THE MISSING MOUSTERIAN - BY HAROLD L. DIBBLE AND SHANNON P. MCPHERRON

 
CONTINUE READING
Current Anthropology Volume 47, Number 5, October 2006                                                                                     777

                                     The Missing Mousterian
                           by Harold L. Dibble and Shannon P. McPherron

               Data from the Middle Paleolithic site of Pech de l’Azé IV (France) indicate the deliberate production
               of small flakes that appear to have been used in an unretouched state. Such materials have not
               typically been recognized as being intended end products, though various techniques for their man-
               ufacture are present in many Middle Paleolithic industries. This raises not only the question of their
               distribution and possible use during that time but also the larger issue of how Paleolithic archaeologists
               recognize lithic classes that are both analytically and behaviorally meaningful.

A significant component of Middle Paleolithic industries                        cance (Debénath and Dibble 1994; cf. Bisson 2000). Thus,
seems to have been overlooked in the current systematics.                       while most Paleolithic archaeologists are willing to accept that
What we attempt to show here is that during this time there                     scrapers, for example, were deliberately manufactured, it is
was a deliberate production of very small flakes that were                      much less certain that all of the some 18 scraper subtypes are
intended for use in their unretouched state. If our findings                    truly distinct (Dibble 1995a and citations therein) or that
are correct, then it is possible that these flakes represent a                  other types, such as pseudo-microburins, notched triangles,
significant aspect of Middle Paleolithic industrial variability                 alternate retouched beaks, rabots, or inverse choppers, were
and behavior even though such artifacts are typically unrec-                    deliberately manufactured to represent specific kinds of ob-
ognized in Paleolithic assemblages and certainly do not form                    jects. It is important, therefore, to begin to develop methods
part of standard descriptive analyses.                                          by which evaluations of existing types or artifact classes or
   In large part, this study is about more than just small flakes,              the proposal of new ones can be made.
since the demonstration of significance raises a larger question
as to how archaeologists recognize and define their basic units                 Methodological Background: Recognizing
of analysis. This is a serious issue in Middle Paleolithic ar-
chaeology, since many if not most analyses proceed either by
                                                                                Deliberate Products in Lithic
comparing frequencies of particular classes of objects (in-
                                                                                Assemblages
cluding named types and other more informal classes) or by                      Before beginning the presentation of data that suggest the
examining the technological processes underlying the man-                       deliberate production of small flakes, it is useful to outline
ufacture of certain classes. In either case a beginning as-                     some of the methodological problems that underlie such a
sumption is that the classes or types under investigation are                   demonstration.
somehow “real” in the sense of being behaviorally meaningful                       There are ways to identify deliberately produced end prod-
or otherwise interpretable because the processes underlying                     ucts in lithic assemblages, but all of them require particular
their production are understood.                                                characteristics or contexts that can be used to set them apart
   In Middle Paleolithic archaeology, most of the typological                   from the general background of knapping debris. For ex-
classes that constitute our basic units of analysis were origi-                 ample, the presence of retouch strongly suggests that pieces
nally defined in the first half of the twentieth century or even                exhibiting it were deliberately selected and their edge(s) mod-
earlier, though it was Bordes (1961a) who standardized many                     ified in order to perform a specific task. By analyzing what
of their definitions. Unfortunately, however, there have been                   characteristics particular kinds of retouched pieces have in
only sporadic attempts to examine critically each of the types                  common and in what ways they differ from nonretouched
to determine whether they have any real behavioral signifi-                     pieces, it is sometimes possible to arrive at some idea of the
                                                                                criteria used by prehistoric flintknappers for their selection.
                                                                                Likewise, the clear and unambiguous presence of use wear on
Harold L. Dibble is Professor of Anthropology at the University of
Pennsylvania (Philadelphia, PA 19104, U.S.A. [hdibble@sas.upenn                 pieces would also argue for deliberate selection for use. Finally,
.edu]). Shannon P. McPherron is Research Scientist in the De-                   by noting that particular forms are found in specific contexts
partment of Human Evolution of the Max Planck Institute for Evo-                (caches or burials, for instance), one can make the argument
lutionary Anthropology in Leipzig, Germany. The present paper was               that they served specific purposes.
submitted 31 III 05 and accepted 16 II 06.                                         All of these approaches, however, especially in the context

䉷 2006 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved 0011-3204/2006/4705-0003$10.00
778                                                                     Current Anthropology Volume 47, Number 5, October 2006

of Middle Paleolithic assemblages, are fraught with problems.       inally excavated from 1970 to 1977 by Bordes (1975). Overall
First, knowing that a blank was selected for retouching or for      he recovered about 90,000 lithic artifacts and another 30,000
use does not necessarily mean that the original blank was           faunal remains (McPherron and Dibble 2000). From 2000 to
purposefully manufactured to have those specific character-         2003 we performed new excavations at the site in an effort
istics—useful flakes can just as easily be selected from a pile     to collect fresh dating samples and to study its geological and
of knapping debris long after their original manufacture            taphonomic history. A final report on this work, including
(McDonald 1991). It is also recognized now that criteria for        dating of the complete sequence, is under way. The data pre-
discard are much more of a factor governing what enters the         sented here are based on the combined collections, and the
archaeological record than are criteria for use. This is one of     correspondence between the old and new stratigraphic se-
the main lessons from the work of Frison (1968) and sub-            quences is shown in table 1 (see Dibble, Raczek, and Mc-
sequent researchers who investigated changes in artifact form       Pherron 2005 for a complete discussion of how these two
due to resharpening and rejuvenation. This ultimately led to        collections were combined).
the recognition of the “finished artifact fallacy” (Davidson           One of the more interesting assemblages to come from this
and Noble 1993), which basically refers to the fact that what       site is one that Bordes named the “Asinipodian” (fig. 1), defined
we find in the ground is not so much what the prehistoric           originally (Bordes 1975) by the presence of very small Levallois
knappers wanted as what they wanted to throw away.                  cores and Kombewa cores (flakes whose bulb of percussion has
   Even objects found in special contexts have to be treated        been removed [see Brézillon 1968; Debénath and Dibble 1994])
with some skepticism. For example, at the site of Fonté-           and the general absence of retouched tools. From our own
chevade (France), the original excavator found many quartzite       analysis of the two collections it is apparent that a third class
cobbles in the sediments of the cave. Because these cobbles         of object, truncated-faceted pieces (Schroeder 1969), can also
could not have originated in the surrounding limestone, she         be considered a significant element in this assemblage. Al-
concluded that they must have been deliberately imported to         though, to the best of our knowledge, the term “Asinipodian”
serve as hammerstones (Henri-Martin 1957). Subsequent               has never been applied to any other assemblage, all of the
analyses have shown that they washed into the cave naturally        elements that make up this particular industry are known to-
through a chimney in the back (Dibble et al. 2006).                 gether or singly in many other Middle Paleolithic contexts.
   The goal here is to demonstrate that the production of a         There are several lines of evidence that suggest that all three of
particular class of objects—small flakes—was an intentional         these classes—Levallois cores, Kombewa cores, and truncated-
and deliberate result and not simply an unintended by-prod-         faceted pieces—were used to produce very small flakes, al-
uct of the knapping of other products. This demonstration           though the specific techniques differed considerably.
is admittedly much harder than the examples given above. In            Two technologies characteristic of the Asinipodian—Le-
the first place, all chipped stone technologies result in flakes    vallois and Kombewa—are critical for this demonstration. By
of all sizes, and a major proportion are relatively small because   its very definition (Debénath and Dibble 1994), Levallois is
of processes that include the preparation of platforms on cores     seen, at least in part, as a way of predetermining, through
and the retouching of tools. Therefore, the simple presence         preparation of the core surface, both the size and shape of
of small flakes, in any frequency, cannot be used to dem-           the resulting flake. Thus, when we find unmistakable Levallois
onstrate that they represent, in themselves, desired end prod-      flakes or cores that are exceptionally small, it is difficult to
ucts. Another problem, as described below, is that the small
flakes discussed here are not retouched and do not show any         Table 1. Stratigraphic Correlations between Bordes’s and
obvious signs of having been used. Moreover, they occur spa-        McPherron/Dibble’s Excavations
tially throughout the assemblage, and a large number of them
are identical to typical knapping by-products. Thus, the ar-                                  Dibble/McPherron’s
gument to be presented here that they were intentionally pro-       Bordes’s Level                   Level                   Global Level
duced will be based on analyses not of these small flakes
                                                                    F1-F3                            3A                           I-A
themselves but of the pieces from which they were struck.           F4                               3B                            I-B
                                                                    G, H1-H2                         4A                          II-A
                                                                    I1                               4B                           II-B
Small Flakes as Desired End Products:                               I2                               4C                          II-C
Data from Pech de l’Azé IV                                         J1                               5A                          III-A
                                                                    J2-J3                            5B                          III-B
The present study is based on analysis of material from the         J3A-J3B                          6A                          IV-A
site of Pech de l’Azé IV, a Middle Paleolithic site located in     J3C                              6B                          IV-B
southwestern France in the department of the Dordogne. Pech         X                                N/A                            –
IV, one of a series of Lower and Middle Paleolithic sites located   N/A                               7                             –
                                                                    Y-Z                               8                             V
close to one another that have been excavated since the mid-
nineteenth century (see McPherron and Dibble 2000; Mc-              Note: Global Level IV-A corresponds to the Asinipodian as described by
Pherron, Soressi, and Dibble 2001), is a collapsed cave orig-       Bordes.
Dibble and McPherron The Missing Mousterian                                                                                 779

