Function, Style, and Standardization: Is the Proximal or Distal End of a Middle Stone Age Point More Variable?

ABSTRACT

Middle Stone Age (MSA) point styles may reflect the development of regional identities of human social groups. MSA people likely used these points for various functions including as weapon tips. Recent work in cultural evolution suggests that functional aspects of technology tend to be more standardized and subject to negative selective pressures, while stylistic aspects are highly visible and tend to be more variable. This study examines the patterns of segmented (proximal vs. distal) standardization for unretouched MSA points from six archaeological assemblages using landmark-based geometric morphometrics to quantify shape variation. Contrary to expectations, our results show that the proximal end has greater shape variation than the distal end. Segregating within point variation provides clues to past function and style patterns in the MSA. This approach also opens new avenues for exploring ethnographic and recent archaeological projectile points.

KEYWORDS:

• MSA points
• geometric morphometrics
• style vs. function
• regional identities
• cultural evolution
• Lithic armatures

Introduction

The human capacity for social learning and sociality is rooted in Africa’s Middle Stone Age (MSA). The archaeological record of the MSA provides some of the earliest evidence for complex technologies such as multi-component weapon systems (Brown et al., 2012; Wilkins et al., 2012), the heat-treatment of stone (Brown et al., 2009), compound adhesives and paints (Henshilwood et al., 2011; Villa et al., 2015; Wadley et al., 2009), and medicinal plant use (Wadley et al., 2011). These kinds of complex technologies provide evidence for increased reliance on social learning because they are unlikely to be repeatedly reinvented by individuals (Wilkins, 2020). Instead, the complex knowledge needed to produce sophisticated technologies require imitation and transmission over multiple generations. Also, during the MSA, there is evidence for the long-distance transport of stone raw materials (Blegen, 2017) and ostrich eggshells (Stewart et al., 2020), suggesting the existence of long-distance social networks. In addition, archaeologists associate the MSA with early evidence for ritual and symbolism, including pigment use (Brooks et al., 2018; Hodgskiss, 2013; Watts et al., 2016), engravings (Henshilwood et al., 2009), beads (Bouzouggar et al., 2007; d’Errico et al., 2005; Hatton et al., 2020), and collected crystals (Wilkins et al., 2021). The expanded distribution of MSA sites across the continent also indicates increased sociality. That is in contrast to Earlier Stone Age hominins who may have been tethered to water (Finlayson, 2014), raw material sources (Sampson, 1985), or a narrow range of habitats (Petraglia et al., 2019). Moreover, MSA populations likely inhabited a wide range of environments (Roberts & Stewart, 2018), including grasslands, arid and semi-arid regions (Dewar & Stewart, 2016; Wilkins et al., 2021), tropical rainforests (Taylor, 2016), and coastal areas (Marean, 2010; Will et al., 2013). Humans may have only occupied harsh environments once extended social networks had been established and maintained (McDonald & Veth, 2011), which is vital during resource shortfalls where a social safety net allows movement to more productive regions (e.g. Wiessner, 1982).

Regional traditions in point form are another line of evidence for increased human sociality during the MSA. Clark (1992) and others (McBrearty & Brooks, 2000) consider the MSA’s diverse styles of pointed stone artifacts as the first evidence for group identities, and thus regional traditions. Today, signals of group identification are everywhere in human society. Groups of cooperative individuals often signal affiliation and actively symbolize social relationships through their material culture (Wiessner, 1983; Wobst, 1977). Archaeologists use projectile point typologies in many regions and periods to define the geographic bounds of technological exchange networks. For example, Paleoindian occupation patterns in North America are based mainly on the spatial distribution of projectile point morphologies (e.g. Bamforth, 2009; Buchanan & Collard, 2007). Stylistic differences in projectile points are also used to track the Neolithic transition in the Near East, including patterns of agricultural intensification, trade, and interaction (Borrell & Štefanisko, 2016). Weapon tips are thought to be good candidates for signaling social information because they are visible by close-contact residential household members as well as by distant social contacts during foraging activities (Tostevin, 2007). Ethnographic research in the Kalahari suggests that recent foraging societies use arrowhead styles to signal social relationships (Wiessner, 1977, 1983, 2002). In Baringo District, Kenya, metal spears’ shape and design reflect conscious aspects of social identity, including age cohort and warrior status (Larick, 1985, 1991). Thus, point form variability in the MSA seems to reflect the social signaling of regional identities.

