The exploration of Arubo in central Luzon conducted in 2001 has again brought up the issue of the limitations of classification and technology for open sites (Pawlik, 2009). Arubo is situated approximately 80 km north of Manila and 300 km south of the Cabalwanian sites of Cagayan in a pasture landscape at the foothills of the Sierra Madre mountain range (Pawlik, 2002 and Pawlik, 2004). Upon its discovery, the site was heavily disturbed by the dredging of a fishpond (Fig. 2). However, that activity brought back the archaeological material to the surface, thus leading to their discovery. Arubo is characterized by the presence of large chert boulders from a Miocene limestone formation that once covered large parts of Luzon, which served as the raw material for the majority of the artefacts. During the 2001 fieldwork, 200 lithic finds were made at Arubo. At least 18 of them could be identified as modified stone tools, mostly cores. Most artefacts were recovered during surveys on the surface, few were found during test excavations on top of a sandstone bedrock. The morphology of the artefacts, especially a handaxe, bifacial cleaver and several pebble tools seems to point to a Lower Palaeolithic age of the site. Also a so-called horsehoof core, similar to those found in Java associated with the Lower Palaeolithic Pacitanian industry (van Heekeren, 1957 and von Koenigswald, 1936), is among the several cores recovered at Arubo (Fig. 3 and Fig. 4). However, no directly datable material could be retrieved.

Arubo provided an assemblage different from the stereotyped “chopper/chopping tool industries” (Movius, 1944). Unlike the Philippine Cabalwanian and also the Pacitanian in Indonesia, it incorporated bifacial technology. Evidence of curation is present, as well as variation in core preparation and core reduction. The artefacts exhibit traces of multiple uses and function. In a functional study of the Arubo artefacts, Teodosio noted use traces from working soft and hard materials of organic and inorganic origin, including bamboo. Not just the handaxe but also some of the used flakes and core tools show signs of multiple uses on various contact materials (Teodosio, 2006). Consequently, Arubo represents a technological concept disctinct from smash-and-grab or expedient strategies common in the region (Mijares, 2002).

Tabon Cave is located at Lipuun Point on the southwestern coast of Palawan. It is currently the most important site for understanding the peopling of the Philippines. Excavations of the National Museum of the Philippines in 1962, 1965 and in 2000 (Dizon et al., 2002, Fox, 1970a and Fox, 1970b) identified two levels of occupancy during the Pleistocene. Human fragments, faunal remains and over 3800 artefacts were discovered (Corny, 2008, Fox, 1970a and Fox, 1970b). Human fossils were found in levels II and III. Indeed, four fragments of jaw and frontal bone, which belonged to three different individuals, were found between 50 and 88 cm depth. Two of them seemed to belong to the same individual (Macintosh, 1978). The other two, however, differ markedly. Recent studies of these fossils reveal that two different modern humans lived at the same time (Détroit, 2002, Détroit et al., 2004 and Dizon et al., 2002). The age of these individuals, which was first associated with the dating of level III, 22,000–24,000 BP, was more recent for the frontal bone with 16,500 ± 2000 years BP and especially for the older jaw (31,000 BP ± 8000–7000 years) and a tibia fragment discovered in 2000 (47,000–10,000 BP ± 11,000 years) (Dizon et al., 2002).

The lithic material from Tabon Cave was divided into five assemblages (flake tool industry I to V). According to Fox, it was an industry of more than 19% chert flake tools, produced by direct percussion with a hard hammer or in some cases by percussion on an anvil. Only 1% of the blanks have been retouched including kuba-scrapers (Fox, 1977). The technological analysis that we conducted on the material of the layer dated to 9,500 years BP, and the study of hundreds of specimens discovered in 2001 relativized the initial results (Jago-on, 2006, Orogo, 2001, Patole-Edoumba, 2002, Patole-Edoumba, 2009 and Patole-Edoumba et al., In press). Chert (probably red jasper) has been used for more than 80% of the artefacts but also andesite occurs as raw material. It is likely that the Panitan and Malatgao rivers, located c. 7–8 km away from the site, have been the main raw material sources (Schmidt, 2008). The selection of the best samples was made at the source by fracturing the rock to assess its potential. Blanks are elongated with laminar flakes (42.3%) and sometimes narrow blades (20.6%) appearing, mainly with thin cross-sections (51.6%). The mean dimensions of the flakes at Tabon Cave are 25.1 mm length, 18.7 mm width and 6.4 mm thickness. Macroscopically visible use traces are located in most cases, along one edge only (90.3%). The angles of active edges are between 40–59° (49.5%) then between 30–39° (15.9%). When the delineation of the edge is straight, the edge angle is usually regular (40–49°) or greater than 70°. When it is convex, the angle is between 40° and 59° (Fig. 5 and Fig. 6). The identification of micropolish on several stone tools discovered in 2001 showed that they were used for the manufacturing of wooden implements (Mijares, 2004).

