A large variety of artefacts and definitions

PSBs include a wide range of objects in terms of morphology, size and weight, ranging from cubical sharp-edged items to smooth perfectly-rounded artefacts. This variety of forms, combined with the multiplicity of (generally) undemonstrated theories about their functions, led to the emergence of multiple definitions. In the literature, their classification varies depending on the archaeological contexts and authors. They are sometimes classified as multipolar or multifacial cores (e.g., [20, 21]), modified battered pieces (e.g., [21]), tools (e.g., [8]), or all of these categories (e.g., [5]). In the past, PSBs were described as hâches celtiques by Boucher de Perthes [22] when he discovered these shaped rounded stones in the Somme Valley (France), generally pecked, made from sedimentary rocks (flint, chalk and sandstone) or granite. Then, depending on the authors, a same PSB can be classified as a bola, spheroid, polyhedron or stone ball, flexible terms incorporating a more or less wide range of morphologies depending on studies. For example, Clark [23] only considered the perfectly rounded items as “stone balls” while Gruet [24] described a 75 cm high pile of “stone balls” in limestone at the El Guettar site (Tunisia), including a great variety of PSBs with cubical to more rounded morphologies, sometimes without clear human modifications. Indeed, some authors include manuports (natural spherical stone without human modification, selected and transported by hominins) in their PSB classification. This is the case of Kleindienst [25] who distinguished three categories of PSBs: (1) missiles: manuports, (2) polyhedrons: facetted items, with visible negatives of removals or (3) bolas: nearly perfectly rounded objects without visible facets, smoothed by battering.

PSBs also used to be named according to their putative functions. At Festons (France), Pittard and de Saint-Périer [26] mentioned spherical pieces (350g to 600g) as casseurs d’os, presuming that they were used for bison and horse bone breakage to access marrow. In addition to these items, they also described approximately 15 limestone and quartz stones rounded by crushing. They called them bolas (with diameters of 57 mm to 63 mm, 240 g to 250 g) and suggested that they were used to capture animals, in reference to South American ethnographic stone or metal balls attached to ropes, composed of one to three pieces. Bolas were used by Amerindians and gauchos from Argentina, Uruguay and south Brazil as projectiles mainly to capture animals, and were also functional for combat. This hypothesis of PSBs used as projectiles convinced several authors (e.g., [5, 26–30] and is still popular today. Louis Leakey [27] made PSBs widely known by encouraging this missile theory. According to him, the discovery of rounded items by groups of three in Olorgesailie assemblages (Kenya), combined with his experimental work, corroborated this theory [31]. This hypothesis has long been based on actualism with comparisons to the South American bolas. Indeed, these ethnographic artefacts gave their name to the prehistoric objects, and recent quantitative studies (volumetric, ballistic and mass distribution) provide objective data that could support the idea of some Palaeolithic PSBs (and manuports) being missiles [29, 30].

In her pioneering classification of the Oldowan industries, Mary Leakey [5] categorised PSBs typologically based on direct observation of their morphological similarity to a sphere and to the smoothness of their surfaces: (1) polyhedrons: “angular tools with three or more working edges, usually intersecting. The edges project considerably when fresh, but, when extensively used, sometimes become so reduced that the specimens resemble subspheroids”, (2) spheroids: “include some stone balls, smoothly rounded over the whole exterior. Faceted specimens in which the projecting ridges remain or have been only partly removed are more numerous”, and (3) sub-spheroids: “similar to the spheroids but less symmetrical and more angular”. This classification is still used by some authors while criticised by others. It has for instance been revised by Sahnouni et al. [9], who also discerned these three types of PSBs: polyhedrons, objects with a minimum of three flaked faces mostly displaying obtuse core angles; sub-spheroids, with more obtuse angles than polyhedrons, widely flaked on three or more faces; and finally spheroids, artefacts flaked over much, if not all of their surfaces with very obtuse angles and a relatively spherical shape.

Rounded shape: A predetermined morphology?

In the literature, two main contrasting interpretations of PSBs are put forward. 1) The first considers spheroid morphologies as forms of exhausted cores, by-products of a debitage process. These pieces would subsequently have been used in battering activities (e.g., [7, 9]). The different types of PSBs would be phases of a debitage reduction process with no preconceived techno-typological purposes. To investigate this first hypothesis, Sahnouni, Schick and Toth led functional and experimental replicative studies [9] on the limestone PSBs from Ain Hanech (Algeria). They applied a hypothetico-deductive approach. Their aim was not to reproduce the original production operative chain of these PSBs, but rather to reduce the volume of pebbles and cobbles as much as possible and generate ‘usable’ flakes [9]. Their conclusion was that spheroids can be exhausted cores resulting from intense debitage, subsequently used as hammerstones. Cobble blanks are big enough at Ain Hanech to generate potentially functional or transformable flakes. However, as De Weyer highlights, this study raises some issues: the flakes produced are in a soft raw material and 2 cm long on average, which seems quite small for the manufacture of tools or to be used in a task. Furthermore, the large majority of flakes used at Ain Hanech are in flint, potentially implying that limestone was not the most suitable material for producing flaked tools [32]. Nevertheless, the conclusion of Sahnouni et al. is in keeping with the earliest experiments of Toth [33, 34] on the PSBs of Koobi Fora (Kenya). The authors conclude that intensive debitage (in order to produce usable flakes) naturally leads to a polyhedral core shape, and this becomes increasingly spherical throughout the reduction process. Moderate reduction would produce unifacial or bifacial choppers [9]. More particularly, Schick and Toth [7] suggested that exhausted quartz cores would have been systematically used as battering instruments, and would have been accidentally increasingly rounded as a result of recurrent percussive activities.

