The presence of both chert and quartzite cores in the
underlying assemblage implies some transport (both
indirect and direct) of these artifacts to the site from
the lithic sources noted above. The form of the major
lithic materials in the region allows a range of possible
careful selection of suitable pebbles for immediate
transport and later on-site reduction, to the systematic
on-source reduction and transportation of pre-prepared cores.
Some indication of procurement procedures and core
preparation in the assemblage can be gained by examining
the blank types on which cores were formed. Because of
the considerable reduction of the cores in both the
quartzite and chert the number which retain enough of
their initial structure for their original blank form to
be identified is low. For instance, only 4 (5%) of the
quartzite cores and none of the chert retained sufficient
cortex to be attributed, with confidence, to pebble
blanks. The proportion attributable to flakes is higher
at 25 (31%) for the quartzite cores and 5 (8%) for the
chert. These were identified on the basis of the presence
of both intact striking platforms and positive ventral
face features. Cores without extensive pebble cortex and
flake features were classified as indeterminate blank
forms.
In quartzite there was no significant change in the proportion of flake to indeterminate blank forms between
the the upper and lower halves of the assemblage (chi2 =
.46, df=l, p=.5) and with mean masses of 76.5 and 75.8gms
respectively there is also no significant size difference
between the two blank forms (MW-U z=-0.43 p=.33). The
chert sample of five artifacts was too small for any such
Both flake and indeterminate quartzite blanks
retained high proportions of cortex. Of the 14 single
platform cores (R o ) made on flakes 11 (79%) retained
cortex while it was present on only 5 (45%) of the 11
multiplatform cores (R n ) on flakes. The difference is
attributable to a significant difference (chi2 = 6.95,
df=l, p<.01) in the proportion of all the multiplatform
cores retaining cortex relative to the single platform and a marginally significant decline in the amount of cortex on the cores through time (chi2 = 3.9, df=l, .05>p>.01).
The proportion of single platform flake blank cores retaining cortex (83%) is substantially higher than that
found generally in the flakes in the assemblage. In
quartzite the primary and secondary decortication flakes average 14.3% of the total flake numbers per spit with no
discernible trend through time (see section 4:6b). This
is also the case for the indeterminate blank forms which
at 74% and 46% for single and multiple platform cores
respectively show no significant difference in cortex
proportion to those found in the flake blanks (chi2 =
.122, df = 1, p>.95).
In the chert sample only multiplatform cores retained evidence of flake blanks, however, the absence of the form
in the single platform cores is not statistically
significant (required sample size of 30 before 0 is
significant). Both 40% of the indeterminate and flake
blank multiplatform chert cores retained cortex. This is
comparable to the percentages in the quartzite core
chert primary and secondary decortication flakes in the assemblage, which range from 7-12% of the total chert flakes per spit.
As the proportion of cortex retention in all core blank types is high relative to the flake population o n site it can be concluded that the majority of the quartzite and chert flake blanks derived from primary and secondary decortication flakes and were reduced without any systematic d e c o r tication procedure.
To test for a size difference between the core flake blanks and the flakes constituting the bulk of the assemblage the dimensions of the b l a n k s ’ residual striking platforms and overall thickness were compared with those from a random sample of flakes taken from squares A and B9 (see section 4.6). Of the 25 quartzite cores on flake blanks intact striking platforms were present on 14 examples while only 2 chert cores retained completely intact platform features. Blank thickness was taken as the best measure of the primary flake dimension because flaking of the core had generally used the ventral surface as the principal striking platform reducing length and w i d t h .
