CHAPTER 1: HISTORY OF FUNCTIONAL ANALYSIS
CHAPTER 2: THE QUANTIFICATION
OF MICROWEAR POLISHES
CHAPTER 3: INVESTIGATING HAFTING
TRACES WITH IMAGE PROCESSING
CHAPTER 4: A MULTI-DIMENTIONAL
APPROACH TO FUNCTIONAL ANALYSIS
SECTION 1:INTRODUCTION SECTION 2:CLEANING SECTION 3:MICROSCOPY
SECTION 4:OBSERVATION RECORDING
SECTION 6: THE FUNCTIONAL SIGNIFICANCE OF THE VARIABLES
SECTION 9: EXPERIMENTAL
REPLICATION
SECTION 10: BLIND TESTS
CHAPTER 5: DRILL BITS FROM KUMARTEPE
CHAPTER 6: THE LIMITATIONS AND
APPLICATIONS OF USE-WEAR ANALYSIS
APPENDIX 1: TESTING FOR EFFICIENCY
BIBLIOGRAPHY
The development of the analysis of the function of stone tools
(hereafter referred to as use-wear analysis) began by the use
of analogy with ethnographic tools. This involved finding ethnographic
tools that resemble, in form, prehistoric tools and assuming the
prehistoric tool was used for the same function as that documented
for the ethnographic tool. These analogies were primarily based
on parallels with American Indian tools because of the similar
morphology of prehistoric and Indian tools. The main problem,
apart from assuming a direct correlation between form and function,
is that many prehistoric tools have no ethnographic analogues,
notably hand axes and burins.
The next stage was to compare wear traces between prehistoric
and ethnographic tools. Gould,
Koster and Sontz (1971) observed that the wear traces on Australian
Aboriginal adzes (purpuna), used for planing hard wood, appeared
similar to those on Quina scrapers, and therefore inferred that
the Quina scrapers were used on hard wood. The wear on the Quina
scrapers was not differentiated from manufacturing traces and
no hard wood was present in the environment from which the scrapers
come, the Wurm glaciation of South West France. So even though
ethnographic analogy is largely responsible for tool types being
called scrapers, borers etc., no direct form equals function relationship
has been demonstrated, or even that similar wear traces on ethnographic
tools can be paralleled with the wear on prehistoric tools. Though
these functional terms are regarded merely as labels for classification
purposes, an example of the misuse of functional names for morphological
types is found with scrapers as, typologically, both concave and
convex scrapers are classed in the same group of tools, but they
could not have been used in the same way because the two kinds
of edges are not mechanically useful for the same task.
Experimentation was used to obtain functional information by first
assuming a function for a tool and then testing that hypothesis
by using the tool in the manner assumed and assessing its viability
and efficiency. For example the Danish experiments in using Neolithic
axes for forest clearance. However, only the assumed function
was tested and possible alternatives were not, so that the efficiency
of the tool for a specific task is tested rather than the functional
capability of the tool. It may well be the case that a tool is
more efficient for an alternative task than the one tested.
For an example of a small experimental programme to test the efficiency
of different types of tools and different edges see
Appendix 1.
Edge analysis involves taking the working edge as the unit of
study rather than the whole tool. Wilmsen
(1968) measured 2,139 artifact edge angles and he claimed that
the analysis showed there to be 3 modes: 26-35 degrees, 46-55
degrees and 66-85 degrees. This is an example of the use of quantitative
methods to analyse data that produced edge angle modes rather
than assuming a function for the tools. The meaning and significance
of the apparent modes remain a matter of interpretation. Wilmsen
interprets the activities that these edge angle modes represent
as follows: 26-35 degree angled edges were used for cutting, 46-55
degree angled edges were used for hide scraping/heavy cutting
and the 66-75 degree angled edges were used for wood and bone
working.
Another example of edge analysis was the work carried out by White and Thomas (1972).