A number of blanks were retouched into borers, scrapers, notches or composite tools (e.g. scraper-burin). The retouching increased the edge angles to prepare the tools for a particular use. The knapping technique applied is usually direct percussion with a hard hammer. Nevertheless, it appears that occasionally a soft hammer (limestone or animal bone) had been used as well. The knapping process was short with no more than two or three sequences. This method fits into the algorithmic type system of alternating debitage surface (SSDA). Here, during each sequence the previous debitage surface is used as a striking platform for new removals (Forestier, 1993). This technological system has been identified on other sites of the same and later periods, e.g. in Musang Cave in northern Luzon (Patole-Edoumba, 2002).

The karst formation overlooking the Pinacanauan River, a tributary of the Cagayan River, near the city of Tuguegarao, is home to the Musang Cave remains, older than 12,000 years. Two successive excavations during the 1970s succeeded in identifying a stratigraphy over a depth of two meters with two occupational levels. The older level contained a lithic assemblage associated with a shell midden, composed mainly of freshwater snails (Thiara scabra) from the river and animal bones (wild pigs, deer and fish). Thiel considered that the site had been used as a temporary settlement by hunter-gatherers approximately 10,500 years ago (Thiel, 1978). Different traceological and technological studies on the lithic material provided some information about prehistoric activities (Barbosa, 1979, Patole-Edoumba, 2002 and Thiel, 1978). The tools were mostly manufactured from andesite (60%), a raw material available at the Pinacanuan river bed, followed by a chert (32%) distributed all over the place, Fig. 7).

The majority of the blanks are thick (55%), elongated and laminar flakes (47%) with an average length of 26 mm. They were used without any further modification of the edges. Straight and convex lateral edges often have angles above 50°. The functional analysis of the stone tools showed that they were used for scraping and cutting wood, bamboo or bone. Only 2.4% of the entire assemblage comprises retouched flakes. Retouching was applied to produce more specific tools such as notches and denticulates or a nosed-end scraper. In some cases, some cores and raw materials were also used for the natural properties of some of their edges.

The knapping technique was again identified as direct percussion with a hard hammer. With regard to the limited access to a better quality raw material, the orientation of the reduction shows a maximum use of the raw nodules although the knapping sequence remains very short with usually two sequences with four removals each.

The analysis of the flaked industries from Tabon and Musang Cave shows that tools forms are not haphazard products but correspond to specific techno-types, linked to an identifiable recurring knapping process.

