Evidence from Liang Bua, a limestone cave on the island of Flores in East Indonesia, provides a unique opportunity to explore the long term relationship between hominins and their environment. Occupation deposits at the site span ῀95 ka and contain abundant stone artefacts, well preserved faunal remains and evidence for an endemic species of hominin: 'Homo floresiensis'. Work at the site included detailed geomorphological and environmental analysis, which has enabled comparisons to be drawn between changes in the occupational intensity in the cave, using stone tool and faunal counts, and changes in the environmental conditions, using the characteristics of the sedimentary layers in the cave and speleothem records. These comparisons demonstrate that 'H. floresiensis' endured rapidly fluctuating environmental conditions over the last ῀100 ka, which influenced the geomorphological processes in the cave and their occupational conditions. The intensity of occupation in the cave changed significantly between 95 and 17 ka, with peaks in occupation occurring at 100–95, 74–61 and 18–17 ka. These correlate with episodes of channel formation and erosion in the cave, which in turn correspond with high rainfall, thick soils and high bio-productivity outside. In contrast, periods of low occupational intensity correlate with reduced channel activity and pooling associated with drier periods from 94 to 75 and 36 to 19 ka. This apparent link between intensity of hominin use of the cave and the general conditions outside relates to the expansion and contraction of the rainforest and the ability of 'H. floresiensis' to adapt to habitat changes. This interpretation implies that these diminutive hominins were able to survive abrupt and prolonged environmental changes by changing their favoured occupation sites. These data provide the basis for a model of human–environment interactions on the island of Flores. With the addition of extra data from other sites on Flores, this model will provide a greater understanding of 'H. floresiensis' as a unique human species.

A small marine pearl was recovered at the Brremangurey rockshelter, on the Kimberley coast, from layers dating to approximately 2000 years ago. In an area famous for its pearls and history of cultured pearl production, public interest centred on whether the pearl was as old as the layer in which it was contained, or whether it was a recent cultured pearl that had infiltrated down from above. The near-spherical shape of the pearl hinted at a possible cultured origin. Owing to the uniqueness and historic cultural significance of this find, non-invasive analytical techniques were used to investigate whether the Brremangurey pearl was cultured or natural. Midden analysis was further used to assess the likely origin of the pearl within the stratified deposits. Analysis confirmed that the pearl is of natural origin and a dense midden lens of 'Pinctada albina' shells is its likely origin.

In the Soa Basin of central Flores, eastern Indonesia, stratified archaeological sites, including Mata Menge, Boa Lesa and Kobatuwa (Fig. 1), contain stone artefacts associated with the fossilized remains of Stegodon florensis, Komodo dragon, rat and various other taxa. These sites have been dated to 840-700 kyr bp (thousand years before present). The authenticity of the Soa Basin artefacts and their provenance have been demonstrated by previous work, but to quell lingering doubts, here we describe the context, attributes and production modes of 507 artefacts excavated at Mata Menge. We also note specific similarities, and apparent technological continuity, between the Mata Menge stone artefacts and those excavated from Late Pleistocene levels at Liang Bua cave, 50 km to the west. The latter artefacts, dated to between 95-74 and 12 kyr ago, are associated with the remains of a dwarfed descendent of S. florensis, Komodo dragon, rat and a small-bodied hominin species, Homo floresiensis, which had a brain size of about 400 cubic centimetres. The Mata Menge evidence negates claims that stone artefacts associated with H. floresiensis are so complex that they must have been made by modern humans (Homo sapiens).

This study examines trends in stone tool reduction technology at Liang Bua, Flores, Indonesia, where excavations have revealed a stratified artifact sequence spanning 95 k.yr. The reduction sequence practiced throughout the Pleistocene was straightforward and unchanging. Large flakes were produced off-site and carried into the cave where they were reduced centripetally and bifacially by four techniques: freehand, burination, truncation, and bipolar. The locus of technological complexity at Liang Bua was not in knapping products, but in the way techniques were integrated. This reduction sequence persisted across the Pleistocene/Holocene boundary with a minor shift favoring unifacial flaking after 11 ka. Other stone-related changes occurred at the same time, including the first appearance of edge-glossed flakes, a change in raw material selection, and more frequent fire-induced damage to stone artifacts. Later in the Holocene, technological complexity was generated by "adding-on" rectangular-sectioned stone adzes to the reduction sequence. The Pleistocene pattern is directly associated with 'Homo floresiensis' skeletal remains and the Holocene changes correlate with the appearance of 'Homo sapiens'. The one reduction sequence continues across this hominin replacement.