Here they were looking for types, as the New Guinea tools which
they were studying did not fit into any established morphological
typology. They asked some New Guinea people who remembered making
and using stone tools to classify their tools. Their classification
was based on function in that the initial differentiation was
made between flakes that were usable and those that were not.
Then they separated the usable flakes into those suitable for
hafting and ones to be hand held. The flakes suitable for hafting
were then separated into ones to be used as vegetable shredders
and ones used for drilling. The hand held ones had a great deal
of variation in their overall morphology and the New Guinea people
said that they chose a tool for a particular task according to
the presence of a suitable edge on the tool. The morphology of
the whole tool only played a marginal part in the decision of
which edge to use for a particular task.
The New Guinea people's classification of the tools was submitted
to statistical analysis, from which it was suggested that the
edge angle of the tools was important in that they tended to choose
tools with a particular edge angle for a specific task, though
not being cognisant of such a process themselves. White and Thomas
then went on to derive a classification of the prehistoric material
based on such attributes as edge angle. This procedure constitutes
one of first attempts to produce a functional classification,
though in this case the intention was to use it to study relationships
between assemblages, rather than for functional reconstructions
of tasks being carried out on the sites.
Cantwell (1979) separated
scrapers from mid-west America as being used on hard wood or hide
by the presence of edge wear. Her assumption was that working
hard wood would produce more edge damage than working hide. Hard
wood was considered the worked material because there were no
bone or antler artifacts present, though preservation was good.
The measurement of the edge angles of these two groups demonstrated
that the wood scrapers had a mean of 61 degrees and the hide scrapers
a mean of 70 degrees. Thus implying a correlation between edge
wear and the angle of edges used for the same activity.
After the translation of Semenov's Prehistoric
Technology in 1964 the emphasis in use-wear analysis centered
on the use of microscopy for studying the effects of use on the
edges of tools. At first this was mainly concerned with low power
(i.e. <100 magnifications) looking principally at edge wear.
Broadbent (1979) with
his colleague Knutsson used microscopy in conjunction with looking
at edges using attributes such as edge angle, profile, and "placement
of edges in respect of tool length and width" (ibid,81).
They studied the tools in order to discover particular attribute
sets based on experiments and then separated the tools into those
used for scraping or cutting. This work was done in relation to
quartz scrapers from Lundfors, Sweden. The experimental programme
consisted of mechanical experiments where they attempted to control
such variables as contact angle (by using pliers as hafts), standardising
stroke length and pressure, and the duration of use was measured
as number of strokes. Low power microscopy (up to 80x) was used
mainly to ascertain how the edge angle changed during use. They
noted that edges tended to stabilise, so that weak (i.e. acute
angled edges) or irregular edges quickly became rounded and edge
angles stabilised after use. Tools having edge angles <55 degrees
that were used on wood stabilised at 70-80 degrees. Tools having
original angles of >80 degrees sustained little edge wear.
Scraping hides with tools having angles >70 degrees "barely
functioned" (ibid,84) as they were not sharp enough to cut
through the tissues. Lower edge angled tools would break too easily
so that edges of 50-60 degrees were ideal. They suggested that
edges were designed by retouching them to the optimum angle, or
chosen from debitage as being suitable for a particular task.
They separated edges into modes suitable for use on:
1) soft pliable materials eg. hide.
2) hard heterogeneous materials eg. wood.
3) very hard homogeneous materials eg. bone.
They suggested that, rather than there being distinctive polishes
according to worked material, edge wear represents a continuum
so that use on different materials can appear the same. For example,
a tool heavily used on hide could produce the same edge wear as
a tool slightly used on bone. But they considered that it only
takes 8-10 minutes of use on a hard material for a tool to exhibit
edge wear that is characteristic of use on a hard material, and
they assumed tools would have been utilised for at least that
length of time, so that they could distinguish between tools that
were used on hard or soft materials.
