A MULTI-VARIATE APPROACH TO THE FUNCTIONAL
ANALYSIS OF STONE TOOLS
GRACE R*, ATAMAN K.*, FABREGAS R*. and HAGGREN C.M.B.*
ABSTRACT :
‘This paper presents a new approach to the interpretation of the function of stone tools. The
assumption that micro-wear polishes are distinetive according to worked material is rejected in
favour of a multi-variate approach. This approach looks at the relationships between the morphology
of tools, the macro- and miero- edge wear and the distribution and level of development of polish.
‘A set of defined variables with specific values has been developed that allows the analyst to
make observations in a systematic way. This scheme makes the observations consistent between
tools and between different analysts, rather than using subjective descriptive terms that vary
according to each analyst’ s perception. The interpretation of the function of the tools is based on the
‘climination of possible functions until the most probable function can be isolated The efficacy of this
‘method is demonstrated by the results of a multi-analyst blind test, and examples are given from the
‘test which illustrate the methodology employed.
RésuME:
Cet article présente une nouvelle approche de T'interprétation de la fonction des outils de
pierre. Lihypothése que les polis de micro-usure sont distinets selon la matiére travaillée est rejetée
en faveur d'une approche multivariée. Cette approche considere les rapports entre morphologic des
outils, macro- et micro-usure des bords et répartition et niveau de développement de poli.
On a développé un ensemble de variables déterminées de valeurs spécifiques qui permet &
Tanalyste de faire des observations d'une maniére systématique. Ce projet rend les observations
consistantes entre outils et entre différents analystes, plutot que utiliser des termes deseriptifS
subjectifs qui varient selon la perception de chaque analyste. Liinterprétation de la fonction des
outils est fondée sur I’élimination des fonctions possibles jusqu’a ce qu'on isole la fonetion la plus
probable. Lefficacité de cette méthode est démontrée par les résultats dun test aveugle multi
Bnalystes, et lon donne des exemples tirés du test qui Mlustrent Ia uiéthodulogie employée.
Within the last few years a good deal of research in the field of stone tool studies has
focused on the development of techniques for the determination of stone tool function
through the microseopic examination of wear traces on the surfaces of stone tools. This
technique is generally referred to as microwear analysis. The practitioners of this
technique have claimed to be able to determine the function of stone tools by the
characterisation of polished areas and striations on the working edges of the tools and to
be able to reconstruct aetivities such as butchering, fresh wood whittling, dry hide
scraping etc. (Keeley 1980, Moss 1983, Vaughan 1981). The basic assumption behind this
technique of high power microwear analysis is that contact between the flint tool and a
specific material like wood, will produce a polish which is distinet from that produced by
contact with some other material like bone. This assumption has recently come into
question (Neweomer, Grace and Unger-Hamilton 1986). The failure to identify specific
wood, bone, antler ete. polishes in blind tests has demonstrated that, rather than being
* Institute of Archaeology, 3&-24 Gordon Square, London WCIH, England
217distinctive, these polishes overlap to the extent that they represent a continuum, not
discrete recognisable entities. This lack of distinctiveness has been further
demonstrated by texture analysis, using image processing techniques, which involves
the measurement of the differences between the polishes created on the tools by contact,
with various materials by mathematically characterising the textures of the polishes
(Grace, Graham and‘Neweomer 1985).
In response to the fundamental doubts that have been raised about the accuracy and
usefulness of microwear analysis as it is currently practised, a new approach which
attempts to standardise the methodology of microwear analysis and to test the limits of its
interpretations has recently been developed at the Institute of Archaeology, University of
London, This project was carried out over two academic terms and involved several
people with varying degrees of experience in microwear. The development of this method
involved many hours of discussion on theoretical as well as practical issues and resulted
in a method which could be used successfully by all the participants. The usefulness of the
method was confirmed by the results of blind tests administered by an independent
experimenter.
‘The method involves the systematic recording of the functionally diagnostic
variables of a tool. These variables are described using a standard vocabulary and the
descriptions can be replicated, so that different analysts can learn to describe the same
tools in similar ways. Combinations of the variables then allow the analyst to eliminate
some of the possible functions of a tool until the most probable function is isolated. In
some cases the elimination of possible functions leaves only one that is consistent with
all the wear traces on the tool. This means that in some cases functional reconstructions
which include the specific material the tool was used to work can be postulated with some
confidence.
‘The variables can be divided into five groups : raw material characteristics, edge
morphology, macroscopic edge damage, microscopic edge damage, and polish
distribution and characteristics. These variables and the recording sheet developed in
the course of this project are illustrated in figure 1. As the method is being constantly
revised and developed, a full detailed description cannot be included here, but such a
description is in preparation. Variables are being tested for their diagnostic value by
using them in blind tests, which also reveal other factors helpful in making functional
reconstructions more reliable. The main purpose of this paper is to present the results of
blind tests based on this multi-variate approach. These results improve on those from
previous tests where identifications were made on the assumption that polishes are
distinctive according to worked material (Newcomer, Grace and Unger-Hamilton 1986,
Unrath et al. 1986). Though it may be considered premature to report the results of tests
before presenting a detailed description of the method employed, we believe that it is
important that this more positive comment on microwear research should be made
available in order to demonstrate our approach to solving some of the problems of
functional analysis.
The observation sheet used in this method enables the analyst to record the
variables in a systematic way and to keep the observations consistent between one tool
and another. The presentation of information in this form also provides primary data
which can be referred to by other analysts when assessing any functional interpretations
made from the data. Most microwear reports provide polish descriptions using subjective
terms which may be interpreted in different ways by different analysts reading the
reports. In addition the cost of publishing often restricts the number of photographs which
can be published in a report, thus further limiting the primary data available to the
2EDGE MORPHOLOGY
EDGE ANGLE 40
LENGTH 45
THICKNESS 5
PROFILE, 0.08 (Profile value represents an almost straight line, O=straight)
SHAPE, 224 (Shape is a measure of the length/breadth ratio of the tool in its
working attitude)
1 1
2 or < 0.5 of the diameter of the field of view through the microscope at 200
magnifications.
Fig. 1 : Observations. Tool no. 50, type Truncated blade