Sub-quadrilateral Sub-triangular Elongate No. Percent 1 11.1 1 11.1 2 22.2 1 11.1 2 22.2 2 22.2 Trimmed face Parti-bifacial Bifacial Dorsal Ventral Spindle Cores (2, 0.7%) Both are bifacial, and the primary form is Dimensions Length Width Thickness W/L T/W Range 11-15 cm. 6-7 cm. 4-5 cm. 47 -55 57-83 Mean 13.0 cm. 6.5 cm. 4.5 cm. 51 70 unidentifiable. Standard deviation 2.8 cm. 0.7 cm. 0.7 cm. 39.3 13.0 Utilized Pieces (83) Polyhedral (1, 1.2%) Dimensions: Length 6 cm.; Width 5 cm.; Thickness 4 cm. Hammerstones (73, 88%) Dimensions Range Mean Length 6-16 cm. 10.3 cm. Width 5 -13 cm. 8.5 cm. Thickness 4-9 cm. 6.2 cm. W/L 58-113 83 T/W 38-114 74 Primary form: 23 chunks (31.5%1); 50 pebbles (68.5%) Standard deviation 2.1 cm. 1.7 cm. 1.3 cm. 11.6 15.6 No. 26 7 1 1 10 35 Percent 32.5 8.8 1.3 1.3 12.5 43.8 No. 6 1 1 1 Percent 66.6 11.1 11.1 11.1 Mode Mode 100 71 v - . - 66 Keller: Montagu Cave in Prehistory Anvils (9, 10.8%) Dimensions Length Width Thickness W/L T/W Range 11-19 cm. 10-14 cm. 4-15 cm. 63-109 54 -125 Mean 14.9 cm. 11.9 cm. 9.1 cm. 84 88 Primary form: 7 boulders (77%); 2 chunks (22%). LAYER 5 DISCUSSION n Layer 5 four occupation horizons were discovered exposed. As discussed above, these horizons or aces appeared as concentrations of tools within a tigraphic unit and were excavated as separate enti- within the band in which they occurred. Layer 5 isted of four bands, in each of which an occupation rface was discovered. The surfaces were numbered aily from top to bottom in the site so that surfaces through VII occur in Layer 2 and surfaces VIII through in Layer 5. A total of 1 468 artifacts were present on surface , comprising 1,405 pieces of waste, 14 cores, 6 Wutilized pieces, and 43 tools. A more detailed break- down appears in figures 28, 29, and 30. Although no Obvious activity areas appear on the plot drawn of this surface (see fig. 31), there is a concentration of waste in squares 30 E and F and a cluster of larger pieces of waste as well as some tools nearby in the four squares 20 and 25 E and F. The assemblage from surface VIII closely approximates that of Layer 5 as a whole, with the exception of the hand-axes; only two band-axes are present on surface VIII, but no great significance is attached to this since the number of tools is fairly small, and thus the presence or absence of a small number of tools makes a considerable differ- ence in the percentages. In the band below surface VIII, surface IX was dis- covered. Of the 1,424 artifacts on this surface, only 1.2 percent (17) were tools. The assemblage resembles that of the layer as a whole less than was the case with surface VIII, but again this is probably owing to the small size of the sample. The assemblage on sur- face IX taken as a whole, however, does generally resemble that of Layer 5 in two characteristics: the presence of the large amount of waste and nearly as many minimally trimmed and broken pieces as hand- axes and cleavers (see figs. 32, 33, and 34). There are several clusters of artifacts evident on the plot, one of large flakes or flake fragments in square 35 F, another of smaller pieces in 30 F, and yet another in 25 F. Unlike most of the other surfaces, in this surface there are more trimmed pieces in the 25 and 30 squares, nearer the front of the cave, than in the back part of the 30 and 35 squares toward the back of the cave (see fig. 35). Surface X had a greater number of artifacts present, 1,812, than either surfaces VIII or IX, but again the assemblage is much like that of the whole layer (see figs. 36, 37, and 38). The percentage of tools present on this surface is greater than on any of the others, 5.8 percent, as opposed to 4.4 percent on surface XI and 1.3 percent on Layer 5 as a whole, and there are slightly more hand-axes and cleavers. There are con- centrations of artifacts in the four squares 30 and 35 F and G and in 20 and 25 G as well (see fig. 39). There was evidence of compaction in Layer 5 and this was particularly evident on surface X. The material in the deposit had been pushed down around large fallen rocks, but there was no evidence of any rearrangement of the artifacts, and the artifacts were plotted as if the surface had been flat. In other words, no attempt was made to make the plot look the way the surface would have appeared if photographed from above as we un- covered it. Instead the measurements were made along the dips and humps of the surface rather than across them. The greatest number of artifacts present on any surface was found on surface XI, where 3,322 were recovered. Here again the assemblage is practically identical with those of the other surfaces, as an exami- nation of figures 40, 41, and 42 will show. There are two very dense concentrations of artifacts, one in square 30 E and one in 35 F, and another concentra- tion, this time of large tools and flakes, occurs next to the wall of the cave in square 25 G. The rest of the surface is more or less evenly covered with arti- facts. Because of the extreme density of artifacts on this surface three plots are reproduced rather than Standard deviation 2.7 cm. 1.4 cm. 3.3 cm. 13.7 22.2 Mode 87 - - - i I 6 7 68 Anthropological Records TABLE 8 Tools and Cores from Layer 5 - ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ C12 X~~~~~~~~~~~ a) ) a) a) a a a) a) _ C) C)C) c ) C ) ) C) c) > ~ ~ ~ ~~~4 4- 4" 14 4C 4 2 d > Z 0 i ? i m m ?l n m m Hn-x an > fk 2 - 22 C0 - 3 0 0 Polyhedrals~~~~~~ - - 0 _ 0 _ _ _- a U la cu a a a) a Hand-axes and hand-axe flakes 2 - 29 1 15 - 21 22 1 27 57 - Cleavers and cleaver flakes 6 1 19 3 13 24 21 1 33 61 2 Knives - - 2 - 2 - 3 4 - 7 8 - Discoids - - 1 - - - 1 - - 1 - - Large scrapers 7 - 17 4 17 1 9 18 2 17 49 - Corescrapers 3 - 1 - 4 2 2 5 - 1 9 - Choppers 3 - 10 - 3 - 6 6 1 3 19 - Polyhedrals - - 1 - 4 - - - - - - - Small scrapers 2 - 17 - 13 - 1 8 1 6 26 - Nosed tools - - - - - - 1 - - 2 - Pointed tools 2 - 1 - - - - - 1 - Burin 1 - - - _ _ _ _ - - - - Unidentified - - - - 10 - - - - - - Minimally trimmed flakes 3 1 13 1 9 8 15 1 13 29 1 Minimally trimmed chips - - 2 - - - - 1 - - 2 - Ainimally trimmed chunks 3 1 19 3 15 - 5 12 1 6 11 1 Biface tips 3 - 8 2 - - 5 8 1 4 34 1 Bifacebutts 2 10 - 8 - 7 3 1 7 29 - Cleaver bits - - - - - - - - - - 2 - Bifaces 5 -p15 1 10 - 4 10 - 4 21 - Chisel-ended bifaces - - 1 - 1 - 1 4 3 2 9 - Pushiplanes 1 1 2 - 2 2 - 6 2 - Amorphous bifaces - - 1 - - 1 2 - - 1 - Beaked bifaces 1 - - 1 1 - 5 10 - 5 15 - Twisted-bit bifaces 1 - 1 - - - - 1 - 1 2 - Gouge-ended bifaces - - - 1 - 1 1 - 3 3 - Hand-axe/ choppers - _ - - 1 - - 2 - - 3 Double-pointed bifaces - - - 1 - - 1 1 - Single-platform cores - - 2 1 1 - - - - - - - Disc cores 8 - 45 5 19 - 12 21 1 8 50 1 Formless cores 2 - - 1 1 - - 4 - 1 6 - Irregular cores - - 4 1 4 - - 1 - - 2 - Biconical cores 3 - 6 4 5 - 13 13 - 11 23 2 Plano-convex cores - - - - - - 5 2 - - 2 - Spindle cores 1 - - - - - 1 - - - - - Totals 57 3 224 29 151 3 137 196 14 167 478 8 Keller: Montagu Cave in Prehistory in order to make the distribution clearer (see 43, 44, and 45). There is always a question about how closely the nple of material from an archaeological excavation resents the original assemblage, and this is parti- ly true in a situation like that at Montagu, where site has known limits and part of the deposit has e removed so that it is impossible to secure brial from all parts of the site. The question id be raised as to whether the material from this avation, which took place at one side of the cave, aurately represents the original contents of the cave. ieel, however, that since the four surfaces described e are so similar to each other and to the Layer 5 kemblage as a whole the material from this excava- is a representative sample. The alternative would to assume that the group or groups that visited the Oe always used one side for a specific activity, and seems unlikely. At least there is no evidence from her Acheulean occupation floors that anything like t kind of specialization occurred. Since there was no significant variation among the races in Layer 5, in figures 45, 47, and 48 the 'er is treated as a unit. The combinations of tool Pes used in discussing Layer 3 are used here as well. Scrapers, taken together, are the most frequently curring tool, but when divided into large and small itegories, this is no longer true. Cleavers then be- pie the single most frequent type, comprising 15.8 rcent of the tools. They are followed in frequency the minimally trimmed pieces, hand-axes, and other aces. Next come the large scrapers and tool frag- Onts, which make up 11.9 percent and 11.2 percent spectively, and those are followed in frequency by the heavy duty tools, 7.1 percent, and the small scrapers, 6.6 percent. Knives, discoids, and other small tools each make up less than 3 percent, the latter two less than 1 percent. Of the total 85,163 artifacts in Layer 5, 98.6 per- cent is waste. It should be pointed out here that two random samples of waste were taken, and the analysis for those units, above surface IX in band 21 and below surface IX in band 21, is based on those samples. The cores, however, were not sampled since they had been described with the tools, so the two samples were taken only of the flakes, flake fragments, chips, and chunks. Sampling was done by the coning and quarter- ing method (White and White, 1964:64). The flakes, and so forth from a square were put on a table and inspected to remove any small tools or utilized pieces, and the material was counted. It was then thoroughly mixed and heaped into a cone. Next it was divided and a portion selected for the sample. Thirty-eight percent (4,690) of the waste from above surface IX and 58.2 percent (5,119) of the waste from below surface IX were the sizes of the two samples. Cores comprise 0.3 percent of the waste, and the frequencies of the different types are shown in figure 49. End-struck flakes comprise 48.2 percent and side- struck flakes 51.8 percent of the flakes. The minimally trimmed pieces have again been included with the tools rather than being put into a separate "modified" category. There are slightly more minimally trimmed pieces than there are hand- axes, but this comparison would be impossible if these trimmed pieces were not included with the tools. Utilized pieces make up a very small part (0.1%) of the total assemblage. f I I t, i. ?L 69 COMPARISONS LAY ER 1 The term "Wilton" has been applied not only to material from South Africa but also to other assem- blages, from Rhodesia, through Zambia, and into the Horn. However, the descriptions of this material in the literature are very unsatisfactory when one at- tempts to make detailed comparisons. Often no arti- fact counts are given, and when they are provided, the "types" in terms of which the counts are given are not defined, so it is virtually impossible to cor- relate lists of tools from one site with those from another. The inadequate state of our knowledge about the later Stone Age has recently been summarized by R. R. Inskeep (1967). The following comparisons will be, perforce, somewhat superficial. Peringuey illustrated a crescent from the Orange Free State in 1911, and later Hewitt (1921) reported crescents from Stillbay. In the same publication, Hewitt reported the occurrence of crescents, small scrapers, ostrich-eggshell beads, and other items from a rock shelter on the farm Wilton in the eastern part of the Cape Province of South Africa. Since that time, Wilton material has been found on many sites, and hundreds of thousands of artifacts have been col- lected. Although much of this material has come from surface sites, several caves along the coast of South Africa have been excavated and produced Wilton assem- blages. The artifacts from Layer 1 at Montagu have their closest affinities with this Wilton material. At Khami, a stratified open site in Rhodesia, Cooke excavated a lower, a middle, and an upper Rhodesian Wilton. The lower and upper units are distinguished from the middle by differences in the technique of manufacture, the presence of "large implements" in the middle Wilton, and differences in the raw material used (Cooke, 1957:37). The three units together total only 115 tools and so the percentages listed for each unit are likely to be misleading. Crescents account for 14.6 percent to 21.9 percent and "thumbnail" scrapers" for 20.8 percent to 30.8 percent of the tools. Backed blades make up from 27.1 percent to 34.6 percent of the tools. An examination of my figure 9 will show that the percentages of tools in these Khami occurrences do not resemble the Montagu occurrence, although in general the tool types present are similar. Wilton material was also reported from two caves in Rhodesia, Pomongwe and Tshangula (Cooke, 1963), but no list or counts of the occurrences are given. Another site, which is closer to Montagu and which contained Wilton material, is the Oakhurst shelter located in the Cape Folded region near George. Oak- hurst was excavated between 1932 and 1935 by Good and reported by him in 1938. He asserted that there were three units present, one above another: Smithfi B at the bottom, above this Smithfield C, and Wilton at the top. Fagan (1960) has reexamined the material and concludes that it is, in fact, homogeneous from bottom to top. Schrire (1962) points out that Fagan had not examined all the original material and con- cludes from her analysis that Goodwin's original assessment was correct. The descriptions provided by Fagan and Schrire are among the most detailed for any of the "Later Stone Age" occurrences in South Africa. Although there are no definitions of the types of tools, gener comparisons can be made. The Smithfield B straturm contained no crescents and so will not concern us. The occurrences that are termed Smithfield C and Wilton respectively at Oakhurst are very similar to one another; scrapers are the most common in both, and both have crescents. Schrire states, however, that the scrapers in the Smithfield C level are signi- ficantly larger than those in the Wilton level and separates the two occurrences on this basis. At Oakhurst in the Smithfield C level, scrapers are the most numerous artifacts, comprising 86.5 percent of the tools, and the same is true of the Wilton level where they make up 69.2 percent of the tools. In Layer 1 at Montagu, scrapers account for 71.9 percent of the tools. Single crescents make up 4 percent of the tools in the Smithfield C and 13.5 percent in the Wilton, whereas at Montagu 4.4 perc of the tools are single crescents. Outils 6caill6s found in both the Smithfield C and the Wilton where they account for 2.1 percent and 3.8 percent of the tools respectively. At Montagu outils 6caill6s make up 11.4 percent of the tools. From this brief comparison we can see that the Smithfield C and the Wilton at Oakhurst are similar to each other and that both are similar to the Mon [ 70 1 i.L I t. ,4 4 I i I t I Keller: Montagu Cave in Prehistory Layer 1 occurrence. More detailed comparisons are sot possible because of the differences in terminology, Out if we compare Schrire's figures for the size of tconvex flake scrapers" with the scraper category at Montagu, we find that the mean length for Smithfield and Wilton is between one-half and one inch, whereas Montagu Layer 1 scrapers have a mean length of 28.4 ,nm. or about one and one-eighth inches. Schrire's na1ysis also used the area of the rectangles that can pe described around a scraper as an important feature Sor distinguishing the Smithfield C from the Wilton. ,Tis information is not presently available for the kiontagu material, so that comparison cannot be made. i general, however, it is clear that the three occur- pences are similar to one another. In discussing the fmithfield, Clark (1959:211) points out that the Smith- "eld C has been called a crescentless Wilton; yet ,Ochrire's use of the term contradicts Clark's state- "ient and further confuses the issue. Another cave site is that of Matjes River, which is Oeported by Louw (1960). Layer C of this stratified site ontained an occurrence including small scrapers and prescents. Louw compares it with