Because the archaeological and palaeontological remains are encased in a hard carbonate breccia, excavation had to be carried out with demolition hammers and sieving was not possible. However, much effort was made to collect all the remains. Stone tools and faunal bones were mapped in situ prior to extraction, while small fragments were bagged by 1 m² units of provenience. The chunks of sediment extracted with the power tools were collected and then broken into small pieces in a separate area. All the finds were restored during the excavation using a pneumatic microhammer and microchisel (Mod. CTS 178) and a normal air scribe (Mod. W 224). Where necessary, bones were consolidated with an acrylic resin (Paraloïd-B72) dissolved with acetone.

Archaeological samples were analysed macroscopically using a stereomicroscope (Olympus SZ61 up to ×45 with a coupled photographic camera Olympus SC30), and geological samples were observed under the stereomicroscope and a petrographic microscope (CARL ZEISS Axiophot Pol up to 200× with coupled photographic camera Sony DXF-S500). Details of the data and methodology used for the analyses of the geological flint and other silicifications collected in the vicinity of the cave have been published elsewhere (Aubry et al., 2016, Aubry et al., 2012, Aubry et al., 2014, Matias, 2016 and Matias, 2012). Macroscopic, non-destructive analyses considered the main distinguishing elements, including texture (Dunham, 1962), sedimentary structures, skeletal and other bioclastic elements, porosity and non-skeletal elements, weathering, cortex type, degree of cortex rounding, and knapping quality. Microscopic analyses of the geological samples also considered the mineralogical content. Metamorphic rocks and minerals, for example quartzite and quartz, were classified according to their colour, transparency, grain size, weathering, cortex type, degree of cortex rounding, and knapping quality.

To evaluate the state of preservation of the lithic artefacts, we conducted an analysis of their surfaces and edges with a portable magnifying lens (10×). In order to be as cautious as possible, the analysis of edge preservation only used unretouched flakes of quartzite and flint. Quartz examples were excluded due to the difficulty in clearly discerning between wear-, excavation- and taphonomic-related edge damage.

A techno-categorical analysis was conducted on the entire lithic assemblage using the operational sequence approach (Barsky and de Lumley, 2010, Boëda et al., 1990, Forestier, 1993, Mourre, 2004, Mourre, 1996 and Tixier, 2012). Unretouched flakes and tool blanks were classified according to technological criteria and retouched tools according to the general characteristics of the retouch. Cores and bifaces were described with the help of diacritical schemes. Cobblestone sites were analysed for the identification of anthropogenic marks (of percussion, heating, etc.).

A total of 27 artefacts was selected on the basis of the presence of macroscopic edge damage potentially attributable to tool use (micro scars, fractures, edge rounding) and detectable at low magnification (up to 80×). Because all the artefacts analysed are made of quartzite, high-resolution casts of the tool edges were made to assess their functionality. This method has proved to be highly effective in the analysis of such coarse, highly reflective rocks (Igreja, 2009). The functional interpretation relied on a multi-stranded approach combining the analysis of macroscopic edge damage with high-power microwear analysis, using a standard reflected light microscope (BH Olympus) equipped with differential interference contrast (DIC) at 100×, 200×, and 400× magnification, following standard use-wear analytical procedures (González and Ibáñez, 1994). Photomicrographs of the use-wear traces detected were taken using a digital camera Canon EOS 600D. The patterns observed on the archaeological materials were compared with an experimental reference collection of use-wear traces on a wide range of rocks (e.g., flint, quartzite, quartz).

The approximately 6 m³ of layer Xb/c that we excavated yielded 393 lithic objects, which translates into an artefact density of 65/m³ (51/m³ considering the flaked material only); by weight, the corresponding values are, respectively, 7.2 and 3.2 kg/m³. A total of seven different types of igneous, sedimentary and metamorphic rocks and minerals is present in the sample (Table 1).

Limestone, lydite and an undetermined igneous rock represent less than 2.5% of the artefacts. Although available nearby, flint is relatively scarce in our sample, where it represents 6.5% of the artefacts, against 51.6% for quartzite and 39.6% for quartz. An anhedral milky quartz variety presents varying degrees of translucency, apparently unrelated to knapping quality. A very fine-grained, green or red variety with excellent knapping qualities represents almost 20% of the quartzite assemblage. Quartzite and quartz cobbles derived from the Iberian Hercynian Massif are found in the sedimentary deposits of the Tagus basin ubiquitously present around the Almonda spring, in which this variety is, however, rarely observed (Fig. 2).