There are multiple views on the nature and role of style in material culture. For example, Wobst (1977) emphasized the active role of stylistic variation in conveying information about group memberships. His classic example relates to Yugoslavian male headdresses that are highly visible from a distance and signal the affiliation of the largest social group to which its wearer identifies. Wiessner (1983) defined two roles for such active style – emblemic and assertive style. From this perspective, emblemic style in Kalahari arrows allowed points made by one’s linguistic group to differentiate from another. Emblemic style is sometimes thought of as ethnic boundaries in the archaeological record, or “group protocols” (Hegmon, 1992). Assertive style may allow individuals within a linguistic group to identify its maker and signify relationships through exchange networks. Objects can carry meaning associated with both emblemic and assertive style. Kalahari arrows carry assertive style as well, and members within a group can identify who made specific arrows. In contrast to an active role of style, Sackett (1985) emphasized the passive role of isochrestic, or technological, variation, which reflects choices between functionally equivalent alternatives made in a cultural context. In this view, material culture does not necessarily transmit information about group identity. Sackett’s (1985) interpretation of the Kalahari arrows described by Wiessner (1983) is that groups are not intentionally distinguishing themselves ethnically. However, these groups produce uniquely functional arrows based on social interaction and historical context. In this way, artifact design is dictated, either consciously (Wobst, Wiessner) or unconsciously (Sackett), by the community of practice (Langley, 2019).

Cultural evolution models have sought to explain stylistic change through time via processes recognized and defined in biological evolution (Boyd & Richerson, 2005; Rigaud & García-Martínez de Lagrán, 2018). From this perspective, negative selection pressures reduce variation in cultural traits, resulting in greater standardization of artifact forms (Lipo & Madsen, 2001). The amount of negative selection acting on material culture may be correlated with the objects’ functional importance. Those artifacts interacting with the environment with observable performance traits are expected to be relatively standardized and fixed because they have a greater bearing on survival (Rogers & Ehrlich, 2008). In contrast, stylistic elements that do not strongly affect functional performance may be selectively neutral, and more significant variation is anticipated (e.g. Dunnell, 1978). Importantly, even within a single artifact type, portions integral to the function of an artifact are unlikely to exhibit stylistic variability in form (Rogers & Ehrlich, 2008). Other portions of an artifact are more likely to exhibit stylistic variability. For example, where form does not have such a strong impact on performance, or where there are other functionally equivalent alternatives.

The regional tradition argument for MSA point style is based partly on the assumption that MSA points were used as weapon tips and are therefore highly visible, frequently used, and often encountered on the landscape. There is support for this assumption, with an important caveat. Functional studies have revealed that MSA points at many sites across the continent were most likely used as spear tips. Evidence includes artifacts recovered from Kathu Pan 1 in South Africa, where point damage is concentrated near the tips. These fractures are consistent with high-impact damage (Wilkins et al., 2012). Similarly, at Gademotta, Ethiopia, several points exhibit macro- and microscopic features consistent with hafting and use as weapon tips (Sahle et al., 2013). At Klasies River, a broken bit of stone was embedded in the vertebrae of a large bovid (Milo, 1998). Points at ≠Gi, Botswana were retouched at the proximal end, perhaps to facilitate hafting into a dart shaft (Brooks et al., 2006). Analyses of MSA points across the continent reveal they are the right size and shape to function effectively as spear tips (Shea & Sisk, 2010; Sisk & Shea, 2011). Experimental research shows that hafting an MSA point to a spear shaft increases its killing power by creating a larger wound (Wilkins et al., 2014). The caveat is that people did not use all MSA points as weapon tips. Functional studies have also indicated that MSA points served as cutting and scraping tools (Douze et al., 2020; Iovita, 2011; Schoville et al., 2016). This is the case, particularly at sites interpreted as residential locales (Schoville, 2016). Nonetheless, many MSA points were used as weapon tips, and as components of weapons, they could hypothetically have played a role in signaling social relationships.