the Wilton material prom Oakhurst and concludes that they are the same, o the Matjes River layer C material is probably simi- Fa to Montagu as well. Clark (1959:209 ff.) discusses the Wilton and describes t "Western Cape" Wilton, the characteristic feature of bich is the double-crescent-that is, a crescent backed jpboth sides to two convex edges. It is interesting to pate that no double crescents occur in Layer 1 at pontagu, and so such implements may not be reliable brkers of a geographic unit after all. As the number of carbon-14 dates increases, the tlater Stone Age" is pushed farther back in time. In 1959 Clark felt that the dates from the Matjes River Cave, which placed the Wilton between 5,758 + 400 B.C., and 3,443 + 250 B.C., were surprisingly early and suspected contamination (Clark:1959:188). Cooke, discussing the date for the Wilton at Pomongwe, says "The Later Stone Age deposit, which on all visual evi- dence and other criteria should be dated during the last 1,000 to 1,500 years, continues to give dates very much earlier" (1963:147). The date for the Wilton at Pomongwe is 7,610 + 110 B.P. Montagu is no exception to this trend, and a date (GRN. 4725) for Layer 1, taken on charcoal from feature 4, is 7,100 + 45 B.P. Recently, H. J. Deacon (n.d.:5) has suggested that the Wilton first appears approximately 8,000-9,000 years before the present, and that it may have been preceded by an earlier "large-tool tradition" with an age of 8,000-13,000 years B.P. From this brief resume, the general affinities of the Layer 1 occurrence at Montagu are clear. In basic makeup and chronology it certainly falls within the Wilton Industry of the "Later Stone Age," but our knowledge of the "Later Stone Age" is so unclear that the assignment of an Industry or Phase name to the Montagu occurrence is premature. There are many questions yet to be answered, such as the relationship of the Smithfield C as represented at Oakhurst to the Wilton. Are they really different, or are they only variants of the same Industrial Complex? Similarly, what kinds of differences are there in the Wilton that are attributable to activity variation? What do the high frequencies of outils 6cailles and scrapers at Montagu mean when compared to the large quantity of crescents at Khami? The recent work of Hilary and Janette Deacon is helping to clarify these issues. LAYER 2 u 1927 and again in 1928 Stapleton and Hewitt pub- ished papers in which they described the artifacts covered from a rock shelter located in the pass led Howieson's Poort near Grahamstown. This regate, which was named Howieson' s Poort, after type site, has never been adequately described has it been found in many other sites. As a result, Fe following comparisons will be necessarily less cific and less extensive than those in the section kaling with the Acheulean from Montagu. Through the gwness of Mr. and Mrs. Hilary Deacon of the Albany fuseum, Grahamstown, I was able to examine the lwieson's Poort type collection, and later Mr. and Mrs. Deacon inspected the material from Layer 2 at Montagu. All of us agreed that the two were the same. Since the type collection is only now being described by Mrs. Deacon, however, it is impossible to demon- strate this similarity, and their virtual identity can only be asserted. Stapleton and Hewitt (1928:407) listed four features that they believed distinguished the Howieson's Poort aggregate from others. These were burins (the first burins described from South Africa are the ones from Howieson's Poort); large crescents; trimmed points, some of which were bifacial and often made on flakes with facetted butts; and obliquely truncated blades. 7 Anthropological Records The crescents were larger than those known from Wilton and other sites, and the blades that they found were also larger than those usually associated with Wilton material. This was interpreted as indicating that the Howieson's Poort was pre-Wilton. The points were felt to have certain affinities with those known from Stillbay and other sites near the Cape Peninsula, and so the Howieson's Poort material was held to belong, with the Stillbay, in the "Middle Stone Age," but also to be in part contemporary with the Wilton. This interpretation was based on typological similari- ties, not on stratigraphic evidence. In 1932 Goodwin and Malan excavated a cave on Cape St. Blaize near Mossel Bay and recovered an aggregate that has been called the Mossel Bay variant of the "Middle Stone Age." This material was described by Goodwin and Malan in 1935, and its assignment to the "Middle Stone Age" was based on its typological similarity to the Stillbay and on the fact that the "Howieson's Poort Variation cut into the Mossel Bay at this site (Cape St. Blaize Cave), much as it cuts into the Stillbay Culture at the Skildegat site" (Good- win and Malan, 1935:138). The relationship of Howie- son's Poort to Mossel Bay at Cape St. Blaize was based on the occurrence of a "single lance-head of Howieson's Poort type" (Goodwin and Malan, 1935: 124). I was able to work through the Mossel Bay collection, which is housed in the South African Museum, Cape Town, and I did not find Howieson's Poort-like material in the excavation unit to which Goodwin and Malan refer. There is an obliquely trun- cated blade present, but it came from an unstratified part of the deposit, so it is impossible to say what its associations were and how it is related to the rest of the occurrence. The Mossel Bay occurrence has been redescribed and its affinities discussed elsewhere (Keller, 1969). Suffice it to say here, there is no evi- dence to show that the Howieson's Poort "cuts into the Mossel Bay." Skildegat Cave on the Cape Peninsula was excavated in the 1920s by two amateurs, a Mr. Peers and his son, who found the following stratigraphy: at the top, Wilton material; below, a "midden" with hammerstones and cores but no diagnostic tools; typical Stillbay material; material that was similar to the Howieson's Poort; a practically sterile zone, and at the bottom, Stillbay material (J. Deacon, pers. comm.). Jolly, in 1948, published the results of further excavation at Skildegat and concluded that the stratigraphic sequence outlined by the Peers was incorrect and that there was no Stillbay material between the Howieson' s Poort and the Wilton, but rather a continuous change from Stillbay through Howieson's Poort to Wilton. At another cave on the Cape Peninsula, near Kalk Bay, Howieson's Poort material was found, but in the upper levels there, were a few artifacts that were felt to be possibly "Later Stone Age" (Goodwin and Peers, 1953). These are the only presently known stratified occurrences Of the Howieson's Poort. Malan (1949) summarized the sequences from the "Middle Stone Age" to the "Later Stone Age" as they were known in the various parts of South Africa and equated the Howieson' s Poort of the Cape Folded region with the post-"Middle Stone Age"/pre- "Later Stone Age" Magosian assemblages. These were named after the type site of Magosi in Uganda, which was thought to contain an industry typologically transitio between the "Middle Stone Age," with its emphasis on the "Levallois" type core, and the "Later Stone Age," with its emphasis on blades. The original Magosi collection has subsequently been shown to be mixed and this transition is not well-documented in South Africa. However, two sites in Rhodesia contain se- quences of occurrences from the "Middle" to "Later Stone Age." It is with the material from these sites, Khami and Pomongwe, that the Montagu material can be compared. The excavations at Khami and the material recove from them were mentioned above. The "Magosian" le produced points, scrapers, and crescents that are 1 than those associated with the "Later Stone Age"-all of which are common to "Magosian" occurrences. Co parisons of tools from Layer 2 at Montagu with thoe from the "Magosian" levels at Khami are difficult to make, since Cooke's definitions are not detailed nor are the captions to his illustrations. All the features that Stapleton and Hewitt felt were diagnostic of the Howieson's Poort, however, seem to be present in "Magosian" levels at Khami. Even had comparisons been possible, detailed comparisons of the tools re- covered from two sites could have been misleading, since the activities carried out at a given site can drastically affect the kind of tools left there, and as a result a single group of people could be represe by two very different tool aggregates at two differe sites. A more profitable way of comparing material is look at the techniques used in manufacturing the t as illustrated by the cores and waste, and to base assessments of similarity or lack thereof on this of evidence. Fortunately, Cooke has recorded not o the kinds of cores found but also the kinds of platf present on the flakes recovered from Khami. Therel are some difficulties in the utilization of this info tion, but they are fewer than those encountered w attempts are made to compare tool types. For pu 72 Keller: Montagu Cave in Prehistory arison, cores can be grouped into three cate- (1) those that have been trimmed radially, e that have more or less parallel flake scars, miscellaneous other cores. Radially trimmed may have been so trimmed in preparation for oval of a single flake (the "Levallois" type of or the trimming may have been done in order ure several flakes. In any case, the main dis- n is between the production of more or less rly shaped flakes, on the one hand, and parallel- bladelike flakes produced by the cores with 1 scars, on the other. Of the 465 cores from osian" levels at Khami, 60.4 percent have trimming and 20.4 percent have parallel trim- In contrast, 52.1 percent of the 614 cores from wieson's Poort layer at Montagu have parallel This would indicate a preference at Khami for d of flake produced by a radially trimmed core s parallel flakes were preferred at Montagu. nongwe, a cave site in Rhodesia also excavated ported by Cooke (1963) as we mentioned above, ned a sequence similar to the one at Khami. again the figures indicate a preference for y trimmed cores; 47.8 percent as opposed to rcent "blade" cores. platform found on a flake is often indicative technique used in detaching the flake from the and also of the kind of core that was used. ed platforms are commonly held to be associated .the "Levallois" technique and very small, almost stent platforms are felt to be the result of blades been detached from cores by the punch tech- The terms "facetted" and "blade," as used by in the Khami report, are taken here to be the ents of "facetted" and "bulb only," as used for ntagu material. Of the 1,301 flakes in the "Mago- levels at Khami, 25 percent had facetted plat- and 27.5 percent had blade platforms. At gu 7.5 percent of the flakes had facetted plat- O and 10.2 percent had "bulb only" platforms. ~nmation about the flake platforms is not given in .omongwe report. The figures from Khami are Westing in that they apparently indicate the reverse he preferences reflected in the core types, although Montagu information agrees with the core types, teas the percentages are much smaller than those in Khami. [t should be pointed out that 24.4 percent of the es with parallel scars from Montagu have facetted lorms and so could have produced flakes with Alar platforms. The problem is that one cannot uime that facetted platforms occur only on radially bmed "Levallois" type cores. The platform of a core intended to produce parallel-sided flakes could be prepared in just the same way as a "Levallois" core. A more fruitful way to compare assemblages would be to look at the interrelationships of flake shapes and platform treatment as well as the platforms and shapes of the scars found on the cores. Unfortunately, this is not possible with either the Rhodesian or the Montagu material. Cooke sees the "Magosian" at Khami as a direct outgrowth of the Stillbay, which precedes it. He says "The Magosian at Khami must be included as a Middle Stone Age industry . . . being it seems the final devel- opment of the Stillbay" (1957:41). Summers has much the same opinion: "The Magosian cores are in the main a development of MSA cores, and Magosian flakes are in the main facetted butt flakes struck from disk cores of a very advanced Levallois type" (Summers, 1957:57). Cooke thinks that the "Magosian" is the result of an autocthonous development on which "impinged a blade- burin industry," but that the "blade-burin" influence was slight, and the preference for "Levallois" cores and their products continued (Cooke, 1957:41). On the other hand, the Howieson's Poort is described as an "almost pure blade-burin culture." From the foregoing review of the Montagu material, it is clear that the situation is far from being as simple as had been supposed. There are more "blade" cores in the Howieson's Poort from Montagu than in the "Magosian" from Khami or Pomongwe; "blade" cores make up almost half the cores from Montagu, and there are more flakes with "blade" platforms from Khami than there are flakes with facetted platforms. The types of tools found in the Howieson's Poort and the "Magosian" are broadly similar, and it would seem profitable to withhold judgment about the relationships of these industries until they have been more thoroughly studied. The problem of chronology was mentioned briefly above, but it warrants further discussion. The various "Magosian" industries have been placed in the "Second Intermediate Stage," between the "Middle and Later Stone Ages," and they were thought to date between 6,000 and 10,000 B.C. (Clark, 1959:169). More recently, however, other dates have been obtained that make these industries much older, and the whole concept of the "Second Intermediate Stage" seems less useful than previously (Clark, 1970:105-137; H. J. Deacon, n.d.). Clark (1970:129) suggests that the "Middle Stone Age" may have begun 35,000 or even as much as 50,000 years ago. The chronology of the "Middle Stone Age" has been summarized by Klein (1971). Carbon from the top of Layer 2 at Montagu gave a date of 23,000 + 180; samples from the middle of the 73 Anthropological Records layer dates to 19,100 + 110 B.P. and 50,800; and samples from the very bottom of the layer gave dates of 45,900 + 210 B.P., and greater than 38,000-the latter date representing the counting limits of the laboratory. Certainly these dates are very surprising, particularly the earliest ones from Montagu, but there is no easy way to explain them away. It is unfortunate that the series of dates from Montagu is inconsistent, since this inconsistency partially obscures the issue raised by these and other surprisingly early dates (Mason, 1969; 1971). Charcoal is absent in Layers 3 and 5 at Montagu and so contamination of samples from Layer 1 by old charcoal is impossible, and in any case contamination usually results in dates that are too young rather than in those that are too old. Either there is a presently unknown mechanism of contamination in C-14 samples, or these early dat are accurate, and the younger dates have come fz contaminated samples. If they are accurate reflect of the age of the artifacts associated with them tb ideas of the Late Pleistocene prehistory of Afri have to be changed. To summarize then, the Howieson's Poort na from Layer 2 at Montagu has been compared wi "Magosian" from Khami and Pomongwe and has found to be generally similar. There are differe however, particularly in the kinds of cores used, these differences seem to be less extreme than and Summers believed. Unfortunately, a clearer of the chronological and cultural relationships Howieson's Poort is unobtainable from the prqs literature. COMPARISONS OF LAYERS 3 AND 5 In the following section the Acheulean material from the lower two artifact-bearing layers in Montagu Cave will be compared with each other and then with other appropriate sites in Africa. The comparison of these assemblages will be more extensive than the compara- tive sections dealing with Layers 1 and 2, first because two layers are available to compare with each other, and second because more