Border Cave in KwaZulu-Natal, South Africa, preserves a long and continuous archaeological record from 227 ka to 24 ka years ago, rendering it a key Middle Stone Age site in southern Africa. It has yielded the skeletal remains of eight anatomically modern Homo sapiens individuals, a lithic sequence that includes MSA 1, MSA 2, MSA 3, and Early Later Stone Age Industries, ochre, marine shells, the oldest burial associated with a personal ornament, and early expressions of complex cognition and innovation. Organic preservation is remarkable, with grass bedding that contains aromatic leaves with insecticidal properties found throughout the sequence. The bedding layers show that it was systematically placed on a layer of ash to deter crawling insects. Comprehending that aromatic leaves and ash deter pests, and using them as tools for delayed gratification following planning and strategizing, and organization that entailed a sequence of events, implies that complex cognition was in place from 200,000 years ago. Charred underground storage organs come from layers dated to 170,000 years ago, making them the oldest known examples of cooked starchy rhizomes. The fact that the rhizomes were cooked implies that the inhabitants of the shelter were able to make fire at will, another indication of complex cognition in early modern humans. A range of organic remains such as ostrich eggshell beads and bone points used as poisoned arrowheads are found in Early Later Stone Age layers starting at around 44,000 years ago, and they represent the earliest examples of modern human behavior as we know it.

Kaingo Sheep Rock Shelter was used by Later Stone Age (LSA) hunter-gatherers between 4370±180 and 170±30 BP. The site has rock art that includes a fine-line painting of a large, fat-tailed sheep, animal finger paintings, and geometric motifs. There are many microlithic end scrapers, a few backed tools, and more than 500 complete, incomplete and broken ostrich eggshell beads, as well as grooved stones and worked bone. By ~170 BP the density of material culture items reduced and the shelter may have been used only occasionally for ritual purposes like rain-making or initiations. Hunters, herders and farmers are represented in one way or another in the shelter, but it is unclear whether residential and non-residential 'time-share' is involved during the contact period. Since the shelter has contemporaneous LSA and Iron Age material culture signatures, there may have been sporadic interaction between the groups. Most ceramics belong to the Eiland facies, but a few fragments of one of the earliest ceramics found in southern Africa, the Bambata facies, were also discovered. Seed and charcoal identifications reveal bushveld vegetation similar to that of today, but possible evidence for mopane trees in the last 1 900 years implies greater diversity of plant life at that time.

Border Cave hosts a rare Middle and Early Later Stone Age sequence of deposits that extends as far back as ca. 250 thousand years (ka). The site's chronology has been built mainly on Electron Spin Resonance (ESR) ages obtained from teeth, conducted at the end of the 1990s, and on radiocarbon dating for the more recent layers. In order to refine the sequence's chronology, several materials were selected for luminescence dating, including 34 siliceous lithic artefacts, and quartz and feldspar grains extracted from 10 sediment samples. Since the radioisotopic contents of the cave sediments are abundant in the volcaniclastic host rock (about 2% K, 10 ppm ²³²Th, 2 ppm ²³⁸U in the sediments) and the ages for the lower layers are over 40 ka, high (>150 Gy) equivalent doses (De) were expected for most samples. The saturated thermoluminescence (TL) signals of the lithic artefact samples suggested either that they had not been heated in the past, or that they were already too old to be dated. The presence in the sediments of 40–63 μm quartz grains dominated by the fast Optically Stimulated Luminescence (OSL) component and with apparent high D0 values (200–1000 Gy) suggested, a priori, that they could be appropriate for measuring high Des. However, a systematic drop of the De values for these high D0 grains was observed and could not be fully understood and corrected accordingly. The recently proposed pIT protocol, together with a global growth curve (GGC) approach, was applied to 63–80 μm feldspar grains extracted from the sediment samples. The data suggest that the IR225 signal was likely not fully bleached at the time of deposition. Nonetheless, two age estimates, assumed to bracket the target age, have been calculated for each sample. These estimates are consistent with the previous ESR ages and the ESR/C14-based Bayesian models. Further improvements can be considered in the future with the pIT - GGC method, which would allow both increased accuracy and precision.

The end of the Middle Stone Age in southern Africa, often called the final MSA (∼40–28 ka), represents one of the most understudied technocomplexes in this part of the world. Researchers have often focused on earlier time periods associated with Marine Isotope Stage 4 or have emphasised the transition between the Middle and the Later Stone Age. Thus, the final MSA has been poorly understood and, at least in KwaZulu-Natal, only a few chrono-cultural markers called hollow-based points are known for it. Since 2016, excavations at Umbeli Belli rock shelter have produced new insights into this period. The site provides one of the most accurately dated sequences for the final MSA, spanning four geological horizons, respectively GH7, GH8, GH9 and GH10, that date to between 29.9 ± 2.3 and 40.3 ± 3.5 ka. Significant technological and typological variations are evident between those horizons, raising questions about the mechanisms behind them. A direct comparative analysis with the final MSA layers Coffee – Espresso at Sibhudu, which date to ∼38 ka, places these results in the regional archaeological context. The analysis shows first that the final MSA encompasses diachronic variability within relatively short time frames at Umbeli Belli. Secondly, it reveals several distinct chronological discrepancies between Sibhudu and Umbeli Belli. A detailed review of the environmental setting of the research area helps to explain these changes.

Besides providing a unique archaeological assemblage that documents the early emergence of complex behaviour in the human lineage, Border Cave (South Africa) is noteworthy for having yielded hominin remains of at least nine individuals, including the partial cranium Border Cave 1. While the exact provenance of Border Cave 1 is unknown, sequence stratigraphy and ESR dating converge towards an age from about 82 ka to 170 ka. Here we present novel information about the brain, braincase and bony labyrinth of Border Cave 1 and discuss related evolutionary implications. We compare Border Cave 1 to specimens of Early and Middle Pleistocene Homo as well as to fossil and extant Homo sapiens. Virtual segmentation techniques were used to reconstruct the brain and bony labyrinth endocasts, assess the distribution of cranial bone thickness, and identify the vascular and sulcal imprints preserved on the inner surface of the braincase. Our results show that the overall morphology of the brain endocast approximates the globular shape of the modern human brain and differs from the long and low brains seen in Middle Pleistocene fossil hominins. The vascular imprints preserved on the right hemisphere indicate that the middle branch derives from the anterior branch, which is a pattern shared with Neanderthals and modern humans. Bone thickness distribution in the Border Cave 1 cranium resembles the patterns seen in Cro-Magnon 1 and Abri Pataud 1, which both share a diffuse distribution of thickened areas over the frontal region. Finally, the relative size and curvature of the semicircular canals of the bony labyrinth conform to the ancestral configuration shared between Early and Late Pleistocene fossil hominins from Africa and the Levant, as well as modern humans, and distinct from the more derived condition documented within Neanderthals. We discuss the implications of our findings for understanding the bioge-ography, evolution, and, to some extent, behaviour of fossil Homo sapiens.