The conversion of the Zeekoe Valley Archaeological Project survey data to a GIS format allows rapid and accurate analysis of this large hunter-gatherer database. During the 16-month survey 13,866 prehistoric Stone Age sites were recorded and plotted on aerial photographs. These site locations and archaeological data can now be analysed in a manner never possible before the conversion. The distribution and abundance of sites spanning over ~700,000 years of occupation demonstrates how human hunting and gathering societies organized themselves spatially on an African landscape. These results show how these different groups positioned themselves in different locations especially in relation to water sources in the semi-desert Karoo. These distributions show flexible patterns of spatial organization through the prehistoric past.

It has been proposed that Paleolithic studies should abandon their focus on groups and turn instead to the individual. If individuals are to emerge from the lithics-dominated Middle Paleolithic record, the best chance of success is to identify the products of learner knappers from those of their mentors. To do so we need a framework of knapping standards by which to measure Middle Paleolithic skill level. Selected measurements on a sequence of 100 subcircular Levallois tortoise core reductions by a knapper of intermediate skill were compared with 25 reductions by his highly experienced instructor. Four measures emerge as potential markers of skill level: total stone consumption during initial core preparation, consumption from the upper and lower core surface, symmetry of the first detached Levallois flake, and failure rate of that detachment by overshooting the core's rim. These markers allow us to discriminate between the work of a modern learner and his mentor, but > 30 percent were misclassified. The learning trajectory is more complex than the mere honing of skills through practice and is punctuated by increasing numbers of mentor-like reductions. It follows that skill-level measures on their own are imperfect discriminators. Personal markers other than those of skill level must be found by which to seek individuals in the Middle Paleolithic record.

Lithic raw material constraints are widely assumed to be a determining factor of flaked stone tool morphology, but this assumption remains largely untested. We conducted a controlled experiment to determine whether a knapper’s growing replication skills would be hindered if the toolstone used was switched from large flakes of an easily worked chert to nodules of less tractable one. Two batches of Preferential Levallois cores were knapped, an earlier series made from standardised large flakes of sediments dominated by chalcedonic quartz followed by a more challenging one using variably-shaped, cortical nodules of microcrystaline quartz that varies in the completeness of quartz replacement of calcite and dolomite. Skill level markers were designed to measure the knapper’s ability to achieve a series of set goals. These were quantified and subjected to statistical testing. In all but one test, significant increases in skill could be detected from the earlier to the later batch of reductions, despite the drop in toolstone quality. Significant improvements in the consistency of the knapper’s output could also be detected. However, the switch to a more challenging, nodular chert did require extra shaping, which resulted in more waste. This masked visible progress towards producing a less costly core. Overall, our results do not support the assumed primacy of toolstone constraints over other factors in influencing the morphology of flaked stone tools.

Kuhn's GIUR is a set of measurements taken on a resharpened stone flake's edge. Its purpose is to estimate the amount of stone removed before the exhausted tool was abandoned. An experimental test of Kuhn's Index led Hiscock and Clarkson to suggest that it is an excellent estimator of flake mass loss. Our own tests, using different experimental procedures, reached much less optimistic conclusions. Here, we apply their own measurement procedures, and successfully duplicate their results. It does not follow, however, that our now-consistently high correlations have a causal connection. As GIUR values increase, they are less able to estimate the actual loss of flake mass. It appears, therefore, that incremental loss of flake mass need not be the sole driver of the GIUR value. If the correlation is an inevitable byproduct of the experiment's design, as it now appears, then analyses based on this high correlation will be misleading. The GIUR remains, however, and excellent gauge of flake edge exhaustion.