recent and detailed work has been done on this kind of material than on artifacts of the kind in Layers 1 and 2. The occurrences in Layers 3 and 5 are very similar to each other. Probably the most striking resemblance is in the amount of waste present: of the total, 99.1 percent is waste in Layer 3 and 98.6 percent in Layer 5, and as we will see, this is a higher percentage of waste than is known from any similar site. Both layers have similar percentages of large cutting-edge tools, that is, hand-axes, cleavers, and knives, 30.1 percent in Layer 3 and 33 percent in Layer 5. The percentages of biface fragments are similar, 14.2 percent in Layer 3 and 12.6 percent in Layer 5. "Other bifaces" amount to 9.7 percent in Layer 3 and 13.9 percent in Layer 5. The percentages of minimally trimmed pieces are less similar, 11 percent and 15.1 percent in Layers 3 and 5 respectively, but this is a distinctive class of artifact, and the fact that they are present at all in both occurrences is important. One of the features in which these two layers differ is the scraper category. Scrapers comprise 30.8 per- cent of the Layer 3 tools but only 18.5 percent of the tools in Layer 5; small scrapers are most common in the upper layer, whereas large scrapers are t frequent type in Layer 5. Heavy-duty tools are common in either layer, but they are less com 3 than in 5; the reverse is true of the "other tool" category. Looking at the waste, we find that although layers have large quantities of waste, there ar ences in the kinds of waste present. Flakes as fragments make up 47.3 percent of the waste i 3, whereas they make up 61.9 percent of the L waste. On the other hand, chips and chunks a8 common in Layer 3, comprising 52.2 percent, I they are in Layer 5, where they represent onI percent of the total. Cores account for less percent of the waste in each layer. - These variations in the waste are tantaliz difficult to interpret. It seems reasonable to that some differences in the kind of chipping done, or in the methods used, have resulted variations, since the raw material used in was the same. Yet, as we have seen, the ki artifacts found were very similar in both la In figures 27 and 49 the types of cores f Layers 3 and 5 are shown. The disc core iJ common type in both layers, but there are more biconical cores in Layer 5 than in La There is a small number of other types of present in both layers; one of the most inte features is the presence of the plano-conver Layer 5. As has been remarked in the desc section, these are very similar to unstruck 74 I bores, which are well known from the Vaal River win (1929:11-12) mentions differences in hand- pe between what we have called Layer 3 and 5. Of his 77 "better specimens" (no waste or hed" tools were saved) in Layer 5, 80.5% are shaped," which appears to be equivalent to our ate, and 19.5 percent are "almond shaped," is similar to our ovate shape. In Layer 3, 16.8 of his 89 "better specimens" are "pear- ','78.8 percent are "almond shaped," and 4.5 are "round." These figures have prompted writers to suggest that the hand-axes at Mon- llustrate a trend toward the Fauresmith, but our s indicate that this tendency is not so strong as In's figures lead one to believe. Goodwin's own s for the Fauresmith and Stellenbosch in general n and Van Riet Lowe, 1929:72) indicate a w/l of 57 for the Fauresmith and of 51 for the Stellen- The measurements were taken on fifty "advanced nbosch" tools from twelve sites and on thirty ced Fauresmith" tools from six sites. Goodwin ted the mean length, width, and thickness for os and then derived an "average ratio" from eans, rather than computing the w/l for each inen and deriving the means of those values as ve done. re recently Glynn Isaac has measured and ned to shape categories all the bifaces in the African Museum from the 1919 excavation at gu, and he has very kindly put his work sheets disposal. Isaac's measurements were made on band-axes from our Layer 5 and 195 hand-axes our Layer 3. This information is summarized in ires 50 and 51, and Table 9. The relationship be- in the hand-axes of the two layers indicated by CIs figures are essentially the same as those sug- ted by our measurements of the 1964 material. The gn length of the Layer 3 hand-axes is about 2 centi- Lers less than the mean length of Layer 5 hand- , Similarly, the mean w/l index is higher for er 3 than for Layer 5. On the other hand, the quency with which the ovate and lanceolate forms 75 occur is different according to Isaac's classification. His "Pyriform" or Hemilemniscate category is the most common in both layers, 74.2 percent in Layer 5 and 61.7 percent in Layer 3, and the ovate forms increase from 5.2 percent in Layer 5 to 19.4 percent in Layer 3. Other forms account for 15.6 percent and 18.8 percent of the hand-axes from Layers 5 and 3 respectively. Clearly, the shift in both the 1919 and 1964 aggregates is the same, that is that ovates occur more frequently in Layer 3 than in Layer 5. However, in the 1919 collection ovates are less common than other forms in both layers, whereas the reverse is true in the 1964 collection. This difference is interest- ing, but the simplest explanation would seem to be that it is a reflection of differences in the classifier rather than in anything else, and this is supported by the simi- larities in mean lengths and w/l ratios. The frequencies of the basic hand-axe shapes in the material from the 1964 excavation from Layers 3 and 5 are shown in figures 52 and 53. In these figures, all the varieties of the lanceolate shape have been combined into the lanceolate category, and the same is true for the ovates. The one exception is the ovate- acuminate form, which is so different from the rest of the ovates that it is included in the "other" category, with such shapes as "double pointed" and "limande." Ovates are the most common shape in both layers, whereas lanceolates make up 15.8 percent of the hand- axes in Layer 3 and 33.7 percent in Layer 5. The incidence of ovate forms is reflected in the w/l ratios for hand-axes since the mean ratio for Layer 3 is 60 and that for Layer 5 is 55. The mean length of Layer 3 hand-axes is 14.7 cm., and the mean length of the Layer 5 hand-axes is 16.6 cm. One feature, then, that differentiates the earlier Acheulean from the later at Montagu is the greater incidence of lanceolate hand- axes in Layer 5; further, the earlier hand-axes are somewhat larger than the later ones. Among the cleavers the most common shape in Layer 3 is parallel sided, 60 percent, and the same is true of Layer 5 where parallel-sided forms amount to 51.6 percent. In both layers the next most common shape is convergent. The mean length of the Layer 3 TABLE 9 Hand-Axes from 1919 Excavation Layer Mean Mean Mean Mean Mean length width thickness w/l t/w D 16.1 cm. 8.9 cm. 4.5 cm. 57 50 F 18.2 cm. 9.4 cm. 5.0 cm. 53 54 Keller: Montagu Cave in Prehistory I k? V 76nthropological Records cleavers in 15.7 cm., and of Layer 5 cleavers, 17.5 cm.; the mean w/l ratio for cleavers is virtually the same, 58 in Layer 3 and 59 in Layer 5. So again, even though the shapes are similar in both layers, the trend for the later tools to be smaller continues. To summarize, the assemblages in Layers 3 and 5 are very like each other. The percentages of con- ventional types of tools, such as hand-axes and cleavers, are similar, and the characteristics that (as will become apparent) are distinctive of Montagu, such as the high percentages of waste, the "other bifaces," minimally trimmed pieces and fragments of tools, are not only present in both layers but also are present in similar quantities. These characteristics are best interpreted as indicating that the cave was used as a factory or work- shop site. Although there are some differences between the two assemblages, these differences are far outweighed by the similarities, and for the purpose of a comparison of the Montagu material with that from other sites, the two layers are discussed as a unit. A question can be raised about the accuracy with which the 1964 collection reflects the total assemblage of the cave. Cooke (1963:120-121) has shown that therl were significant variations in the assemblage from o part of the cave to another within a single layer at Pomongwe, and this might apply equally to Montagu. One could ask, might not one side of the cave have been used primarily for workshop activities? This is possible, but not very probable. The likelihood that a site would be occupied intermittently over a long p of time, with one period of abandonment lasting pr ably several thousand years, and that exactly the kinds of artifacts would be left in the same part of cave, with no trace of the artifacts from the other parts of the cave, seems very small. To my knowle nothing comparable has ever been demonstrated or even suggested for any other site in the world. In addition, Isaac's figures for the 1919 collection sug- gest that the density of tools in the other parts of cave was about the same as in the area we excavate COMPARISONS WITH OTHER SITES Hand-axes and cleavers from gravel deposits have been collected for many years in South Africa. They were discussed by Peringuey (1911), and later Good- win and van Riet Lowe (1929) referred to them as the Stellenbosch industry, after a site near the town of Stellenbosch in the Cape. The first Pan-African Con- gress on Prehistory recommended that the term "Acheulean" be substituted for the term "Stellenbosch," which had been confined to the Republic of South Africa (Leakey:1952). An aggregate that lacked hand-axes and cleavers but was thought to be contemporary was called the "Hope Fountain" after a site in Rhodesia. But the work at Isimila and at Olorgesailie and Kalambo Falls (Howell and Clark, 1963; Clark, 1964a) has shown that the small tool units are, in fact, part of the industrial complex that contains hand-axes and cleavers. The excavation of Acheulean occurrences in undis- turbed archaeological contexts has shown that there is a considerable range of variation in their composition and these variations have been described in detail (Kleindienst, 1961; Howell and Clark, 1963; and Clark, 1964a). As a result, more precise comparisons can be made between the Acheulean occurrences from Montagu and from other sites than was possible with the occurrences from Layer 1 or 2. Kleindienst (1961), and later Howell and Clark (11 have outlined three, or perhaps four, types of aggrel that contain tools resembling those found in Layers and 5 at Montagu. Type "A" (Howell and Clark, 196 is distinguished by a high frequency of hand-axes/ cleavers/knives, and the frequency of waste is low. The frequencies of large cutting-edge tools on flo that have type "A" aggregates range from 58.9 per on Olorgesailie surface 7 to 77.5 percent on Isim. J 12. Waste frequencies range from 40.5 percent floor 7 at Olorgesailie to 71.6 percent on Isimila The Montagu collections fall considerably outside ranges, as an examination of figures 25 and 47 w. reveal. A second type is " B," which contains low perce of large cutting-edge tools, a high frequency of s tools, and "substantial" frequencies of waste. Fr quencies of large cutting-edge tools range from none Olorgesailie surfaces 12 and 13 to 20.5 percent on Isimila K 18. The frequencies of small scrapers other small tools combined on these floors range 56.3 percent at Broken Hill to 100 percent on Olo sailie surfaces 12 and 13. Waste frequencies rang from 75 percent on Olorgesailie surface 11 to 35. percent at Broken Hill. Once again, an inspection II PI 76 I s 25 and 47 will show that the Montagu occur- ts fall outside these ranges. "C" is characterized by a high percentage of -duty tools, and one well-documented example is at Isimila, where heavy-duty tools account for percent of the total number of tools. Another site an aggregate of this kind has been found is in Angola where the artifacts occur in river s, and here chopping tools make up 96.5 percent total (Clark, 1963:94). Once again the Montagu *1 does not approach these figures. fourth kind of aggregate is intermediate between and "B" in that large cutting-edge tools and small are present in about equal frequencies. This ly does not apply to Montagu (see figs. 25 and 47). floors at Kalambo Falls have not been men- d in this discussion, but the information about floors published by Clark (1964a) indicates that are no more like Montagu than the others sum- zed above. Material from Acheulean floor 8 at -bo approaches that from Montagu in the amount aste, 91.5 percent, and has a moderate number ge cutting-edge tools, 45.5 percent, but it also 36.4 percent small tools, and this is quite unlike The situation is the same with Acheulean 6B, which has about equal amounts of large -edge tools and small tools. rom these comparisons it is clear that the Mon- Acheulean occurrences do not conform to the sification from other sites in Africa. These com- ons have been made in terms of the artifact ories that Montagu shares with these sites, but mention has been made of the more distinctive res in the Montagu occurrences. First, Montagu a higher frequency of broken bifaces than do the r sites. The frequencies of broken bifaces at Olor- ailie range from 0 percent to 10.7 percent and at Lia from 0 percent to 8.4 percent whereas Montagu er 3 has 14.2 percent and Layer 5, 12.6 percent. Another distinctive characteristic at Montagu is the ally trimmed pieces. As stated in the descriptive on, these are artifacts, often flakes, that have had all amount of trimming but that have not been ked into any standardized shape, nor has any special of working edge been fashioned. Often the bulb and form of the flake have been removed or reduced. Facts of this kind are nearly as common at Mon- as hand-axes or cleavers, yet few, if any, pieces this kind are reported from the other sites with h Montagu is being compared. In fact, Kleindienst , "The characteristics of the sites for which the assification is designed largely eliminate the prob- 77 lem of dealing with 'unfinished' tools or 'roughouts'll (1962:84). The problem of what is " finished" and what "un- finished" is difficult for the archaeologist to solve since it requires knowledge of the intentions of the makers of the tools. For this reason I have named this category "minimally trimmed objects" instead of something like "roughouts." Nevertheless, the number of hand-axes and cleavers, as well as other tools, that are made on flakes on which the bulb and plat- form have been removed or reduced makes one sus- pect very strongly that these minimally trimmed pieces represent the first stages in the process of manufacturing some kind of bifacial tool and are not end products in themselves. The presence of these forms suggest that Montagu was used as a factory site. Another feature of the Montagu material is the "other biface " category. In the introduction I gave my reasons for following Kleindienst's typology, and this has been done with considerable ease except for the "other biface" category. Clark (1960:315) has made the distinction between formal tools, such as hand- axes and cleavers, and informal tools, such as small scrapers and other small tools. The distinction is based on the recurrence of a constellation of attributes -such as size, plan, and cross section-in a single combination that produces the great similarity of the formal tools over a wide area for a long time period. Informal tools are so named because they do not ex- hibit a recurring combination of traits but rather have an enormous variation in form and edge. At Montagu there are a number of tools, which are combined in the charts under "other biface " category, that are less amorphous than the informal tools but less regular than the formal tools. In the descriptions these tools are split into a variety of categories, such as twisted- bit bifaces, chisel-ended bifaces, push planes, beaked bifaces, and others. There are so few examples in each category that they become statistically meaning- less. Yet when all are lumped together, they exhibit so much variation in form, type