Several islands have provided stone implements that were attributed to Palaeolithic cultures. The first discoveries were made in the Cagayan Valley, northern Luzon. During the 1950s, when Ralph von Koenigswald worked on several Southeast Asian sites and assemblages such as the Pacitanian in Java, he decided to conduct surveys at Espinosa Ranch in Cagayan Valley where fossils of a Pleistocene megafauna, including stegodon and rhinoceros bones, had been discovered several years ago by H. Otley Beyer. He collected a set of modified pebbles and fragments of quartzite and sandstone on the surface which he presented as a new lithic tradition named “Cabalwanian” after the Cabalwan region where Espinosa is situated. Over a decade later, the National Museum of the Philippines organized several archaeological campaigns in Cagayan as a research project on the first Filipinos. Fox and Peralta made a systematic survey of the Cagayan Valley. Of the sixty-eight sites recorded in the region of Kalinga-Apayao (west of the Cagayan Valley), twenty-one were identified as knapping workshops or butchering sites, belonging to the Pleistocene Awidon Mesa formation. These test pits were supplemented by the excavation of three open sites: Espinosa 1–4, Madrigal Wanawan 1 to 12 and 1, 3 and 5 (Bondoc, 1979, Fox, 1978 and Fox and Peralta, 1974). This fieldwork resulted in a collection of one hundred stone tools and animal fossils. A selection of sixty-three tools was used to create a classification (Fig. 8). Two more campaigns in 1976 and 1980 complemented the corpus of pebbles and flakes (Shutler and Mathisen, 1979; National Museum paper, 1981). All assemblages were dated only by typological comparison with other industries known in Southeast Asia even though this was a highly unreliable method. Von Koenigswald associated the pebbles with the Middle Pleistocene faunal remains found in the same area and concluded that they were produced by Homo erectus around 400,000 to 500,000 BP, contemporaneous with the finds from Java and Choukoutien in China (von Koenigswald, 1958). For von Koenigswald the choppers were made by Homo erectus around 400,000 to 500,000 BP because their features were very similar to those from Choukoutien in China. The presence of extinct animal fossils was used as another argument for such an old age. Von Koenigswald's hypothesis was then taken over by Fox and Peralta who insisted on the status of a cul-de-sac of the Cagayan Valley which would have resulted in the concentration of extinct species such as elephant, Stegodon, buffalo and rhinoceros from China (Fox and Peralta, 1974). However, there is at present no site in Cagayan (or elsewhere in the Philippines) where lithic artefacts and Middle Pleistocene faunal remains have been found together and in context. The assumed old age of the Cabalwanian remains, therefore, entirely speculative (Dizon and Pawlik, 2010). Furthermore, a taxonomic comparison is insufficient to establish a date of lithic material without any stratigraphic position and absolute dating, especially since a working classification system is lacking for the entire Southeast Asian region. Moreover, taxonomic analysis shows only morphological differences whereas a techno-functional approach addresses the process by which a tool is created.

The island of Luzon has yielded other sites and artefacts outside the Cabalwan region. In the 1970s, Peterson identified several choppers, chopping tools and flakes in an assemblage collected in 1926 by Beyer during the building of a dam for Manila's water supply (Beyer, 1947, Peterson, 1979 and Ronquillo, 1998). More recently, the open site of Arubo in central Luzon delivered a lithic assemblage including a bifacial handaxe and a cleaver (see above). In their paleontological surveys in Solana (Cagayan) and on the island of Cabbaruyan (in Pangasinan), Bautista and De Vos discovered Pleistocene fossil remains in association with flakes and sharpened pebbles that they compared to Cabalwanian (Bautista and de Vos, 2002a and Bautista and de Vos, 2002b).

Other stone tools with a Lower Palaeolithic “appearance” come from the islands of Palawan and Mindanao. In Palawan, a dolomite handaxe was collected in a secondary deposit of Ile Cave, in El Nido (Fig. 9; Archaeological Studies Program, 2007: 13; Dizon and Pawlik, 2010). On the northern coast of Mindanao in Cagayan de Oro, a team from the University of the Philippines Archaeological Studies Program recovered five pebble tools at the open site of Huluga (Fig. 10, Neri, 2006).

But all these assemblages are surface finds and without direct dating the age remains uncertain. Their lack of context has been stated by a number of researchers (Mijares, 2002, Patole-Edoumba, 2002, Patole-Edoumba, 2006, Pawlik, 2009 and Pawlik and Ronquillo, 2003).

Absolute radiometric dates for the earliest human occupation of the Philippines at present go back to app. 66.7 ka for Callao Cave in northern Luzon (Mijares, 2008 and Mijares et al., 2010) and around 50,000 years BP for Tabon Cave in Palawan (Détroit et al., 2004, Dizon, 2003 and Dizon et al., 2002). So far, no artefacts have been found in association with these ancient human fossils. In Callao Cave, about seven cervid bones have been observed to have cut marks but no stone implements were recovered (Mijares, 2007 and Piper and Mijares, 2007). The earliest stratigraphic layers containing stone material are dated 22,000–24,000 years BP at Tabon Cave and 26,000 years BP at Callao Cave (Détroit et al., 2004, Dizon, 2003, Dizon et al., 2002, Fox, 1970a, Fox, 1970b, Mijares, 2008 and Mijares et al., 2010).

Some authors have shown a technological evolution (Leroi-Gourhan, 1943, Leroi-Gourhan, 1945 and Simondon, 1989) and that each artefact can be seen as a stage of one kind of technological evolution (named lineage). For André Leroi-Gourhan, this evolution is linked to a search for functional efficiency. Simondon thinks that major technological lineages are governed by the law of concretization. Indeed, he has shown that there is a structural determinism of the artefacts independent of their environment impact: “there is a primitive form of a technical subject, the abstract form as an absolute, complete perfection inherent in requiring, for its operation, to be incorporated in a closed system…” (Simondon, 1989, p. 21).