The comparison between the flake blank dimensions and those of the flake sample for the quartzite are given in table 4.6. The table shows there to be substantial differences in the size values of the core and flake samples (MW-U test results of z=10.8 for thickness, 5.92 for platform width and 6.1 for platform thickness, all of
TABLE 4.6: Quartzite Flake and Flake Blank Dimensions (mm) X C v R n P l a t f o r m c o r e s 3 2 . 6 3 0 . 9 % 2 2 - 5 9 14 w i d t h (P w ) f l a k e s 1 1 . 6 5 4 . 4 % 2 - 3 3 171 P l a t f o r m c o r e s 1 5 . 8 2 3 . 4 % 1 0 - 2 4 14 t h i c k n e s s f l a k e s 4 . 4 7 2 . 7 % 1 - 1 7 171 P l a t f o r m c o r e s 9 5 . 4 1 4 . 2 % 7 2 - 1 2 5 14 a n g l e f l a k e s 7 3 . 2 1 8 . 3 % 4 0 - 1 2 6 1 7 0 T h i c k n e s s c o r e s 3 1 . 5 2 2 . 0 % 2 1 - 4 4 14 f l a k e s 5 . 5 5 8 . 2 % 1 - 1 8 171
which are significant at p<.00003). These data confirm that the core blanks were drawn from a population of
substantially larger flakes than those represented on
site. The core blanks also exhibit higher platform angles
and in all four variables retain absolutely less
variability.
Of the quartzite flake blank sample only 2 (14%)
examples retained cortical platforms of the remainder 11
(78.6%) had plain and 1 (7.1%) combined platform flaking
and cortex. This variation in platform type shows there
to have been a variety of approaches to the initial
platform preparation on the rocks from which the blanks
were derived. This may be expected given the irregular
sub-rounded to angular forms in which the quartzite occurs
in the landscape. Plain platforms are, however,
predominant, suggesting the systematic removal of
unsuitable cortical surfaces from the primary core before producing the large flakes which served as core blanks. The high percentage of cortex retention on the single
platform quartzite cores (78%) also suggests that
systematic quarrying of unweathered material from the
outcrops of vain quartzite in the region was not a
significant feature of quartzite procurement in the
underlying assemblage.
In the case of the quartzite some indication of the differences between the process producing core blanks and that of the larger flake sample can be gained by examining the relationships between platform variables and flake
thickness. It has already been seen that the original platform angles on the core blanks are on average 20° larger than those of the flake sample. The comparison of platform thickness and flake thickness ratio (t/Pt) also shows the core blanks to be substantially thicker in relation to platform thickness (t/Pt: x=1.9, Cv=29%) and less variable than the flake sample in which the thickness/platform thickness ratio is closer to unity (t/Pt: x=1.2, Cv=40%). The shape of the striking platform also tends to be squarer in the core blanks (Pt/Pw: x=.51, Cv=25%) compared to the flakes (x=.4, Cv=48%). The blanks also exhibit a different set of best platform predictor variables for flake thickness. Because of their structural similarity it would be expected that there should be a strong correlation between platform thickness and flake thickness in the core blanks as is the case in the flake sample (r=.72, p<.01). This, however, is not the case with the blank sample which shows no significant correlation between the two variables (r=.03, p>.2).
The breakdown in the expected correlation of platform thickness and flake thickness in the core blanks may reflect their relative uniformity of size and is also in accord with Dibble and Whittaker’s (1981:295) experimental finding that increases of platform angle caused the relationship between platform thickness and flake thickness to become less predictable, although the production of larger flakes still resulted. Dibble and Whittaker (1981:295) also found that as platform angle increased the values of the platform thickness which would
produce a flake become more restricted, which may account for the substantial amount of overhang removal present on
some of the core blanks and the high t/Pt ratio. These
data are consistent with attempts to maximize the size of
flake produced from a given core. This is also supported
by a marginally significant negative correlation in the
blank sample between the platform thickness and platform
angle (r=-.51 ,05>p>.01) which, as will be discussed in
detail in section 4.7b, is also consistent with the
presence of high input forces and relatively low core mass .