of working edge, and size that the group is not typologically very convincing. That is, the differences appear to outweigh the simi- larities, yet these types resemble each other more than they resemble other types. These "other bifaces" comprise 9.7 percent of the tools in Layer 3 and 13.9 percent of the tools in Layer 5; thus, they are nearly as common as hand-axes. I found considerable diffi- culty in applying Kleindienst's terms to these tools. An examination of Kleindienst' s typology shows that she does include categories such as "twisted-bitted chisels," "end-notched bifaces," and "various bifacial Keller: Montagu Cave in Prehistory I Anthropological Records tools," but the definitions of these terms are very brief. Tools of this kind are lumped together in the "other large tool category" by Kleindienst (1961:49-50), and the frequency of this category ranges from 0 per- cent to 2.1 percent at Olorgesailie and from 0 percent to 9.8 percent at Isimila. The frequency of "other large tools" at Kalambo Falls ranges from 0 percent to 1.9 percent (Clark, 1964a:101). Initially then, it is possible to say that "other bifaces" occur more fre- quently at Montagu than at the other sites. Since so little material from these sites has been illustrated, it is difficult to make direct comparisons of these "other bifaces" between Montagu and the other sites; nonetheless, a few comparisons are possible. The similarity between the beaked bifaces at Montagu and picks from other sites was mentioned in the descrip- tions, and a comparison of the tools illustrated herein on plates XLVI:3 and LII:1 with Clark (1964b) plate 1:4 and Clark (1963) Plate 6:3 will show these similarities. The Montagu tools are smaller than the others, however, and beaked bifaces illustrated in plates XXII:3 and 6, XXXVI:3, and XLIV:2 will make it clear that these resemble picks only in that the points are heavy and thick. There are no tools at Montagu that resemble the large pick from Angola illustrated by Clark (1963, Plate 6:4). Kleindienst's pick category has subsequently been subdivided into picks and core-axes, and the latter form has been described by Clark (1964a:96). Other tools have been called hand-axe/choppers by Clark (1964b, Plate I:1; 1963:52 and Plate 6:1 and 2). These illustrated tools resemble some of the Montagu "other bifaces," and the core-axe illustrated by Clark (1964b, Plate II:2) resembles in plan those pieces from Montagu illustrated in plates XLI:1 and XLVII:5. How- ever, the Montagu tools lack the characteristic core- axe trimming described by Clark (1964a:96) and con- sequently have been placed in the hand-axe/chopper category. This question is confused even further because picks, core-axes, and hand-axe/choppers from other sites that have been discussed above are all part of the post- Acheulean Sangoan or Sangoan/Lower Lupemban assem- blages, whereas the Montagu hand-axe/choppers come from the earlier Acheulean, Layer 5. In discussing core-axes, Clark has said, "Considerable difficulty has been experienced in finding a suitable term to describe the numerous types of bifacial core tools associated with the later Pleistocene cultures in the Congo Basin . . . . Considerable variation in shape, thickness, nature of secondary working and working edges is apparent from the literature" (1963:50). Obviously the situation is much the same with regard to the "other bifaces" of the Acheulean. One problem is that there are not many of these "other bifaces" on any one site. According to the figures in Howel and Clark (1963), there are 15 of these tools at Isis 5 at Olorgesailie, 8 at Kariundusi, 8 at Lochard, 6 the collection from Homestead on the Vaal River, al 16 from Kalambo Falls (Clark, 1964a:101), a total a 58 tools. This is obviously a small number of ite on which to base a typological category, and the p lem is increased because the tools are so varied form. At Montagu, however, there are 214 tools have been placed in this typological category. The were significant numbers of similar tools in the tions (which I had the opportunity to examine) fr Amanzi near Port Elizabeth and Mulder's Vlei ne Paarl, both in the Cape Folded Belt. Clearly ther needs to be a thorough comparative study made of these tools from as many sites as possible so tha the ranges of variation can be defined and useful butes for description and comparison discovered. Because of differences in terminology and, in respects, differences in the basic approach to ar logical material, less precise comparisons are with the collections described by Revil Mason fr northern part of South Africa. However, some i ing information can be derived from Mason's fi Mason (1962:221) cites hand-axe length figures Earlier, Middle and Later stages of the Acheule show no clear trend. These are summarized bel table 10. The mean lengths for all stages are le the means for either of the Montagu Acheulean TABLE 10 Hand-Axes from the Transvaal Earlier Stone Stage Mean length Mean w Early 12.1 cm. 60' Middle 13.2 cm. 54 Late* 12.5 cm. 63 *These figures are a mean of means from four occurrences. In terms of the w/l ratios, also summarized in 10, again there is no clear trend but the hand-a from the Later Acheulean are slightly broader fi length than the Earlier ones. The Montagu Layer axes resemble most closely the Transvaal Middl lean, and the Layer 3 hand-axes resemble either Earlier or Later Acheulean. Mason (1965:5) has statement about the frequency with which end-sti and side-struck flakes occur. This question has discussed elsewhere (Isaac and Keller, 1968); br it is apparent that the Montagu waste flakes res the waste from East Africa more than they resei 78 I Keller: Montagu Cave in Prehistory Transvaal waste with regard to the occurrence of -struck and side-struck flakes. It is possible that form in which the raw material occurs-some in in the Transvaal contrasted with boulders at u and Kalambo Falls, for instance-contributed e production of end-struck flakes. The percentages of artifact types shown by Mason 2:218) remain virtually the same from the Earlier he Later Acheulean, with the exception of four new 38es added in the Later Acheulean. In fact these .mblages show much less variation through what be an enormous time span-if in fact they do fesent the Acheulean from Earlier to Later-than relatively contemporary assemblages from the East zcan Late Acheulean. This conservatism is difficult explain in the current state of knowledge. The p1es on which the descriptions are based are onably large, but all but one of the assemblages e come from scree or gravel deposits, and it is ible that noncultural factors influence the content liese aggregates. None of the contexts from which on's material has been gathered is comparable to Montagu or East African sealed primary contexts, this detracts from the validity of the comparisons. For all of these difficulties and differences, there similarities between the Montagu and Transvaal leulean, and Mason's illustrations of later Acheulean Ifacts (Mason, 1962: figs. 94-97, pp. 110-120) make e clear. For instance, his hand-axe, figure 94:3, lld certainly have been placed in my "other Ice" category, as would figure 95:2, and the ge cores, figures 118:3, 119:1 and 2, and 120:2, Oe very similar to the "plano-convex" cores of wer 5 at Montagu. To summarize, then, the occurrences in the two yers, Layers 3 and 5 at Montagu, resemble each other very closely and much more than they resemble comparable aggregates from other sites. This is true not only for the frequencies of categories, such as hand-axes and cleavers or waste, which Montagu shares with these other sites, but also for those high-frequency categories that are characteristic of Montagu, that is, minimally trimmed pieces, other bifaces, and biface fragments. None of the four types of aggregates that have been defined from other sites fits the Montagu material, and since these variations have been inter- preted as the manifestations of different activities, Montagu must reflect an activity that is not represented by material from any of the other sites. This makes it difficult to answer the question of whether there is spatial variation within the Acheulean of Africa, since one possible hypothesis would be that those items that are characteristic of Montagu are connected with the special activity represented there and, therefore, would not be expected to be present at other sites representing other activities. This is cer- tainly a justifiable position as far as the high quantities of waste, broken bifaces, and minimally trimmed pieces are concerned. But the presence of the "other bifaces" in relatively large numbers at Montagu and the presence within the same region of similar tools at other sites that lack large quantities of waste and of minimally trimmed pieces would suggest that regional differences will be found in the "other biface" category. This is not at all improbable if the variations possible within the formal tools were rigidly delineated by the makers and the informal tools were allowed to be so amorphous as to defy a stricter organization. It is, then, to what might be called "semiformal" tools that we should look for regional differences, and yet it is just this class of tool which has been least well described and about which least is known. 7 9 CONCLUSIONS In the preceding pages the setting of Montagu Cave and its stratigraphy and contents have been described as well as the methods of excavation and analysis used in this study. The assemblages from the four tool-bearing layers have been compared with other assemblages and some of the problems that thwart such comparisons discussed. In the following pages the conclusions reached as a result of this work are summarized. The assemblage from Layer 1 falls within the range of the "Later Stone Age" both typologically and chronologically, but to make a more precise designa- tion is difficult. As pointed out above, the Montagu Layer 1 assemblage would have been called "Wilton" on the basis of the crescents found there. However, Schrire's work indicates that there are crescents in the "Smithfield 'C"' as well, and, as we have seen, in one characteristic at least-the length of the scrapers-the Montagu assemblage resembles the "Smithfield 'C"' more than it does the "Wilton." But neither the "Smithfield 'C"' nor the "Wilton" is really well described, nor has either been described from enough sites to make possible a clear idea of the range of variation that exists. For these reasons, then, the assemblage from Layer 1 at Montagu should be referred to simply as an occurrence of the South African "Later Stone Age" Industrial Complex. After additional comparable material has been described, it should be possible to make a more precise designation. The situation is somewhat more clear-cut with re- gard to the assemblage from Layer 2. As was pointed out, this assemblage is virtually identical with that from Howieson's Poort. However, the exact chronological relationship of the Howieson's Poort material to the other aggregates is not at all clear. The carbon-14 dates from the Howieson's Poort shelter and from Montagu Layer 2 are earlier than one would have pre- dicted. Nor is the relationship clear of the Howieson's Poort material to other aggregates with which it has been asserted to be typologically equivalent. The evi- dence for regarding the Howieson's Poort as represen- tative of a "pure" blade and burin industry is not con- clusive by any means. The Howieson's Poort does appear to be identical to the "Magosian" as it is from Rhodesia. Consequently, pending more absolu dates, more descriptions, and more evidence of th stratigraphic relationships of the later "Middle Sto Age," this assemblage from Layer 2 should be c an occurrence of the Howieson's Poort Industry. It was indicated earlier that it is possible to more about the Montagu Acheulean from Layers 3 5 than from the other assemblages present in the- because more work has been done on the Acheule than on other industries in Africa. It has been de strated that the aggregates from Layers 3 and 5 very similar to each other and different from the other known Acheulean material. The Montagu A lean is the result of workshop activities, the site visited and occupied repeatedly for the exploitatio raw material in the stream below the cave. The tence of such sites has been assumed from the E African evidence (Kleindienst, 1961:46). J. Wymer- recently excavated an Acheulean butchering site ae Hopefield in South Africa, and the indications the are that the tools had been manufactured elsewhe and brought to the site (pers. comm.). One of the original objectives of the project wS investigate what regional variations might be pre in Acheulean assemblages, but the problem was c plicated by the discovery of a previously undesc activity variant. In addition to the purely worksho elements present at Montagu, there is a broad of tool, the "other bifaces, " which seem different the East African tools. Although it is not possible say definitely whether these "other bifaces" are of a factory assemblage, I do not believe that the There is sufficient cause to refer to the Layers 3 5 aggregates as the Montagu Phase of the Acheul Industrial Complex because they represent a pre undescribed activity variation. It seems probable as other assemblages from Acheulean sites in the' Cape Folded Belt are described, more evidence O regional variation between East and South Africa be discovered. [ 80 ] o 4 'i4 I I; -r 1, APPENDIX I A DESCRIPTIVE CLASSIFICATION FOR THE EAST AFRICAN LATE ACHEULIAN ASSEMBLAGE I. ARTIFACTS OF STONE A. T OOLS 1. Shaped Tools e final form depends upon human workmanship. general groupings are broadly applicable to ized tools as well.) Large implements the general form of which is ntrolled by a primary linear rock mass, usually large flakes struck off unprepared cores. ) Implements characterized by cutting edges, of any size, having unifacial or bifacial trimming. Rarely less than 10 cm. (4 inches) or more than 30 cm. (12 inches) in greatest dimension (length). The average length:width:thickness ratio approximates 4:2:1 (width:length .50; thickness: width .50; thickness:length .25). a) Hand-axes (bifacial) and pointed flakes (unifacial). Characterized by a cutting edge around the entire circumference of the tool, or more rarely around the entire circumference with the exception of the butt. The emphasis in manufacture, if distinguishable, seems to have been upon the point and both edges. Usually bilaterally symmetrical, and more- or-less biconvex in major and minor sections (i.e., along the major and minor axes). Points range from exceedingly acuate to linguate. There is large variation in size, degree and quality of workmanship, and plan-view. Classi- fied on the basis of variation in plan-view, primarily according to the curvature of the edges, the length:width ratio, and the place- ment of the greatest width relative to the length of the tool. 1/ Lanceolate. Edges straight or slightly con- cave from greatest width to apex; greatest width no higher than 1/3 of the distance from the butt, relative to the length of the tool. Length:width ratio usually greater than 2:1 (maximum length is over twice the maximum width). 2/ Narrow Lanceolate. Edges convex-a much attenuated long ovate shape-with greatest width higher than 1/2 of the distance from the butt. Length:width ratio markedly greater than 2:1. 3/ Long Ovate. Edges convex, with greatest width higher than 1/3 of the distance from the butt. Length:width ratio approximately 2:1 or slightly greater. K' M. R. Kleindienst, "Components of the East African Fheulian Assemblage: An Analytic Approach," Actes Ive Conorbs Panafricain de Prehistoire et de ude, Section III, Pre- et Protohistoire, 1962, pp. 401. 4/ Asymmetrical Long Ovate. Not bilaterally symmetrical-one edge straight, the other convex. Length:width ratio 2:1 or slightly greater. 5/ Pointed Long Ovate. Long ovate in overall form, with slight concavity of the edges near the point, producing an attenuated point which causes the length:width ratio to ex- ceed 2:1 6/ Ovate. Edges markedly convex. Greatest width higher than 1/3 of the distance from the butt, and often about mid-section. Length:width ratio less than 2:1. 