Prior considerations of regional diversification in MSA points are based on overall point shape differences on a continental scale, using the stone point exclusively. The stone point is the only component of the weapon recovered archaeologically. Unfortunately, wood and other organic remains used for binding do not preserve as well as stone in the archaeological record. When a person hafts a point to a shaft, the binding covers a significant portion of the base. Therefore, most of the base is not visible (Figure 1). Thus, a point is composed of two distinct modules distinguished by visibility. The base or proximal end functions to fit in the organic haft and is often obscured by hafting. The spear tip or distal end functions to pierce the prey and is visible enough to signal social information, potentially. This modular role of points has implications for evaluating the argument for regional traditions in the MSA. Different selective pressures on the proximal and distal ends may influence the likelihood of these components conveying stylistic information and meeting functional requirements for hafting, piercing, cutting, etc. For instance, Scerri’s (2013) analysis of North African Aterian points shows that the morphology of the proximal “tang” across the region is “remarkably homogenous” relative to overall tool form that is quite variable. In the case of Aterian tanged tools, the unique hafting design may necessitate greater standardization (Scerri, 2013). However, to our knowledge, no previous studies of point form have explicitly examined this segmented role of MSA points.

Figure 1. MSA point form variation showing that proximal variation would be visually obscured when point is hafted to a handle.

In this study, we investigate shape variation in the proximal and distal portions of one type of MSA point – the convergent flake-blade, using landmark-based geometric morphometrics (GM). GM is an efficient characterization of object form compared to linear measurements. GM has seen a rapid uptake in lithic analysis over the last decade. However, few studies have quantified the base (proximal) and tip (distal) shape variability in points. In biology, such divisions of modules within anatomical form provide essential information about the interactions working on shape variation within an organism, such as between the mandible and maxilla. Within lithic artifacts, the modules are expected to covary but also reflect the unique pressures of hafting standardization, cutting ability, penetration performance, and stylistic constraints. Shott and Otárola-Castillo’s (2022) 3D analysis of experimental Clovis points investigated how the stem and blade change shape independently due to multiple resharpening episodes as a means for quantifying the impact of allometry on Paleoindian point form. Identifying evident modularity of the base and tip in known projectile points provides support for analyzing shape variability within those two portions in points of more uncertain function in the MSA.

GM is a powerful tool for obtaining high-resolution quantitative data on point shape and has been applied effectively to MSA points in previous studies (Archer et al., 2015; Iovita, 2011; Schoville & Otárola-Castillo, 2008; Wilkins et al., 2012). While 3D GM has advantages by capturing a more complete set of shape dimensions (Shott & Otárola-Castillo, 2022) and will be used in future analyses, 2D approaches can also be fruitfully applied in some situations and is an appropriate choice in this case because MSA points are thin in profile view, and variation in shape is only visible at any distance across two dimensions. Additionally, we included published images of points in the analysis, which we could not have done with 3D analyses.

Our sample includes 680 MSA points from the southern coast archaeological sites of Klasies River Main, Pinnacle Point Cave 13B, and Die Kelders Cave 1, and the interior sites of Florisbad, Kathu Pan 1, and Ga-Mohana Hill North rockshelter (Figure 2 and Table 1). These sites were chosen because they cover a range of environmental contexts and rainfall regimes and include a large dataset of available convergent MSA flake-blade point images for 2D geometric morphometric shape analysis. For comparison, we also include a sample of 50 points produced experimentally by a single knapper using Table Mountain Sandstone Formation quartzite. We draw from Rogers and Ehrlich’s (2008) analysis of Polynesian canoes from early explorers and museum collections as a frame of reference. They show that the functional aspects of canoe design (i.e. those that affected sailing success) were under negative selective pressures and did not vary significantly across islands. In contrast, the stylistic aspects of the canoe design that did not impact sailing success showed higher variability and rates of change.