It is widely believed that the change from discoidal flake production to prismatic blade-making during the Middle–Upper Paleolithic transition in Europe led to enhanced technological efficiency. Specifically, blade-making is thought to promote higher rates of blank production, more efficient and complete reduction of the parent core, and a large increase in the total length of cutting edge per weight of stone. Controlled replication experiments using large samples, computer-assisted measurements, and statistical tests of several different measures failed to support any of these propositions. When resharpened, the use-life of flake edges actually surpasses that of blades of equivalent mass because the narrower blades are more rapidly exhausted by retouch. Our results highlight the need to replace static measurements of edge length that promote an illusion of efficiency with a more dynamic approach that takes the whole reduction sequence into account. An unexpected by-product of our replications was the discovery that real gains in cutting-edge length per weight of stone are linked to surface area. There is now a need to test the proposition that all the perceived advantages currently bestowed upon blades only occurred during the shift from macroblade to bladelet production. If our results are duplicated in further experiments, the notion of “economical” blades will have to be rejected and alternative explanations sought for their appearance in the early Upper Paleolithic. While Aurignacian bladelet (Dufour) production could signal the advent of composite tool technology (wooden handles or shafts with bladelet inserts), this does not help to explain why macroblades were also produced in large numbers. We may need to reexamine the notion that macroblades were of more symbolic than functional significance to their makers.

Conventional wisdom has it that ceramic technology reached southernmost Africa with or just ahead of the so-called Iron Age, Bantu migrations of ca 2000 years ago. A review of the evidence suggests that the earliest ceramics in the subcontinent are thin-walled and smooth surfaced vessels, technologically quite distinct from the first thick-walled, coarse surfaced “Iron Age” ware of the subcontinent, and predating the latter by two to four centuries. There is no published evidence of a thin-walled ware to the north of the Zambezi, although undated examples are known from coastal Angola. It seems unlikely that the thin-walled wares in southernmost Africa represent a residue of some mass human migration in the distant past. It is more likely that the art of making fired clay pots reached the subcontinent through archaeologically invisible infiltrations by small groups, perhaps peripatetic artisans; or it may have been invented locally.

Middle Archaic earthen mound complexes in the lower Mississippi valley are remote antecedents of the famous but much younger Poverty Point earthworks. Watson Brake is the largest and most complex of these early mound sites. Very extensive coring and stratigraphic studies, aided by 25 radiocarbon dates and six luminescence dates, show that minor earthworks were begun here at ca. 3500 B.C. in association with an oval arrangement of burned rock middens at the edge of a stream terrace. The full extent of the first earthworks is not yet known. Substantial moundraising began ca. 3350 B.C. and continued in stages until some time after 3000 B.C. when the site was abandoned. All 11 mounds and their connecting ridges were occupied between building bursts. Soils formed on some of these temporary surfaces, while lithics, fire-cracked rock, and fired clay/loam objects became scattered throughout the mound fills. Faunal and floral remains from a basal midden indicate all-season occupation, supported by broad-spectrum foraging centered on nuts, fish, and deer. All the overlying fills are so acidic that organics have not survived. The area enclosed by the mounds was kept clean of debris, suggesting its use as ritual space. The reasons why such elaborate activities first occurred here remain elusive. However, some building bursts covary with very well-documented increases in El Niño/Southern Oscillation events. During such rapid increases in ENSO frequencies, rainfall becomes extremely erratic and unpredictable. It may be that early moundraising was a communal response to new stresses of droughts and flooding that created a suddenly more unpredictable food base.

