On the earliest evidence for habitual use of fire in Europe

For most of prehistory until ∼10,000 y ago, hominins all over the world subsisted as hunters and gatherers, with a generally highly mobile lifestyle. Until the later phases of the Paleolithic, these mobile foragers invested very little in camp layout, in dwelling structures, or in the construction of formal hearths (14). The overwhelming majority of their traces consist of stone tools, the debris of stone tool production, and often but not always food remains. With few exceptions (see below), preserved fireplaces are simple clusters of burnt materials occurring as flat lenses or in small hollows. The most common proxies for fire use consist of various find categories that display traces of having been submitted to heating: the reddened sediments on which a fire was built, heated stone artifacts, charred bone fragments, and pieces of charcoal. The problem is that other processes than human activities can create such findings too. Volcanic eruptions can be an important factor in driving the natural fire regime in volcanically active areas. Lightning strikes—with an average global flash rate of an estimated 44 ± 5 flashes per second (15)—or spontaneous combustion cause natural fires that can burn archeological camp sites. The remains of such grass and forest fires can become associated with archeological find materials. Hence, charcoal, charred bone material, and heated flints do not necessarily indicate anthropogenic fires (2). At open-air sites, direct evidence for human fires, such as charcoal, can be easily removed by natural processes, including erosion by water or by wind (16). Thus, the context of possible fire indicators is important. If burnt bones, heated artifacts, and charcoal occur in caves or enclosed sites where the deposits are demonstrably in situ and are not reworked by slope wash or debris flow entering the cave, then they can be considered good indicators of anthropogenic fires. Much of the debate on the history of human control of fire relates to the problem of the correct interpretation of possible fire indicators, especially in the absence of modern excavation techniques and detailed studies of site formation processes. In this last domain, significant progress has been made by the application of micromorphological techniques to the study of combustion features in rock shelters and caves (17–20).

The earliest traces of hominin presence in Europe come from its southern parts and date to more than one million y ago (21). Recent data from the English site Happisburgh 3 suggest that hominins may already have been adapted to the challenging environments of the boreal zone in the Early Pleistocene, more than 800,000 y (800 ka) ago (10). Fire would have been an important tool in such environments. However, surprisingly, evidence for use of fire in the Early and early Middle Pleistocene of Europe is extremely weak. Or, more exactly, it is nonexistent, until ∼300–400 ka. Our review of the early European sites (Dataset S1) shows that the earliest possible evidence of fire comes from two sites dated to ∼400 ka, Beeches Pit in England and Schöningen in Germany. At Schöningen, the evidence consists of some heated flints (although mostly natural pieces) (22) and charred wood, including a wooden tool, with the studies of possible remains of former hearths still in progress (23). At Beeches Pit, dated to Marine Isotope Stage (MIS) 11 (Dataset S1), the evidence consists of heated lithics and heated sediments (24, 25), interpreted as the remains of hearths. Terra Amata (France) and Vérteszöllös (Hungary) also provide credible evidence of fire, but estimates of their age vary from MIS 11 to 9.

Of direct relevance are sites that do not have traces of fire before MIS 9–11. It could be argued that early open-air sites, such as those in the Orce region in Spain (26) or Isernia (27) and Venosa Notarchirico in Italy (28), have been affected by water transport, which removed traces of former fires. The primary context Acheulian site of Boxgrove (29) in England likewise did not yield any evidence for the presence of fire use, a few scattered charcoal particles aside. The absence of heated material from these and other Lower Paleolithic open-air sites could be ascribed to the short-term occupation character of these sites and/or their sedimentary history. However, open-air sites from comparable settings do yield traces of the former presence of fire from MIS 11 onward (e.g., Maastricht-Belvédère and Neumark-Nord 2; Dataset S1).

