The oldest direct evidence of stone tool manufacture comes from Gona (Ethiopia) and dates to between 2.6 and 2.5 million years (Myr) ago1. At the nearby Bouri site several cut-marked bones also show stone tool use approximately 2.5 Myr ago2. Here we report stone-tool-inflicted marks on bones found during recent survey work in Dikika, Ethiopia, a research area close to Gona and Bouri. On the basis of low-power microscopic and environmental scanning electron microscope observations, these bones show unambiguous stone-tool cut marks for flesh removal and percussion marks for marrow access. The bones derive from the Sidi Hakoma Member of the Hadar Formation. Established 40Ar–39Ar dates on the tuffs that bracket this member constrain the finds to between 3.42 and 3.24 Myr ago, and stratigraphic scaling between these units and other geological evidence indicate that they are older than 3.39 Myr ago. Our discovery extends by approximately 800,000 years the antiquity of stone tools and of stone-tool-assisted consumption of ungulates by hominins; furthermore, this behaviour can now be attributed to Australopithecus afarensis.
The Dikika Research Project area is located in the Lower Awash Valley (Ethiopia) and is bordered on the north by Gona and Hadar and on the south by the Middle Awash research areas (Fig. 1). Work there (led by Z.A.) began in 1999 and has focused on survey in Hadar (>3.8 to 2.9 Myr ago) and Busidima Formation (2.7 to <0.6 Myr ago) deposits, both of which are exposed in their entirety within the project area3,4. This work has resulted in the discovery of a diverse and well preserved fauna, the discovery of several hominin fossils including a nearly complete juvenile Australopithecus afarensis (DIK-1-1) and a complete definition of the hominin-bearing Hadar Formation3,4,5,6.
In January 2009, the Dikika Research Project systematically collected fossils from localities just opposite the DIK-1 locality in the Andedo drainage, which predominantly exposes the Sidi Hakoma (SH) Member of the Hadar Formation (3.42–3.24 Myr ago; Fig. 1). Archaeological survey was conducted simultaneously in these same localities. In the course of this work, four fossils were identified with surface modifications which, based on field observations, resembled stone-tool cut marks7. These fossils were subsequently studied with optical and environmental scanning electron microscopy (ESEM) (see Methods and Supplementary Information). Secondary electron imaging (SEI) and energy dispersive X-ray (EDX) spectrometry data show that the marks on two of these fossils (DIK-55-2 and DIK-55-3) formed before fossilization. Optical and ESEM observations show that the marks lack the morphology indicative of trampling and biochemical marks, and that these two specimens have modifications clearly indicative of stone tool use, including cutting and percussion.
Both bones were found on the surface at the same locality: DIK-55. Stratigraphically this locality, an area of approximately 25 m × 50 m, can be placed into the section described previously for the nearby DIK-1 locality. It is below a low ridge that exposes only the lowermost sediments of the SH Member and below the level of a limestone marker (SH-lm) with a stratigraphically scaled age of 3.39 Myr ago, providing a minimum age for the site (Fig. 1 and Supplementary Information). Nowhere in the entire Andedo drainage are sediments above the lacustrine Triple Tuff 4 (TT-4) marker (3.24 Myr ago) exposed, providing a minimum age for the entire section. Specimen DIK-55-2 was found on the slope below the SH-lm marker and DIK-55-3 was found on the flats just in front of this slope. Fossils from this locality lack adhering matrix, indicating that they derive from a ∼1.5-m-thick sand bed that outcrops here. This sand is unique compared to many of the fossil-bearing sands of the SH Member (such as at DIK-1) in that it is not strongly cemented and thus its fossils lack adhering matrix.
DIK-55-2 (Fig. 2 and Supplementary Information) is a right rib fragment of a large ungulate, probably size 4 (cow-sized) or larger. Marks A1 and A2 are perpendicular to the cortical surface, V-shaped in cross-section with internal microstriations and diagnosed as high-confidence stone-tool cut marks. Mark B is a more obliquely oriented mark that shaves off the bone surface within which are microstriations, all consistent with a high-confidence stone-tool-inflicted mark from cutting, scraping and/or percussion. An indentation (mark C) with microstriations and crushing of the bone surface is a high-confidence hammerstone percussion mark described in Supplementary Information.
DIK-55-3 (Fig. 3 and Supplementary Information) is a femur shaft fragment of a size 2 (goat-sized) young bovid. The surface is densely marked (Fig. 3a). Mark A is perpendicular to the cortical surface and has clear microstriations running out one end (Fig. 3b, c), diagnosed as a high-confidence cut mark. Mark D (Fig. 3d–f) is a dense cluster. One prominent mark within D (Fig. 3d) has crushing of the bone surface with microstriations and is diagnosed as a high-confidence percussion mark. Mark E (Fig. 3g, h) is obliquely oriented, shaves off surface bone, has microstriations and a shouldered edge highly consistent with a stone-tool cut mark. Marks H1 and H2 overlap. H1 has clear microstriations, is associated with the broken edge of the bone and swirls in a way typical of a percussion mark. H2 shaves off bone surface and has clear microstriations, consistent with stone-tool cut marks and a scraping motion. DIK-55-3 has other high-confidence stone-tool-inflicted marks, and there is at least one mark (mark I) of uncertain agency (Supplementary Information). This specimen does not have any notches of the type that are sometimes associated with hammerstone percussion on long bones8,9,10, but this may be owing to post-depositional breakage of the edges that removed such notches.