                             Figure 1. Examples from Pech de l’Azé IV of the lithic classes described
                             in the text. 1–4, Levallois cores (2 and 3 made from a small knob of raw
                             material); 5–7, truncated-faceted pieces; 8–10, Levallois flakes; 11, Kom-
                             bewa core, 12, Kombewa flake. All objects from Level IV-A.

argue that, despite their size, their production was anything       to have two “interior” surfaces, one being the interior surface
other than deliberate. Likewise, Kombewa technology is de-          of the flake used as a core.
scribed (Brézillon 1968; Tixier, Inizan, and Roche 1980) as a         Truncated-faceted pieces have achieved recognition only
deliberate method of flake production in which the naturally        relatively recently and have previously gone under a number
convex surface of the flake’s interior surface at the bulbar end    of different names (Dibble 1984; Nishiaki 1985). They are
was exploited as a surface for the removal of a small flake.        flakes that exhibit a truncation on one edge which in turn
The resulting flakes are clearly identifiable because they appear   served as a platform for the removal of even smaller flakes
780                                                                     Current Anthropology Volume 47, Number 5, October 2006

                                                                    cores suggests that the same goal could be ascribed to the
                                                                    latter classes as well.
                                                                       Further evidence for a functional relationship among these
                                                                    classes is the fact that in the stratigraphic sequence of Pech
                                                                    IV, the percentage of Kombewa, truncated-faceted pieces, and
                                                                    small (less than 30 mm in maximum dimension) cores, in-
                                                                    cluding Levallois cores, show the same spike in occurrence
                                                                    in the Asinipodian (fig. 3). It could be that these classes pos-
                                                                    sessed complementary functions during the time of that oc-
                                                                    cupation (small flakes from Levallois cores being used for one
                                                                    thing while Kombewa cores or flakes and truncated-faceted
                                                                    pieces were used for others). However, what these elements
Figure 2. Frequency distributions of maximum scar lengths of        essentially share is the evidence of similar kinds of removals,
Levallois cores, Kombewa cores, “other” cores, and truncated-       which reached a peak during this particular time.
faceted pieces from Level IV-A.                                        There is no evidence that the flakes produced by these
                                                                    various techniques were intended to be fashioned into re-
                                                                    touched tools. While it is virtually impossible to recognize
(“facets”) from either the interior or the exterior surface         flakes removed from a truncated-faceted piece, examination
(Schroeder 1969; Debénath and Dibble 1994). They have been         of the Levallois and Kombewa flakes shows that they were
variously interpreted as a means of basal thinning for hafting      not generally retouched (only 68 out of 884 Levallois flakes
(Coon 1951), as tools in themselves (Leakey 1931, 99–100;           and 3 out of 21 Kombewa flakes). This is not surprising,
Dibble 1984), or as cores (Schroeder 1969; Solecki and Solecki      however. Many studies (Geneste 1985; Meignen 1988; Dibble
1970; see also Debénath 1988; Goren-Inbar 1988; Hovers             and Holdaway 1990; Dibble 1995a, 1995b) have shown that
2006; Dibble and McPherron 2006).                                   in the Middle Paleolithic flakes with retouch are generally
   While both Levallois and Kombewa are considered tech-            among the larger flakes in the assemblage, and data from the
niques for producing flakes, it is more difficult to assume that    Asinipodian and other layers from Pech de l’Azé IV (fig. 4
truncated-faceted pieces are also used to produce flakes rather     and table 2) continue to support this. Given that the average
than being tools in themselves or pieces exhibiting hafting         lengths of the flake removals presented earlier represent, on
modification. However, at Pech IV there do appear to be at          average, only half the length of the retouched tools, the small
least two aspects of underlying similarity for all three artifact   sizes of these flakes, including those from the truncated-fac-
classes. The first is the distribution of sizes of the flakes re-   eted pieces, probably made them unlikely to be selected for
moved from these pieces, and the second is their frequency          subsequent retouch. The fact remains, however, that they were
in the sequence of levels of Pech IV.                               manufactured, which is a strong argument that the intent
   All three classes of objects—Levallois and Kombewa cores         must have been to use them in an unretouched state.
and truncated-faceted pieces—exhibit negative flake scars rep-
resenting flakes that were removed from them, and the length        Potential Alternative Hypotheses
of these scars (the longest if there were several) was recorded
(on Levallois cores the measured scar represents the length         While it is clear that small flakes were being produced, it is
of the principal, or preferential, removal). The data on max-
imum flake scar lengths are presented in figure 2, and two
things are important to emphasize here. While the average
maximum flake scars on Levallois and other cores are statis-
tically significantly larger than those on truncated-faceted
pieces and Kombewa cores (t p 11.159, d.f. p 1935, P !
.00001), there is a high degree of overlap, and all of these
techniques produced very small flakes. In fact, the median
scar length on the Levallois cores is only 24.9 mm, which
means that fully half of the maximum scar lengths—and
therefore the length of the final flakes produced from them—
were less than this. The median flake scar on Kombewa cores
is even smaller, 19.63 mm. Thus, if both Levallois and Kom-
bewa cores were designed for the deliberate production of           Figure 3. Relative proportions of Kombewa flakes, truncated-
flakes, then the flakes they produced were quite small. By          faceted pieces, and small cores (less than 30 mm in length)
inference, the similarity in the distribution of scar lengths       through the Pech de l’Azé IV sequence. The spikes in all three
between them and the truncated-faceted pieces and “other”           occur in the Asinipodian Level IV-A.
Dibble and McPherron The Missing Mousterian                                                                                              781