Duinefontein 2 (DFT2) preserves at least two buried land surfaces within a 10-m thick dune plume on the Atlantic Coast of South Africa, about 35km north of Cape Town. Optically stimulated luminescence dating indicates that the sands enclosing the upper surface accumulated around 270ky ago, while the sands between the two surfaces were accumulating about 290ky ago. Excavation so far has focused on the upper buried surface, which is now exposed over 480m2. Historically, there was no body of fresh water nearby, and the surrounding vegetation was a variant of the regional fine-leafed shrub or fynbos. Pedogenic alteration of the sands and bones of water-loving mammals and amphibians indicate, however, that the bones and associated Acheulean artifacts accumulated near the edge of a large pond or marsh. In addition, the principal mammal species (buffalo, wildebeest, and kudu) imply a sharply different regional vegetation in which grass and broad-leafed bush were much more common. The artifacts are distributed across the upper buried surface in no apparent pattern, but the large mammal bones tend to occur as clusters of skulls, vertebrae, ribs, and other axial elements, often in near anatomical order. Limb bones are mostly missing, and the clusters appear to mark carcasses from which the limb bones were selectively removed. The bones rarely show marks from stone tools, but marks from carnivore teeth are common. Together with numerous hyena coprolites, the abundant tooth marks suggest that hyenas and perhaps other carnivores were largely responsible for carcass disarticulation. The human role appears to have been insignificant, which suggests that local Acheulean people obtained few large mammals, whether by hunting or scavenging. Among the small number of other Acheulean ‘carcass’ sites for which bone damage observations are available, the best-documented sites suggest the same limited human ability to acquire large mammals, but many additional sites will be necessary to determine if this was the Acheulean norm. Greatly expanded excavation of the lower buried surface at DFT2 can provide an additional, high-quality data point.

Abundant amphibian remains were encountered at Duinefontein 2, a late mid-Quaternary fossil site with Acheulean artifacts. The amphibians are distributed in concentric bands that suggest shoreline lags around the southeast rim of an interdunal pond, part of which has been exposed by very extensive excavations. The pond community was dominated by ranids and pipids, both requiring less than a meter depth of open water. Burrowing species such as toads, rain frogs, and sand frogs were present but rare, as were reed frogs. This suggests a thin rim of damp but poorly vegetated sand. Frequency and size changes suggest that the pond became shallower at a time when large mammal remains were proliferating on the adjacent land surface. The evolutionary status of the eight recovered amphibian taxa cannot be addressed since the osteology of most of their modern analogs still awaits systematic study.

In the semi-arid Karoo of South Africa's central plateau, tortoises are commonly preyed upon by raptors, some of which roost on ledges in the backs of small rock shelters. In the past, any suitable ledge could be occupied only when the shelter was not in use by Bushman hunter–gatherers. Thus, both agents could have contributed tortoise elements to the faunal deposits which accumulated in the shelter fills over the past two millennia. Element survival and breakage rates of tortoise remains in a recent shelter roost accumulation show reversed frequencies to those for tortoise carcasses left at raptor kill sites: what is abundant at the roost (skeletal elements, particularly from the neck and head) are scarce at the kills. When the roost sample is compared with the Bushman foodwaste sample beneath it, the two are readily distinguished by their element composition and condition. A surface sample from a neighbouring shelter without roosting ledges is also rich in skeletals, but lacks the characteristic cranials and vertebrae. Small carnivores are the suspected non-human agents. Although radiocarbon dates indicate an earlier hiatus in this sequence, no such raptor-like tortoise assemblages are detectable at this level.

Bifacial arrowheads are very rare elements in Later Stone Age sites from the South African interior and Lesotho highlands. While differences in arrowhead design across this vast region may serve as group markers, their potential remains uncertain because local variability has not been studied. Here, the design variables of a large sample of arrowheads from one small area of the Upper Karoo are described. Arrowhead production is so diverse that prospects for their use as ethnic/social markers look dim. The most variable arrowheads are those with double-backed tangs. They are more numerous than the bifacial arrowheads and may also be more widely distributed.

John Vogel has been an active participant in the Zeekoe Valley Archaeological Project for more than a decade. During those years he generated for us a radiocarbon chronology of the past two millennia, which allows us, at last, to begin to understand some of the complex events that took place in the upper Karoo during recent prehistory That John's contribution has provided the very backbone of this project is to understate the obvious. It is a pleasure and a privilege to offer this brief overview and interpretation of our findings in his honour.