Among the unflaked cobbles, quartzite and quartz appear in the same percentage as among the artefacts, flint is absent, and sandstone (which may be either an orthoquartzite or a slightly metamorphosed quartzite) is present. One relevant characteristic of these finds is the recurrent presence of fractures and other imperfections that would make knapping difficult. Although it cannot be excluded that some of these cobbles represent a component of the surrounding soils washed-in by natural processes, most are manuports that bear stigma of use (Table 2).

Due to a high level of desiliconization or necrosis (Bressy, 2003 and Masson, 1981), the flint is physically fragile, opaque, and white in colour (Fig. 3.1), with rose-red and grey zonations and frequent, isolated iron oxide pellets. When present, the external surface is well-rounded and features a thin (∼1 mm) neocortex with iron precipitates (attributable to the iron-enriched soils of the Tagus basin; Fig. 3.2). In a few artefacts, it was possible to identify fossils, which, because of recrystallization, are generally difficult to recognise. Monoaxonic spicules are frequent, as are other undetermined bioclastic fragments; rare and poorly preserved foraminifer, ostracod, algae and bivalves are also present. Euhedral macro-quartz crystals replace the original carbonate of the fossil shells, providing for good contrast with the siliceous, cryptocrystalline sedimentary matrix (Fig. 3.3–3.8). Even though geode formations are frequent in Cenomanian flints, no relevant imperfections were observed in our sample.

Four types of flint have been identified in the vicinity of the Almonda system (Aubry et al., 2014, Matias, 2016 and Matias, 2012) (Fig. 2). Bajocian (Jurassic) formations < 5 km from the cave contain medium-to-good quality flints in primary and sub-primary positions. More distant Upper Cenomanian (Cretaceous) formations contain very good-to-excellent quality flints, while Tertiary silcrete occurs in secondary position in lower Miocene deposits > 10 km to the SW, alongside Hercynian rock cobbles (quartzite, quartz, etc.). Good quality Oxfordian (Jurassic) flint in primary and sub-primary positions is documented some 20 km to the NE. Our sample, however, only contains Upper Cenomanian flint types. Their nearest known area of occurrence is indicated in Fig. 2.

Flint artefacts present an extensive patina, ranging from a whitened exterior surface (36%) to partial or total desiliconization (12%), but quartzite lithics are well preserved: their dorsal ridges show little crushing, while edge damage caused by syn- or post-depositional processes, or resulting from the use of power tools in the excavation process, found on 70% of the pieces, is relatively limited (28% of pieces show some, 19% show more extensive but localized, and only 23% show generalized edge damage). Rolled surfaces suggesting significant displacement or washing-in are found among neither quartzite nor flint pieces. Nine quartzite artefacts (eight cobbles and one flake) have a reddish surface possibly indicative of damage inflicted by thermal alteration; however, naked-eye observation is unable to provide definitive confirmation. These observations show that the assemblage did not undergo substantial post-depositional alteration, and this conclusion is supported by the preservation of micro-wear traces (cf. infra).

Bearing in mind the small size of the 2013–2015 trench, the vertical and horizontal plotting of the archaeological remains suggests a random distribution across the total thickness (∼1 m) of our excavation of layer Xb/c. This can be clearly seen in Fig. 4: neither the horizontal (Fig. 4.1) nor the vertical plots (Fig. 4.2–4.4) show any signs of discrete accumulations as lenses or patches, even though the lithics seem to be more numerous towards the top and the base of the unit. Only two flakes could be refitted.

As a rule, edges and surfaces are well preserved, facilitating use-wear analysis in reliable analytical conditions, in particular with regards to the identification of microscopic use-wear traces. Of the 27 artefacts examined (11 bifaces, 15 flakes and one core), 16 presented recognizable use-wear evidence (Table 6 and Supplementary Figures 1–6). The nature of the worked material can be inferred for five bifaces, four flakes and the core.

In some cases, the degree of development of the use-wear traces and their features were insufficient to securely determine either the nature of the worked materials or that of the working motions. For four bifaces and two flakes we could only make inferences about the hardness of the contact material — these pieces present use-wear associated with work on hard materials, such as wood, antler or bone — and for two bifaces and one flake it was not possible to determine the type of working motion. Where it was possible to identify worked materials and motions, use-wear traces are mostly suggestive of wood processing activities, and of several types of motion: percussion, longitudinal, and transversal.