Figure 2. Topographic map of site locations in South Africa with rainfall zones from Roffe et al. (2021) indicated. Abbreviations as indicated in Table 1.

Table 1. Sample of MSA and experimental points and approximate age of the sample.

Site	Sample	Approximate age
Die Kelders 1 (DK1)	83	∼70 ka
Klasies River (KRM)	91	∼125–60 ka
Florisbad	56	∼121 ka
Ga-Mohana North (GHN)	17	∼105 ka
Pinnacle Point 13B (PP13B)	315	∼162–90 ka
Kathu Pan 1 (KP1)	69	∼500 ka
Experimental	50	NA
Total	681

Within this framework, we test the following three hypotheses:

Hypothesis 1: The proximal end of MSA points is more standardized than the distal end. Morphological constraints on the proximal end for hafting, which leads to greater standardization on the proximal end would be consistent with Hypothesis 1. This would be particularly true if attaching the proximal end to a handle strongly affected the functional performance of the point. Less functional constraints on the distal portion allowing for greater stylistic variation would also be consistent with greater distal variation and consistent with Hypothesis 1.

Hypothesis 2: The distal end of MSA points is more standardized than the proximal end. This hypothesis is consistent with the performance characteristics of points placing constraints on the distal end because of the need to penetrate targets when used as armatures and/or increased proximal variation reflecting greater stylistic design latitude.

Hypothesis 3: The proximal and distal ends of MSA points do not reflect any significant differences in variability, either because they are not functionally segmented or because the functional and stylistic pressures on each tend to cancel out any apparent differences.

Materials and Methods

The MSA Point Sample

As defined by Goodwin (1929), a typical MSA point is characterized by a triangular shape with convergent dorsal ridges. Many MSA points also show evidence of platform preparation. Retouch is generally minimal and mainly unifacial, if present, sometimes occurring on the edges and near the tip, but proximal thinning flakes are also present. There are multiple ways of producing pointed flakes and flake-blades that researchers classify as MSA points (McBrearty & Brooks, 2000). Even within individual assemblages at sites such as Kathu Pan 1 (Wilkins, 2018), Porc-Epic Cave (Pleurdeau, 2005), and Kudu Koppie (Wilkins et al., 2010), multiple methods of reduction were used to produce points. At Kathu Pan 1, Levallois reduction (sensu Boëda, 1995) with preferential and recurrent flake removals, and non-Levallois flake and blade cores were all used to produce points. This led Wilkins (2018) to argue that toolmakers appear to be focused more on producing pointed tools through emulative learning rather than imitative copying of the reduction process. Here we expand on this perspective by focusing on within-point variability rather than production variability.

We included points from six sites (Table 1) in the analysis covering a range of environments (e.g. coastal and interior) and rainfall regimes (winter, year-round, and summer) in South Africa (Figure 2). Klasies River Main site (KRM), Pinnacle Point Cave 13B (PP13B), and Die Kelders Cave 1 (DK1) are located on the southern coast of South Africa, whereas Ga-Mohana North rockshelter (GHN), Florisbad, and Kathu Pan 1 (KP1) are in the semi-arid interior.

The interaction of core preparation and retouching processes determines the overall shape of MSA points. Some points in our sample are minimally retouched. This means that the degree of imposition of form will be less than bifacially reduced points that appear in the Still Bay, Aterian, and later industries. Retouched points also go through greater reductive modification through use that we are able to side-step by focusing on unretouched and minimally retouched points from these six MSA assemblages (Figure 3). Allometric changes in shape as points are reduced to present a critical consideration for analyses of projectile points. However, those concerns are irrelevant to this study because retouch is minimally present on the MSA points in our sample.