Although the cores on pebble blanks tend to be larger than those on the flake blanks with a mean mass of 157gms
compared with 77.7gms this is not statistically
significant (MW-U U=12 p>.05). The pebbles selected for
direct reduction do, however, appear to be more variable
in size (ranging from 70-370gms) compared to the flake
blank’s range of 22-190gms. Due to the reduction of the
cores the relation of these masses to those of the initial
selection criteria is, however, problematic. Examination
of the manuported material in the deposit shows that masses of up to 5-600gms were carried onto the site in the form of poor quality quartzite pebbles which is comparable
to the mass of the largest core of 633gms. This core
retains substantial pebble cortex, and although not
classed as a definite pebble blank, may stand at the upper
limit of the pebble size considered suitable for
Of the pebble based cores, 3 (75%) retained cortical platforms from which flakes had been removed. This is substantially higher than the 12.4% recorded for the total core sample. A binomial test shows that the probability of obtaining this difference by chance is 0.0067 suggesting a significant retention of cortical platforms on the pebble blanks in the initial stages of reduction. This is consistent with the selection of pebbles with surfaces suitable for use as striking platforms without further modification.
In summary, although the number of the quartzite cores attributable to an original blank form is low, the evidence of these cores is that pebbles or surface material, in a range of sizes, were used as sources for the cores in the assemblage and that the type of reduction procedure used in the p r e p aration of these was principally determined by the size of the piece selected. If large and without suitable platform surfaces, cortex was removed to prepare a plain platform which was then used for the striking of a large flake suitable for use as a core. In some cases considerable overhang removal was carried out to raise the platform angle and produce the reduced platform thickness necessary for the production of flakes of the required size. The resulting flake may have been minimally trimmed to reduce transported mass but there appears to have been no systematic decortication of the core blank. Pebbles <600gms with already suitable striking platforms also appear to have been simply selected and transported with little further modification.
4.4: Core Discard and the limits of Reduction
4.4a: Threshold conditions controlling reduction
The following analysis will examine the threshold
conditions which limit reduction and test for their effect
on the core sample. The factors will be examined in two
parts: those which restrict individual platform use and
those which limit overall core reduction.
As the variables controlling flake formation have
been extensively discussed in the literature (see
Cotterell and Kamminga 1979, 1987; Cotterell, Kamminga and
Dickson 1985, Dibble 1981, Dibble and Whittaker 1981,
Faulkner 1972, Hiscock 1979, Lawn and Marshall 1979,
Phagan 1976, Speth 1972, 1974, Tsirk 1979) it is not
necessary to precede the analysis with an extensive review
of the variables and their interaction. In summary,
Phagan (1976:9-10) has grouped these factors into three
categories: core variables (material, platform surface,
point of force application [PFA] and core geometry), force
variables (angle, amount and duration) and interaction
variables (relative masses and hardness). Of these, three
have been seen as primary determinants of flake form: core
geometry (Bucy 1974:6, Cotterell and Kamminga 1987:698,
Faulkner 1972:127, Hiscock 1979:51), PFA location
Whittaker 1981, Hiscock 1979:51) and the amount of energy available for fracture (Dibble and Whittaker 1981, Hiscock
1979:51, Speth 1974:15,27,31).
Of critical importance to flake production is the
dynamic interaction of these variables with the mass of
the core. Where the core mass (as a measure of the
reactive inertia force) is low relative to the applied
force, a proportion of the force will be converted into
the kinetic energy of core motion, reducing the amount of
energy available for the formation of flakes. At its
simplest, the effect of the decreased energy available on
fracture is either the rapid termination of the process
resulting in step or hinge fractures (Cotterell and
Kamminga 1987:700-701, Phagan 1976:25) or the failure to
initiate fracture altogether. As will be seen in the
discussion of the internal relations of flake formation
(section 4.7) the knapper’s response to these conditions is complex, but for the purposes of this discussion, these limitations will be sufficient.
The amount of force required for flaking is directly proportional to the PFA location and the geometry of the
core. The further from the free face of the core (ie.,
the greater the P t ) and the higher the platform angle (Pa)
the greater the energy required for fracture (Cotterell
and Kamminga 1987:700-701, Dibble and Whittaker 1981:295). The point at which these variables reach values requiring greater energy input than the reduced core mass and
increased proportion of the available force is being
converted into core motion) has been termed the "inertia
threshold" by Hiscock ( 1979:fig.2:4 ) .