Tasks potentially related to percussion on wood are observed on two bifaces (Fig. 6.8, 7.1) in which polish is associated with crushing and abrasion of the used edge (Fig. 9.8 and 9.11–12). A further three bifaces (Fig. 6.2, 6.9) present use-wear on their lateral edges indicative of wood cutting (Fig. 9.3–9.4, 9.9–9.10). Two bifaces (Fig. 6.3–6.4) present evidence of the processing of hard materials; for these two artefacts the working motion remains undetermined (Fig. 9.5–9.7).

Five flakes used for woodworking present numerous scars. Two of them present scars along both sides of the active edge, consistent with longitudinal motions such as cutting/sawing (Fig. 9.16). Three other present scars along just one side of the used edge (Fig. 7.4, 7.7–7.8), typical of a transversal motion such as scraping (Fig. 9.15, 9.17–9.19). For one flake presenting wood polish (Fig. 7.12), the nature of the gesture could not be inferred (Fig. 9.22–9.23). Two flakes (Fig. 6.7.3, 7.11) presenting scars on both sides of the used edges were employed to cut hard materials (Fig. 9.13–14 and 9.20–9.21).

One core (Fig. 5.1) was examined to determine whether it would also have been used as a tool. On a section of the edge, it revealed evidence of contact with wood in a scraping motion (Fig. 9.1–9.2). This evidence suggests that cores could have been opportunistically used as tools.

With the exception of one biface, which presents evidence of use on various edges, only one edge was used per tool. Rots and Van Peer (2006) and Beyries (1987a) have reported hafting in late middle Pleistocene and Middle Palaeolithic contexts but, given the size of the stone tools in our assemblage and the fact that no traces of the practice were found, it seems most likely that they were hand-held.

Gruta da Aroeira is one of the few middle Pleistocene cave sites known in Europe that contain a well-preserved and in situ Acheulean toolkit as well as faunal and human remains. All other Acheulean sites in the Atlantic façade of Iberia are found in fluvial or lacustrine contexts, mostly river terraces and palaeolakes (Cunha-Ribeiro, 1999, Méndez-Quintas et al., 2008, Raposo and Santonja, 1995 and Santonja and Villa, 2006). Most of these open-air sites have not been directly dated, but recent studies conducted on the terraces of the Tagus and Douro rivers have begun to provide a reliable chronological framework (Cunha et al., 2016 and Silva et al., 2016).

In other parts of the Iberian Peninsula, most sites are also found in fluvial contexts with varying degrees of preservation. Some of the better preserved sites are located in low energy deposits. This is, for example, the case with Torralba, Ambrona and Áridos (Santonja and Pérez-González, 2006, Santonja and Querol, 1980, Santonja and Villa, 2006 and Villa, 1990). Karst sites containing industries of a similar age are rare in SW Europe: Galería and Gran Dolina in Atapuerca (García-Medrano et al., 2015), Cueva de Santa Ana (Carbonell et al., 2005), Cueva del Angel (Barroso Ruíz et al., 2011), and Caune de l’Arago (Barsky, 2013). While Arago, Santa Ana and Cueva del Angel functioned as living spaces, Galería is a natural trap accessed by both humans and carnivores to exploit herbivore carcasses.

The Gruta da Aroeira lithic assemblage is broadly in line with those reported from other Iberian middle Pleistocene industries (Santonja and Pérez-González, 2010). The provision of raw materials is local in the case of quartz and quartzite, while flint artefacts come from > 10 km away and were imported as retouched tools or bifaces. Centripetal cores with a bipyramidal organization, sometimes set-up on flake blanks, and the unipolar, on-anvil reduction are also found at coeval sites (Barsky, 2013 and García-Medrano et al., 2015). As elsewhere, we found no Levallois flakes or cores. Whether the small size of the bifaces relates to functional considerations, raw-material economy, or deliberate choice, remains to be clarified, but suffices to exclude comparison with the LFA (Large Flake Acheulean) facies as recently described in Galicia (Méndez-Quintas et al., 2018).

A particular aspect of the Gruta da Aroeira assemblage, rarely documented in middle Pleistocene industries, is the proximal thinning of scrapers. Marks et al. (2002) also provide examples of bifacial preparation or sharpening of scrapers, following a method well-known in the Keilmessergruppen of central Europe, but we found no comparable items in our much smaller excavation. Given the limited number (eight) of such items and bearing in mind the difference in the size of the excavated areas (60 m² in 1997–2002 vs. 6 m² in 2013–2015), sampling bias is the parsimonious explanation for this discrepancy.