2) The second hypothesis proposed by Texier and Roche [8] groups polyhedrons, sub-spheroids, spheroids and bolas as three segments of the same preconceived shaping operative chain (Fig 1), from the angular polyhedron shaped in a more spherical form (sub-spheroids, spheroid), and finally pecked into a perfectly smoothed rounded stone (bola). The aim of the process would be to purposefully obtain a spherical artefact according to reasoned organised final shaping, in order to use it for a predefined (unknown) function [8, 35]. Since PSBs can display a totally battered surface, Texier and Roche [36] exclude the possibility that they were simple hammerstones, since the latter generally only comprise one or two battered faces. This is contradicted by the conclusions of Assaf and Baena [37] and Assaf et al. [12], who proved that the spheroids from Qesem, Israel, could have been used as hammers for bone breakage. Assaf and Baena [37] also suggest that the large removals on the surface of these pieces were not due to use, but rather the result of an intentional shaping process to obtain the desired objects. Contrary to Sahnouni and Toth’s conclusions [9, 33], experiments by Texier and Roche [8], based on the archaeological material from Isenya (Kenya), show that removal organisation is different from the volume management of a simple core, suggesting a global view of the volume. They highlight that a rounded volume tends to reduce the possibilities of extracting flakes. After the analysis of more than 500 PSBs from Isenya, Texier and Roche [8] noted that hominins selected blanks for the production of PSBs, choosing pieces with a natural spherical shape and thus reducing the first phase of the operative chain, the most productive in flakes. Once again, this would contradict the theory of the by-product core. Another argument of Texier and Roche is that at Isenya, only a few flakes derive from the operative chain of polyhedron production, and these flakes are neither usable nor transformable by shaping or retouch (poor quality raw material and inadequate flake morphology). However, this last statement does not contradict Sahnouni’s experimental conclusions [9] since Sahnouni did not aim to reduce blocks in order to obtain a spherical morphology but aimed to produce flakes. Thus, both models can be demonstrated but are not compatible with all sites. Knapping experiments of Toth [33] and Texier and Roche [8] concluded that intentionally obtaining PSBs is a complex process. Texier and Roche [8] highlight that the difficulty lies in removal organisation, which totally differentiates this reduction process from bifacial shaping and from debitage.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Hypothesis of the linear operative chain of PSB manufacture (based on Texier & Roche 1995).

https://doi.org/10.1371/journal.pone.0272135.g001

Intermediary scenarios have been proposed between these two opposed interpretations. Mora and de la Torre [19] base their analyses on Schick and Toth’s study of quartz blocks possibly rounded by percussive activities [7]. Like Jones [18], they concluded that the PSBs from Olduvai were rounded during battering activities. In contrast with the main assumptions, some authors also argued that polyhedrons could be purely coincidental or the consequence of a knapper’s lack of experience [33, 38].

Willoughby [16] underlines that PSBs appear at Olduvai at the same time as anvils, suggesting that these PSBs could be hammers combined with anvils in pounding activities. In this case, PSBs would not be deliberately rounded but would have acquired this spherical shape by repeated use as hammers [17]. She also interprets the pecked cupulas on some anvils as the result of the preparation of the anvil before use. On the contrary, Texier and Roche [8] suggest that these cupulas could result from use. Based on this potential correlation between PSBs and anvils at Olduvai [16], Texier and Roche [8] argue that producing PSBs with a hammer on a hard surface would allow for better control of the percussion process. This would result in bipolar percussion, generating multiple battered zones that would round the PSB. This point of view coincides with innovative studies led by Clark in 1955 [23], who proposed that PSBs were worked on an anvil until they obtained their spherical shape. Clark [23] suggests that battering ridges would be a deliberate part of the manufacture process to remove protruding edges. However, he does not exclude the possibility that PSBs could have been hammerstones involved in nut crushing or knapping, although his main idea was that PSBs were missiles.

Polyhedrons, spheroids and bolas: Three stages of a same reduction process?