The cut marks demonstrate hominin use of sharp-edged stone to remove flesh from the femur and rib. The location and density of the marks on the femur indicate that flesh was rather widely spread on the surface, although it is possible that there could have been isolated patches of flesh. The percussion marks on the femur demonstrate hominin use of a blunt stone to strike the bone, probably to gain access to the marrow. The external surfaces of ribs have thin sheaths of flesh, so the scraping marks on the fossil rib suggest stripping off of these sheaths.
The presence of stone-tool-modified bones and by implication the use of stone tools at Dikika by 3.39 Myr ago greatly increases the known antiquity of this behaviour. The earliest demonstration of stone tool production known thus far is after 2.6 Myr ago at several localities in Ethiopia and Kenya1,11,12,13,14. It is not possible to demonstrate from the modified bones whether the stone tools were knapped for this purpose or whether naturally occurring sharp-edged stones were collected and used. No stone artefacts or sharp-edged stones were found in association with the bones at DIK-55. However, stone tool production and consequently archaeological accumulations are not expected at this locality given the sedimentary environment characterized by the palaeo-Awash River emptying into a nearby lake3,4. In this relatively low-energy depositional environment, clasts suitable for stone tool production are not present (few particles larger than fine gravel, 8 mm diameter). Within the exposed SH Member, the distance from DIK-55 to cobble-sized raw materials (>64 mm) is ∼6 km (at Gorgore; Fig. 1). Thus, in this instance the absence of evidence for stone tool production in the immediate vicinity of the cut-marked bones may reflect landscape-level raw material constraints.
The bones presented here are the earliest evidence for meat and marrow consumption in the hominin lineage, pre-dating the known evidence by over 800 kyr2. Pending new discoveries, the only hominin species present in the Lower Awash Valley at 3.39 Myr ago to which we can associate this tool use is A. afarensis5,15. Whether A. afarensis also produced stone tools remains to be demonstrated, but the DIK-55 finds may fit with the view that stone tool production pre-dates the earliest known archaeological sites and was initially of low intensity (one-to-a-few flakes removed per nodule) and distributed in extremely low density scatters across the landscape such that its archaeological visibility is quite low16. The evidence presented here offers a first insight into an early phase of stone tool use in hominin evolution that will improve our understanding of how this type of behaviour originated and developed into later, well recognized, stone tool production technologies.
Bone surfaces were examined under ×8–80 magnifications with adjustable incident light from a bifurcated light source. ESEM was used to further document marks and to collect SEI/EDX data (see below) but was not needed for diagnosis and identification. Here (and in Supplementary Information) the nested terminology of Gifford-Gonzalez17 is used to draw inferential distinctions between the actors responsible for producing the marks, the effectors used to make the marks and causal action. A mark is considered high confidence in its diagnosis to effector if it has all the criteria defined in the literature for that mark. A distinction is made in mark diagnoses between general (stone tool, tooth, or unidentifiable) and specific (cut mark versus percussion mark), with the caveat that specific identifications are more tenuous given the overlap between percussive and cutting damage indicated by these specimens.
Modern collection damage was assessed visually and chemically. The patina inside the surface marks resembles the surface patina and not the lighter colour of the interior fossil bone made visible by some modern damage on the ends. The elemental composition, measured with EDX spectrometry, of the marks and adjacent surfaces indicates that fossilization occurred after mark formation (Supplementary Information). EDX spectrometry was also applied to a rock fragment, probably of igneous origin, embedded in one mark (Supplementary Information). The marks were assessed for criteria described as indicative of biochemical damage18,19 and of trampling20,21 and found to lack key criteria (Supplementary Information). Finally, we used well known and described morphological criteria9,18,19,22,23,24,25,26,27 to distinguish between cut marks, percussion marks and tooth marks. Further comparisons were made to experimentally generated stone cut-marked, percussion-marked and carnivore-tooth-marked comparative specimens. Identifications were blind tested for correspondence between three experienced taphonomists and zooarchaeologists who examined the specimens under the same light-microscope conditions. The results showed a high correspondence and agreement that most marks were stone-tool inflicted (Supplementary Information).
We thank the Authority for Research and Conservation of Cultural Heritage, the National Museum of Ethiopia, the Ministry of Tourism and Culture and the Afar regional government for permits and support; C. Mesfin, Z. Bedaso, T. Gebreselassie, Mesfin Mekonnen, H. Defar, A. Zerihun, G. Senbeto, Mogues Mekonnen, W. Aberra, T. Yifru and the people of the Dikika area for field assistance. We also thank the administration of Adaytu town and members of the Ethiopian armed forces. Funds for the 2009 field season were provided by the California Academy of Sciences. Travel expenses for D.G., S.P.M., D.R. and J.G.W. were covered by their respective institutions. C.W.M. and H.A.B. acknowledge the assistance of the research professionals in the John M. Cowley Center for High Resolution Electron Microscopy, LE-CSSS, ASU in conducting the ESEM imaging, and J. Thompson and S. Lansing for participating in the blind test. Z.A. thanks P. Mollard and K. Berge for assistance during fieldwork preparations.
This file contains Supplementary Information comprising: 1 Survey Methodology; 2 Faunal list; 3 Stratigraphic and palaeoenvironmental context; 4 Surface modifications on the fossil specimens; Supplementary Figures 1- 23 with legends; Supplementary Tables 1- 3 and References.