7/ Asymmetrical Ovate. Not bilaterally sym- metrical-one edge straight, the other con- vex. Length:width ratio less than 2:1. 8/ Pointed Ovate. Ovate in overall form. Slight concavity of the edges near the point may produce an attenuated point which slightly exceeds the 2:1 ratio. 9/ Broad Ovate. Length:width ratio approaches 1:1, but distinct point present. 10/ Discoidal. Length:width ratio 1:1, without a point. 11/ Cordiform. Definite basal shoulders, rounded and not higher than 1/4 of the distance from the butt. Edges straight or slightly convex from greatest width to apex. 12/ Elongate Lanceolate. Edges slightly concave near the attenuated point, without definite break in the plan-outline. Length:width ratio markedly greater than 2:1. 13/ Ovate-acuminate. Ovate with definite upper shoulders (break in the plan-outline), and an extended point-either linguate or acuate. Both edges convex from butt to shoulder, concave from shoulder to apex. 14/ Single-shouldered Ovate. Ovate with definite shoulder on one edge; opposite edge is con- vex. 15/ Single-shouldered Narrow Lanceolate. Narrow lanceolate with distinct shoulder on one edge. 16/ Ovate with twisted point. Extremely asym- metrical ovate, single-shouldered in some cases, with the point angled to one side of the long axis. 17/ Triangular. Approximates equilateral tri- angle in plan-view, with definite angular basal shoulders. Butt nearly perpendicular to the long axis. Length:width ratio less than 2:1 . 18/ Elongate triangular. Approximate isoceles triangle, with angular basal shoulders. Butt nearly perpendicular to the long axis. Length: width ratio 2:1 or greater. 81i 11 82nthropological Records 19/ Sub-triangular. Approaches triangular form, often that of obtuse triangle with butt at an oblique angle to the long axis. Length:width ratio less than 2:1. 20/ Double-pointed. Pointed butt; no definite shoulders. Convex edges, with greatest width about mid-section. 21/ Diamond. Pointed butt, with definite shoul- ders 1/2 to 1/3 of the distance from the butt. 22/ Truncated Diamond. Diamond with lower (butt) apex truncated. Definite shoulders about 1/3 of the distance from the butt. Length:width ratio usually less than 2:1. 23/ Limande. "Double-pointed" form in which both ends are equally rounded (linguate) or squared off. Greatest width near mid-sec- tion. Convex edges. Length:width ratio less than 2: 1. 24/ Elongate Limande. Length:width ratio 2:1 or greater. 25/ Asymmetrical Limande. Limande in which the long axis describes a curve. 26/ Untrimmed Butt ("Micoque-type"). Well- worked point with straight to concave edges. Butt minimally trimmed or untrimmed. Approaches triangular plan-view. (This type is not primarily classified according to the plan- view). 27/ Various. Occasional combinations of form, or extreme asymmetry. b) Cleavers (bifacial) and cleaver flakes (unifacial). Characterized by worked butt and/or edges, and a bit at one end more-or-less perpendicu- lar to the long axis of the tool. The bit is formed by two intersecting positive flake sur- faces, by intersecting positive and negative flake surfaces, or even by flake surface and cortex, giving a long, sharp, cutting edge which is rarely retouched. There is large variation in size, workmanship, minor cross section, and plan-view-in part due to the technique(s) used in producing the original flake, and in part to subsequent trimming. Cleavers made on nodules are rare. Classi- fied according to the trend of the bit, the position of the edges relative to the long axis, and by the shape of the butt. The following classification is a simplification of the large number of varieties theoretically possible in such classification. Arbitrary standards for determining guillo- tines, convergence, and divergence were estab- lished to insure uniformity in typing: A guillo- tine bit departs 150 or more from a line paral- lel to the minor axis when the implement is oriented along the major axis; a straight bit is thus more-or-less perpendicular to the long axis. An ultra-convergent cleaver is defined as one in which the length of the bit is 1/3 or less of the greatest width of the tool. The margin of convergence is defined as a line midway between the ultra-convergent margin and a line parallel to the long axis. The mar- gin of divergence is set an equal distance out- side this parallel line. (These standards were set after study of several hundred cleavers. If the class still occurs in significant propor- tions, having thus allowed a bias for parallel- edged cleavers, deliberate shaping for conver- gence or divergence can be assumed). U-shaped or rounded butts, V-shaped or pointed butts, and squared butts more-or-less perpendicular to the long axis occur in all categories. Bit varieties such as concave, convex, pointed, straight, or irregular could be noted as secondary features. 1/ Parallel-edges, straight or guillotine bit. Edges parallel to the long axis of the im ment from greatest width of the tool to ends of the bit. On the "high" side of a guillotine bit the edge may curve in sharl 2/ Asymmetrical parallel, straight or guillot bit. The long axis describes a curve rath than a straight line. Two varieties occur: one with a straight edge opposing a conca edge; the other with a convex edge oppos a concave edge. 3/ Convergent edges, straight or guillotine b Edges converge from the greatest width the tool toward the ends of the bit. 4/ Ultra-convergent edges, straight or guillo bit. The degree of trimming is similar that of hand-axes but a narrow transver81 bit is retained, whose length is 1/3 or les of the greatest width of the tool. Shoulder ultra-convergent varieties occur. 5/ Asymmetrical convergent, straight or guil tine bit. The long axis describes a curve. Edges converge. 6/ Shouldered-convergent, straight or guilloti bit. Definite break in the plan-outline (8h ders) on both edges. Edges usually convez from butt to shoulders, concave from shoa ders to ends of the bit. 7/ Divergent edges, straight or guillotine bi Edges diverge from the butt to the ends the bit, which is the widest point on the tool. Edges straight or convex. 8/ Splayed edges, straight or guillotine bit. The edges diverge from the butt to the e! of the bit, with a marked concavity near bit. 9/ Side Cleaver. Bit on one side of the im ment, more-or-less parallel to the long 10/ Burin-blow Bit. A form of cleaver not fined according to plan-outline. At least face of the bit is formed by a flake scart in which the flake was driven off the bit end in the manner of a burin. The bit is at an angle to the horizontal plane, and relatively narrow. A. ade on chunks or nodules. c) Knives. Characterized by having one side, or part one side, blunted or "backed" while the op ing side, or opposing side and one end, has sharp cutting edge. The backing may be an original surface-cortex or a fracture plane in the raw material; it may be the striking platform of the flake, plain or facetted; or may be a deliberately trimmed surface. T cutting edge may be untrimmed, formed by intersecting flake surfaces, unifacially tr' or bifacially trimmed. If trimmed, it is and sharpened. The backed edge is marke thicker in minor section than the opposing cutting edge. 1/ Pointed. One end is pointed; the cutting? edge extends up to, or around the point. Fine examples grade into a single-shoul hand-axe, and this tool might be conside a subtype of hand-axe not classified acc ing to plan-view. I 82 I 2/ End-and-Side. The cutting edge extends along one side and around one end. This might be considered a subtype of cleaver, not classified according to plan-view. 3/ Disc. The form varies between a square with rounded corners and a discoid. These tend to have a wedge-shaped cross section in one direction and a plano-convex or bi- convex section in the other direction. 4/ Various. Implements which have little con- sistency in plan-view, but which have in common a cutting edge on one end, or on one end extending down the side(s). These may be fashioned on pebbles, slabs, chunks, etc., and like other knives have the feature of a blunt back opposing a cutting edge. d) Elongate bifacial tool with cutting ends. Minimally worked, large, bifacial implements with quadrilateral minor sections at mid-sec- tion, and well-worked cutting edges on one or both ends. te) Round-bitted Biface (bifacial) and round-bitted flake (unifacial). Well trimmed cutting edge at one end, rounded in plan-view, with slightly trimmed or untrimmed butt. More-or-less limande shaped in total plan- view. Implements characterized by scraping edges, over 10 cm. (4 inches) in greatest dimension. "Flake tools" in that the implements retain their primary artifact form-usually that of a flake-and were used in that form, or had only the edges modified in such a manner as not to obliterate the original form. Some are made on slabs or chunks. Trim- ming produces a unifacial or bifacial scraping edge; if more than one edge is worked, the trim- ming may be on opposing faces. Edge trimming can be described as "shallow," "blunt," or "steep," in order to express the angle of flaking and the resulting thickness of the edge relative to the size of the tool. The classification is largely based on edge form, and not on the overall form of the tool. a) Side Scrapers The working edge(s) is more-or-less parallel to the long axis. One or both sides are trimmed. Edge forms are similar to small scrapers (cf. below), but the variety found is more limited. b) End Scrapers The working edge(s) is more-or-less perpendi- cular to the long axis. One or both ends are trimmed. Edge forms are similar to small scrapers (cf. below), but the variety found is more limited. c) Combined End and Side Scrapers. The working edge occurs on one or both sides, and one or both ends; it may not be a contin- uous edge (cf. below). d) Various Forms. Forms classified by overall shape (cf. below). 83 3) Other tool types with uncategorized edges, usually exceeding 10 cm. (4 inches) in greatest dimension. Workmanship varies, but is often minimal. Classi- fied on overall form. a) Chisels. Minimally-trimmed, bifacial tools with a sharp bit at one end which is sometimes hollow or gouge-like. The bit may be trimmed, or formed by intersecting flake scars. Both major and minor sections are thick, relative to the size of the tool: the major section may be wedge- shaped. 1/ Straight-bitted. The bit is in the horizontal plane. 2/ Twisted-bitted. The bit is at an angle to the horizontal plane. b) Pushplanes. Implements which tend to be flat-based and highbacked, with an emphasis upon one end. These may bear a sharp, unworked edge on one end, and scraping edges along the sides. Unifacial or bifacial trimming. Usually made on thick flakes from which the bulb has been removed. 1/ Shallow-nosed. The working end is thin, has shallow trimming or lacks trimming. The butt is thick in minor section, and usually trimmed. The nose has a "dished- out" appearance in major section. 2/ Bitted. Thick minor cross section over the whole length of the tool. Steeply trimmed or untrimmed sharp bit on one end, set at approximately a 450 angle to the base. The butt is unifacially or bifacially worked. c) Discoid. Bifacially worked, discoidal implement with biconvex cross section. Secondary trimming on the edge and, often, utilization, distinguish these from cores. d) End-notched Biface. Variously shaped in plan-view, but having trimmed and/or utilized concavity at one end. e) Various bifacial tools. Bifacially worked implements which can only be described according to plan-view, as "rec- tangular" or "elongate." Characterized by a minimum of trimming, asymmetry, and thick cross sections. b. "Heavy Duty Tools." A group of implements which have in common a minimum of trimming and less standardization in form than do the implements in preceding categories, but which are a constant component of an East Afri- can Late Acheulean assemblage. Large tools, usually exceeding 10 cm. (4 inches) in greatest dimension, but occasionally as small as 5 cm. (2 inches), which are made on compact rock masses-thick flakes, nodules, or chunks. Classified according to overall form. 1) Picks. Sturdy tools with a minimum of overall trimming, but with emphasis upon a point as such rather than upon edge retouch. Individual pieces tend to Keller: Montagu Cave in Prehistory Anthropological Records be distinctive. Width:length ratio in the cutting tool range (c. 1:2 or .50); thickness:width exceeds .80; thickness:length exceeds .40. a) Trihedral. Point triangular in minor section, with dorsal median ridge extending from point toward butt for most of the length of the tool. The ridge may not occur on the butt, which is slightly trimmed or untrimmed. The base tends to be flat and untrimmed, but the point may be worked on all three faces. b) Bifacial. Bifacially trimmed, with a tendency to a lenti- cular minor cross section, especially at the point. The butt may be heavy and little trimmed. Tend to be either long ovate or elongate tri- angular in plan-view. c) Roughly trimmed butt. Butt slightly trimmed, and heavy, with a greater degree of attention paid to the point. The cross section at the butt tends to be quadrilateral, at the point, lenticular. Unifacial or bifacial trimming. d) Untrimmed butt. Unifacially or bifacially trimmed point, often well-worked, with a heavy untrimmed butt. If made on cobbles, cortex is retained on the butt, and in unifacial examples, on the ventral surface. e) Flat-based, high-backed. Quadrilateral in minor section over almost total length of tool. Trimming on the dorsal face may be from both the base and the top; the ventral face may or may not be worked. f) Flat-based, beaked. Both sides trimmed, upward from the base and downward from a median dorsal ridge. Point is worked to a greater degree and is trimmed back into the base, so that it is perpendicular to the plane of the base. g) Point on cobble. Elongate, spiky point trimmed on a cobble. Cortex is retained on the butt. Tend to be about 15 cm. (6 inches) in length. h) Block. Point fashioned on a chunk or slab of rock by a minimum of trimming. i) Spindle. Picks, generally of small size-under 15 cm. (6 inches) in length-with minimal gross bi- facial trimming; narrow relative to their length, double-pointed, with a characteristic spindle "twist," resulting in a more-or-less diamond- shaped minor section. May be worked from the median ridge, as well as radially. j) Core. Minimal radial trimming on one or both faces. Convex ventral surface; dorsal median ridge. k) Various. A number of picks must be individually described. 2) Core Scrapers. High-backed tools characterized by steep trimming, from a flat surface along some segment of the circumference. The steeply trimmed edge is of scraper type, but tends to be "undercut" by step- flaking (cf. "trimming stones" in van Riet Lowe 1952:42). Usually large tools, but may range down to 5 cm. (2 inches) in greatest dimension. a) Flat-based, steep. Fits the class specifications. Examples with the greatest degree of trimming tend to have a high "domed" appearance in section. The working edge may be convex, straight, or con- cave. b) Flat-based, blunt. Trimming is less steep, with less tendency undercut the edge. to c) Bevelled-base, steep. Similar to a), but the base is formed by two intersecting planes, with the edge trimmed against one. d) Bevelled-base, blunt. Trimming is less steep. e) Keeled. High-backed, flat-based, with a well-worked dorsal ridge on top. f) Double-edged. A cross-section through the working edges forms a segment of a circle. The arc is trimmed from two flat surfaces, forming two scraping edges, each against a flat base. g) Core. Appears to be primarily some variety of core,' on which a scraping edge has been trimmed. 