Both main hypotheses about the nature of PSBs (predetermined tools vs. exhausted cores) agree that polyhedrons, spheroids and bolas are stages of the same technological trajectory. Texier and Roche [8, 36] demonstrate this for the assemblages of Isenya, as does Sahnouni at Ain Hanech [9, 39]. Nevertheless, this linear reduction between PSBs has been contested in several cases with morphometric assertions (as polyhedrons are smaller than spheroids in some assemblages) and because in some sites, polyhedrons, spheroids and bolas are respectively made from different raw materials, which according to Jones [18], are clues that polyhedrons and spheroids do not belong to the same sequence. Indeed, this difference in raw materials can be observed in several sites, such as at Isenya where bolas are made from quartz, whereas polyhedrons and spheroids are mostly in nephelinic phonolite (less frequently in quartzite and quartz). This is also the case in Beds I and II of Olduvai, where polyhedrons are made of lavas whereas spheroids and bolas are almost all in quartz [18]. Based on diacritic schemes, De Weyer [11] also argues that at the site of Ounjougou (Mali, Oldowan cultural facies attribution), polyhedrons, spheroids and bolas stem from different operative chains.

Functional assumptions

The functional interpretation of PSBs has given rise to theories and debates for decades. These pieces often present macro-traces on their protruding ridges, generally described as crushing, pecking or battering traces (e.g., [4, 40–43]. As discussed above, a popular theory considers PSBs as missiles or bolas [5, 26–30]. Authors presenting PSBs as hammerstones also abound (e.g., [44]), some arguing that they were rounded without predetermination in battering activities [18], or exhausted cores secondarily reused as hammers for tool manufacture [7, 9]. According to the recent experimental and use-wear study of Assaf et al. [12] and Assaf and Baena [37], the spheroids from Qesem Cave were used as hammers in bone breakage to access marrow. Isaac [45] shares this hypothesis of bone breaker tools. Others assume that some PSBs were involved in pounding activities [23], or used as club heads [28]. For her part, Willoughby [6, 17] does not exclude any of these scenarios, but is more sceptical about the use of PSBs as throwing weapons since some of them are quite heavy (several kilograms). She also proposed that PSBs could have been vegetal processors [16]. Thus, in the same way as manufacture modes, the debate about the functions of PSBs still remains open and no single scenario can be extended to all sites.

This uncertainty surrounding PSBs is partly due to the scarcity of functional studies, since they are barely mentioned in the literature and even more rarely described or illustrated. The lack of homogeneity regarding their classification and denominations can lead to misunderstandings when illustrations are lacking. Recent interest in these objects and a few focused studies have emerged over the past few years [10–13, 29, 30]. For instance, Titton et al. [13] carried out a study of the five PSBs from the Oldowan site of Barranco León (Spain), combining diacritical and 3D geometric morphometric analysis with an evaluation of the raw material (limestone) and of percussive use-wear (types and localisation). De Weyer [11, 32] proposed an analysis of the PSBs from Ounjougou with technological, techno-functional and morphometric approaches, comparing these pieces with PSBs from other sites such as Olduvai Gorge and Ain Hanech. The most extensive and significant works on PSBs may be attributed to Willoughby (e.g., [6, 17]) for her techno-functional analysis, to Sahnouni, Schick and Toth for their experimental and morpho-technical studies (e.g., [9, 33, 46]), and to Texier and Roche for their experimental and technological approaches [8, 36, 47].

Quantities of PSBs in assemblages

A synthesis of the Dunn test results regarding quantities of PSBs according to regions and chrono-cultural attributions is presented in Table 1. In a nutshell, there are significantly fewer PSBs in European assemblages, especially compared to East and North Africa and to the Levant (for PSBs in general, but also respectively regarding polyhedrons, spheroids and bolas) (Table 1, Figs 3 and 4). PSBs are also scarce in Central Africa, which can partly be due to the paucity of documented prehistoric sites in this region, but not as much as for West Africa or Southeast Asia.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Distribution and quantities of PSBs in Africa, Eurasia and the Levant according to chrono-cultural attributions.

(A) In the Oldowan (dark-grey site labels) and Core and Flake (light-grey site labels). (B) In the Acheulian. (C) In the MSA (dark-grey site labels) and Middle Palaeolithic (light-grey site labels).

https://doi.org/10.1371/journal.pone.0272135.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Quantities of polyhedrons, spheroids and bolas in assemblages according to chrono-cultural attributions and regions.