According to this last theory, prehistoric artefacts are developed from the abstract to the concrete in the direction of maximum optimization of production capacity of a pebble. For example, the chopper corresponds to the degree 0 of fracturing to get a sharp edge. From this initial stage, two technological lineages are formed: debitage, which is aimed at getting a sharp edge on a blank and the sharpening (façonnage) which is based on the creation of the sharp edge on one core. The sharpening evolution is based on the best way of core reduction to get a desired shape. The debitage optimization is based on the core managing to get the maximum number of blanks. In this context, the SSDA core-type (mentioned earlier) is at the beginning of a lineage of which the best result is the Levallois core-type, which requires a high ability of anticipation (Boëda, 1997 and Boëda, 2005). The integrated structure of the laminar core is the most advanced and most practical in the European and Mexican Neolithic because the volume is fully exploited to produce sometimes up to a hundred blanks (Boëda, 2005). The identification of lineages does not mean that each prehistoric culture is destined to develop all the production systems of the same lineage; the identification of the Levallois method does not mean that the SSDA existed previously.

We can apply this last analysis to the prehistoric Philippines as it has been proposed for Chinese Pleistocene lithic assemblages (Boëda, 2005). In the Philippines, we observe the coexistence of two lineages of lithic traditions at the sites mentioned above: a lineage represented by bifacial artefacts found in Luzon and Palawan and a unifacial lineage at all other sites. Since the early 2000s, three handaxes have been discovered in the archipelago. The principle is based on a double symmetry: side axial and bifacial (Texier, 1996). It also implies a biconvex volumetric design. For us the Arubo specimen at this later stage belongs to the category of “blanks on bifacial artefacts” of which the distal end was modified into a scraper (Boëda, 1997, Boëda, 2001 and Pawlik, 2002). The bifacial asymmetry and the cortex reserve put this specimen at the beginning of the technological lineage of the bifacial production system.

The artefacts found in Cagayan de Oro in 2008 and at Ille Cave, Palawan have quite similar characteristics but have not been retouched. By these techno-functional criteria they are still close to handaxes found in the Bose valley in southern China dating from 800,000 years BP (Feng, 2008, Huang, 1989, Leng and Shannon, 2000, Schick and Zhuan, 1993, Xiang, 1990, Xie, 1990, Xie and Bodin, 2007 and Yamei et al., 2000). Furthermore, whereas the bifacial phenomenon has been identified in Africa for at least 1.7 million years, in Europe 500,000 years ago and now in Asia for about 800,000 years, the technological stages of Asian artefacts do not support the hypothesis of technological expansion from the center of hominin origin. The Asian handaxes from China and possibly from the Philippines are not linked to those from Africa (and can be seen as local inventions: Boëda, 2005).

If we apply the same logic to analyze unifacial artefacts found at different sites, we find that several modes of production existed. Besides the choppers and chopping tools from Cagayan Valley and from the Huluga open site in Mindanao as the most “primitive” artefacts, handaxes and cleavers at Arubo, core-scrapers and two hemi-choppers in Mindanao and Cabalwan appear. Among the 68 tools identified by Robert Fox for the Cabalwanian, the so-called D-type possesses an asymmetrical design with a plano-convex shape. The retouching was performed on the more convex side and from the partially cortical flat surface. We can identify different sequences of flaking: first preparation of the tool's shape, then pre-sharpening. The morphology of the D-type artefact is very close to the so-called “Sumatraliths”, the “type fossil” of the Hoabinhian technocomplex of mainland Southeast Asia (Colani, 1927 and Patole-Edoumba, 2006). The modification of the surface can be regarded as an optimization of the potential of the peripheral edge. This type of hemi-chopper is probably the final result of the unifacial lineage. Also, we propose that the lithic assemblages of Luzon, Palawan and Mindano contain tools at different technological stages.

The technology-based re-analysis of historic collections also gives a new picture of the evolution of the knapping process between 25,000 and 10,000 years ago. Although a technological homogeneity has been proposed up to now, it seems that several methods of operating sequences existed.

The assemblages II and III from Tabon Cave, excavated by Fox and dated between 22,000 and 23,000 years BP, yielded a large amount of stone tools. A recent study recognized four different operating sequences. Two methods show a surface exploitation of the core; the other two used a volumetric exploitation of the core. Although the algorithmic SSDA system (system of alternating debitage surface) is the most representative in these assemblages (Tabonian facies), a discoid reduction sequence was also identified (Arzarello et al., 2008).