3) Trimmed Pebble or Chunk. Minimal, unifacial, usually steep trimming against a flat surface at one end or side of a piece, whic forms a steep scraping edge. 4) Choppers. Implements, usually with bifacial trimming, chara terized by a chopping edge. These are usually made on compact rock-masses-chunks or nodulei or cobbles-and only occasionally on thick flakes. Classified according to the form of the raw mate or on the overall form of the tool. These may be secondarily classified according to the placement, or extent, of the working edge relative to the lo axis. a) Pebble. Chopping edge worked on the side or end of a waterworn piece. The major portion of the implement surface retains the original cortex. iI j I i j .I .11 11 kl 84 Keller: Montagu Cave in Prehistory b) Chunk. [ Similar, but worked on a chunk or nodule. c) Flake. * Thick flake, with chopping edge. d) Core. Piece which has complete bifacial radial trimming, with a chopping edge on much of the circumference. These may have been cores, t but chopper status is indicated by a markedly I sinuous edge, and in most cases by utilization as well. ,e) Side. Crescent-shaped in plan-view, with a sinuous chopping edge around the arc, and an un- sharpened back along the chord of the arc. In some cases the back was deliberately trimmed and blunted. f) Discoidal. More-or-less round in plan-view and biconvex in section, with a chopping edge around much of the circumference. Secondary edge work- manship indicates that these are primarily implements and not cores. Stone Balls. Artifacts which fit a general category of more-or- ess deliberately shaped, more-or-less spheroidal forms. These have been variously called "stone balls" (Clark 1955), "bolas stones" (Leakey 1953: 58-59), or "polyhedral stones" (van Riet Lowe 1952:171). Classified on the bases of type and gdegree of workmanship. a) Missile. Roughly spherical artifacts which have been largely shaped by nature, but which show some signs of having been subsequently shaped by man. These may approach the polyhedral in form by having been facetted over a portion of the surface. b) Polyhedral. Roughly spherical artifacts which have been shaped by man over most or all of their sur- ^ face area. These are facetted by intersecting negative flake scars. c) Bolas. Artifacts which are nearly spheres, and which have been pecked or battered to a nearly smooth surface over most or all of their surface area. 69l Implements. plements generally under 10 cm. (4 inches) in atest dimension, which fall into various categories which are characterized by recurrent techniques manufacture, as well as by size and form. Small implements characterized by scraping edges, fashioned on flakes, chunks, or other small pieces of raw material, which retain the original form except on the working edge(s). Edge trimming may be shallow, blunt, or steep. The classification is based primarily on the form of the modified edges, rather than on overall form; in this respect it is comparable to the classification used by Clark (1950:79 and 81) for the Northern Rhodesian "Hope- Fountain." Scrapers may be unifacial or bifacial, and if more than one edge is worked, may be trimmed on opposing faces-the latter being a secondary characteristic. a) Side Scrapers. Single or double; working edges are more-or- less parallel to the long axis of the tool. In double forms, any combination of edge form may theoretically occur. 1/ Convex. Worked edge has marked convexity. 2/ Straight. Worked edge is more-or-less straight. 3/ Concave. Worked edge has marked concavity. 4/ Notched. Worked edge has two or more small concavities or notches, which are few in number relative to the size of the tool. 5/ Denticulate. Worked edge has many small notches, which are numerous relative to the size of the tool. 6/ Rounded. Worked edge is continuous around approximately 1/2 of the circumference of the tool, without an angular break in the plan-outline. 7/ Pointed. Worked edge extends around a point. 8/ Irregular. Worked edge is irregular in out- line. b) End Scrapers. Single or double; working edges are more-or- less perpendicular to the long axis of the tool. In double forms, any combination of edge form may theoretically occur. 1/ Convex. Worked edge has marked convexity. 2/ Straight. Worked edge is more-or-less straight. 3/ Concave. Worked edge has marked concavity. 4/ Notched. Worked edge has two or more small concavities or notches, which are few in number relative to the size of the tool. 5/ Denticulate. Worked edge has many small notches, which are numerous relative to the size of the tool. 6/ Rounded. Worked edge is continuous along part of both sides and around the end with no angular break in the plan-outline. 7/ Pointed. The worked edge extends around a point. 8/ Irregular. Worked edge is irregular in out- line. c) Combined End and Side Scrapers. A trimmed edge on one or both sides in com- bination with a trimmed edge on one or both ends. Any of the edge forms may theoretically occur. d) Various scraper forms, classified according to overall shape. 1/ Angled. Two worked edges meet at an acute angle, with a sharp break in the plan- outline. 2/ Nosed and Winged. A form with a rounded, protruding nose produced by concavities or I f, 85 Anthropological Records broad notches on either side. Base may or may not be trimmed. 3/ Round (unifacial) or discoidal (bifacial). Worked edge extends nearly or completely around the circumference of the tool. More-or-less round in plan-view. 4/ Steep. Steeply trimmed, high-backed scrapers-analogous to large core scrapers, but lacking an undercut edge. Worked edge may occupy a large portion of the circum- ference. 5/ Thumbnail. A very small scraper, under 3 cm. (1.2 inches) in greatest dimension, with a rounded working edge. Highbacked relative to the size of the tool, and often bearing a minature graver point as well. 6/ Pointed. Two scraping edges meeting in a point. 7/ Triangular. Triangular in plan-view, with three worked edges. 8/ Various. Forms not otherwise classified. 2) Small implements with various types of edges, fashioned on flakes, chunks, and other forms of raw material. Most bear points or bits on a miniature scale. These are classified according to overall form. a) Protoburin. One or two flakes driven off an end, in the horizontal plane, to form a sharp edge (Clark 1959: in press) considers these a type of core, but in many examples the flake(s) removed is very small, and the resulting edge is utilized. b) Burin. Tools which have a typical burin blow-more- or-less in the vertical plane, producing a sharp, transverse edge. The tool may be trimmed and/or utilized on other edges as well. c) Point (bifacial) or pointed flake (unifacial). Made on flakes generally, thin in cross sec- tion relative to the size of the tool if com- pared with a small hand-axe. Trimmed from the edges. Comparable to "points" in later stone industries. d) Chisel-ended. Tools which have in common bits formed by a single flake scar running longitudinally along the tool. In some cases this may be accidental; in others the flake has been driven off the bit end. A high-backed, flat-based variety is com- mon. The tool may be trimmed and/or utilized on other edges. Usually unifacially trimmed. e) Elongate Bifacial Tool. Bifacially worked small tool, with rather indif- ferent point, radial trimming, and irregular plan-outline due to gross minimal trimming. These lack cutting edges. Signs of utilization, if present, occur on the pointed end. (Chisel ends also occur.) f) Bifacial Tool with Bevelled End. Similar to the above, but with a marked bevel on one face at the pointed (or chisel) end. g) Discoid. Small, round, bifacially worked implemen h) Borer. Small, generally under 3 cm. (1.2 inches) greatest dimension, with a sharp point. In some, the butt has been retouched on the opposing surface to the point trimming. i) Pointed Tool. Various forms on which a point has been pared. A flat-based, high-backed variety il common. j) Various. Forms not otherwise classified. 2. Modified Tools Trimmed, and often utilized, but fit no shaped-tool form category. May be separated into large (excem 10 cm. [4 inches] in greatest dimension) and small (under 10 cm. in greatest dimension) size ranges. Classified according to the type of trimming and tU form of the raw material used, such as flakes, co slabs, etc. a. Unifacially trimmed. b. Bifacially trimmed. c. Variously trimmed. For example, flakes trimmed only on the flake4 surface. 3. Utilized Tools Identified only by signs of use. May be categorize according to edge types and grouped with the app priate shaped tool category. a. Hammerstone. Various stones, usually natural chunks or cobb which show signs of battering and bruising on stricted areas of the surface, and which are th fore interpreted as due to human agency. b. Anvil. Various large chunks, which may be naturally shaped or roughly trimmed, on which the arete have been heavily battered and bruised. c. Utilized waste products which show signs of u on one or more edges. May be classified by a range, by type of artifact, and/or by distinctiv kinds of utilization such as notching, nibbling, jagged edges, etc. 1) Utilized Cores. 2) Utilized Flakes. 3) Utilized Chips and Chunks. 4) Utilized Natural Pieces. B. WASTE PRODUCTTS Presumed to be waste products produced in the i cess of manufacturing tools. Show no observable of trimming, modification, or utilization (although 86 I in purposes the utilized waste products may be ned in the category). 1. Cores cts assumed to be the waste products left after orkman had struck off the desired flakes. In part system devised by Paterson and Fagg (1940:16-19) Paterson (1945:8-12) has been followed. eble. es were struck off a pebble, using prior nega- ie flake scars for subsequent flakes. A major por- n of cortex is retained, compared to the area of lar, but a chunk or nodule rather than a water- orn piece of raw material was used. orless. piece which has had flakes removed from most r all of the surface area. Flakes have been struck f in all directions, using the negative flake scars platforms for subsequent flakes. ramidal. core which approximates a pyramid in side-view. kes, some of which were flake-blades, were ck off from one direction, around the basal ircumference of the core. The base may be a tural surface, or formed by one or more nega- ye flake scars. [iconical. iconical in side-view. May be conceived as bifacial, nd more-or-less round in plan-view. Flakes were truck off the periphery, forming a more-or-less nuous edge around the core. Flake scars converge ward central foci on both faces. The flakes may struck alternately, or one face completely worked fore the other. Trimming is described as radial. evelled-base Biconical. One face is formed by two intersecting planes, with radial trimming on the obverse. Discoidal. Lenticular, or biconvex, in side-view. More-or-less round in plan-view. Flakes struck off as in a bi- conical core, with radial trimming. Struck. Representatives of the "Levallois" or "facetted- Platform" technique, in which the core was prepared in order to remove one or several comparatively large, well-shaped flakes. Platform preparation minimal, if present at all. These cores are recog- nizable as prepared cores only because they have been struck; unstruck examples, although so intended by the maker, would probably not be identifiable as different from biconical/ discoidal cores. Radial r trimming. i Double- struck. ' Two flakes struck off parallel to each other. Angle. kFound at Kalambo Falls where slabs of raw material ,-were used. Flake-blades or short quadrilateral flakes were removed from two faces at approximately right angles to each other, by using prior negative flake scars as a platform. 87 k. Large Cores. The cores from which the flakes were struck for the making of large implements are present at Isimila, but not at Olorgesailie nor in the collection from Kalambo Falls in the Rhodes-Livingstone Mu- seum. These are known to reach 60 cm. (24 inches) in length at Isimila. From them large side-struck flakes were obtained, with a minimum of prior pre- paration, if any. The technique has been described as "Tachenghit" by Goodwin (1933:111-116), and has been used elsewhere to describe cores in an East African assemblage (van Riet Lowe 1952:38). How- ever, it is now known that a rather specialized form of prepared core was used in the Tabelbalat-Tachen- ghit region of the Sahara (Tixier 1957:919), and the term "Tachenghit" might better be reserved for that form. 2. Debris Assumed to be the waste products produced either in striking flakes off cores, breaking up raw material, or in the process of manufacturing shaped or modified tools. a. Flakes. Struck off cores, or in the process of shaping tools. 1) Large Flakes. Over 10 cm. (4 inches) in greatest dimension generally. Presumably produced in the process of trimming cores, and in the size range from which large tools might have been made. 2) Small Flakes. Under 10 cm. in size. Either struck off cores, or from shaping tools. In the size range from which small implements could have been made. Tool - trimming flakes, tool re sharpening flakes, and core-trimming flakes can be partially differ- entiated. 3) Levallois Flakes. Flakes, which from the number of radial dorsal scars, type of platform-generally right angle, sometimes facetted-and general shape probably were struck off prepared cores. 4) Flake-blades. Those flakes in which the length:width ratio is 2:1 or greater. Often show parallel dorsal scars. 5) Short Quadrilateral Flakes. Those flakes which have parallel dorsal scars and a rectangular or quadrilateral shape, but in which the length:width ratio is less than 2:1. 6) Cleaver-edged Flakes. Flakes in the large size range which have a cleaver-bit edge, but which have not been secon- darily trimmed. Of the type on which cleavers or other large implements could have been made. Presumably, the shape is due to the type of core used. 7) Pointed Flakes. Similar flakes, but pointed, either fortuitously or because of the type of core used. b. Chips and Chunks. Various debris resulting from the trimming of cores and tools. May be classified according to size. Keller: Montagu Cave in Prehistory Anthropological Records II. WOOD, BONE, OTHER ARTIFACTS Artifacts of materials other than stone: wood, bone, other. These are unfortunately not yet a classificatory problem, although wooden implements occur at Kalambo Falls (Clark 1954:55-56). III. EXOTIC MATERIALS Materials not of local origin-"exotic"-presumed to have been introduced by human agency; imported stoi fauna, vegetal remains, etc. Depends upon the type o site and its locality. May also include fauna and vegi tal remains which were not imported by early man, but which are assumed to have formed part of his di DISCUSSION The foregoing descriptive classification constitutes the East African Late Acheulian assemblage on the basis of artifactual material excavated from occupa- tion sites in archaeological context. Almost all of the various tool types (underlined), and the majority of subtypes, occur at all of the occupation sites. Although studies are not yet at a stage where it is possible to conclude definitely which subtypes have most signifi- cance for temporal or spatial or cultural interpreta- tions in the East African region, some conclusions as to the usefulness of consistent quantitative analy- sis can be drawn from the preliminary work on assemblages from Isimila and the other sites. Types and classes of artifacts can be correlated wi differences in the type of raw material used in the occul tion areas where a variety of raw materials was availa -primarily at Isimila and Kalambo Falls. At Isimila large cutting-edge implements, large scraping-edge in ments, and other large tools primarily manufactured o4 large flakes are made of several varieties of fine-graxi silicious cataclasites, but also in quartz, quartzite, an granite. Heavy-duty tools, also hammerstones and anvi tend to be made in such massive, durable materials as quartz, quartzite, and granite. Small implements-and most utilized flakes-tend to be made of the most homo geneous material which was available, i.e., quartz. I I 88 APPENDIX II A PROVISIONAL INTERPRETATION OF THE SEDIMENTARY SEQUENCE FROM MONTAGU CAVE (CAPE PROVINCE), SOUTH AFRICA By Karl W. Butzer (The University of Chicago) ha ction. ontagu Cave is located in the side of a small cut through the Table Mountain Sandstone, near town of Montagu. Facing ENE the cave seems to veloped near a structural contact in low-grade imorphics, where shear and tension have favored ter friability and removal of cementing minerals lution. Sediments with archeological materials found in the larger and outer of 2 chambers, with Duth width of c. 11 m, a maximum length of c. 17 lada vault elevation of as much as 13 m. The site .