Ass.: Assemblages.

https://doi.org/10.1371/journal.pone.0272135.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Statistically significant differences in quantities of PSBs according to regions and chrono-cultural attribution.

https://doi.org/10.1371/journal.pone.0272135.t001

PSB morphologies appeared in Africa during the Oldowan, where they are frequent. Meanwhile, they are scarce in Eurasia and the Levant at slightly later Core-and-Flake-type sites (Tables 1 and 2, Fig 3). The production of these morphologies became more common outside Africa during the Acheulian (diffusion out of Africa? Technological convergence?), and tend to be scarcer again in the MSA and the Middle Palaeolithic (Table 1, Fig 3). Graphically, the frequencies of these objects in lithic assemblages are lower in European assemblages, Middle Palaeolithic, MSA and Core-and-Flake-type sites (Fig 5). In Asia, although our data are partial, we note that there are less Middle Palaeolithic sites with PSBs (n = 5 assemblages) than Early Palaeolithic sites (n = 10 assemblages). However, one Asian series attributed to the Middle Palaeolithic, Xujiayao (China) yielded more than a thousand PSBs, which strongly influences the mean for this period (Table 2), when PSBs are generally scarce. In the Levant, we did not identify any Middle Palaeolithic site with PSBs (Fig 4). Polyhedrons are way more common than more rounded forms (Fig 4), and are the only category of PSBs for which analyses of quantities provided significant results when only considering sites that yielded the type of PSB of interest (here, considering only assemblages with polyhedrons) (Table 1). Fig 6 shows a higher frequency of polyhedrons in early chrono-cultural complexes (Oldowan and Acheulian) than in later ones (Middle Palaeolithic and MSA), and on the contrary, higher frequencies of spheroids and bolas in more recent chrono-cultural complexes. This is a tendency at the scale of the cultural techno-complexes, not at the scale of the site: polyhedral and spherical pieces are present in all the techno-cultural complexes discussed here, but in various proportions, that seem to evolve with time. Over time, PSBs seem to become scarcer and more rounded (Fig 6). This could suggest a standardisation of these objects over time.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Proportions of PSBs in lithic assemblages according to regions and chrono-cultural attributions.

The width of each column is proportional to the number of assemblages considered. Assemb.: assemblages. Outliers not displayed in the construction of the graph.

https://doi.org/10.1371/journal.pone.0272135.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Frequencies and quantities of polyhedrons, spheroids and bolas in assemblages according to chrono-cultural attributions.

Oldow.: Oldowan. Co.-Fl.: Core-and-Flake. Acheul.: Acheulian. Mid. P.: Middle Palaeolithic. Ass.: assemblages.

https://doi.org/10.1371/journal.pone.0272135.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Statistics regarding quantities of PSBs per site, according to regions and chrono-cultural attributions.

https://doi.org/10.1371/journal.pone.0272135.t002

Types of raw materials of PSBs

PSBs made on hard stones represent the vast majority of the PSBs of the corpus (Table 3). The hardness of a stone is based on its response to a physical constraint: on one hand, the ability to withstand moderate shocks by transfer of energy between grains or inside the matrix, and on the other hand, the ability to facilitate the breakage or the burst. The silicifications (e.g., chert, flint, millstone) will fragment more easily, contrary to more resistant raw materials. This is more related to the way the silica crystallised than to the content of silica of the stone, with on one hand amorphous silica or chalcedony for flint and chert, and on the other hand, when there is some, a siliceous cement consolidating crystals or minerals. There is also a part of qualitative appreciation based on the knapping experience. Many PSBs are also made from quartz. All the following counts of assemblages are minimum since data may be partial:

• Limestone (mostly hard types): 32.38% of PSBs (n = 600/1853), in n = 18/150 assemblages.
• Quartzite: 18.24% of PSBs (n = 338/1853), in n = 41/150 assemblages.
• Quartz: 11.87% of PSBs (n = 220/1853), in n = 62/151 assemblages.
• Phonolite: 10.90% of PSBs (n = 202/1853), in n = 4/150 assemblages (remark: 3 of these 4 assemblages are from the site of Isenya, Kenya).
• Andesite: 9.93% of PSBs (n = 184/1853), in n = 15/150 assemblages (remark: 13 of these 15 assemblages are from Kaletepe Deresi 3 site, Turkey).
• Basalt: 7.39% of PSSBs (n = 137/1853), in n = 32/150 assemblages.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Minimum counts of assemblages with at least one PSB in the raw material of interest, and PSB frequencies according to raw materials and regions.

https://doi.org/10.1371/journal.pone.0272135.t003

PSBs are rarely made of siliceous materials such as flint or chert (Table 3, Figs 7 and 8). Among the 169 assemblages, only three assemblages yielded PSBs in chert: 1 polyhedron at Swartkrans SWT-M2 in South-Africa (n = 1/12 PSB of this assemblage) [59], and both ‘Ubeidiya III-20 and III-22 present polyhedrons in chert and fine-grained limestone (while spheroids are in limestone) [6, 60–62]. At least ten assemblages yielded PSBs in flint (n = 48/1853 PSBs are in flint), including five European assemblages, three from the Levant and two from Africa. More precisely, sites with only flint PSBs (n = 3 assemblages, minimum) are exclusively located in Europe, and they comprise only few PSBs (one to four PSBs), and the rest of these assemblages are totally (or nearly) made of siliceous materials. This could mean that siliceous rocks were not selected or not suitable for the production of PSBs. Furthermore, siliceous stones are predominant in Northwest European basins, where PSBs are rare. Could this scarcity of PSBs be linked to a Northwest European tradition? Furthermore, can this tradition be partly correlated to this abundance of siliceous materials, which could somehow have influenced lithic production? In this case, this availability of siliceous materials would have impacted technological traditions, resulting in a toolkit without PSBs. In some sites, PSBs are the only non-flint items of the series, as at Qesem Cave [12]. In total, out of the 33 assemblages (some of which are part of a same site, at a different period) of the corpus that yielded lithic artefacts in flint, we observe:

• Eighteen assemblages with no PSBs in flint. In five of these 18 assemblages, PSBs are the only limestone pieces among series of artefacts almost exclusively made of flint, as at El Guettar, Tunisia [24], Revadim Quarry, Israel [63], Evron Quarry, Israel [64], or in US 22 of Jonzac, France [65]. At La Quina, France, PSBs are the only Turonian limestone pieces in an assemblage made mostly of Coniacian flint and quartz, with a few objects in chalcedony. The only flakes in limestone here are in Coniacian limestone and PSBs are not concerned. In the 13/18 other assemblages, PSBs are not made from siliceous materials in assemblages including flint artefacts. For instance, at Ain Hanech, PSBs are all made on limestone whereas the rest of the lithics are mostly in very good quality black flint and limestone [39, 46]. At North of Bridge Acheulian (NBA), a locality near Gesher Benot Ya’aqov site, Israel, six spheroids are made in limestone and three in basalt (the raw material of the three pieces described as “sub-spheroids” is not given), and the rest of the lithic assemblage is in basalt and flint [66]. At Latamné, Syria, spheroids (n = 14) are in limestone except one in basalt, the only basalt piece of the assemblage. Spheroids are among the rare limestone objects (n = 74 items in limestone in the whole assemblage) out of 1759 pieces in flint [67, 68]. At Barranco León, Spain, PSBs are in limestone while the rest of the lithic assemblage is in flint and limestone [13]. The Caune de l’Arago, France (Units D, E, G, H1-3) and Zhoukoudian 15, China, yielded PSBs nearly exclusively made in quartz in series with pieces in flint [69, 70]. At Duclos, France, and Bañugues, Spain, PSBs are in quartzite, as are most of the rest of the assemblages, that also include pieces in flint [71, 72].
• Two assemblages yielded a very small minority of PSBs in flint: Qesem Cave, Israel, with one spheroid in flint (different from the flint of the rest of the assemblage) and the 28 other PSBs are the only pieces made from hard limestone, among tens of thousands of artefacts in flint [12]. According to Assaf et al. [12], these PSBs were not manufactured in the cave but rather recycled by hominins who found it already shaped in the vicinity of the site, and brought it to the cave to be used in bone breakage to extract the marrow [12]. Tighennif I, Algeria, also yielded only one polyhedron in flint associated with 52 PSBs in other raw materials (39 in quartzite, six in sandstone and seven in limestone) [73].
• Four assemblages yielded PSBs in flint and in other raw materials in approximately the same proportions: Sterkfontein Member 5, South Africa [17], Joubb Jannine II, Lebanon [74] and Festons, France [26, 75], where the flint PSBs were the least rounded pieces of the assemblages; and Sablière Rambour, France [76], where on the contrary, polyhedrons in flint are more regular than the PSBs in sandstone.
• Three European sites where PSBs are only made on flint, and in very low quantities: Ca’Belvedere di Monte Poggiolo, Italy, with two polyhedrons and 93.24% of the assemblage made of local flint [13, 77]; stratum C of La Noira, France, with one polyhedron and the rest of the assemblage mainly in millstone (siliceous stone) and some pieces in flint, chert, quartz, sandstone and quartzite [78]; and finally Tourville, France, with four PSBs in an assemblage in local flint [79].
• Finally, five assemblages where there are items in flint but the raw material of PSBs is not given.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Counts of PSBs in each raw material per assemblage, per region and chrono-cultural attribution.

For more details about counts: see Tables 3 and 4. Igneous und.: unspecified igneous rock. Siliceous und.: unspecified siliceous rocks. Sedimentary: breccia and ironstone. Ass.: assemblages.

https://doi.org/10.1371/journal.pone.0272135.g007

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Raw materials of PSBs in Africa, the Levant and Eurasia.