In Pilanduk Cave in the Isugod area, app. 11 km north of Quezon, Palawan, an excavation conducted in 1969 by Jonathan Kress revealed four cultural layers (I–IV) containing a small flake assemblage (Fig. 11). The layers were dated between 25,000 and 18,000 BP. However, no mention of the origin of these dates, nor their laboratory numbers was made (Fig. 11). Additionally, some disagreement can be observed in the assessment of the artefact material. Whereas Fox (1978) regarded the material as practically identical to Tabon Cave, Kress did not confirm a relationship between the two sites. He assessed that the chert flakes from Pilanduk are smaller and proportionately more retouched than those in recent levels in Tabon Cave, comparing them instead to the Hoabinhian (Kress, nd, 3; Kress, 1977 and Kress, 2000).

Stone implements from Callao Cave in Peñablanca, northern Luzon appearing in a layer radiocarbon-dated to 26,000 BP are mainly chert flakes manufactured by simple direct percussion. However, the recovery of more blade-like flakes in the Late Pleistocene period in Callao Cave could point to a certain variation in Philippine lithic tradition through time. The possible evidence of two blade-like flakes being used as spear- or arrow-points hints at a more formal lithic technology.

Very recent evidence for the use of unretouched flakes as hafted projectile points and the manufacture of composite tools comes from the lowest layer of Ille Cave, Palawan (Pawlik, In press). Radiocarbon-dates delivered an age of app. 14,000 cal. BP (Paz et al., 2008). A microwear analysis of its small flake assemblage identified on a triangular flake the presence of characteristic impact scars with hinge- and step-terminations together with polish spots and longitudinal striations on the tip of the tool. The flake's basal part displays on the interior surface polishes which are not use-related but do conform to what is expected from minor movements of a tool against its haft (Cahen et al., 1979: 681). Together with such polishes appear blackish residues, obviously remains of organic resin used as hafting mastic. Another tool, a drop-shaped endscraper exhibits similar hafting polishes and blackish-reddish resin at its base. This kind of adhesive appears to be very similar to resinous residues found on projectile points made of bone and stingray spine from the West Mouth of Niah Cave in Borneo, dated to 11,700–10,690 cal. BP (Barton et al., 2009). The resins have been identified as deriving from either Shorea spp., Agathis spp. or Canarium spp. These trees and their resins are also common in the Philippines and Palawan, and have been found in the Neolithic layers of Ille Cave as well, used there as appliqués on shell disk beads (Basilia, 2010). The use of unretouched lithic flakes as hafted implements for multicomponent tools and projectiles at Ille Cave is a unique find for the region and points to a technological concept that is beyond traditional morphological and categorical models. It is another indication that seemingly simple and “backward” lithic technologies can nevertheless be a part of complex sequences of action.

Unfortunately, there is yet no evidence in the Philippines for stone points made with platform preparation techniques as reported by Ian Glover for the Late Pleistocene site of Leang Burung 2 in South Sulawesi, or the bifacial techniques reported by Bellwood for the Tingkayu industry from Sabah, or the backing and serrating techniques used in the Holocene Toalian industry of South Sulawesi (Bellwood, 1988, Glover, 1977 and Glover, 1981). Callao Cave is also the only lithic assemblage in Luzon dating to c. 25,000 BP found so far. Its size is also small, thus limiting our analysis and interpretation. It is, therefore, absolutely necessary to verify this lithic technology in other cave sites of the same time period.

The proliferation of human occupation during the Late Pleistocene in different parts of the Philippines seems not to have significantly affected the manufacturing system of stone tools. All industries known at present display similar characteristics. Variations are linked with environmental and/or economic constraints (Patole-Edoumba, 2002 and Patole-Edoumba, 2009). For example, the method of lithic production used in Musang Cave remains very close to what was identified for Tabon Cave. This manifests itself by the frequency of some types of tools (14% blanks are retouched to chisels, scrapers and multiple tools in Tabon versus 2.4% in Musang Cave, mainly scrapers, notches and denticulates) and the nature or the preference of the raw material. These criteria are primarily economic constraints such as a difficult access to raw material sources, rather than environmental, with a different use of a more open environment in Cagayan than in Palawan.