                                                                       no evidence that these processes were any different during
                                                                       the occupation of Level IV-A.
                                                                          Third, many of the small Levallois cores from the Asini-
                                                                       podian levels were made on very small knobs of flint struck
                                                                       from larger nodules (see fig. 1, 2 and 3) and are not, therefore,
                                                                       the end products of continued reduction of originally large
                                                                       nodules.
                                                                          Finally, while it is possible that some of the Levallois cores
                                                                       may have produced larger flakes in earlier stages of reduction,
                                                                       this is not the case for Kombewa cores and truncated-faceted
                                                                       pieces, which almost always retain the overall form of the
                                                                       original flake.
                                                                          While there is no evidence that the emphasis on small flake
                                                                       production was a consequence of raw materials or a high
                                                                       degree of intensity of utilization of lithic resources, an alter-
                                                                       native possibility is that the removals were taken from Kom-
Figure 4. Average lengths of tools (columns) and flakes (solid line)   bewa cores and truncated-faceted pieces in order either to
from the Pech de l’Azé IV sequence.                                   prepare those pieces for a specific use or as a means of thin-
                                                                       ning them for hafting. Arguments against hafting of the trun-
nonetheless relevant to ask whether their small size was some-         cated-faceted pieces have already been made (see, e.g., Dibble
thing that was desired in and of itself or a consequence of            1984, 29). In part these are based on the fact that truncated-
something else. For example, the size of stone tools can be            faceted removals occur randomly with regard to specific clas-
affected by raw-material size and the degree to which cores            ses of tools (scrapers, points, notches, etc.) and that many
are reduced. Four lines of evidence suggest that neither of            truncated-faceted pieces exhibit multiple removals—for ex-
these is playing a significant role in the Asinipodian, however.       ample, from the proximal and distal ends or from both lateral
   First, the overall sizes of flakes and tools in the Asinipodian     margins. These same results hold true for the material from
level are, on average, not smaller than in levels that do not          Pech de l’Azé IV, where 50 out of 196 truncated-faceted pieces
show similarly high percentages of these small core techniques         exhibit such multiple removals. Having such removals from
(see fig. 4). Four of the assemblages in the Pech IV sequence          opposite margins of the piece is inconsistent with normal
have greater average flake lengths (II-B, IIC, III-A, III-B),          hafting modification. Likewise, there are no visible traces of
while six have small averages. In other words, in Level IV-A,          hafting wear or residues on these pieces or on the Kombewa
where a significant number of small flakes were being pro-             cores, though more thorough microscopic examination of the
duced by these three techniques, larger flakes were also being         pieces would be critical in addressing this problem. None-
produced in sufficient numbers and sizes to maintain a rel-            theless, the presence of the small removals from Levallois and
atively high average length of all flakes in the level. Unfor-         other cores cannot be explained by hafting or other use of
tunately, there is no independent means of assessing raw-              the cores.
material availability and size during the various occupations.
However, given that the overall average flake size in Level IV-        Table 2. Summary Statistics for the Lengths of Complete
A is similar to that seen in the other levels, it would not            Retouched Tools and Unretouched Flakes
appear that raw-material size or variability was any different
during the occupation of that level.                                            Retouched Tools       Unretouched Flakes
   Second, there is no evidence that the overall intensity of
utilization of the lithic resources was significantly higher dur-              Mean     N     S.D.   Mean      N      S.D.       T          P
ing the Asinipodian occupation. There are two measures of              I-A     43.22   266   13.05   33.49    3,066    9.57   ⫺10.08    !   .0001
overall intensity of utilization—the ratio of blanks to cores          I-B     42.05   520   13.46   34.36    4,246   10.22   ⫺11.66    !   .0001
(how many flakes are removed from a core) and the ratio of             II-A    50.79   117   14.48   35.56      406   11.92   ⫺13.96    !   .0001
tools to flakes (how many flakes are converted into retouched          II-B    53.76    84   14.13   38.52      192   12.11   ⫺14.26    !   .0001
                                                                       II-C    58.52   615   16.08   39.82    3,344   12.56   ⫺47.93    !   .0001
tools). The Asinipodian expresses values of both of these ratios
                                                                       III-A   56.08   108   13.67   37.72      759   11.91   ⫺18.29    !   .0001
the same as or even lower than those for the other assemblages         III-B   51.66   286   16.01   38.85    1,293   12.69   ⫺22.94    !   .0001
at the site (fig. 5 and table 3), which suggests that cores were       IV-A    43.88   336   13.49   36.79    4,056   11.34   ⫺12.36    !   .0001
not being more reduced at that time and that more of the               IV-B    40.04   157   11.07   34.23    1,395    9.43    ⫺4.25    !   .0001
larger flakes were not being retouched into tools, leaving be-         V       48.42   287   14.08   36.56    1,837   11.42   ⫺18.21    !   .0001
hind only small flakes. While both of these ratios can be              Note: Length is taken from the point of percussion to the most distal
affected by other processes (see Dibble 1995c), especially im-         end of the piece. In all of the assemblages, retouched pieces are signif-
port and export of material into and from the site, there is           icantly larger than unretouched ones.
782                                                                          Current Anthropology Volume 47, Number 5, October 2006