They considered re-sharpening, in which case the edge wear reflects
the last use of the tool. But as re-sharpening almost always increases
the edge angle the re-sharpening would produce edges more suitable
for use on a hard material, so that a hide scraper would not be
re-sharpened but another tool would be made in order to continue
hide scraping, the blunt one being abandoned. They investigated
the edge angles of the tools and considered that the edge angles
did cluster at 70-85 degrees, which they interpreted as being
used on hard materials, and 55-65 degrees which they considered
suitable for use on medium to soft materials (n=553).
They also examined the edge contours (i.e. profiles) of their
archaeological sample, putting them into the categories convex,
straight, and serrated. Concave was excluded because there were
only 9 examples. Convex and straight edges had similar edge angles
of 70-80 degrees, whereas serrated (and notched) edges tended
to have lower angles, 60-70 degrees. The convex edges had more
hard material wear than the straight ones.
In the paper by Tringham
et al. (1974) experiments were carried out to test the following
hypothesis: "A tool made of a specific raw material, whose
edge is activated in a specific direction across a specific worked
material will develop a distinctive pattern of edge damage of
a kind that is recognisable on the edges of prehistoric tools"
(ibid,178). This was tested by .mechanical experiments where it
was attempted to control such variables as direction of use, pressure
and contact angle. The duration of use was determined by a set
numbers of strokes. The problem with mechanical experiments is
that mentioned in the last part of their hypothesis; do these
mechanical experiments produce edge wear that "is recognisable
on the edges of prehistoric tools", especially when, "The
efficiency of such edges in performing a given task was not tested"
(ibid,178). So that if they were using edges totally inappropriate
to the task (i.e. inefficient), such edge wear which might result
would not occur on prehistoric tools and would be irrelevant to
the problem of identifying edge wear on archaeological tools.
They attempted to deal with the problem of separating retouch
from use-wear by looking for patterning in edge damage. "The
distinction of flaking which results from deliberate retouch from
that caused by wear from usage has been the source of much confusion.
In our opinion the distinction should be made on the basis first
of size and second of patterning of the scars" (ibid,181).
The evidence for the patterning of scars was produced by their
mechanical experiments, in which they considered the effects of
such morphological aspects as edge profiles (ibid,180). But this
kind of information is not relevant to prehistoric tools, as such
different profiled edges would not have been used for the same
task since they are not equally efficient.
Odell (Odell and Odell-Vereeken
1981) is the only practitioner of the low power approach to
do a blind test and the results were such that he accepts that
the technique is limited to identifying the hardness of the material
rather than specific materials. (For a discussion of Odell's blind
test see Section 10 of Chapter
4).
A major aspect of low power microwear analysis is the classification
of fractures. Kamminga
(1982) devised 6 types of fractures in his classification scheme:
bending fractures, feather fractures, hinge fractures, retroflexed
hinge fractures, step fractures and clefts (ibid,6). He describes
various sources for the origin of these fracture variations: hardness
of material, yielding or resistant effect of material, angle of
tool to material (i.e. contact angle), edge angle of the tool,
direction of use and the raw material of the tools (flint, quartzite
etc.). He mentions an earlier comment of his that "The variations
of these fractures might be so complex and unpredictable that
fracture patterns would be elusive and not task specific"
. He then claims that this statement was "unduly pessimistic",
in that the condition of activities varies sufficiently for their
use-wear to be separated. He compares fleshing of a fresh hide
with the adzing of dense wood. His experimental research demonstrated
that for the efficient use of tools certain variables must be
controlled, for example the edge angle and profile of the tools.
This indicates that tool selection by prehistoric people (or design
by retouch as mentioned by Broadbent
1979) would have been very important and in fact necessary
in order to carry out the task, that is, selecting a suitable
edge for the job in hand.