García-Medrano et al. (2015) have recently proposed an evolutionary scheme for the technological characteristics of lithic industries at the site of Galería (Atapuerca). By comparison, the Gruta da Aroeira industry would seem to be most similar to that in Galería's level GIIIa, in which, however, a smaller proportion of bifacial tools (< 5% of the artefacts) is observed. In both contexts, the bifaces are imported as finished tools. However, there is one striking difference: neither in our sample nor in that from the 1997–2002 work (Chabai et al., 2001 and Marks et al., 2002) have flake cleavers been found at Gruta da Aroeira.

In the Caune de l’Arago sequence, technological change through time differs from that described in Galería. For example, bifaces, numerous and finely worked in level P (assigned to MIS 14), are absent from levels K–H, and reappear in level G, but with more irregular forms (Barsky, 2013). Centripetal and Clactonian flaking increase in level G, which also sees a diversification of tool types. With a mean age estimate of 438 ± 31 ka, this Caune de l’Arago level G is broadly coeval with Gruta da Aroeira.

In the Iberian Acheulean, the use of large flakes as blanks for bifaces and cleavers has been interpreted as representing diffusion from the North African Acheulean via the Strait of Gibraltar (Santonja and Pérez-González, 2010 and Sharon, 2011). However, that use is variable, as can be seen in the Galería sequence, in the basal levels of which bifacial tools are exclusively made from cobbles. According to García-Medrano et al. (2015), the use of large flakes as blanks in the upper levels can be considered a successful adaptation to local raw materials. The technological decisions taken by Acheulean populations would certainly have been the result of multiple environmental, functional, and cultural factors, and examples of local adaptation to a specific mineral environment are documented at Grotte de l’Observatoire, where the large flake technology is recorded (Porraz et al., 2014). Indeed, technological convergence rather than diffusion has been proposed to explain aspects of the recently described reduction sequences of Victoria West and Tabelbala-Tachenghit (Lycett, 2009b, Mourre et al., 2016 and Sharon and Marder, 2016).

The absence of flake cleavers in the Gruta da Aroeira assemblage is remarkable given that their distribution usually coincides with that of bifaces. Santonja and Villa (2006) claim, however, that their presence is more common in regions with abundant large quartzite cobbles, while Santonja and Pérez-González (2010) and Sharon (2011) argue that their distribution in Europe may be related to migrations from Africa. Flake clavers are present in the fluvial terraces of the Tagus (Cunha-Ribeiro, 1999), and they are found in the late MIS 5, Middle Palaeolithic levels of Gruta da Oliveira, another site of the Almonda karst system (Deschamps and Zilhão, 2018). The absence of flake cleavers in the Gruta da Aroeira assemblage thus cannot be explained by raw material constraints and, given the area and volume of the Gruta da Aroeira excavation, it cannot be explained by sampling bias either.

It has been suggested that cleavers were manufactured for a specific function (Freeman, 1969 and Freeman, 1966), while other authors propose that they were used in a wide range of activities (de la Torre et al., 2014 and Rios-Garaizar, 2012). The few use-wear analyses of cleavers point to power-hitting and chopping actions in the case of Acheulean specimens (García-Medrano et al., 2014 and Ollé, 2003), and to cutting or percussion in heavy-duty activities in the case of Middle Palaeolithic ones. Experimental analyses have shown that cleavers are likely to have been hafted and used to fell trees and split carcasses (Claud et al., 2015). Since the overall features of the archaeological context suggest that Gruta da Aroeira was a living place, it is unlikely that such kinds of activities were not carried out in the context of the occupation of the site. Hence, the lack of cleavers in the stone tool assemblage can hardly be explained by functional factors either.

At this stage of our research, the parsimonious explanation would therefore seem to be that cleavers were not in use during the phase of the Acheulean tradition represented in the Gruta da Aroeira assemblage. This hypothesis is in line with the fact that recent dating work conducted at Galería suggests that its Acheulean industries are more recent (between ∼363 and 220 ka) (Demuro et al., 2014 and Falguères et al., 2013) than Aroeira's.

The Gruta da Aroeira assemblage features a level of raw material diversity comparable to that seen in the Middle and Upper Palaeolithic sites of the Almonda karst system (Aubry et al., 2016, Aubry et al., 2014, Gameiro et al., 2008 and Matias, 2016). This diversity mirrors the petrographic composition of the siliciclastic deposits of the Tagus basin, from where raw material was procured, as clearly illustrated by the well-rounded cortex surfaces associated with an epigenetic mineralization attributable to the basin's iron-rich soils (Fig. 3.2).