Human occupations without fire also occur at six Lower Paleolithic caves: Treugol’naya (Russia), Kozarnika (Bulgaria), Visogliano (Italy), Sima del Elefante (Spain), Arago (France), and Gran Dolina (Spain) (Dataset S1). The latter four sites also yielded hominin remains. The evidence from three sites, Treugol’naya, Kozarnika, and Sima del Elefante, is not yet sufficient to support a hypothesis of hominin habitation, however. To date, Sima del Elefante has yielded only 32 artifacts (21). The early layers of Treugol’naya and Kozarnika have a complex history of human and carnivore occupation. The first site has yielded a small number of artifacts [395 (30)]; faunal remains represent natural mortality or carnivore accumulations, and there is evidence of stream transport of both lithics and bones (31). The lowest levels at Kozarnika (32) have yielded a large number of artifacts, but contextual and taphonomic data are sparse. In contrast, the rockshelter of Visogliano has yielded 2,000 artifacts together with large numbers of cervids (33). Arago and Gran Dolina contain long sequences and large quantities of lithic and faunal remains, subjected to taphonomic analyses (34–36). Their settings are comparable to the ones that, in later times, have often provided strong evidence of fire, such as Bau de l’Aubesier, Grotte XVI, and Lazaret in France; Bolomor Cave in Spain (Dataset S1); and Middle Paleolithic/Middle Stone age caves in Israel and in South Africa. Traces of fire have been found in the upper part of the sequence at Arago, in layers younger than 350 ka. No charcoal, no burnt bones, nor any other evidence of fire have been reported from any of the assemblages from the lower levels (dated to MIS 10–14). No charred bones or heated artifacts have been reported from the Gran Dolina sequence (TD4–TD10). Rare charcoal particles have been found in thin sections of the TD6 sediments, but these sediments originate from the exterior of the cave, and there is evidence of low-energy transport (37); thus, the charcoal may not be in situ. However, the high density of human, faunal, and lithic remains as well as their state of preservation and refitting studies (38, 39) clearly indicate an occupation in situ with little postdepositional disturbance. The absence of any heated material from the long sequences of Gran Dolina and Arago, both documenting hominin occupations over several hundred thousand years (36, 40), is striking. This is a strong pattern, which can be tested by future work at other hominin habitation sites. We suggest that the European record displays a strong signal, in the sense that, from ∼400 to 300 ka ago, many proxies indicate a habitual use of fire, but from the preceding 700 ka of hominin presence in Europe, we have no evidence for fire use.

From around MIS 9, we find extensive traces of the presence of fire in cave settings as well as in the open (Dataset S1). The evidence from the multilevel site of La Cotte de Saint Brelade (Jersey, UK), for instance, shows that fire was extensively used throughout the MIS 7–MIS 4 occupation periods. There are indications of burning in every occupation layer here, with high densities of burnt material and a predominance of burnt bone over wood charcoal. Apparently, in addition to wood, bone was used as fuel as well (41). Spatial concentrations of heated flint artifacts and/or heated stones and charred faunal remains are a common phenomenon on Middle Paleolithic open-air sites all over Europe (Dataset S1).

It is also from the second half of the Middle Pleistocene onward that we can observe some spectacular cases of pyrotechnological knowledge in the production of hafting materials (42, 43), which some consider a proxy for complex, modern cognition (44) (Fig. 1 and SI Text). Significantly, several Middle Paleolithic cave and open-air sites have multiple successive levels representing a long time span with clear evidence of fire. Consider the 187 combustion structures in the many levels of Abric Romani in Spain, the 29 combustion structures of Ksiecia Jozefa in Poland, the multiple levels of El Salt, St. Marcel, Esquilleu Cave, Peyrards, La Combette, La Quina, St. Césaire, Oscurusciuto, and the fireplaces of Roca dels Bous (Dataset S1). Les Canalettes, France, is a well-documented Middle Paleolithic site where lignite was repeatedly imported as fuel from a distance of 8–10 km (45). Fig. 2 and Table 1 show a threefold increase in the number of sites with good evidence of fire during MIS 5 and a steady increase in MIS 4 and 3. Stone-lined or stone-delimited fireplaces are not as common as in the late Upper Paleolithic (46–49), but they have been documented at a number of Middle Paleolithic sites: Vilas Ruivas, Les Canalettes, La Combette, Bolomor layer XIII, Port Pignot, Abri du Rozel, Pech de l’Aze II, Grotte du Bison, and Abric Romani (Dataset S1). Hearth-centered activities have been suggested at the rock shelter site of Abric Romani (50), at the open-air sites of La Folie (France) and Ksiecia Jozefa (Poland) (Dataset S1), and in Western Asia at the site of Tor Faraj (51). We conclude that Middle Paleolithic Neandertals did not have to wait for lightning strikes, meteorite falls, volcanoes, or spontaneous combustion: they had the ability to make, conserve, and transport fires during successive occupations or at different sites, like ethnographically documented recent hunter-gatherers, a pattern comparable to that documented in the Upper Paleolithic.