Figure 3. Example of MSA point from the six archaeological and one experimental assemblage. Klasies River main site: pointed flake-blade no. 18 from MSA1 layers redrawn from Singer and Wymer (1982, p. 56). Florisbad: Upper F Layer, point 1008 (photo by. P. Chiwara-Maenzanise). PP13B: Lot 419 (Eastern Area Roof Spall-Upper), plotted find no. 30619. GHN: Lot 282 (DBSR), plotted find no. 4281 (photo by. P. Chiwara-Maenzanise). DK1: Lot 8324 (Layer 14), ID no. 1587. KP1: Stratum 4a, SPC2373 (photo by J. Wilkins). Experimental: produced by Kyle S. Brown on Table Mountain Sandstone quartzite.

Located near Cape St Francis in the year-round rainfall zone on the south coast, KRM is one of the original yardsticks of the MSA by which other sites are compared. There have been multiple excavations at KRM since the 1960s, first by Singer and Wymer (1982), then by Deacon in the 1980s (Deacon et al., 1986; Deacon & Geleijnse, 1988), and currently by a team led by Wurz et al. (2018). The MSA component of the site has been dated between ∼125 and 60 ka (Wurz, 2000). The sample of points included here was digitized from Singer and Wymer’s (1982) excavation monograph and includes 91 worked and un-worked points. Important to keep in mind with this sample of points is that they potentially represent a biased selection of points deemed fit for publication, rather than the complete assemblage as at other sites included here. However, given the importance of KRM for understanding variability in the MSA, it is crucial point of comparison to include here, albeit with that caveat in mind.

PP13B is a relatively deep coastal cave located near the town of Mossel Bay on the south coast of South Africa. The site was excavated by Marean and colleagues (Karkanas & Goldberg, 2010; Marean, 2010) and has MSA occupation dated to between ∼162 and 90 ka (Jacobs, 2010). The MSA points were part of the technological analysis completed by Thompson et al. (2010) and then examined for edge damage and macrofractures by Schoville (2010) and consist largely of quartzite from the Table Mountain Sandstone formation that is locally available. The PP13B assemblage consists of the largest sample of points in our analysis (n = 315).

Excavations at DK1 began with Schweitzer and colleagues in the 1970s targeting LSA deposits (Schweitzer, 1970), and resumed in the 1990s by a team led by Graham Avery expanding the MSA sample (Avery et al., 1997; Marean et al., 2000). The sample of points from DK1 (n = 83) are from the 1995 excavation of layers 6–12 originally reported by Thackeray (2000). The MSA layers at DK1 are not yet well-dated but may be ∼70 ka (Schwarcz & Rink, 2000).

GHN is located north of the town of Kuruman in the Northern Cape and was recently reported by Wilkins et al. (2020, 2021). The raw-material used at GHN is unique in the area, consisting of a locally available silicified tuff, as well as Banded Ironstone Formation, and chert. The sample of points reported here come from the DBSR, dated to ∼105 ka.

Florisbad is an open-air site located near a spring in the Free State, South Africa, near the town of Bloemfontein (Brink & Henderson, 2001; Kuman et al., 1999). The site has a well-preserved fossil mammal assemblage as well as a hominin cranium recovered in 1932 (Dreyer, 1935). The sample of MSA points used here are from the top of Unit F in the Eastern Eye. Unit F is dated to 121 ± 6 ka through ESR (Grün et al., 1996). A total of 56 points are included here from Florisbad.

KP1 is located near the town of Kathu in the Northern Cape, South Africa. The points from KP1 Stratum 4a are consistent with the earliest evidence for use as stone-tipped spears (Wilkins et al., 2012; Wilkins & Schoville, 2016), and are by far the oldest assemblage included here – dated to ∼500 ka through combined U-series/ESR dating (Porat et al., 2010). While the points from 4a are part of the “transitional” Fauresmith lithic industry, they are consistent with later MSA industries analyzed here. The points were produced with a diverse range of reduction strategies and retouch on different blank types produced similar pointed tools (Wilkins, 2018).