                                                                       artifacts may simply have been gripped in the hand. Recent
                                                                       studies of Neandertal hand morphology have shown that their
                                                                       hands were as capable as our own in terms of the range of
                                                                       movement (see Niewoehner, Weaver, and Trinkaus 1997; Nie-
                                                                       woehner 2001 and citations therein) but also show traits be-
                                                                       yond modern variation, including “unusually hypertrophied
                                                                       hand musculature, significantly increased mechanical advan-
                                                                       tages across many joints, unusually broad fingertips, and un-
                                                                       usual shapes and orientations of some of their CMC [car-
                                                                       pometacarpal] joints” (Niewoehner 2001, 2979). There is
                                                                       continued debate, however, about what kinds of behaviors
                                                                       would have generated these patterns. Generally, these traits
Figure 5. Ratios of blank to core (the number of complete and          are interpreted as adapted for a greater power grip (gripping
proximal flakes and tools divided by the number of cores) and          a hammerstone, for instance). However, when the direction
tool to flake (the number of complete and proximal tools divided       of force is considered along with its magnitude, it seems that
by the number of complete and proximal flakes). Although both          Neandertals were not as well adapted for the kind of oblique
of these measures can reflect other processes, they also monitor       joint reaction forces that are involved, for instance, in holding
the intensity of utilization of the lithic resources through either
                                                                       the handle of a hammer (Niewoehner 2001). Thus, Nie-
increasing core reduction or increasing tool production, and both
can result in smaller average sizes of the unretouched flakes. The     woehner concludes (p. 2980) that, despite some evidence of
fact that the Asinipodian level does not show relatively high values   hafting that would imply oblique power grips, Neandertal
of either suggests that these processes were not responsible for       hands were actually better suited to use “either with hand-
the emphasis on small flake production at that time.                   held stone flakes or with flakes hafted into the distal rather
                                                                       than the lateral aspects of handles” (see also Villemeur 1994).
   There are many reasons, then, to conclude that these As-            While these interpretations should not be taken as conclusive
inipodian levels from Pech de l’Azé IV reflect an emphasis            evidence that small flakes were hand-held, they are presented
on the production of very small flakes and that these flakes           here to show that there is no a priori reason to conclude that
were produced in a variety of ways. There is virtually no              the small size of these flakes would have prevented their use.
question that Levallois is a means of flake production, and               There is, however, a very important question as to the
the same is probably true—though not as certain—for the                function of these very thin and sharp flakes. There are sug-
Kombewa technique. These two techniques share important                gestions (Niewoehner 2001) that small flakes may have been
characteristics with each other and with truncated-faceted             used for woodworking; however, preliminary microwear anal-
pieces, and together all three classes resulted in very small          yses of 15 pieces from Pech IV yielded only two that showed
removals. Moreover, the small size of these removals does not          signs of having been used for short-term cutting of soft ma-
appear to be related to other factors, such as raw-material            terials (W. Banks, personal communication), and this may
size or degree of core reduction, and there is no evidence that        reflect a more expedient production and use of such flakes.
they represent hafting or other deliberate modification of the         This is clearly an area of research that should be pursued
original piece. Our conclusion is, therefore, that their small         further. However, a lack of understanding on our part of the
size was, in itself, a desired feature and that they were pro-
duced for use in their natural, unretouched state.
                                                                       Table 3. Counts of Complete Cores, Retouched Tools, Un-
                                                                       retouched Flakes, and Total Blanks (Retouched or Not)
Discussion
One possible objection to the arguments presented here is              Level         Cores          Tools         Flakes         Total Blanks
that the small flakes appear to be too small to have been used.        I-A            182            351           3,999             4,350
There is, of course, the possibility that they were hafted, per-       I-B            216            655           6,025             6,680
haps into composite tools, and there is evidence for hafting           II-A            18            141             590               731
in the Middle Paleolithic (see Churchill 2001 for a review).           II-B            16            103             301               404
                                                                       II-C           166            769           5,044             5,813
These studies are based on larger flakes (for example, Levallois
                                                                       III-A           46            132           1,091             1,223
points) or other retouched tools, however, and there are no            III-B          163            358           2,222             2,580
systematic studies of Middle Paleolithic artifacts of the sizes        IV-A           693            522           7,736             8,258
being emphasized here. However, examination of these pieces            IV-B           287            224           2,708             2,932
revealed no signs of hafting modification or residues.                 V              126            360           3,214             3,574
   Hafting of the small flakes is not, however, the only way           Note: Counts of both tools and flakes are based on complete or proximal
they could have been used. Alternatively, as Moncel (2003,             pieces (which accounts for the number of platforms present) and do not
46) notes in the context of a different kind of industry, small        include medial or distal fragments.
Dibble and McPherron The Missing Mousterian                                                                                   783

way Middle Paleolithic implements were used is not in itself       an additional purpose for their manufacture and transport.
a compelling argument against those pieces’ having been a          Our point, though, is only that there are many possible tech-
significant part of the general technological component. There     niques that can be used to produce small flakes, and this
is, for example, still considerable debate on the functions of     opens the door for new interpretations of particular tool types.
most Paleolithic implements, from bifaces, which could have           Moreover, it is possible that small flake production may
been tools or cores or something else entirely (e.g., Davidson     represent an important aspect of variability in the Middle
and Noble 1993; Kohn and Mithen 1999; Roberts and Parfitt          Paleolithic. As has been noted for some time (Dibble 1988a;
1999), to scrapers and so-called points (Gordon 1993; Dibble       Dibble and Rolland 1992), most Middle Paleolithic assem-
1995a) and notches and denticulates (which can also reflect        blages contain more or less the same range of major classes
taphonomic disturbances [Bordes and Bourgon 1951; Mc-              of tool types and technologies but differ in the relative fre-
Brearty et al 1998; Nielson 1991]). A better understanding of      quencies of those classes. Likewise, all of the techniques de-
the spatial and particularly the environmental distribution of     scribed here occur in most if not all Middle Paleolithic as-
small flake technologies could help address this, and exper-       semblages but vary in their relative proportions. In this regard,
imental studies would be useful to address the relative merits     the Asinipodian appears to represent an extreme emphasis
of small, thin flakes versus the edges exhibited on other Mid-     but probably only one end of a continuum of variability
dle Paleolithic elements, be they large unretouched flakes,        among Middle Paleolithic assemblages. Thus, it is likely that
scrapers, or notches.                                              such production, along with the production of other re-
                                                                   touched tools, bifaces, and particular technologies, represents
                                                                   a significant axis of assemblage variability and (as is still true
Further Implications
                                                                   of the others) one that definitely needs explanation. Having
If further research supports the conclusion that Middle Pa-        said that, we do not concur with Bordes (1975) in elevating
leolithic hominins deliberately manufactured and used small        the status of the Asinipodian to the level of another named
flakes, it would have significant implications for the inter-      facies of the Mousterian. Rather, for whatever reason, activities
pretation of Mousterian assemblage variability. In the Asi-        at this particular time and place were suited for this kind of
nipodian, the production of small flakes was based on three        production to a degree not previously observed.
particular techniques, and, to the best of our knowledge, in          Aside from what it means for Middle Paleolithic behavior,
this regard this particular assemblage appears to be unique.       the production of small flakes also raises important meth-
However, it is known that many other Middle Paleolithic            odological issues in terms of techniques of excavation as well
industries exhibit one or more of these techniques. For ex-        as the analysis and description of Middle Paleolithic assem-
ample, Moncel (2003 and citations therein) reports on other        blages. It is common practice to provenience, number, and
Middle Paleolithic industries that exhibit small cores, and        analyze lithic objects only above a minimum size—usually 2
truncated-faceted pieces clearly exist (under a variety of         to 2.5 cm. The flakes that were being deliberately produced
names) in assemblages from Europe (Turq and Marcillaud             by the techniques described here were often much smaller
1976; Roth, Lenoir, and Dibble 1995), the Levant (Schroeder        and thus easily overlooked. Even if such cutoffs are lowered,
1969; Solecki and Solecki 1970; Crew 1976), the Zagros (Dib-       the problem of identifying deliberately produced small flakes
ble 1984; Dibble and Holdaway 1990), and Africa (Leakey            among the debris produced through core reduction and tool
1931). Kombewa technique is also widespread through Eu-            production still remains. While flakes removed from Levallois
rope and Africa (Owen 1938; Dauvois 1981) though it is             and Kombewa cores are generally recognizable, it is less clear
virtually absent in the Zagros or Levantine Mousterian. Like       for those coming from truncated-faceted pieces and perhaps
many other aspects of Middle Paleolithic industries, small         other techniques of small flake production as well.
flake production may be very widespread, though it may be
associated with several different techniques.
                                                                   Conclusion
   Furthermore, there may be other, still unrecognized tech-
niques for small flake production. One distinct possibility is     One of the most important contributions to Paleolithic ar-
Clactonian notches. Rather than the notch itself, the flake that   chaeology was the standardization of the typology for lithic
was removed from the notch may have been the intended              artifacts put forward by Bordes during the middle of the last
end product, since it would possess many of the same char-         century (Bordes 1961a). His typology of flake tools consisted
acteristics as the small flakes made with the other techniques     of 63 named types, most of them defined by the type and the
described above. Another possibility is handaxes. While most       placement of retouch on flakes and some by the technology
have focused on these artifacts as tools with heavy-duty cut-      used to create them (Debénath and Dibble 1994). His second
ting edges, it is also widely acknowledged that they are a ready   major contribution (see Sackett 1981) was to apply this ty-
source of sharp flakes whose morphology would appear to            pology to Mousterian assemblages of northern and south-
be quite similar—that is, small, thin, and extremely sharp.        western France, and on that basis he developed the assemblage
While we are not concluding that handaxes existed for the          systematics—the various Mousterian facies—that are still rec-
sole purpose of flake production, it could very well represent     ognized and used to this day. The significance of these con-
784                                                                     Current Anthropology Volume 47, Number 5, October 2006