A recent study provides evidence of early modern humans at the site of Pinnacle Point in Southern Africa regular use of heat treatment to increase the quality and efficiency of their stone tool manufacture process 164 ka ago (13). The authors infer that the technology required a novel association between fire, its heat, and a structural change in stone with consequent flaking benefits that demanded “an elevated cognitive ability.” They also suggest that, when these early modern humans moved into Eurasia, their ability to alter and improve available raw material may have been a behavioral advantage in their encounters with the Neandertals. However, this interpretation ignores that Neandertals used fire as an engineering tool to synthesize birch bark pitch tens of thousands of years before some modern humans at Pinnacle Point decided to put their stone raw material in it. In more general terms, a greater control and more extensive use of fire is sometimes (12) seen as one of the behavioral innovations that emerged in Africa among modern humans and favored the spread of modern humans throughout the world. As stressed by Daniau et al. (52), if extensive fire use for ecosystem management were indeed a component of the modern human technical and cognitive package, one would expect to find major disturbances in the natural biomass burning variability associated with and after the colonization of Eurasia by modern humans. In their study of microcharcoal particles from two deep-sea cores off of Iberia and France, spanning the 70- to 10-ka period of biomass burning, the authors did not recover any sign that Upper Paleolithic humans made any difference: either Neandertals and modern humans did not affect the natural fire regime, or they did so in comparable ways.

The European signal for fire use aligns well with what we know from other continents, where indications for habitual use of fire are present from the second half of the Middle Pleistocene onward. As in Europe, fire seems to have become a maintainable technology between ∼400 and 200 ka, as illustrated by the evidence from the cave site of Qesem in Israel. Here, micromorphological studies suggest that recrystallized wood ash was a major part of the cave deposits over a long time period and that, at least in the upper part of the sequence, repeated use of fire was common (17). Another signal for habitual fire use in the 300- to 200-ka period comes from a study of heated flints from the Tabun site, also in Israel (53). Many later Middle Paleolithic sites have yielded a wide range of evidence for fire, in the form of the large combustion areas from Kebara or the hearths from Hayonim (54, 55). There are a few claims for earlier use of fire in Asia, however, with the case of Zhoukoudian Locality 1 being the most contested one. Although this site was considered for a long time to be the fire-lit home of Homo erectus, studies by Weiner and colleagues (56, 57) showed that the “ashes” supposedly reflecting former hearths were not ashes at all: samples from Layers 10 through 3 showed extensive water deposition of fine silt-sized material (reworked loess), including fine-grained organic matter. The dark, organic-rich unit in Layer 10, often cited as some of the earliest evidence of fire, is in fact a water-laid accumulation. Much of the fine-grained sediment was derived from outside Locality 1. The 4- to 6-m accumulation of supposed ashes in Layer 4 is interpreted as subaerial water-laid silt deposits derived from the loess-covered hill slopes surrounding the site. With such a depositional environment, the presence of charred bone fragments in the sequence can hardly be considered as direct evidence for in situ burning (contra ref. 58).

There is however, one well-established case for earlier repetitive fire use in western Asia: the Acheulian site of Gesher Benot Ya’aqov (GBY) in Israel, dating to ∼780 ka ago (59–61). Most of its many find levels might be correlative with MIS 19, i.e., twice as old as the oldest sites with fire indicators in Europe. How strong is the evidence? GBY has both charred plant remains and heated microartifacts (≤2 cm) occurring in localized concentrations in various levels throughout the sequence. Five superimposed GBY assemblages contain clusters with frequencies of burned flint microartifacts of 3.7–5.8%. The spatial distributions and the frequencies of burned microartifacts are comparable to those occurring at much younger Magdalenian sites in Western Europe, where knapping near a fireplace was a common occurrence, leading to the accumulation of lithic debris near the hearth (62). The GBY excavators argue that burnt microartifacts and other charred materials are the best indicators of the position of formerly ‘invisible” hearths, better than the distribution of macroartifacts that are often rejected away from the combustion area.

For the African record, claims for early fire use from ∼1.6 million y ago onward have been made. These claims include 270 charred bones from Swartkrans (5, 63) and red patches of sediments from Chesowanja, Kenya, and from Koobi Fora FxJj 20, thought to be the oxidized remains of 1.5- to 1.6-million-y-old hominin fires (3, 4, 64, 65). These sites do demonstrate that fire-altered materials are associated with early hominin activity areas but not that hominins were involved in the production and/or the use of such fires. When comparing the records of Africa and Europe, one also has to envisage that natural fires were very probably a much more important phenomenon over much of Africa than at northern latitudes. More lightning occurs near the equator than near the poles because of differences in the degree to which daily sunshine heats up the land surface. A recent satellite study shows that ∼78% of all global lightning occurs between 30°S and 30°N latitudes, which encompasses most of the African continent (15).