The upper Karoo of central South Africa is a semi-desert region supporting several species of chelonia, and cultural deposits in local rock shelters also yield abundant tortoise bones. In the Late Holocene fill of Haaskraal rock shelter, remains of at least four kinds of tortoise and a terrapin have been identified. However, there is an anomalous absence of the giant adult form ofGeochelone pardalis, which is highly visible in the modern fauna. Frequencies of the two dominant speciesHomopus femoralisandH. boulengerido not vary during the two millennia of accumulation, in spite of the major changes in veld cover reported in local pollen and micromammal records. Instead, changes in tortoise size and skeletal composition reflect increasing admixture of remains derived from non-human predators, as occupation by humans dwindled during historical times. Comparisons with other rock shelter samples are needed, but investigations of tortoise remains from modern raptor nests and killing grounds are also an urgent priority.

Several rock shelter excavations in the Seacow River Valley of the upper Karoo region of central South Africa confirm that backed microlith production persisted there until at least AD 1800. It is widely assumed that they were used as arrow tip inserts by the prehistoric ancestors of Bushman hunter-gatherers. Microlith reduction strategies are reconstructed from excavated samples covering the last two millennia. Backed bladelets and point production became increasingly based on the partial backing of larger bladelets, from which the desired form was snapped, leaving a concave-backed stub. Production errors gave rise to a wide variety of partially backed fragments.

Early travelers in the southwest and south of the Cape Colony, and later explorers in the north, saw Khoikhoi pastoralists making and using large, reddish or black, coil-built cooking vessels with shoulder lugs and incised necks with everted rims. In these, they boiled meat and used some as drums. Smaller serving bowls were also seen. Travelers on the east and north frontiers of the Colony saw Bushman hunter–gatherers using flat-bottomed cooking bowls tempered with grass and decorated with punctate motifs. They boiled meat, soups, bones, skins, locusts, and seed mush in these, converted some to drums, and used others in gift exchanges. Few later sherd collectors made full use of these ethnohistoric sightings, but developed their own labeling systems. Most thought that the Bushmen of the interior learned pot-making through contact with coastal Khoikhoi. Why the two wares differed in every respect, however, could not be explained. Recent multidisciplinary studies in the northeast frontier area verify those differences, but also show that both wares were introduced together before 700 A.D., by herders. Thereafter the use of Khoi ware dwindled, then disappeared when the herding economy collapsed, leaving only grass-tempered bowls in general use. Thus “Bushman” pottery in the northeastern Cape appears to have its prehistoric roots in ancestral Khoi technology.

Ceramic vessels from Later Stone Age sites in the Seacow river valley, South Africa, were heavily tempered with C4 grasses. However, most δ13C values of the temper fall between the known δ13C ranges for C4 and C3 grasses. Among possible C3 additives that could skew the signal, ground bone, animal dung, alpine grass, Karoo scrub ash, and humic acid contamination are eliminated. The sherd surfaces yield fatty chars with C3‐like values. Although these are not enough to account for all the skewing, they represent only the final cooking event. They were preceded by many such events in which fat was absorbed into the C4 grass then charred to soot by later periods in the fire. Thus organic carbon with the isotope composition of the fat built up while the vessel remained in use.

Three ordination methods and three clustering methods were compared, using the same data from proton induced X-ray emission (PIXE) analysis of two visually distinct pottery types found in the same area of the upper Karoo region in South Africa. Petrographic analysis of these two Later Stone Age ceramic traditions indicate that geologically distinct clays and temper were used for each. The PIXE data show, however, that the separation is not as perfect as appearances suggest—some “hybrid” specimens are present. To determine whether these hybrids were a spurious byproduct of our statistical manipulation, we applied correspondence analysis, principle components, multidimensional scaling, and three versions of Q-mode clustering. The same hybrids showed up in most applications. Correspondence analysis achieved the most complete separation, but the real benefit was from the combined insights afforded by all techniques. Although a weak separator, Ward’s minimum variance method of Q-mode clustering proved more useful for discriminating spatially discrete subgroups of vessels within the 2000km2study area.