tributions is undeniable given the tremendous amount of re-         suggestions. Funding for the excavation and analysis of Pech
search that they stimulated concerning the factors that             de l’Azé IV was provided by the Leakey Foundation and the
underlie variability in the types themselves (e.g., Dibble 1995a;   National Science Foundation (Grant No. BNS0073856).
McPherron 1994, 1995) and the assemblages that are com-
posed of them (e.g., Bordes 1961b; Binford and Binford 1966;
Mellars 1969, 1996; Rolland 1981; Rolland and Dibble 1990).
Much of this work has also focused on fundamental anthro-
pological concerns, using these same lithic types as evidence       Comments
for the evolution of language (e.g., Holloway 1981; Gowlett
1984; Dibble 1989), culture (Holloway 1969; see Gibson and          Nick Ashton
Ingold 1993), and intelligence (Wynn 1979, 1985).                   Department of Prehistory and Europe, British Museum, 56
   As is true in any empirical discipline, research tends to be     Orsman Road, London N1 5QJ, UK (nashton@thebritish
focused on units of analysis defined on the basis of particular     museum.ac.uk). 28 IV 06
criteria (see papers in Ramenofsky and Steffen 1998). Thus,
in the case of Paleolithic archaeology, for a long time these       Dibble and McPherron have identified some interesting pat-
units were largely retouched pieces, and with the exception         terns in the lithic data from Pech de l’Azé IV. Their arguments
of Levallois flakes and cores most unretouched pieces were          centre on the distinctiveness of Level IV-A, in particular the
related to the status of “waste” or by-products. After all, it      production of Levallois and Kombewa cores and truncated-
was assumed for many years that retouched tools represented         faceted pieces, and the idea that these pieces represent the
the desired end products, and so naturally they were the focus      intentional production of small unretouched flakes. This
of most research. This situation has changed over the past          opens up questions relating to Neanderthal power grip or
two decades, however, as interest in the underlying technology      hafting and the effects of a previously unrecognized category
of blank production has increased (Marks and Volkman 1983;          of tool on interpretations of Mousterian assemblage vari-
Boëda 1986; see papers in Dibble and Bar-Yosef 1995). None-        ability. The data that they present indicate that all cores are
theless, even in a technological approach, where the emphasis       smaller in Level IV-A, flake scars from intended products
is on reconstructing the processes that resulted in particular      generally vary between 10 and 35 mm in maximum length,
flake forms, the need still exists to understand what the in-       average flake lengths are similar for all levels at Pech de l’Azé
tended results were (Bar-Yosef and Dibble 2005). To put it          IV (means 33–40 mm), there is less conversion of flakes into
simply, it is irrelevant to know how certain things were made       tools in Level IV-A, and there are fewer flakes per core in
unless we know what their producers were trying to make.            Level IV-A. Dibble and McPherron argue that because of the
In a similar vein, while we desperately need to know more           similarity in flake sizes between the different levels, raw-ma-
about the functions of these objects, such studies also require     terial size was also similar. However, the less intense working
that potentially informative classes of material be identified.     of the cores in Level IV-A despite their smaller size suggests
Thus, this paper is not at all about whether a new “type”           that nodule size was also smaller. The size of flakes removed
should be added to Bordes’s type list but focuses on the prob-      from a core will depend on the size of the original nodule,
lem of being able to recognize meaningful units in the ar-          the intensity of working, and the type of technology deployed.
chaeological record.                                                Therefore the question is to what extent the observed pattern
   What we have tried to do here is to present arguments            is based on the use of different technology, which itself may
based on several lines of evidence that Middle Paleolithic          have been determined by differences in the size and, impor-
hominins deliberately produced small flakes for specific func-      tantly, the shape of the nodules.
tions. If this conclusion is true, then this fact has been over-       A further problem lies in how certain it is that the removals
looked for virtually the entire history of the discipline. It       that have been measured were intended products. With Le-
represents, then, a component of Middle Paleolithic tech-           vallois, this is commonly assumed, but when some of these
nology and behavior that is, in effect, missing from our con-       (3–4%) were as small as 10–15 mm (fig. 2) doubts must arise
cepts of what constitute behaviorally significant aspects of        about the intentions of the makers. Could some of these cores
Middle Paleolithic industries. While future work may clarify        have been in the process of repreparation when a decision
the specific role that theses flakes played in Middle Paleolithic   was made to discard them? From a Lower Palaeolithic per-
adaptation, that they have been overlooked for nearly a cen-        spective, knapping frequently seems to go beyond the point
tury and a half reminds us once again of how little we truly        of utility. One example is at High Lodge (Suffolk, UK [Ashton
understand about those past lifeways.                               et al. 1992]), where a refitting knapping scatter shows the
                                                                    reduction of a partly worked core through the removal of 30
                                                                    further flakes. Several flakes are missing from the middle of
Acknowledgments                                                     the knapping sequence, but most of the final flakes have been
We thank several colleagues, including the seven anonymous          discarded with the scatter, presumably as waste products. It
reviewers of this manuscript, for their helpful comments and        would seem that the core was rapidly flaked and discarded
Dibble and McPherron The Missing Mousterian                                                                                          785