The experimental sample was produced by K. Brown for use in projectile experiments over the course of several years (e.g. Schoville et al., 2016; Schoville & Brown, 2010) and are included here as a comparative sample. Raw-material was collected from cobbles and outcrops of Table Mountain Sandstone near Mossel Bay, South Africa. The goal of these knapping sessions was to produce convergent MSA points for armature tips, and the questions addressed in this manuscript were not provided to the knapper. Therefore, no specific focus on maintaining specific proximal or distal shapes was employed. For instance, in subsequent experiments, the wooden hafting elements were cut and prepared for each individual point rather than attempting to produce points that fit a predetermined shaft morphology.

GPA and Procrustes Variance

GM is the study of variation in shape and how shape covaries with other variables (Bookstein, 1997). To quantify shape, landmarks are placed at homologous locations on every specimen and around the outline of each specimen. These landmarks are superimposed and shape variables are generated through a generalized Procrustes analysis (GPA; Slice, 2007; Zelditch et al. 2004). This superimposition method removes size, rotation, and location differences in specimen images and leaves only the shape information. However, size information may be used later if desired. We identified three technologically homologous landmarks: the platform’s corners and the point’s distal tip (Figure 4(A)). We then placed 23 sliding semi-landmarks placed along the tool outline to identify the outline of the tool shape between the homologous landmarks. These were divided into 13 contiguous landmarks that constitute the distal end of the tool, centered on the tip of the point, and 13 contiguous landmarks centered on the platform center on the proximal end (Figure 4(B)). Landmarks were digitized with tps dig2 (Rohlf, 2006). We performed GPA using R and the GeoMorph package (Adams & Otárola-Castillo, 2013).

Figure 4. (A) Photographs of points were taken on a 1 × 1 cm grid and landmarks placed around MSA point perimeter using tpsDig2; (B) Type 1 landmarks indicated as red circles, with proximal (gray) and distal (black) sliding semi-landmarks indicated for each portion of the point.

To measure shape variability and standardization, we quantified the morphological disparity among Procrustes aligned specimens from each site (Webster & Sheets, 2010). We used geomorph’s morphol.disparity() function to accomplish this. To compute morphological disparity of the proximal and distal halves, we used morphol.disparity() on each module at every site. We then computed the disparity difference between the two for each site. To assess whether differences between proximal and distal disparity values were statistically significant, we conducted a permutation test. The test compared the observed morphological disparity difference to differences from a null distribution. The null distribution of differences was calculated by randomly shuffling the specimen labels, then computing morphological disparity differences on each specimen of the shuffled set. We conducted this procedure 10,000 times, creating a null distribution of differences. We then asked how typical or extreme our observed values were when compared with the null distribution. Changes in shape due to size (i.e. allometry) were accounted for by using the Procrustes variance calculated from the residuals of linear regression on centroid size as a proxy measure of point size. These analyses were performed in R with the GeoMorph and RStudioApi libraries. All code is available through FigShare (10.6084/m9.figshare.21210995).

Results

The Procrustes variances of landmarks around the overall mean point shape, or disparity, are shown in Figure 5(A).

Figure 5. (A) Overall MSA point landmark dispersion around mean after generalized Procrustes analysis. (B) Distal landmark dispersion showing less variability compared to (C) proximal landmark dispersion.

Collectively, the proximal disparity (D = 0.202) is greater than distal disparity (D = 0.014), indicating that MSA points as a whole show greater standardization in the distal portion (Figure 5(A,B)). A permutation analysis of the D-values indicates this difference is extreme (p = 0.013) as shown in Figure 6. Difference in D-values equal to or greater than the observed difference occurs in less than 1.3% of permutations run 10,000 times.