However, the exclusive use of Cretaceous flint (and possibly of silcrete, found in the same secondary deposits and, therefore, in the same areas) contrasts with the flint procurement patterns seen in the Middle and Upper Palaeolithic sites of the Almonda system. Both at the Magdalenian site of Lapa dos Coelhos and the Mousterian site of Gruta da Oliveira, local and regional Jurassic flints (Bajocian and Oxfordian), as well as Cretaceous flint, were employed, albeit in different proportions. The same pattern is also observed at the regional level. For instance, at the Gravettian open-air site of Terra do Manuel, located close to Cretaceous flint sources, Jurassic flint sources lying at distances of up to 40 km (Bajocian) and > 60 km (Oxfordian) as the crow flies are also represented (Aubry et al., 2012).

Thus, while the Mousterian and Upper Palaeolithic occupants of the Almonda system procured their flint in both the lowlands of the Tagus basin (containing quartz, quartzite, silcrete and Cretaceous flint) and the highlands of the adjacent Limestone Massif (where the Jurassic flint sources are located), the Gruta da Aroeira Acheuleans only used sources in the former. In Portugal, a pattern of flint being little used and mostly imported in the form of retouched tools tends to be found whenever it occurs neither locally nor regionally and has to be imported from distant territories, as exemplified by a number of Upper Palaeolithic sites (Aubry et al., 2016, Aubry and Mangado Llach, 2003a, Aubry and Mangado Llach, 2003b, Aubry and Mangado, 2006, Aubry et al., 2004 and Aubry et al., 2012). This is not the case with Gruta da Aroeira. Here, given the site-to-source distances involved (< 10 km), the pattern suggests that raw material procurement was embedded in daily subsistence activities and that these activities remained restricted to the lowlands of the Tagus plain (Binford, 1980 and Binford, 1979).

Lower Palaeolithic heavy-duty tools such as bifaces have frequently been linked with woodworking tasks in which they would have been used as adzes or shaving tools (Anderson-Gerfaud, 1990, Aranda et al., 2014, Beyries, 1987a, Beyries, 1987b, Beyries, 1988, Beyries, 1993, Hardy and Moncel, 2011, Márquez et al., 2001, Ollé and Vergès, 2008, Ollé et al., 2014 and Rots, 2013). In the Middle Palaeolithic, however, bifaces were used primarily for butchery (and, occasionally, as retouch hammerstones), as shown at the site of Jonzac (Claud, 2012); in this later period, therefore, bifaces do not seem to have been functionally distinct from unretouched flakes, whose sharp edges have been linked to cutting motions, in particular to the processing of soft materials during butchery (Claud, 2015, Clemente-Conte et al., 2015 and Márquez et al., 2016).

Bearing in mind the limited size of the assemblage, our preliminary results for the Gruta da Aroeira bifaces are consistent with the Acheulean rather than the Middle Palaeolithic pattern, even though, dimensions-wise, our sample resembles that latter because of the many small specimens. The Gruta da Aroeira bifaces were also used in woodworking tasks, and the range of activities otherwise identified at Gruta da Aroeira by use-wear analysis is also dominated by wood processing. However, the faunal remains are anthropogenic, so some of the tools for which the material worked on could not be diagnosed beyond “hard” may well have been used on bone.

The fragmentation of the faunal remains, the presence of cut marks in the bone assemblage, the number of stone tools retrieved and the identification of on-site reduction sequences suggest that the cave was used as a residential space and suggest continuous occupation of the area through the time frame represented by the deposit.

The paleoanthropological and archaeological records are inherently noisy: the uncertainty intervals of dating methods are wide, and the units of stratigraphy we work with are for the most part palimpsests compressing unknown amounts of time and unknown amounts of behavioural diversity. This situation complicates any discussion of possible links between human biology, hominin taxa, cognitive capacities, stone tool technology, symbolic behaviour, and the consumption of natural resources. Even so, many scholars have proposed that, at several key-points in the human evolutionary process, the changes or innovations seen in the archaeological record go hand-in-hand with changes in the biology, or the identity, of its makers. Examples thereof concern the emergence of stone tool making in Africa (e.g., Harmand et al., 2015 and McPherron et al., 2010), the human taxonomy of the bearers of the Acheulean tradition (McPherron, 2000, Stout et al., 2014 and Stout et al., 2015), or the emergence of the Upper Palaeolithic in Europe (Benazzi et al., 2015 and Peresani et al., 2016).

