A second problem results from the treatment of very different
quantities of material of any given pottery type as a single score on the
matrices. Types of minor significance in terms of numbers of sherds are
elevated to the same importance as major types. This raises problems of
interpretation particularly where high-status sites such as towns, military
bases, or well-appointed villas are concerned, for these tend to have a wide
range of exotic ‘kitchen wares’ occurring in very small quantities that
are far more common on sites in different locations. Thus, Canterbury has
revealed one or two sherds of first-century West Kent shelly wares (abundant
west of the Medway), and Richborough’s pottery includes a single Dales
Ware jar (Bushe-Fox 1928, no. 147), the only site south of Caistor St.
Edmund to do so (Loughlin 1977, 109, fig. 6). A high-status site in one
distribution zone may be computed to have a much higher degree of similarity
with a site in a second zone than visual inspection of assemblages would
suggest.
VI. THE DEMONSTRATION OF SPATIAL TRENDS
The distribution map has long been a primary technique for
presenting archaeological data and facilitating the interpretation of
those data. The methodology of interpretation was one of the main aspects
of the development of archaeological theory and method in the 1970s (e.g.
Hodder and Orton 1976; Clarke 1977, with references) invoking and adapting
a wide range of techniques developed by other disciplines, in particular
geography and plant ecology. One of the main stimulants to this
development was the recognition of the fallibility of subjective
assessments of distributions: underlying structure may not always be
easily discernible to the human eye, and the recognition of only
those patterns and structures which the researcher wishes to see is a
distinct possibility. These problems have been discussed by Hodder and
Orton
|
|
(1976, 1—10) with particular reference to the
archaeology of prehistory. Two aspects of this discussion may be focused
upon: the plotting of both positive and negative data, and the distinction
of random from non-random distributions.
Fox’s classic study of The Personality of Britain (revised
edition 1943) is criticised by Hodder and Orton (1976, 3) for failing to
take account of the effect of differential patterns of site destruction
and fieldwork intensity upon the recovery of archaeological material. The
problem of interpreting blank areas on the map may be to some extent
alleviated by plotting negative occurrences of relevant material on
contemporaneous sites. This approach was applied to studies of
Romano-British pottery by Hodder (e.g. 1974b), and has been adopted by the
present author on the distribution maps of pottery types. However, the
establishment of contemporaneity of sites and assemblages in itself can
cause problems. In his studies of, for example, Savernake ware (Hodder
1974c) and Rowlands Castle wares (Hodder 1974a), Hodder assumed that all
sites utilised the respective wares simultaneously; he thus plotted and
discussed the gross distribution patterns of positive and negative finds.
It is often the case that pottery of a particular type was not current in
the same areas throughout its overall period of usage —
the example of BB2 jars, mentioned in the preceding section and
plotted on Figs. 30, 45, and 47 may be cited —
and it should not be assumed that a type was in simultaneous use
throughout its area of distribution at a certain time in its history. The
example of Oxfordshire colour-coated wares indicates that this was not
always the case; briefly, in the late fourth century these wares were
eclipsed in East Sussex by ‘Pevensey’ ware (Green 1977, 177—8) but
expanded their distribution in East Anglia (Drury 1977, 40). A cumulative
distribution pattern may suggest
|