Figure 6. Result of permutation of overall difference (D) between proximal and distal disparity. The observed difference (D = 0.0061, red line) is seen in fewer than 1.27% of permutations.

A comparison of proximal and distal disparity values by site are provided in Table 2 and Figure 7. At each site and including the experimental collection, the proximal disparity is greater than the distal disparity.

Figure 7. Permuted difference in disparity for each assemblage.

Table 2. Results of GPA and disparity analysis.

Site	Distal	Proximal	Difference	p-Value
PP13B	0.0142	0.0187	0.0045	0.0289
Experimental	0.0127	0.0147	0.002	0.1126
Florisbad	0.0151	0.0229	0.0078	0.0187
DK1	0.0127	0.0196	0.0069	0.0064
KP1	0.011	0.0161	0.0051	0.0186
KRM	0.0086	0.0116	0.003	0.0359
GHN	0.0121	0.0135	0.0014	0.2506

Note: Proximal disparity significantly (at p = 0.05) greater than distal are indicated with bold.

Within five of the archaeological assemblages, PP13B, Florisbad, DK1, KP1, and KRM, a permutation analysis of the observed difference between proximal and distal disparity indicates significantly greater proximal disparity (Figure 7). In other words, the distal end shows greater standardization relative to the more variable proximal end of the point. However, the difference between proximal and distal disparity is not significantly extreme in two of the samples examined here: GHN and the experimental assemblage.

Discussion

Our results best support Hypothesis 2 that the distal end of MSA points has greater shape standardization compared to the more variable proximal end. Given the role that negative selection plays in maintaining cultural traits that are functionally important, this may be explained by constraints on the distal end due to their functional role, e.g. use as projectiles or cutting tools depending on site context and land-use strategies. That the distal end shows less variability than the proximal end is inconsistent with hafted MSA points serving an emblemic role and communicating group membership.

It could be argued that one explanation for the distal end being more standardized than the proximal is that the proximal end exhibits more variation due to stylistic factors, perhaps due to hafting configuration variability. However, the proximal end is not visible in hafted armatures and is thus less likely to signal information to a social group. In this way, the analysis of proximal and distal variation problematizes Tostevin’s approach to identifying social transmission that emphasizes tool “visibility” to potential makers. Proximal variation could reflect less intentional aspects of style in that there are multiple strategies for point production and hafting that are functionally equivalent, and decisions about what strategies to employ occur in a social environment. Through this process shape variation in the proximal segment of the point can be imbued with social meaning, even though it is not visible once hafted.

The results of this analysis indicate that the distal end of MSA convergent points is often more standardized than the proximal end. While this might be expected across different point types (e.g. Still Bay, Aterian), it is interesting that even within one type of pointed artifact found in the MSA – the convergent point – has greater proximal variation than on the distal end of the point. This was found at cave sites on the south coast in the winter and year-round rainfall regions, as well as interior open-air sites in summer rainfall zone, across the time range of sites examined here (∼500–60 ka). By focusing on the relative standardization apparent in the proximal and distal portions of points within assemblages, we are able to look at the overall technological choices at six MSA sites. At five of these, the distal end is more standardized. Whether the modular role of proximal and distal parts of convergent points changed over time will require more assemblages with finer-dating resolution than is afforded by the sites analyzed here.

Another possibility relates to the descriptive typologies used by analysts in classifying points to begin with. Whether distal variation is minimized because of the way archaeologists classify MSA convergent flakes is an important avenue for future research. However, the results from the experimental knapping and GHN suggest that lithic taxonomy does not explain the observed patterning entirely.

In contrast to the five assemblages that show a more standardized distal end, points reproduced by a modern expert knapper do not fit this pattern – the proximal and distal end show equivalent disparity. This may support the notion that MSA points were preferentially produced with a more standardized distal end in mind – a potential “mental template” – that was not part of the decision-making calculus for a modern knapper. For this experiment, the goal was to produce triangular convergent flakes typologically like those found at MSA sites. No instructions were given about minimizing shape variance, so this provides a useful reference point for an assemblage of points produced without specific design criteria in mind other than to produce a convergent MSA point.

