Search

Approximately 74 thousand years ago (ka), the Toba caldera erupted in Sumatra. Since the magnitude of this eruption was first established, its effects on climate, environment and humans have been debated. Here we describe the discovery of microscopic glass shards characteristic of the Youngest Toba Tuff—ashfall from the Toba eruption—in two archaeological sites on the south coast of South Africa, a region in which there is evidence for early human behavioural complexity. An independently derived dating model supports a date of approximately 74 ka for the sediments containing the Youngest Toba Tuff glass shards. By defining the input of shards at both sites, which are located nine kilometres apart, we are able to establish a close temporal correlation between them. Our high-resolution excavation and sampling technique enable exact comparisons between the input of Youngest Toba Tuff glass shards and the evidence for human occupation. Humans in this region thrived through the Toba event and the ensuing full glacial conditions, perhaps as a combined result of the uniquely rich resource base of the region and fully evolved modern human adaptation.

The origins of complex projectile weaponry provides insight into cultural and biological changes associated with the origins and spread of modern human populations. Middle Stone Age backed pieces are often thought to be components of such armaments, however our limited understanding of their functional characteristics as projectiles precludes understanding the adaptive problems they may have solved. Despite acknowledgment of raw material differences and intra-assemblage variability, whether variability in backed piece form reflects functional, economic, or stylistic variation has a paucity of empirical support. Here, the functional differences in backed piece form (size and shape) while hafted transversely and obliquely as high-velocity complex projectile armatures are examined. If there are performance tradeoffs simply in how backed pieces are arranged at the end of armaments that can influence effectiveness, then identifying the archaeological arrangement can provide insight into what variables were being prioritized in prehistoric technological systems. How variation in backed piece size, elongation, and hafting arrangement influences complex projectile performance is tested using experimental and actualistic projectile replications with a calibrated crossbow against animal and ballistics gelatin targets. The results of this study show that, within the size and shape variation of silcrete backed pieces examined, tool form plays a relatively limited role in their performance as projectile armatures. However, hafting orientation has very different performance characteristics for complex projectiles shot at ballistics gelatin compared to animal targets. We demonstrate that transversely hafted tools have more lethal internal wounds, but obliquely hafted backed pieces have greater puncture reliability. These functional differences represent different technological design emphasis: transversely hafted tools create large, deep wounds, while obliquely hafted arrows and darts create a puncture more reliably. Although obliquely hafted armaments cause less internal trauma, they are more likely to penetrate the hide of ungulate prey. Variability in MSA hunting tactics may have played a role in the design of weapon systems to optimize these performance tradeoffs. Despite similarities in shape with ethno-historic technologies, based on these results, MSA-sized backed pieces hafted as projectile armatures were unlikely to have been used with small, low-powered bows - but would have been lethal with a high-velocity delivery system.

•Novel method to identify heat treated silcrete using 3D microscopy.•Surface roughness measurements are lower in heated silcrete than unheated silcrete.•Silicon peels precisely replicate the surface roughness of silcrete flakes.•Our Bayesian model predicts heat treatment with an average accuracy of 84%.•The model performed better in a blind test with averaged roughness data (90%).