Kamminga looks at fractures in terms of overall size (e.g. large/
small/ micro) and the depth of fractures. More detailed quantification
of use-wear fractures has been attempted by other workers (Tomenchuk 1983, Akoshima
1987). But to measure each individual fracture by depth, length,
width and classify them into one of the 6 types as used by Kamminga
would be as time consuming as using high power microscopy. Without
this quantification the fracture classification would be subjective
and would vary between analysts, especially as they often use
different classification schemes.
Kamminga admits that the problem in classifying fractures is that
there is no radical difference between the fracture types (Kamminga 1982,5). They are
defined by continuous variables because fractures are a continuum
in terms of size and form, and the different categories of fractures
can merge into each other so that rather than discrete groups,
a continuum is being separated into arbitrary groups. He states
that some fractures deny classification because they have features
of more than one type (e.g. merging step and hinge termination
fractures). However a hinge termination is a discrete attribute
produced by the mechanical process of its manufacture and its
presence defines a step fracture (see Figure 1).
step fracture conchoidal fracture with hinge termination
A hinge termination is created when the force of the blow turns
outwards producing a distinct mechanical feature, defined by Crabtree (1972,93) as having
an abrupt right angled break at the point of truncation. Kamminga
says "I seriously doubt that anyone identifying step fractures
adheres to this stringent definition" (Kamminga
1982) and describes a hinge fracture as a sort of smoother
step fracture which he says represents a continuum with feather
fractures. So there are feather fractures that, as they begin
to end more abruptly, become hinge fractures and then step fractures
that have a 90 degree termination. The termination angles are
not defined and so it becomes a matter of judging the termination
angle by eye on very small fractures. This will vary between analysts.
The division between conchoidal fracture, where the force of the
blow continues through the stone, and step fractures where the
force turns outwards to create a second fracture plane, is based
on the fracture mechanics of flint rather than any morphological
difference between the resulting flake scars. Therefore, the difference
is mechanical and the two types represent distinct entities rather
than parts of a continuum. The problem with the morphological
classification of fracture scars is illustrated by considering
step and scale flaking as in Quina retouch. When are these flakes
feather flakes with hinge terminations, and when are they step
fractures which always have a hinge termination?
High power microwear analysis was developed by Keeley from Semenov's
work (1964). High power microscopy involves using magnifications
of 100 plus, normally characterising use-wear traces at 200 magnifications,
but occasionally using up to 400 magnifications. The extra information
gained through the use of higher magnifications centers on polishes.
Polish is defined here as the visible alteration of the flint
surface so that the reflectivity of the flint surface is increased
when viewed through the microscope. Keeley carried out a series
of experiments using various tools and he claims to have recognised
that specific materials produce distinctive polishes, so that
we have bone polish, wood polish, hide polish, etc. The evidence
for these distinctive polishes is presented in the descriptions
of certain polish characteristics and illustrated with photographs
(Keeley 1980).
Keeley originally stated that these polishes are distinctive when
certain variables are controlled, in particular the raw material
of the tools, so that a prerequisite of any microwear study is
an experimental programme using simulation experiments with similar
stone, preferably from the same source as that of the archaeological
material. (Simulation experiments attempt to simulate activities
assumed to have been carried out in prehistory by using suitable
tools in the most efficient manner, rather than mechanical experiments
where certain conditions are controlled and which become investigations
into fracture mechanics.) If the same raw material is not available,
then the raw material should be at least of the same type and
grain size. However, since then some microwear analysts have gone
further and suggested that such an experimental programme is not
necessary, but that the polishes are so distinctive that information
from any experimental programme can be used. For example, Vaughan (1985), though agreeing
with Keeley, uses his experimental reference collection made of
Greek flint and of Mesolithic morphological types to study flint
tools from Cassegros, a Magdalenian rock shelter in southern France.
The main problem with high power microwear analysis is that the
descriptions of the distinctive polishes are subjective and largely
unusable by independent workers. The following chapter describes
an attempt to quantify microwear polishes and to test the assertion
that the polishes are distinctive and attributable to a specific
worked material.