GHN also does not fit the pattern observed at the other archaeological sites in this study. While the absolute disparity is lower in the distal end, the observed difference is not as extreme as at the other sites, with observed difference in disparity greater than or equal to the observed difference occurring in 26% of permutations. Although the sample of points from GHN is the fewest of the sites examined here, it is worth noting that the observed proximal and distal disparity are both quite low relative to the other sites examined. In other words, the difference in disparity is not significantly different because both proximal and distal disparity is quite standardized. The functional analysis of points from GHN are still ongoing, therefore it’s difficult to say whether this pattern is due to functional differences. As Gravel-Miguel et al. (2021) show, MSA points are not discarded homogeneously on the landscape but reflect resource availability, extraction activities, and mobility patterns. Whether points at GHN served a different functional purpose, were discarded at different stages of production, or reflect different stylistic pressures may relate to the site’s geographic setting.

The spatial heterogeneity of point forms across Africa in the MSA may be manifestations of assertive style in terms of individual reciprocity rather than as emblemic objects used as indicators of ethnic affiliation (Wilkins, 2010). Two lines of evidence are used to support this critique. The first is that the regional scale of homogenous MSA point form exceeds the spatial extent of interacting adjacent bands. For instance, Aterian tanged points are present across much of North Africa, for perhaps 100,000 years (Richter et al., 2010). Using Whallon’s (2006) heuristic of hunter-gatherer social interaction, the spatial extent of adjacent interacting regional bands is ∼325 km – an order of magnitude less than the East–West distribution of Aterian points. Second, Wilkins (2010) notes that at the regional scale, much of the variation in point form is hidden within the hafted end. This point is illustrated in Figure 1, where differences between a stemmed or tanged Aterian base and a bifacial Still Bay point are obscured by the mastic and lashings by which they would be attached. The results of this study also support this argument – within convergent MSA points, there is greater proximal variation than distal. Whether the distal end is consciously standardized or is more akin to Sackett’s isochrestic variation is unclear. While this study examined six sites in South Africa, greater intra-regional distances and temporal resolution may show structured variability not examined here.

Conclusion

In this study, we problematize the notion that MSA points exhibit shape variation consistent with a role in conveying stylistic information. By examining the proximal and distal shape variation observed within one type of MSA point – the convergent flake-blade, we find that the more visible distal end is less variable than the proximal end. Cultural evolution models propose that cultural traits with stylistic meaning tend to be more variable than traits that are functional, thus our result is not consistent with distal point form having stylistic meaning. We examined proximal vs. distal shape variation in MSA points from six archaeological assemblages and one experimentally reproduced set of points. It was anticipated that the distal portion of MSA points would have greater variability than the proximal end because of the visibility of the distal end when hafted as a tool because points have been argued to reflect regional identities in the MSA, and because of functional constraints on hafting the proximal portion to handles or shafts. However, within all assemblages, the proximal shape variability is greater than the distal. This result is not consistent with uniform stylistic meaning imbued in the distal shape of MSA points. With additional sites from a greater range of environments and time periods, an important question remains whether this pattern holds across greater spatial and temporal scales. A greater understanding of intra-point shape variation in ethnographic and museum spears will also aid in interpreting temporal and spatial changes in point shape.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

B. J. S. is supported by the Australian Research Council Discovery Project DP220100167 and the Everick Foundation. P. C. M. is supported by GENUS (DSI-NRF Centre of Excellence in Palaeosciences), University of Cape Town Human Evolution Research Institute, and IFAS. Funding for this project was awarded to E. O. C. by the Purdue University College of Liberal Arts, the Office of Undergraduate Research Scholar Program, and the Margo Katherine Wilke Undergraduate Research Internship. J. W. is supported by an Australian Research Council Discovery Early Career Researcher Award (DE190100160) and Griffith University.