(A) Oldowan (dark-grey site labels) and Core-and-Flake-type industry (light-grey site labels). (B) Acheulian. (C) MSA (dark-grey site labels) and Middle Palaeolithic (light-grey site labels). Siliceous und.: unspecified siliceous rock. Igneous: mainly igneous rocks for which quantities were unspecified in the literature, including mostly basalt. Sedimentary: breccia and ironstone. For a few assemblages, the proportions represented are higher than in tables for igneous and sedimentary rocks: raw materials were in a few cases grouped here in more general categories to show quantities on the map. For example: Melka Kunturé Garba IV yielded PSBs in volcanic rocks including basalt, but maybe also trachyte or tuff according to d’Andrea et al. [80]. Since we have the total numbers of PSBs in this assemblage, but not the exact proportions for each raw material, this assemblage is not considered in “Freq. PSB” of Tables 3 and 4. However for the map, raw materials of these PSBs were grouped under “Igneous” in order to show information in the figure. The list of these changes and approximations is available in S2 Text.

https://doi.org/10.1371/journal.pone.0272135.g008

As we can see on the maps (Fig 8) there are similarities in PSB raw materials between relatively geographically close assemblages. In particular, we note that quartzite is mostly used in the region of the Gibraltar Strait during the Acheulian, on both the North African (n = 53/77, 68.83% of North African Acheulian PSBs, 3/3 assemblages) and Southwest European sides (Fig 8). PSBs in limestone are also very common in North Africa (n = 525/601, 87.35% of North African PSBs, n = 4/6 assemblages) and the Levant (n = 58/251, 23.11% of the Levantine PSBs, n = 9/27 assemblages), and very prevalent in Southwest Europe as well (Fig 8). PSBs in quartz are widely dominant in South Africa (n = 132/273, 48.35% of South African PSBs, n = 15/20 assemblages), and PSBs made of igneous rocks are the most common in East Africa (n = 339/467, 72.59% of East African PSBs, n = 35/43 assemblages) (Fig 8). In Central and West Africa, PSBs are rare but nearly exclusively made of sandstone (n = 29/31, 93.55% of Central and West African PSBs, n = 4/4 assemblages).

PSBs in limestone are dominant in Core-and-Flake assemblages (n = 17/27, 62.96% of Core-and-Flake PSBs, n = 2/7 assemblages) in Europe and the Levant. In terms of quantity, 71.07% of the Oldowan PSBs (n = 457/643 Oldowan PSBs) of the corpus are in limestone, but they only come from one North African assemblage, Ain Hanech. They are also quite numerous in Acheulian sites from the Levant (n = 46/239, 19.25% of Acheulian Levantine PSBs, n = 8/26 assemblages). Finally, limestone PSBs are also predominant in Middle Palaeolithic assemblages in Europe (n = 12/18, 66.67% of Middle Palaeolithic European PSBs, n = 2/4 assemblages) and at El Guettar, Tunisia.

PSBs in quartz are present in Acheulian assemblages from Africa (n = 57/419, 13.60% of African Acheulian PSBs, n = 20/33 assemblages) and Asia (n = 11/50, 22% of PSBs the Asian Acheulian PSBs, n = 4/9 assemblages), as well as in MSA sites (n = 55/120, 45.83% of PSBs in the MSA sites, n = 6/13 assemblages), especially in South Africa. They are also quite present at the Oldowan in terms of the number of assemblages that yielded at least one PSB in quartz (n = 8/21 sites).

Quartzite is also one of the main raw materials used to manufacture PSBs in the Oldowan assemblages of the corpus (n = 86/653, 13.17% of Oldowan PSBs, n = 7/21 assemblages), however it largely comes from assemblages from Olduvai Gorge. PSBs in quartzite are also quite common in the Acheulian sites from Africa (n = 68/419, 16.23% of Acheulian African PSBs, n = 9/33 assemblages), Europe (n = 93/131, 70.99% of Acheulian European PSBs, n = 8/13 assemblages) and Asia (n = 29/50, 58% of Acheulian Asian PSBs, n = 6/9 assemblages). They are also present in MSA assemblages (n = 12/120, 10.00% of MSA PSBs, n = 4/13 assemblages) and Asian Middle Palaeolithic sites (n = 6/9, 66.67% of Middle Palaeolithic Asian PSBs, n = 2/4 assemblages).

Finally, hard igneous rocks (particularly basalt) are often used for the production of PSBs, especially in Africa during the Oldowan (n = 71/653, 10.87% of Oldowan PSBs, n = 18/21 assemblages) and the Acheulian (n = 269/419, 64.20% of Acheulian African assemblages, n = 23/33 assemblages). This is also the case for Levantine Acheulian sites (n = 187/239, 78.24% of the Levantine Acheulian PSBs, n = 17/26 assemblages), but this high proportion is due to Kaletepe Deresi 3 site, which yielded 13 assemblages with PSBs all made of andesite. According to our data, PSBs in igneous rocks are nearly absent during the Middle Palaeolithic and MSA (Table 4). Indeed, as we saw above, over both of these recent periods, PSBs were scarcer and tended to be more rounded, compared to sites from earliest techno-complexes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Minimum numbers of assemblages with at least one PSB in the raw material of interest, and frequencies of PSBs according to raw materials and chrono-cultural attributions.

https://doi.org/10.1371/journal.pone.0272135.t004

Polyhedrons, spheroids and bolas can respectively be made of different raw materials (Figs 9 and 10). On graphs (Figs 9 and 10), the height of columns depends more on the amount of available data in the literature than on the real quantity of PSBs. Polyhedrons tend to be made in igneous rocks more frequently than spheroids and bolas. On the contrary, spheroids and particularly bolas are more often made in quartz than polyhedrons (Figs 9 and 10). Furthermore, in a same assemblage, spheroids and/or bolas can be in quartz while polyhedrons are in other raw materials. Thus, generally, the more rounded the PSBs, the more they tend to be made on quartz, especially in African and Asian assemblages.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Counts of polyhedrons, spheroids and bolas per raw material in assemblages, in each region.