A general model of the threshold has been described by Hiscock (1979:54) as a hyperbolic curve when inertia is
plotted against some measure of the required force (eg.,
platform angle) and the direction of increase is toward
the origin when plotted on the x axis. The function is
more accurately drawn as a power express ion, as increased
platform angles require a greater core mass if the problem
of low core inertia is to be avoided. The function’s form
is governed by the requirement that the measure of
required force be dimensionally equivalent to the fracture energy of flake formation which is a square function of
flake dimension (ie., E = ML2T~2 ). The model is, however,
only partially correct in its correlation with the
mechanics of flakes formation. Under its conditions it is
theoretically possible to remove flakes from platform
angles greater than 90° and in theory from surfaces
greater than 180°. The model assumes that the mechanics
of fracture would be unaffected by such changes which in
practice is not the case.
The principle control mechanism of conchoidal and
bending fracture is flake stiffness (Cotterell, Kamminga
and Dickson 1985:220, Cotterell and Kamminga 1987:698),
which is possible near the edge of cores where the applied force has an outward component supplying tensile stress
sufficient to initiate crack formation (Cotterell and
Alt h o u g h Cotterell and Kamminga (1987:692) point out that platform angles >45° are important in controlling initiation, the upper boundary of this angle is not discussed (see also Lawn and Marshal 1979:71). In an earlier discussion of conchoidal fracture (Cotterell, Kamminga and Dickson 1985) the maxi m u m platform angle assumed by their beam theory model is 90°. Extending this model it may be seen that as platform angle increases above 90° the flaking process becomes rapidly over stabilized by flake stiffness to the point where energy requirements exceed the physical capacities of the knapper to impart to the core. Angles greater than 90° create threshold conditions which affect all cores regardless of mass. This is in general accord with the conditions noted in accounts derived from the literature of experimental knapping (see Bordaz 1971:24).
Although high platform angles and core inertia can be seen as separate threshold conditions, both interact during reduction to restrict the process. Unless they are braced, all cores will experience some movement while being struck as some of the available energy is lost in core motion. This loss becomes more critical as core mass is reduced relative to any increase in the force which is required to remove a flake, i e . , as platform angles approach 90° or due to poor core face geometry. As cores become smaller the inertia threshold conditions also intervene to reduce the critical platform angle to sub-90° values in the manner described by H i s c o c k ’s model. Although the point at which the critical conditions of
this model become significant have yet to be clarified Hiscock’s (1979:145) investigation suggests that masses of less than 10-6gms will be affected at increasingly lower platform angles depending on material type.
As Hiscock (1979:54) points out, the form of the
inertia threshold may not be a sharp boundary because the conditions governing the interaction of the core and the applied force may not be constant within a general
reduction sequence. The form of the threshold will also
depend on the independent variable used to measure force
requirement. The problem may also be encountered at any
stage of core reduction with an increasing likelihood of its conditions ending the process as reduction continues.
Hiscock (1979:51) has suggested that platform
rejuvenation, rotation, reduction in flake size, bracing
the core and bipolar reduction may be solutions to the problem of low core inertia.
As flaking is a reduction process, which by the
nature of the flakes produced tends to reduce the core platform at a greater rate than the body of the core
(Kamminga 1982:89), the threshold conditions represented
by high platform angles and low core mass are inherently
more characteristic of the process than the comparatively unstable conditions generally associated with optimal flake production.
Before proceeding to the next section some
explanation of the concept of unstable and stable core
percussion can occur only within a limited range of core states which are not inherently stable given the shape of flakes removed and the knapper’s control of variables like
material and input force consistency. As reduction
extends the core geometry approaches more natural states
of stability which are associated with high platform
angles, restricted platform arcs, the increased
concentration of core mass to produce stable geometric forms and the onset of low inertia threshold conditions. The unstable conditions favourable to extended reduction do not derive naturally from the process but have to be
maintained in the face of the approaching
overstabilization which limits reduction.
4.4b: Platform Threshold Conditions
To test for the presence of threshold conditions in
the discarded cores in the underlying assemblage the cores