and several flakes were then selected for use elsewhere. Despite    Table 1. Levallois and Kombewa Flakes, Retouched Flakes,
a different level of planning and production in Levallois tech-     and Cores from Level IV-A at Pech de l’Azé IV
nology, there is no reason that the same process in terms of
automated, procedural knapping and tool selection (either                              Flakes           Retouched Flakes             Cores
blanks or unretouched flakes) was not taking place. In other
                                                                    Levallois            816                   68                      80
words, just as we have to beware of the “finished artefact          Kombewa               18                    3                      35
fallacy,” we should also be aware of the “finished core fallacy.”
   With the other techniques there must also be doubts about        Note: Source: Dibble and McPherron’s figures 2 and 3 and text.
intention. The apparent removal of Kombewa flakes can also
be a by-product of the knapping process, where a single blow        a topic I myself have been studying for many years, though
can remove both the flake and a flake resembling a Kombewa          in a totally different geographical, chronological, and cultural
flake from the same point of impact. This is potentially the        context (Bouzouggar 1997a, 1997b; Bouzouggar, Kozlowski,
case in figure 1 (11 and 12). In terms of the original and          and Otte 2002). I especially welcome their work because the
broader definition of Kombewa, in which flakes were removed         data concerning this technology come from Pech de l’Azé IV,
primarily from the lateral edges and ventral face of another        a cave with a long stratified record of human activity and
flake, there is a reasonable certainty of their having been         well-preserved organic remains. The site was recently exca-
intentionally knapped (Owen 1938). With truncated-faceted           vated by Dibble and McPherron, and I had the opportunity
pieces, as Dibble and McPherron acknowledge, it is not              in 2002 to observe the high level of precision of field recording
known whether they were tools or cores.                             using the total-station system—a system I have subsequently
   It would have been interesting to see more figures for each      adopted in my excavations in Morocco.
of the levels, such as average lengths for each major artefact         However, there are a number of points arising from their
category. For example, how do the average lengths of Levallois      paper that require further explanation and clarification. To
and Kombewa flakes tie in with the figures obtained from the        begin with, the Asinipodian is still not fully explained in terms
cores? One intriguing statistic that can be derived from the        of its technological, chronological, or palaeoenvironmental
figures presented is the apparent production of up to 11 Le-        contexts. As far as the technology is concerned, for Dibble
vallois flakes from each core, including roughly one per core       and McPherron the Asinipodian is a technology exhibiting
that is retouched (see my table 1). This in itself is perhaps       basically Levallois and Kombewa techniques and the trun-
not surprising, but it underlines the problems of identifying       cated-faceted method. Each of these techniques produces
intentional Levallois products as opposed to by-products. In        small flakes. It is equally important, however, to know the
contrast to this there is a deficit of 40% (14) in the number       function of these small flakes. Secondly, it would be inap-
of Kombewa flakes in relation to Kombewa cores. Have these          propriate to extrapolate the definition of the Levallois to the
flakes been taken away, or does this reflect a problem of           small Levallois cores without controlled experiments. I am
identification?                                                     not suggesting creating a new term but simply pointing to
   The suggestion that unretouched flakes (perhaps sometimes        the need to verify whether the same definition fits the small
small) were used is not in doubt, but whether there was             cores or not. Another area for discussion concerns the dis-
planned production of small flakes remains unanswered. In           tinction based on technological criteria between the small
the absence of use-wear, any attempt to quantify the extent         Levallois flakes and those originating from truncated-faceted
to which unretouched flakes were used is clearly problematic.       cores (McPherron, Soressi, and Dibble 2001, 21). Here again
As I am sure Dibble and McPherron would agree, this also            replication studies are needed, as well as analysis of the use
draws attention to the problem of using simply quantitative         wear on both flakes and cores (especially Levallois and Kom-
differences in tool types between assemblages when attempt-         bewa cores) of this technology. Larger nodules are not abun-
ing to identify cultural groupings in the Mousterian.               dant around Pech de l’Azé (Dibble 1985, 392) but a persistent
                                                                    problem concerns the independent means of estimating past
                                                                    raw-material availability, an issue raised by Dibble (1991)
                                                                    many years ago. However, very little evidence exists for the
Abdeljalil Bouzouggar                                               occupational intensity of the site. The increasing number of
National Institute of Archaeological Sciences and Heritage,         small cores and flakes during this occupation could be con-
Av. John Kennedy, 10 000 Rabat, Morocco (bouzouggar@                nected with the fact that the site was a potential source of
menara.ma). 29 IV 06                                                residual raw material, as has been suggested for the Middle
                                                                    Paleolithic of layer 10 at Roca Dels Bous northeastern Spain
Despite the wealth of evidence for the Middle Palaeolithic in       (Mora, de la Torre, and Moreno 2004), and probably exploited
France, surprisingly little is known about the Asinipodian          for making expedient tools.
(Bordes 1975), and Dibble and McPherron’s paper is helpful             Without seeking connections between the different worlds
in redressing this imbalance. It represents a fresh opportunity     of European Neandertals and North African Homo sapiens,
to discuss and share information about small cores and flakes,      small cores also exist in the Aterian (a subdivision of the
786                                                                     Current Anthropology Volume 47, Number 5, October 2006