Igneous und.: unspecified igneous rock. Siliceous und.: unspecified siliceous rocks. Sedimentary: breccia and ironstone.

https://doi.org/10.1371/journal.pone.0272135.g009

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. Counts of polyhedrons, spheroids and bolas in each raw material per assemblage, according to chrono-cultural attribution.

Igneous und.: unspecified igneous rock. Siliceous und.: unspecified siliceous rocks. Sedimentary: breccia and ironstone. Oldow.: Oldowan. Co.-Fl.: Core-and-Flake. Acheul.: Acheulian. Mid. P.: Middle Palaeolithic. Ass.: assemblages.

https://doi.org/10.1371/journal.pone.0272135.g010

Generally, sources of the raw materials used to make PSBs are local, in the same way as the other raw materials of the lithic assemblage. For instance, the tools from Erg Tihodaïne are made on local stones, from 0 to 5 km away [81]. At Isenya, most PSBs are made from local phonolite of Kapiti, and the other materials (quartz, quartzite, nephelinite) are available less than 7 km away [82, 83]. There are however some exceptions. For instance, at Olorgesailie (Kenya), PSBs are mostly in basalt or other types of lava, whereas rare large spheroids are in quartz imported from northern Tanzania, 43.48 km away [84]. This special attention for quartz could point to the hypothesis of a special selection of quartz for the production of PSBs at this site.

In some cases, PSB raw materials are different from the rest of the lithic assemblage, suggesting a selection by hominins of specific stones for these pieces. For instance, we can mention the Clay and SC Units of Olduvai HWK-EE, Tanzania, where items described as “sub-spheroids” are the only PSBs exclusively in quartzite [85], Levels A and B of Olduvai SHK Main Site where spheroids are the only type of objects made exclusively from quartzite [21], Vlakkraal Thermal Springs, South Africa, where PSBs are the only pieces not made from indurated shale [86], El Guettar, Tunisia, with a large majority of lithics in flint but PSBs in limestone [24], Dingcun, China, where 94.6% of the assemblage is made from fine-grained hornfels while PSBs are mainly in limestone and in quartz, green sandstone and quartzite [42], or many Levantine sites where PSBs are the only items in hard limestone among more fragile flint lithics. When PSBs are not the only objects of the series in a specific raw material, they are often made from the same stones used to make the other HDTs of the assemblage (more likely than LDTs or cores), still in different proportions. According to our data, PSBs are more rarely made from the same material as hammers. This may be due to a bias in the available information, as hammerstone raw materials are rarely mentioned. Even when PSBs are made from the same materials as other artefacts in the assemblage, they can sometimes be distinguished by the special morphology and size of the often-ovoid blank, suggesting selection by hominins (e.g., [13, 72]).

Information is often not specified in the literature for PSB blank type, (n = 88/169 assemblages without this information). When it is mentioned and possible to assess (for pieces with cortical patches), they are widely made from pebbles and cobbles (n = 58/79 assemblages). Most of the assemblages with PSBs made on slabs (n = 13/15 assemblages) are from Kaletepe Deresi 3 site, Turkey, attributed to the Acheulian chrono-complex. Few sites yielded PSBs made on blocks, mostly in Africa (n = 8/30 assemblages), and very rarely on big flakes (n = 2 assemblages).

Functions of sites with PSBs

Occupation types and the activities carried out on sites could be completely or partially documented for 57 assemblages of the corpus. Butchery activities are the most frequent (n = 33/57 assemblages minimum), but this is also the case in sites without PSBs. Butchery activities are also more frequently recorded as they are easier to identify at first sight (e.g., presence of intentionally broken and fragmented bones, remains of carcasses). Use-wear analyses have not been performed for many sites, perhaps explaining why wood and plant processing are rarely documented in the sites of the corpus. Most of the sites are also described as habitats or settlements of various durations. At least 12 sites indicate knapping activities and three point to hunting.

Environment and climate

Assemblages with PSBs do not reveal a specific general environmental context. They are located near a water source, generally with an arid to semi-arid climate. In most cases, the landscape was open with sparse vegetation, or forested, sometimes tropical humid, rarely swampy. However, these parameters may be too general to draw any conclusions.