This paper presents the results of the recent technological analysis of the Early Pleistocene stone assemblage from Mata Menge in the Son Basin of Flores, Indonesia, the oldest Palaeolithic stone assemblage recovered from a well-dated stratified context in Southeast Asia. The various methods and techniques used by hominins to reduce stones at the site are discussed, as well as evidence for the deliberate transport of flaked stone artefacts around the Son Basin landscape.

Recent excavations at the early Middle Pleistocene site of Mata Menge in the So'a Basin of central Flores, Indonesia, have yielded hominin fossils attributed to a population ancestral to Late Pleistocene 'Homo floresiensis'. Here we describe the age and context of the Mata Menge hominin specimens and associated archaeological findings. The fluvial sandstone layer from which the in situ fossils were excavated in 2014 was deposited in a small valley stream around 700 thousand years ago, as indicated by 40Ar/39Ar and fission track dates on stratigraphically bracketing volcanic ash and pyroclastic density current deposits, in combination with coupled uranium-series and electron spin resonance dating of fossil teeth. Palaeoenvironmental data indicate a relatively dry climate in the So'a Basin during the early Middle Pleistocene, while various lines of evidence suggest the hominins inhabited a savannah-like open grassland habitat with a wetland component. The hominin fossils occur alongside the remains of an insular fauna and a simple stone technology that is markedly similar to that associated with Late Pleistocene 'H. floresiensis'.

Song Gupuh, a partially collapsed cave in the Gunung Sewu Limestones of East Java, Indonesia, contains over 16 m of deposits with a faunal sequence spanning some 70 ka. Major changes in the range of animals represented show the impact of climate change and humans. The Terminal Pleistocene and Early Holocene was a period of maximum biodiversity. Human use of Song Gupuh and other cave sites in the region also intensified significantly from ca. 12 ka, together with a new focus on exploitation of small-bodied species (macaque monkeys and molluscs), the first evidence for import of resources from the coast, and use of bone and shell tools. Human activity, especially after the onset of the Neolithic around 2.6 ka, subsequently contributed to a progressive loss of many species from the area, including tapir, elephant, Malayan bear, rhino and tiger, and this extinction process is continuing. We conclude by discussing the biogeographical significance of Song Gupuh in the context of other sites in Java (e.g. Punung, Wajak) and further afield (e.g. Liang Bua).

The stone technology from Mata Menge on Flores, Indonesia, is described, providing the first detailed analysis of the largest stone artefact assemblage from a stratified and securely dated Middle Pleistocene site in Southeast Asia. Technological analysis indicates a reduction sequence based on the centripetal, or 'radial', reduction of transported blanks. The implications for early hominin behaviour on Flores are considered.

Time-averaging is a process that affects almost every form of archaeological deposit. The conflation of two or more units from different time periods masks the true temporal span of units which is hidden by postdepositional processes. The implications of this are obvious as archaeological material found in close stratigraphic association may differ in age by hundreds or thousands of years. Some sites have a greater tendency towards the effects of time-averaging, with shell middens being one of the more susceptible. Conventional approaches to midden excavation or analysis, however, do little to tackle the issue of time-averaging. Using amino acid racemisation (AAR), an intensive relative dating programme was undertaken on shell midden excavated from a potentially time-averaged midden deposit. This approach revealed temporally distinct units that had been con flated into one deposit resulting in shell specimens temporally separated by up to 6000 years being found in close stratigraphic association. The application of AAR allowed us to define the temporal parameters of the various comingled deposits, and in doing so isolate temporal units which showed very different depositional patterns. These contrasting units imply different depositional behaviours and in turn changes in site use through time. This new application of AAR offers a way to approach shell midden archaeology to expose instances and repercussions of time-averaging that were previously hidden.