Middle Paleolithic/Middle Stone Age in North Africa) [An-           2000), and the production of composite tools with hafted
toine 1938 and Tixier 1958–59]), and I have suggested a pat-        microliths is generally thought to be part of the modern hu-
tern of exploitation based on experimental knapping in which        man behavioral “package” that was brought into Europe in
it appears that small pebbles were broken into two pieces with      the hands of Homo sapiens. Documenting microlithic tech-
the anvil technique before the preparation of the Levallois         nology among Neandertals (the presumed makers of the
core surface (Bouzouggar 1997a, 1997b). In the Aterian, some        Mousterian at Pech IV) would contribute to our understand-
of the small flakes were transformed into pedunculate pieces        ing of their cognitive abilities and adaptive strategies and
that were probably hafted but left no wear traces (de Igreja        remove from consideration one of the technological behaviors
Araujo, personal communication). Another idea that needs            that seemingly separates modern humans from Neandertals.
to be verified is whether small cores (especially Levallois ex-     Thus the discovery of mircolithic technology in the French
amples) were themselves made as tools because the scar of           Mousterian would be of great significance. But here the old
the preferential removal would allow easy manipulation with         saw about extraordinary claims’ requiring extraordinary evi-
the thumb (Antoine 1938, 19).                                       dence applies, and the evidence in this case does not seem
   I am in basic agreement with Dibble and McPherron in             up to the claim.
that they do not attempt to consider the Asinipodian as a              Dibble and McPherron recognize that small flakes are a
new cultural facies, since it is likely that the Mousterian of      ubiquitous by-product of lithic reduction and therefore their
the Old World is probably a highly flexible technology. Much        presence in an assemblage is not evidence in and of itself that
of the interest in the Old World Mousterian lithic technology       small flakes were the desired end product of knapping. Bona
has tended to concentrate on the very narrow thematic issue         fide microlithic tools of the Upper Paleolithic and Mesolithic
of retouched tools rather than considering a broader set of         of Europe, even of the earliest Upper Paleolithic, are recog-
variables such as the presence of small flakes and cores.           nizable end products of reduction because they show evidence
   As mentioned above, Pech de l’Azé IV continues to offer         of use in the form of either retouch or edge wear (e.g., re-
rich sources of multiproxy data for palaeoenvironmental and         touched Dufour bladelets in the basal Aurignacian). Dibble
chronological studies (McPherron, Soressi, and Dibble 2001).        and McPherron are not able to present such direct evidence
An examination of the detailed palaeoenvironmental evidence         in the Pech de l’Azé case. The evidence they do cite—the
should allow fuller consideration of questions regarding the        presence of very small Levallois cores, Kombewa cores, and
rates and scales of climatic and landscape change and how           truncated-faceted pieces—is problematic. As they note, the
these might have impacted human behavior. This is probably          truncation and faceting of flakes may represent basal thinning
an important gap in our knowledge concerning the Asini-             to facilitate hafting, and the Kombewa “cores”—flakes that
podian, and I hope it will be filled with new studies now in        are considered “cores” on the basis of a single flake removal—
progress.                                                           are also parsimoniously interpreted as flakes subjected to basal
                                                                    trimming. The suggestion that truncated-faceted flakes and
                                                                    Kombewa cores served for the production of microliths would
                                                                    be more compelling if all of these pieces were systematically
Steven E. Churchill                                                 examined and found to be free of traces of hafting resins or
Department of Biological Anthropology and Anatomy, Box              wear polish. In both cases one has to wonder why knappers
90383, Duke University, Durham, NC 27708, U.S.A.                    would have gone to the trouble to produce flakes from which
(churchy@duke.edu). 30 IV 06                                        they then struck one or two microliths rather than simply
                                                                    driving small flakes off of small nodules. The interpretation
Dibble and McPherron are to be commended for an inter-              of Kombewa flakes is equally problematic, and more evidence
esting article on the possible deliberate production of very        is needed to demonstrate that they are not simply debitage
small flakes in the Mousterian at Pech de l’Azé IV and a           from reduction of the bulb of percussion (3 of 21 Kombewa
worthwhile discussion of the difficulties inherent in identi-       flakes show signs of retouch, which is consistent with the
fying the correct units of analysis when interpreting the lithic    intensity of retouch on Levallois flakes from the assemblage
record. They focus on the importance of recognizing small           and may support the claim that they are intentionally pro-
flake production in the Mousterian as a prerequisite for un-        duced microliths, but what percentage of the Kombewa flakes
derstanding assemblage variability, but I would say that its        evinces use wear?). Finally, the existence of small Levallois
importance goes beyond the way we interpret Mousterian              cores is not in and of itself compelling evidence for microlith
assemblages. When it comes to the European record, the pro-         production. Dibble and McPherron argue that the generally
duction and use of microliths—Mode 5 technology as iden-            small size of the flake scars on these cores reveals a desire on
tified by Desmond Clark (1968)—has been confidently iden-           the part of the knapper to produce small flakes. But what
tified only in Upper Paleolithic assemblages. Relatively large      matters is not the size of core preparation scars (which simply
microliths (2.0–2.5 cm in length, similar in size to the possibly   reflects the average size of debitage) but rather the length of
purposeful small flakes at Pech IV), appear late in the Middle      the privileged flake. Judging from the cores illustrated in figure
Stone Age of southern Africa (see McBrearty and Brooks              1, the smaller of the two cores illustrated would have produced
Dibble and McPherron The Missing Mousterian                                                                                 787

privileged flakes on the order of about 3 to 4 cm in length—       combined distribution of Kombewa and truncated-faceted
small but not microlithic. It does raise the question why Nean-    pieces and small cores throughout the Pech IV sequence (fig.
dertals at Pech de l’Azé were interested in relatively small      3), higher percentages of small flakes occur in other assem-
flakes (Dibble and McPherron make clear that it is not a case      blages in the sequence (fig. 4). Again, there is no direct cor-
of limitations in the size of the raw materials as it is, for      relation between distributions of small cores and small flakes.
example, in the pebble-derived Italian Pontinian [Kuhn                Finally, the idea of identifying any universal “function” for
1995]), but the evidence as presented does not seem to war-        all morphologically similar lithic pieces seems faulty. On the
rant the claim that there is an important microlithic com-         contrary, one can suggest functional variability within artifact
ponent to the Mousterian that we have been missing all these       classes in temporally and geographically different archaeo-
years. The suggestion of a microlithic component to Mous-          logical contexts. For example, in the Zagros Mousterian as-
terian assemblages is an interesting one, and further inves-       semblages of the southern Caucasus the truncated-faceted
tigation of this possibility may ultimately lead to the kind of    technique was used to accommodate hafting of the bases of
direct evidence required to substantiate the claim.                points and other tools, while in the eastern Gravettian of the
                                                                   Russian Plain this method was exploited to rejuvenate the
                                                                   cutting edges of Kostenki knives. In Russian Paleolithic ar-
                                                                   chaeology, the truncated-faceted technique was originally
Vladimir Doronichev and Lubov Golovanova                           identified as a “corelike method of thinning.” Also, the Kom-
Institute for the History of Material Culture, Dvortsovaya         bewa technique, though rarely discussed in Russian archae-
nab. 18, 191186 St. Petersburg, Russia (lprehist@rol.ru). 28       ology, is generally considered a method of reducing the bulbar
IV 06                                                              part of a flake to accommodate hafting. Regarding small Le-
                                                                   vallois cores one can suggest that some of them were reduced
Dibble and McPherron’s main goal is to demonstrate that the        centripetal cores, especially in industries with no large Le-
production of small flakes was a deliberate result and not         vallois cores or flakes such as the Asinipodian level of Pech
simply a by-product of the knapping or retouching of other         IV. In true Levallois assemblages with a variety of large, me-
artifacts. Given the impossibility of drawing conclusions from     dium, and small Levallois flakes and cores, the latter were
an analysis of the small flakes themselves, they examine the       apparently preferentially used for removing larger flakes of
pieces from which small flakes were struck—small Levallois         predictable morphology. In any case, the study of small tools/
cores, small Kombewa cores, and truncated-faceted pieces.          flakes is related to two general problems: (1) the existence of
They accept a priori that these three artifact classes were used   special (hence culturally specific) lithic technologies for pro-
to produce small flakes while differing from one another in        ducing small flakes or the deliberate exploitation of small
the mode of production. The main problem here is that they         flakes made from scarce high-quality raw material and (2) the
represent either desired or undesired/reduced core types, de-      absence of most small lithics in most old excavations and the
sired or undesired/unfinished tool types. Originally Levallois     poor preservation of many Paleolithic collections.
and Kombewa were described as special types of large flakes           In general, Dibble and McPherron’s approach is a very
and then their definitions were applied to particular flaking      important attempt to rejuvenate Paleolithic research para-
techniques for producing them. Similar morphology does not         digms and analytical procedures.
mean that small Levallois and Kombewa pieces are desired
cores or represent the same deliberate method of flake pro-
duction as large Levallois or Kombewa cores. This is best
understood when we turn to the third group, truncated-fac-         Metin I. Eren
eted pieces. Contrary to the case with the previous groups,        Department of Anthropology, Southern Methodist
the term “truncated-faceted” was originally applied to small       University, Box 750336, Dallas, TX 75275-0336, U.S.A.
artifacts variously interpreted as results of thinning for haft-   (meren@smu.edu). 17 IV 06
ing, as tools in themselves, or as cores.
   From the analysis of the Pech de l’Azé IV material Dibble      Dibble and McPherron’s lithic analysis at Pech de l’Azé IV
and McPherron make two observations that contradict their          shows how detailed reduction models can delineate prehis-
hypothesis that small Levallois, Kombewa, and truncated-fac-       toric knapping behaviors. Without use-wear or residue anal-
eted pieces were core types deliberately made to produce small     ysis, proving that small, unretouched blanks are intentional
flakes. (1) As figure 2 shows, there is a similarity in the dis-   products is no simple task, and they should be applauded for
tribution of scar lengths between these three groups and           their efforts. Rather than using a static typology of assumed
“other” cores, suggesting that small flakes were struck from       end products, they employ a dynamic typology of process that
many if not all cores in this assemblage and therefore that        identifies an overlooked Middle Paleolithic knapping strategy.
there is no direct correlation between scar length and any         Their study has implications for how lithic analysts recognize
core type. (2) Although the Asinipodian Level IV-A has the         meaningful type categories, and their methods should be
highest percentage of these small cores, including 50% of the      emulated.
788                                                                     Current Anthropology Volume 47, Number 5, October 2006