Composition of assemblages with PSBs

Generally, as for most Palaeolithic sites, assemblages with PSBs yielded big mammal remains. The absence of such remains is generally due to taphonomic processes, for instance high soil acidity that dissolved organic materials, as is the case at Ounjougou, Mali [11], or in the majority of sites from Rajasthan, India. When large faunal remains are present in the corpus, they are mainly intentionally anthropically broken bones, except in the Levant, possibly to extract marrow (Table 5). For example, at Isenya (level V and VI), Kenya, Roche et al. [82] suggest that bone breakage was probably anthropical, with broken long bone epiphyses and diaphysis, as is common in East African sites. Pante and de la Torre [87] also record intentional bone breakage for the extraction of marrow at the East African site Olduvai HWE-EE (SC and Clay Unit, Bed II). This is the case in European sites as well, for instance the Caune de l’Arago (D Unit) [88], Tourville [79] and Jonzac, France. The latter site may possibly have been a workshop for the primary treatment of carcasses [89]. At La Quina and Festons [26] in France, in situ fracturing of reindeer carcasses to extract marrow was also reported [90, 91]. Such occurrences can also be attested in China with for example, Hsuchiyao site [92], layer 11 and the lower part of layer 10 of the kill-butchery-site of Lingjing, [93], or the horse-kill site of Xujiayao [94], and finally, in the Levant with the example of Qesem Cave [12].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Percentages and minimum numbers of assemblages with PSBs and intentionally broken bones, according to regions.

https://doi.org/10.1371/journal.pone.0272135.t005

At in-situ assemblages, PSBs can be spatially associated with intentionally broken bones, for instance at the Oldowan site of Barogali, Republic of Djibouti, where an elephant carcass was discovered in a wetland surrounded by PSBs and other lithic artefacts [95]. Barogali yielded one bola and six polyhedrons, all in diabase. The bola was found near the epiphysis of a big bone with the polyhedrons, associated with choppers, hammerstones and intentionally broken bones (ribs, epiphysis, vertebrae, big bones from limbs). At the Oldowan hippo butchery site of Gadeb 8F, Tanzania, according to Assefa et al. [96]: “scapula, long bone and rib fragments lay together with two sub-spheroids and a spheroid which had been used to break up the bones to get at the marrow”. In Ethiopia, Melka Kunturé Gombore II (Locality 2) yielded a polyhedron and two bolas discovered among bone fragments and near a basalt cleaver [97].

Types of debitage are not exhaustive in our database (n = 79/169 assemblages with no data at all). The rare data show that debitage is discoid (n = 55/90 assemblages), bipolar (n = 30/90 assemblages), unipolar (n = 28/90 assemblages), and bipolar on anvil (n = 18/90 assemblages). The available data do not provide evidence of PSB association with specific types of debitage. The series reflect the technological trends observed over time and space, regardless of the presence of PSBs.

Other types of artefacts are obviously discovered in assemblages with PSBs, as in most prehistoric lithic series: cores, flakes, debris, cobble tools (especially in East and South Africa, East and Southeast Asia, Europe and the Levant, but not so much in the MSA and the Middle Palaeolithic). PSBs are thus not isolated tools but are part of a large tool kit composed of heavy and light-duty tools. Scrapers, the most common flake tools, are also often present in East Africa, East Asia, Europe and the Levant, in Acheulian and Middle Palaeolithic periods. Hammerstones were often found associated with PSBs in the Acheulian, in European and East Asian sites, as well as in Africa in the Oldowan. Regarding LDTs, denticulates, notches and knives are also common in Acheulian assemblages with PSBs, in East Africa and the Levant. Points, notches and burins are present in Middle Palaeolithic sites. Unsurprisingly, bifacial industry (bifaces, handaxes, cleavers), as well as core-scrapers and picks are present in a large majority of Acheulian sites with PSBs, particularly in East Africa, followed by East Asia and the Levant. The presence of HDTs in assemblages with PSBs is less frequent in Europe. Interestingly, as PSBs, cleavers (in particular cleavers on flake) are very scarce in Northwest Europe, but present in the Southwest (e.g., Iberic Peninsula, Aquitaine Basin, Mediterranean Basin) [98, 99], where there is a higher diversity of raw materials and less siliceous rocks. European cleavers are generally made from hard materials (mostly quartzite), and are nearly absent in regions where populations preferred flint raw materials, as also underlined by Capdevielle [100]. This is a common point between cleavers and PSBs: they are lacking in Northwest Europe, where the tradition of production is mostly directed toward flint. Anvils are nearly exclusively mentioned in assemblages from East Africa (n = 13/43 East African assemblages, all from Olduvai and Melka Kunturé). Of the fifteen assemblages with anvils, only one, la Quina, also yielded bolas (n = 3 bolas). This apparent lack of anvils can also be related to recording methods in the field or data availability. Nonetheless, it contrasts with the hypothesis that PSBs were worked on an anvil, although this is still possible (especially for sites with anvils). Manuports are also very rarely mentioned.