-first excavated in 1919, with a second excavation rtaken 1964-65 by Charles M. Keller (University flinois) (Keller, 1966). Since the site contains two kor archeological levels and two quasi-sterile strata, predating 50,000 B.P., Montagu Cave potentially p8 the little-known transition of Middle to Upper Btocene in South Africa. For this reason the writer lied seven samples (see Table 1), collected by Ir in 1965, attempting a broader interpretation of rsediment sequence. Ehn addition to macroscopic examination of structure consolidation, color (dry) was determined by the Soil Color Charts, texture by hydrometer and -sieve analyses, carbonate content uy the Chittick 8od, and pH (electrometrically) in distilled water e Butzer and Hansen, 1968, appendix A). Textural Woes are given according to both the Wentworth and fD.A. classifications. Subsequent microscopic scan- of sand-sized residues included a semi-quantita- w estimate of quartz-grain frosting and rounding. p-quartz aggregates in the sand fraction were found be soluble in 20 percent hydroxide. These analyses re carried out in the Paleo-Ecology Laboratory tropology Department) of the University of Chicago, 1i the assistance of Daniel C. Bowman. 'Since the samples themselves were too limited in ope to allow an adequate description of the sedimen- q strata, a good series of detailed color slides was Kdied, as explained and amplified by Dr. Keller, and supplemented by the descriptions earlier provided by Keller (1966). Although the overall results are not con- clusive, they do show that the strata convey potential paleo-environmental data. The very fact that Montagu is situated within the Cape Folded Ranges, at the edge of the semi-desert Karroo, places it in a climatically sensitive transition.' Bedrock. The bedrock exposed in Montagu Cave is a white, coarse-grained sandstone that has been moderately metamorphosed. Although it does not deserve the desig- nation "quartzite," the quartz grains have been suffi- ciently deformed and partially fused to merit the label of a "metamorphosed sandstone." Grain-size is 50 percent in the 200-500 micron size-range, and the individual grains are quite angular. All the sand-sized quartz grains of the cave sediments fall within the shape and type of this local sandstone, and differences in degree of microscopic rounding are almost negligible. The primary quartz grains show a degree of atypical frosting, which is somewhat more prominent in the cave sediments, particularly in the lower beds. This micro-pitting is almost certainly due to chemical or biochemical agencies operating during and after sedimentation in the cave (see Butzer and Gladfelter, 1968), and offers no paleo-environmental clues. There is no cementing substance in the sand- stone, so that the "rock" is quite friable, breaking down during even gentle sample pretreatment. However, the differences of sand size components among the cave sediments depend primarily on the degree of fractiona- tion of these quartz grains-which can be attained by modest weathering or by prolonged stirring at high speed in a blender. Unfortunately, this property of weak cohesion precluded any systematic study of coarse, 'Montagu itself is at 223 m elevation, and has a mean annual precipitation of only 312 mm, concentrated in the transitional seasons. The Koppen climatic classi- fication is BSk (cool-steppe climate), that of Thornth- waite DB2d (semiarid mesothermal, with little water surplus). 89 I Anthropological Records detrital components in the sediments. The bedrock is weathered, in that there is a small clay component of almost 5 percent (compared with 5-12 percent in the cave sediments) and that the rock matrix (presumably silica and sesquioxides) has been at least partly re- moved. Finally, the bedrock lacks organic components other than intrusive, carbonized roots. Simple breakdown of the sandstone should provide approximately 90 percent sand-sized quartz, with a minor component of silt and clay. Consequently the quartz sand content of the successive cave strata pro- vides a useful index to the extraneous "complications." On this basis, the limited number of samples suggests that materials introduced by man, animals, and inor- ganic agencies account for as little as 10 percent and as much as 60 percent of the different layers. A last point of interest is the black, finely laminated precipitate found on parts of the cave wall and tenta- tively attributed to hyrax urine (Keller, personal com- munication). This precipitate has not been identified but contains at least some ferric components. Small grains of identical material (in the 0.6 to 6.0 mm grade) were found in 3 samples (1713, 1714, 1715), probably indicating that rockfalls had introduced it to accumulating floor sediments. Levels 7 and 6. The basal accumulations of the cave have been described by Keller (1966) as (a) unsorted sand, with rare artificial materials-level 6; and (b) fine-grained sandy clay, with decomposing fragments of bedrock, presumably a weathering residual-level 7. The basal horizon is about 5 cm thick, the succeeding accumula- tion varies from 15 to 60 cm. No samples were avail- able for study. Level 5. The thickness of the Acheulian level 5 increases rapidly from about 30 cm near the front of the cave to some 150 cm farther inside. The lower boundary is smooth and abrupt, the upper wavy and abrupt; both clearly mark discontinuities in sedimentation. The internal bedding is very pronounced, with undulating, alternating bands of 5 to 12 cm dark, organic sands and white, clean sands. There also are occasional, short lenticles of clean sand. Rock debris, angular and ranging from 1 to 25 cm in diameter, is common and dispersed throughout the bed. Two samples were analyzed. The "organic" facies represents a weakly structured reddish gray, sandy loam, with abundant diffuse humus and a trace of macroscopic organic matter. The reddish color comes from red silts introduced to the cave and now adher- ing in part to the autochthonous quartz sands. There are no extraneous quartz or silcrete grains, so th the source of the silt is obscure-man, fissure-wa or eolian dust. Tool debitage (gray quartzite and i quartzitic "chert" with biotite veins) in the 2-5 grade accounts for 2 percent of the sample; there 4 traces of such biotite in samples 1710 and 1712. The "inorganic" facies is a white loam with nex no organic matter, and slightly greater compaction (coarse, sub-angular blocky structure). Tool d6 is completely absent, supporting the inference of occupation. At least 4.0 percent of the material ce sists of sesquioxide-stained silcrete aggregates ia 25-500 micron size-grade. These aggregates are extraneous, and are most probably of eolian origi% This explanation is compatible with the well-strati or laminated, wedgelike character of the lense in such aggregates abound. It would also be a reason process during times of cave-abandonment. None even allowing that some of the finer quartz graina blown in after short transport distance, eolian co nents constitute only a small part of the total se Weathering or fissure-wash must have continued supply silts and some clay. Level 4. The so-called sterile horizon, level 4, includes lower, occupation zone-with many diagnostic pro of Level 5-and a massive, upper horizon, rather ferent in character.2 The lower level has a maximum thickness of 20 towards the cave interior, and consists of well-St fied lenses of undulating aspect, interspersed with rockfall debris. The material is a dark brown, lo sand, weakly structured, with moderate concentra of diffuse humus and a trace of macroscopic, org matter. About 4.5 percent of the sediment consists 2-5 mm tool d6bitage, although artifacts as such very rare. Silcrete grains are absent, although e eous red silt is indicated. The upper level, separated by a clear, wavy co ranges from 75 to 90 cm in thickness. It is rather homogeneous and the stratification is horizontal a undisturbed, with limited amounts of relatively s roof debris. No sample was available for study, b is expected that the material lacks organic inclus or appreciable weathering products. Eolian compo are a probability. Level 3. The second major Acheulian deposit, Level 3, the lowest sand content of the entire cave sequen instead, there is unusually abundant silt (50-60 pe 2Dr. Butzer has included the band that includes s VIII in layer 4 rather than in layer 5 as I have done I 90 Keller: Montagu Cave in Prehistory b basal surface is smooth and abrupt, the upper intact wavy and abrupt, in part coinciding with a Njor rock collapse and suggesting disconformities. Ftckness of the stratum varies from 30 to 60 cm, * internal bedding is undulating, as is also charac- stic of Level 5. There are basal and intermediate bOrganic" horizons. The normal, "organic" facies is a weakly structured, y dark brown, silt loam, with abundant diffuse humus, 1%e macroscopic organic matter, and traces of sil- ete aggregates. Micro-debitage (2-5 mm) accounts ?about 1 percent of the sample. Probably a half of extraneous material (including soil sediment) can safely assigned to human activities, and at least a pd part is fine mineral ash. The "inorganic" bands consist of light gray silt L with greater compaction (very coarse, subangular ky structure), and with traces of both diffuse and lcroscopic humus. Some micro-debitage and occasional Facts further indicate that occupation was reduced intensit but not entirely absent. Silcrete aggregates West eolian activity, but the general prominence of anu soil sediment is nonetheless enigmatic. sel 2. The Middle Stone Age stratum, Level 2, has Groningen liocarbon dates (from top to bottom) of 23,200, W6OO, greater than 50,800, and 45,900 B.P. (GrN 4726- 18), inferring a time-depth of almost 30,000 years. nature of the deposits does not contradict this: the te1 has a thickness varying from 25 to 160 cm and, spite good stratification in detail, is disrupted by a Ober of major rockfalls. Given the careful pretreat- pt of GrN radiocarbon samples in general, and the kelihood of pure charcoal's ever being too old, it hd seem imperative to explore all other possibilities lore concluding that the C14 dates are substantially trrect. The deposit, it goes without saying, is suf- Oently complex to allow for a long period of succes- re accumulations. hThe basic sediment is a weakly structured, very ii gray, sandy loam, with abundant diffuse and macro- epic humus. Silcrete aggregates are moderately promi- lt. Micro-d6bitage (1-5 mm) accounts for 2.2 percent. uibly the abundance of large and small sandstone fritus contributes to the high overall sand content. wived soil sediments are present. Inlike the lower occupation strata, Level 2 includes ~n distinct "surfaces" (i.e., horizontal concentrations irrtfacts) with hearths (Keller, 1966); similarly, char- t1 in different size grades is fairly abundant through- out. However, the undulating nature of the strata, build- ing up from the cave entrance, indicates substantially the same mode of cultural deposition as for 5, 4 (lower), and 3. The only difference appears to be partial decom- position of the related organic material (e.g., matting, bedding, etc.) from the Acheulian levels. Level 1. The thin surface level, varying from 0 to 30 cm in thickness, had been partially removed by early guano hunters. It consisted of a "brown sand" littered with surface rubble, and containing artifacts and traces of hearths with LSA. A C14 date of 7100 B.P. (GrN-4725) was obtained. No sample was available for study. General nature of the cave sediments. Perhaps the most striking aspects of the cave se- quence can be numerated as follows: (1) The extremely acid pH (3.0-3.5) exceeds any of the surface "Greyish- brown to dark brown soils" developed on Table Moun- tain Sandstone (see v. d. Merwe, 1963). The high acidity must be attributed in good part to prolonged and re- peated cave occupance. (2) The high organic content -whether macroscopically visible or decomposed to amorphic humus-as well as the weak structure and rather dusty nature of the bulk of the sediments are rather apparent. This reflects on the lack of cohesive clays, the basic sandy grade of the autochthonous sedi- ments, and the predominant silt size of all extraneous matter. (3) The strongly undulating and conspicuous nature of the stratification was noted by the writer in several LSA accumulations near Plettenberg Bay and appears to be peculiar to organic middens of the type associated with many LSA sites (see Deacon and Deacon, 1963; Wells, 1965; Deacon, 1969). In part, this may re- flect differential compaction as organic components rot out, in part it bears upon the exceptional stratification caused by grass matting, etc. (4) The small yet con- spicuous component of sesquioxide-stained silcrete grains in the 25-500 micron grade is noteworthy. It certainly appears to be eolian, yet it would be difficult to prove this point by rigorous criteria. (5) The variable silt component of the strata, associated with extraneous soil sediments of some sort, remains difficult to explain. The relative proportions introduced by man, by fissure- wash, and by eolian activity cannot be determined, and their identification would be crucial to a firm paleo- environmental interpretation. Montagu Cave will require field examination and further sample collection by a Pleisto- cene geomorphologist, and pollen analyses should be re- warding, if not essential to a more detailed interpretation. 9 1 I Keller: Montagu Cave in Prehistory TABLE 1 Sediment Data from Montagu Cave Sample Color Wentworth U. S. D. A. Silcrete %Quar (dry) Textural Textural CaCO3 pH Aggregate over Number Class Class (%) (%) micro 1 (No sample) 2 1715 10 YR 3/1 Silty Coarse Sand Sandy loam 0.0 3.4 1.0 67 (v. dark gray) 3 (Organic) 1714 10 YR 2/2 Med.- sandy silt Silt loam 0.0 3.0 0.5 3B6 (v. dark brown) 3 (Inorganic) 1713 10 YR 7.5/2 Med.-sandy silt Silt loam 0.0 3.5 2.5 30, (lt. gray) 4 (Upper) (no sample) 4 (Lower) 1712 10 YR 4/2 Silty coarse sand Loamy sand 0.0 3.2 0.0 71 (dk. brown) 5 (Organic) 1710 5 YR 5/2 Silty coarse sand Sandy loam 0.0 3.3 0.0 (red. gray) 5 (Inorganic) 1711 7.5 YR 10/2 Silty coarse sand Loam 0.0 3.3 4.0 6 (white) 6 (No sample) 7 (No sample) Bedrock 1716 10 YR 8/1 Coarse sand Sand 0.0 4.8 0.0 | (white)llll l REFERENCES Butzer, K. W., 1973, "Spring Sediments from the Acheulian Site of Amanzi (Uitenhage District, South Africa)," Quaternaria, 16. In press. Butzer, K. W. and Gladfelter, B. G., "Quartz-grain Micro-morphology," in Butzer and Hansen, 1968, pp. 473-481. Butzer, K. W. and Hansen, C. L., 1968, Desert and River in Nubia. Madison: Univ. of Wisconsin Press, 562 pp. Deacon, H. J., 1969, "Plant Remains from Melkhout- boom Cave, South Africa," preprint of paper read at Pan-African Congress of Prehistory, Daker (1 967), 7 pp. Deacon, H. J. and Janette Deacon, 1963, "Scotl A Late Stone Age Site in the Gamtoos Vail. Cape Prov. Museum, 3, pp. 96-121. Keller, C. M. n.d., "Archaeology of Montagu C Unpublished Ph.D. Dissertation, Universityi fornia, Berkeley, 1966. Merwe, C. R. v. d., 1963, Soil Groups and Sub of South Africa. Pretoria: Government Print of Agric. Techn. Serv., Chem. Series No. 165) Wells, M. J., 1965, "An Analysis of Plant Re" from Scott's Cave in the Gamtoos Valley," Arch. Bull. 20, No. 78, pp. 79-84. 