Based on current evidence, bifaces did not appear in Europe until 550–650 ka, and widespread, continuous occupation of the European continent probably did not happen until that time. Several hypotheses have been constructed to explain this watershed (Hublin, 2009). For example, Carbonell et al. (1999) suggest that the arrival of the Acheulean in Europe can be related to demographic growth in the Rift Valley regions of Africa. This would imply either the arrival in Europe of populations of African origin making the so-called “Developed Mode 2” industries, or the establishment at this time of regular channels of cultural diffusion linking Europe with Africa. Based on the Italian evidence, Villa (2001), however, argues against two different migration events at the origins of the human settlement of Europe (i.e. initially, by bearers of the Oldowan, later on by bearers of the Acheulean), and claims that the main reason for the variability seen in European Lower Palaeolithic industries lies in the adaptive flexibility of hunter-gatherer populations.

More than 80 years since the migration vs. diffusion dichotomy emerged in the context of attempts to explain the difference between the Lower Palaeolithic industries of central Europe and the Russian Plain (without bifaces) and those of western Europe (with bifaces), the debate is still far from being resolved (Lycett and Bae, 2010, Movius, 1944). Due to the similarities between North African and western Mediterranean assemblages, most authors have supported an African origin for the Acheulean of Europe, leading to the proposition that this technological complex reached southern Europe via the Strait of Gibraltar, despite the absence of any direct evidence that such a crossing was feasible or did occur (Bar-Yosef, 2006, Goren-Inbar and Sharon, 2006, Santonja and Pérez-González, 2010, Santonja and Villa, 2006 and Sharon, 2011). Others have argued that the handaxe was reinvented in Europe (Nicoud, 2013, Soriano, 2000 and Villa, 2001), and that cross-Mediterranean contacts remain unproven until recent prehistoric times (Derricourt, 2005, Muttoni et al., 2010 and Straus, 2001).

The relationship between the Acheulean and the putative dispersal into or local evolution of H. heidelbergensis in Europe is another point of contention. The middle Pleistocene archaeological record of Europe gives no indication of major movements of people, let alone of massive African immigration. The physical anthropological evidence suggests that European populations of the period were morphological diverse, or that, at this time, humans from both Africa and Europe formed a single clade (Rightmire, 1998 and Stringer, 2012).

The number of European sites in which clearly diagnostic human remains are associated with Acheulean lithics being scarce (Arsuaga et al., 2014, de Lumley, 2015, Falguères et al., 2004, Pereira et al., 2015, Stringer and Hublin, 1999, Stringer et al., 1998 and Wolpoff, 1980), this issue remains unsettled. As discussed by Daura et al. (2017), however, it has nonetheless been possible to divide the human fossils of the middle Pleistocene of Europe into two groups: one, represented by Atapuerca/Sima de los Huesos (SH) and Swanscombe, belongs to the Neanderthal clade (Arsuaga et al., 2014, Meyer et al., 2016 and Stringer, 2012); the other, represented by the Arago hominins, has been attributed to either an incipient stage of Neanderthal evolution (Dean et al., 1998), to H. heidelbergensis (Stringer, 2012), or to Homo erectus tautavelensis (de Lumley, 2015).

The Aroeira-3 cranium (Daura et al., 2017) adds complexity to this picture because of its contrasting mosaic of features, suggesting that human variability was at this time significantly greater than envisaged by current models. More to the point, the Acheulean technology is associated with the two groups of middle Pleistocene Europeans represented by the Arago and Sima de los Huesos fossils, as well as with the Gruta da Aroeira specimen. This evidence suggests that the middle Pleistocene technological innovations associated with the Acheulean — controlled use of fire, prepared core reduction, projectile weaponry — spread across the European continent irrespective of human palaeontology's perceived taxonomic boundaries. In addition, the Acheulean contexts that have been considered to present a seal of African origin (e.g., the Galician sites reported by Méndez-Quintas et al., 2018) are significantly later than Gruta da Aroeira, which predates them by > 100,000 years and lacks the corresponding indicators (large bifaces made on large flake blanks, presence of flake cleavers). The Gruta da Aroeira evidence is therefore inconsistent with models postulating African migration through the Strait of Gibraltar as the explanatory mechanism for the emergence of the Acheulean in Europe.