   Still, I have a few concerns. Although they mention that         small flake production intensity is similar between Levels II-
“there is no independent means of assessing raw-material            A, II-B, and IV-A, and import and export processes are indeed
availability and size during the various occupations” at the        similar among all the levels as stated, why do Levels II-A and
site, raw-material availability and transport may present larger    II-B have high core-to-blank and blank-to-tool ratios (fig. 5)
problems than they concede. Small flake production may in-          while IV-A does not? Discussion of other levels may provide
deed have been intentional (after all, small flakes were pro-       insight and context for the patterning seen in Level IV-A.
duced), but intentionally producing a small flake because              It remains uncertain whether truncated-faceted pieces are
one’s hand is adapted to holding it and intentionally pro-          cores. Dibble points out elsewhere that there is no consensus
ducing a small flake because one is running out of raw ma-          on whether they were used for producing small flakes or for
terial are very different.                                          producing a particular kind of edge (Débenath and Dibble
   Although there is a large number of small cores in Level         1994, 123). One wonders if use-wear or residue analysis on
IV-A (fig. 3), the average tool and blank lengths are not the       such pieces might distinguish between these two possibilities.
smallest when compared with other levels. The authors ex-           Finally, Dibble and McPherron state, “Many of the small Le-
plain that “larger flakes were also being produced in sufficient    vallois cores from the Asinipodian levels were made on very
numbers and sizes to maintain a relatively high average length      small knobs of flint struck from larger nodules.” How many
of all flakes in the level [IV-A].” It would be good to know        exactly? This has implications for how many Levallois cores
where these large flakes came from. Are there large unex-           were reduced from larger sizes and thus how many small flakes
hausted cores in Level IV-A, and, if so, how many? The pres-        are possibly by-products.
ence or absence of large cores may strengthen or weaken the
argument. Three scenarios illustrate the point:
   Scenario 1: Large cores are present. Limited raw-material
availability is not responsible for small flake production, since   Bruce L. Hardy
the option for creating larger flakes exists.                       Department of Anthropology, Kenyon College, Gambier,
   Scenario 2: Large cores were subsequently transported off-       OH 43022, U.S.A. (hardyb@kenyon.edu). 30 IV 06
site. Though these produced enough large flakes to increase
the overall blank size, the cores were still large enough to be     Dibble and McPherron are to be commended for thinking
taken away for future use. If this is the case, the blank-to-       beyond the normal constraints of typology. Although typology
core ratio will be even lower than figure 5 indicates, again        has certainly been useful in many ways, it has also served as
implying that limited raw-material availability is not respon-      blinders for Paleolithic research. While it is a necessary evil,
sible for small flake production.                                   it is not an end in itself.
   Scenario 3: If large cores are not present in Level IV-A and        While they provide technological evidence for production
were not transported off-site, the only possible explanation        of small thin flakes from truncated-faceted pieces, Levallois
for the presence of large flakes is that large cores were reduced   flakes, and Kombewa flakes, the only information they present
into small cores (not including Kombewa or truncated-faceted        relative to functional analyses is that “preliminary mirowear
pieces). This implies raw-material shortage, indicating that        analyses of 15 pieces from Pech IV yielded only two that
many of the small flakes were probably by-products of large         showed signs of having been used for short-term cutting of
flake production and only some of them were intentionally           soft materials.” To be fair, they recognize that this is insuf-
produced toward the end of the reduction sequence.                  ficient, stating that “this is clearly an area of research that
   While there is no way to detect scenario 2, the presence or      should be pursued further.” However, their case for the “miss-
absence of large cores coupled with an analysis of large blanks     ing Mousterian” would have been greatly strengthened had
may help distinguish between scenarios 1 and 3. Of course,          some form of functional analysis (either use-wear or residue
there is probably some degree of overlap among them.                analysis) been conducted.
   The authors state, “Most Middle Paleolithic assemblages             Citing Niewohner (2001), Dibble and McPherron suggest
contain more or less the same range of major classes of tool        that Neandertal hand anatomy would not have been efficient
types and technologies but differ in terms of the relative fre-     at oblique power grips and that Neandertals therefore would
quencies of those classes.” If this is indeed the case, it would    have been more efficient with hand-held than with hafted
be interesting to know more about small flake production in         tools. How efficient does our anatomy have to be in order to
other levels at Pech de l’Azé IV. While the actual amount of       perform a task? Our own modern human hands may be less
small flake production in other levels is obviously much less       efficient than Neandertal hands in performing tasks involving
than in Level IV-A, how does the intensity of production            power grips, but this does not preclude our performing tasks
compare? For example, Levels II-A and II-B exhibit only 34          with a power grip. Despite the presumed inefficiency of Nean-
cores between them, but if, say, 30 of these 34 cores are small     dertals’ oblique power grips, numerous researchers have
cores used for producing small flakes and the intensity of          found evidence for hafting in the Mousterian (e.g., Anderson-
small flake production in Levels II-A and II-B is comparable        Gerfaud 1990; Beyries 1988; Boëda et al. 1996; Hardy et al.
to that in Level IV-A, new questions arise. Hypothetically, if      2001; Shea 1988). Furthermore, since Dibble and McPherron
You can also read