92 Keller: Montagu Cave in Prehistory Note. [C.M.K.] The following table indicates the results of spectographic analyses of samples of cave sediment. The provenience of the samples is: Sample A Sample B Sample C Sample D Sample E Bedrock Layer 4 Layer 2 Layer 3 (inorganic) Layer 5 (organic) The values for Fe are excessive and probably in error, but there has been no opportunity to reanalyse the sample. A series of samples was examined by X-ray diffraction. Samples A and B appeared to be virtually pure quartz. The other samples, notably Sample D, produced a dif- fraction pattern that resembles that of cristobalite. This mineral usually occurs in metamorphic and igneous rocks, particularly andesite. Since rocks of this kind are not present in the area a more probable explanation is that some mineral or mine- rals that are isomorphic with cristobalite are present. Finally, the following values for organic and inorganic constituents were derived after drying and ashing the samples. Sample Percent Organic Percent Inorganic A 0.40 99.60 B 1.46 98.54 C 8.90 91.10 D 2.68 97.32 E 22.10 77.90 I I t at Sample Wavelength Approximate Concentration (%) Approximate Wavelength Concentration (%) Wavelength Approximate Concentration (% ) 4.0 0.5 0.5 1.0 0.005 0.003 0.000 (4.6 x 10-5) 0.050 0.0005 2.60 51.0 11.0 7.5 0.035 0.0 35 x 0.26 x 0.0002 0.0044 0.0032 0.0100 0.0034 1.0 4.0 3.0 6.0 0.9 3088 3088 3088 3088 0.40 0.50 1.50 0.05 0.7 27.0 80.0 A B C D E A B C D E A B C D E A B C D E A B C D E A B C D E E 2942 2942 2942 2942 4227 4227 4227 4227 4227 2937 2937 2937 2937 2779 2779 x 2779 x 3774 3774 3774 3774 3774 2568 256 8 2568 2568 2568 3373 2973 2973 2973 0.05 71.0 9.0 6.0 3000 3000 3000 2852 2852 2852 2852 2852 2652 2652 2652 2652 2652 x 0.08 0.04 x 0.01 3.0 3.0 1.0 21.0 0.2 I llt Sample Wavelength Exp. Concen. (%0) Prob. Concen. (%0) A 2443 x 100 (x - 0) %T or no background and, or line) B 2443 143 50-80 C 2443 88 70-80 D 2443 216 60-80 E 2443 16 93 L APPENDIX III ANALYSIS OF BOTANICAL SPECIMENS FROM FEATURE 3 In the description of feature 3 in Layer 1, reference was made to the charred floral material present on the bottom of the feature. A section of this material roughly four inches square was coated with Glyptal, a cement, and lifted intact. The specimen was sub- mitted to Michael J. Wells of the Botanical Research Institute, Albany Museum, Grahamstown, South Africa, for analysis. Mr. Wells has done the analysis of botanical remains from Scott's Cave, a "Later Stone Age" site, and has assisted with the work on botani- cal specimens from the Acheulean site at Amanzi. The following factual information is derived from his work. From three ounces of the sample it was possible to isolate 84 "clearly recognizable" fragments of corm or bulb scales in addition to carbonized small twigs and two fragments of what are possibly seeds. The corm scales isolated from the Montagu Cave are, in my opinion, almost certainly those of a species of Moraea or Homeria, two closely related genera of the Iridaceae or Iris family. The coarse "herringbone" reticulations of the scales are an excellent match of those found on herbarium speci- mens of some of the smaller Moraea spp., e.g., M. papilionacea, M. fimbriata, M. agusta, M. setacea, M. edulis, and homeria spp. such as H. collina. No other taxa have been found with reticulate scales which could be confused with those from Montagu Cave (Wells, pers. comm.). Wells refers to Watt and Breyer-Brandwijk, The Medi- cinal and Poisonous Plants of Southern and Eastern Africa (1962), who say that most species of Moraeal and Homeria are poisonous but that the corm of Mor edulis is edible and tastes like a boiled chestnut. The indication that many of the Moraea and Home species are poisonous raises the question of whethe feature 3 might have been used for the preparation arrow poison rather than of food. But the analysis has been done of ethnographically known arrow poil does not suggest that anything like the Moraea or meria corms were ever used for this purpose (Shaw et al., 1963). The evidence, then, suggests strongly that featur was not used for preparing poison. Wells says, Because selective preservation favours the ext tough corm scales above most seeds, the prese or absence of seeds may or may not be signifi However, I agree that there is a considerable that the pit was used for pounding or bruising . . . With the available evidence I would favour likelihood that the corm scales belong to an e species, or to a relatively inoffensive species for medicinal, mastic, or other purposes (pers. comm.). Hopefully as more excavations are conducted at "Later Stone Age" sites, additional information a the utilization of floral resources will come to lig and the uses of such features as feature 3 will be' better understood. [ 94 ] I f BIBLIOGRAPHY ;"Bruner, Jerome S., Jacqueline J. Goodnow, and ZGeorge A. Austin 1956 A Study of Thinking. New York: Wiley and Sons. Butzer, Karl 1971 Environment and Archaeology. 2nd edition. Chicago: Aldine-Atherton. .Clark, J. Desmond 1958a "Certain Industries of Notched and Strangu- lated Scrapers in Rhodesia, Their Time Range and Possible Use," South African Archaeological Bulletin, Vol. XIII, No. 50, pp. 56-66. 1958b "Some Stone Age Woodworking Tools in Southern Africa," South African Archaeolo- gical Bulletin, Vol. XIII, No. 52, pp. 144-152. 1959 The Prehistory of South Africa. Harmonds- worth: Penguin Books Ltd. 1960 "Human Ecology During Pleistocene and Later Times in Africa South of the Sahara," Current Anthropology, Vol. 1, No. 4, pp. 307-324. 1963 Prehistoric Cultures of Northeast Angola and Their Significance in Tropical Africa. Publicacoes Culturais No. 62 da Companhia de Diamantes de Angola, Museu do Dundo, Lisbon. 1964a "The Influence of Environment in Inducing Culture Change at Kalambo Falls Prehistoric Site," South African Archaeological Bulletin, Vol. XIX, No. 76, Part IV, pp. 93-101. 1964b "The Sangoan Culture of Equatoria: The Implications of its Stone Equipment," Mono- grafias IX, Diputacion Provincial del Bar- celona, Instituto de Prehistoria Y Arguelogia (Miscelanea en Homenaje al Abate Henri Breuil, R. Perello (ed.), Barcelona), pp. 309-325. 1970 Prehistory of Africa. New York: Praeger. luark, J. Desmond and E. M. van Zinderen Bakker 1964 "Prehistoric Culture and Pleistocene Vegeta- tion at the Kalambo Falls, Northern Rhodesia," Nature, Vol. 201, No. 4923, pp. 971-975. tlarke, David L. k 1968 Analytical Archaeology. London: Methuen and Co., Ltd. Cole, Monica 1961 South Africa. New York: E. P. Dutton and Co., Inc. Cooke, C. K. 1957 "The Waterworks Site at Khami, Southern Rhodesia: Stone Age and Proto-historic," Occasional Papers of the National Museums of Southern Rhodesia, No. 21A, Vol. 3, pp. 1-43. 196 3 "Report on Excavations at Pomongwe and Tshangula Caves, Matopo Hills, Southern Rhodesia," South African Archaeological Bulletin, Vol. 18, No. 71, Part III, pp. 75-151. Deacon, H. J. n.d. "A Review of the Post-Pleistocene in South Africa. " Deacon, H. J. and Janette 1963 "Scott's Cave: A Late Stone Age Site in the Gamtoos Valley," Annals of the Cape Pro- vincial Museums, Vol. III, pp. 96-121. Fagan, B. M. 196 0 "The Glentyre Shelter and Oakhurst Re- examined," South African Archaeological Bulletin, Vol. XV, No. 59, pp. 80-94. Fitzgerald, Walter 1961 Africa: A Social, Economic and Political Geography of its Major Regions. London: Methuen and Co., Ltd. Goodwin, A. J. H. 1929 "The Montagu Cave: A Full Report of the Investigation of the Montagu Rock-shelter," Annals of the South African Museum, Vol. XXIV, Part I, pp. 1- 16. 1938 "The Archaeology of the Oakhurst Shelter, George," Transactions of the Royal Society of South Africa, Vol. XXV, Part III, pp. 229-257. Goodwin, A. J. H. and C. van Riet Lowe 1929 "The Stone Age Cultures of South Africa," Annals of the South African Museum, Vol. XXVII, pp. 1-289. Goodwin, A. J. H. and B. D. Malan 1935 "Archaeology of the Cape St. Blaize Cave and Raised Beach, Mossel Bay," Annals of the South African Museum, Vol. XXIV, Part III, pp. 111-140. 95 1 Anthropological Records Goodwin, A. J. H. and B. Peers 1953 "Two Caves at Kalk Bay, Cape Peninsula," South African Archaeological Bulletin, Vol. VIII, No. 31, pp. 59-77. Heizer, Robert F. 1958 A Guide to Archaeological Field Methods. Palo Alto: The National Press, 3rd rev. ed. Hewitt, John 1921 "On Several Implements and Ornaments from Strandlooper Sites in the Eastern Province," South African Journal of Science, Vol. 18, pp. 454-467. Howell, F. Clark and J. Desmond Clark 1963 "Acheulian Hunter-Gatherers of Sub- Saharan Africa," African Ecology and Human Evolution (F. Clark Howell and Francois Bourli'ere, eds.). Viking Fund Publications in Anthropology No. 36, Chicago: Aldine Pub- lishing Co., pp. 458-533. Inskeep, R. R. 1967 "The Late Stone Age in Southern Africa," Background to Evolution in Africa (W. W. Bishop and J. Desmond Clark, eds.). Chi- cago: The University of Chicago Press, pp. 557 -582. Isaac, Glynn L. 1969 "Studies of Early Culture in East Africa," World Archaeology, Vol. 1, No. 1. Isaac, Glynn L. and Charles M. Keller 1968 "Note on the Proportional Frequency of Side- and End-struck Flakes," South African Archaeo- logical Bulletin, Vol. XXIII, Part I, pp. 17-19. Jackson, Stanley P. 1961 Jolly, Kei Climatological Atlas of Africa, Commission for Technical Cooperation in Africa South of the Sahara, Joint Project No. 1, Pretoria. ith 1948 "The Development of the Cape Middle Stone Age in the Skildegat Cave, Fish Hoek," South African Archaeological Bulletin, Vol. III, No. 12, pp. 106-107. Keay, R. W. J. 1959 Vegetation Map of Africa South of the Tropic of Cancer, London: Oxford University Press. Keller, Charles M. 1969 "Mossel Bay: A Redescription," South African Archaeological Bulletin, Vol. 23, No. 4, pp. 131-140. King, L. C. 1942 South African Scenery, A Textbook of Geo- morphology, London: Oliver and Boyd, Ltd. Klein, R. 1971 "Problems in the Study of the Middle Stone Age of South Africa," South African Archae logical Bulletin, 25: 127-135. X Kleindienst, Maxine R. I 1961 1962 Krynine, 1949 1950 Leakey, 1952 Levyns, 1964 "Variability within the Late Acheulian Asseng blage in Eastern Africa, " South African Archaeological Bulletin, Vol. XVI, No. 62, pp. 35-52. "Components of the East African Acheulian Assemblage: An Analytic Approach," in G. Mortelmans and J. Nenquin, Actes du IVe Congres Panafricain de Pr6histoire et de l'Etude du Quaternaire, Section III. Pre- 2 et Protohistoire (Musee Royal de l'Afrique Centrale, Tervuren, Annales, Serie in-80, Sciences Humaines, 40), pp. 81-112. Paul D. "The Origin of Red Beds," Transactions of the New York Academy of Sciences, Series II, Vol. 11, No. 3, pp. 60-68. "Petrology, Stratigraphy, and Origin of the Triassic Sedimentary Rocks of Connecticut,' State Geological and Natural History Survey, Bulletin No. 73, pp. 5-247. L. S. B. Proceedings of the Pan-African Congre o Prehistory 1947. Oxford: Basil Blackwell. M. R. "Migrations and Origin of the Cape Flora," Transactions of the Royal Society of South Africa, Vol. XXXVII, Part II, pp. 85-107. Louw, J. T. 1960 Prehistory of the Matjes River Rock Seltei Memoir No. 1, National Museum, Bloemfonteiq Lowe, C. van Riet 1946 "The Coastal Smithfield and Bipolar Tech- nique," South African Journal of Science, Vol. XLII, pp. 240-246. Malan, B. D. 1949 "Magosian and Howieson's Poort," South African Archaeological Bulletin, Vol. IV, No. 13, pp. 34-36. Mason, Revil 1962 Prehistory of the Transvaal. Johannesburg: Witwatersrand University Press. 1967 "Analytical Procedures in the Earlier and. Middle Stone Age Cultures in Southern Afl Background to Evolution in Africa (W. W. Bishop and J. Desmond Clark, eds.). Chic The University of Chicago Press, pp. 737- 96 Keller: Montagu Cave in Prehistory 1965 "Makapansgat Limeworks Fractured Stone Objects and Natural Fracture in Africa," South African Archaeological Bulletin, Vol. 20, No. 77, pp. 3-17. 1969 "Tentative Interpretations of New Radiocarbon Dates for Stone Artifact Assemblages from Rose Cottage Cave, O.F.S., and Bushman Rock Shelter, Transvaal," South African Archaeo- logical Bulletin, Vol. 24., No. 2, pp. 57-59. 1971 Multivariate Analysis of Cave of Hearths Middle Stone A Artefact Assemblages. Occasional Papers No. 7, Department of Archaeology, University of the Witwatersrand. hI)ey, Kenneth P. 1957 "Earliest Use of Fire," Third Pan-African Congress on Prehistory (J. Desmond Clark, ed.) London: Chatto and Windus, pp. 385-386. Peringuey, L. 1911 "The Stone Ages of South Africa," Annals of the South African Museum, Vol. VIII, pp. 1-218. chrire, C. 1962 "Oakhurst: A Re-examination; and Vindica- tion," South African Archaeological Bulletin, Vol. XVII, No. 67, pp. 181-195. haw, E. M., P. L. Woolley, and F. A. Rae 1963 "Bushman Arrow Poisons," Cimbebasia (SWA Research), No. 7, pp. 2-41. herman, G. Donald 1952 "The Genesis and Morphology of the Alumina- rich Laterite Clays," Problems of Clay and Laterite Genesis (Symposium at Annual Meet- ing of the American Institute of Mining and Metallurgical Engineers, 1951), New York, pp. 154-161. Stapleton, P. and J. Hewitt 1927 "Stone Implements from a Rock-shelter at Howie- son's Poort near Grahamstown," South African Journal of Science, Vol. XXIV, pp. 574-587. 1928 "Stone Implements from Howieson' s Poort near Grahamstown," South African Journal of Science, Vol. XXV, pp. 399-409. Summers, Roger 1957 "Notes on the Possible Origin of Magosian Cultures in Africa," Occasional Papers of the National Museums of Southern Rhodesia, No. 21A, Vol. 3, pp. 56-60. Van Houten, F. B. 1948 "Origin of Red-Banded Early Cenozoic Deposits in Rocky Mountain Region," Bulletin of the Ameri- can Association of Petroleum Geologists, Vol. 32, No. 11, pp. 2083-2126. van Zinderen Bakker, E. M. and J. Desmond Clark 1962 "Pleistocene Climates and Cultures in North- Eastern Angola," Nature, Vol. 196, No. 4855, pp. 639-642. Watt, J. M. and M. G. Breyer-Brandwijk 1962 The Medicinal and Poisonous Plants of Southern and Eastern Africa. Edinburgh: E. and S. Living- stone, 2nd ed. Well s, M. J. 1965 "An Analysis of Plant Remains from Scott's Cave in the Gamtoos Valley," South African Archaeological Bulletin, Vol. XX, No. 78, Part II, pp. 79-84. White, P. and C. 1964 "Surface Sites in South Africa," South African Archaeological Bulletin, Vol. XIX, No. 75, Part III, pp. 64-66. 97 ABBREVIATIONS Layer 1 plain simple facetted facetted "bulb only" removed shaped utilized waste crescents backed blades obliquely truncated blades thumbnail scrapers hollow scrapers scrapers outils 6caill6s chisels burins discoids points trimmed flakes trimmed chips pointed tools choppers core scrapers flakes flake fragments chips chunks long quadrilateral short quadrilateral triangular irregular ha clv knv d s hd lg flk sing plat disc frm irr bic tri ov ss ost mtr frg obi sm flk plno cnx spl stk splt cbl pbl lanc 0 Layers 3 and 5 hand-axes cleavers knives discoids large scrapers heavy duty large flake single platform disc formless irregular biconical triangular ovate small scrapers other small tools minimally trimmed pieces fragments other bifaces small flake plano- convex split pebble struck cobble pebble lanceolate other Layer 2 multiple hollow scraper strangulated scraper trimmed chunk nosed tool point fragment pestle un 40~ 11, `o '." A 1* o, & * .b _ h _ D ? 0 0 I LI 1 25 30 FEET SURFACE I Fig. 17. -I 35 I I . s 4 a I 1 \,- 4 -T lt . .0 I O', Q I GH K~j r/ / zA v U.@t9 , 0 O ELI 20 25 3 FEET SURFACE 11 0 Fig. 18. I" 4Z., b I I %~, \ o ~\ - Om I I. _ I %Ip H 1s- \y5 1 FEET I SURFACE III Fig. 19. -1 -I q t 0 " ? oF" 1? . E L 20 Ct, calo rj,? - p 0 rj i a A, yo Q t% - q, c? %C? 0 9 1 25 30 35 7 F L lS I u6 20 25 30 1 1 1 FEET I SURFACE IV Fig. 20. 0?1 - I - ? ?, N47 25 30 1 1 1 FEET I SURFACE V Fig. 21. -1 -1 V l E 20 35 25 I I ISrFACE I V I SURFACE VI Fig. 22. FL 15 EL 20 30 FEET w w U. La. C,) ,Jo CY IL In -? CY I 99.1 (66,816) .8 . SHPD UTZD WST Fig. 24. Artifacts from Layer 3. SING PLAT (563) Fig. 25. Tools from Layer 3. (66,180) 34 CORES LG SM FLK FLK FLK FRG Fig. 26. Waste from Layer 3. (110) DISC CNX Fig. 27. Core Types from Layer 3. CHP CNK CBL (1468) 96.5 1 2.9 SHPD UTZD WST Fig. 29. Tools from Surface VIII FLK FLK FRG Fig. 28. Artifacts from Surface VIII. l Fig. 30. Waste from Surface VIII. Rb (I " 0>l r . 0=s