assemblage need be studied in detail, featureless body sherds need not be
fabric-sorted except to facilitate the search for clearly-defined wares. Rim
sherds are placed on a chart with concentric arcs of preset radii divided
into degrees of a circle, so that the portion of a complete circular vessel
represented by each sherd is expressed as a percentage figure. In practice
it is difficult to measure the diameter of sherds of less than 18° (5 per
cent) of the rim; these are consequently omitted from the quantification.
It will be obvious that either the full assemblage recovered or
a random sample drawn from that assemblage (Orton 1978; Redman 1979; Vince
1977b) is required for quantified techniques to produce results that reflect
accurately the recovered material. These conditions can be ascertained to
occur only rarely in museum collections, owing to the lack of archived notes
by the excavators and receiving curators. Hodder (1974b) faced this problem
when he sought to deduce marketing patterns from pottery collections; he
argued that whilst collection and retention procedures might favour visibly
attractive sherds, producing a bias in assemblage proportions towards fine
wares, there was no evidence to suggest that these procedures treated one
coarse or ‘grey’ ware differently to another (Hodder 1974c). It was
legitimate, therefore, to use sherd-counts of coarse wares to provide data
from which their marketing patterns might be hypothesised. These counts are
translated into percentage expressions of relative frequency to produce
comparable figures for inter-site analysis.
The present author has not favoured Hodder’s method, as it is
not compatible with the vessel rim equivalence analysis (Orton 1975; 1980,
156—67) conducted on suitable assemblages, and the assumptions that it
makes concerning collection and retention would seem to represent a
fundamental weakness that could not justify the application of the method on
a large scale. The number of assemblages
|
|
that have been analysed by the ‘equivalent number of vessels’ technique
is not large, and it is unfortunate that this technique has only recently
received publicity, for as a consequence of this there are few published
analyses available for comparison with the present author’s data. Those
that have been consulted are listed in Appendix One. The ‘minimum numbers
of vessels represented’ technique has been applied to a small number of
assemblages from the north-west of Kent and Greater London (e.g. Philp 1973;
Tyers 1977; Tyers and Marsh 1978). Orton’s statement (1975, 31) that the
results of this method cannot be compared with one another severely
undermines the usefulness of these publications. However, in some instances
sherd counts have also been published (e.g. Philp 1963a; 1973) providing a
basis for inter-site and intra-site comparison particularly within the
north-west.
4. ‘Fine’ and ‘Coarse’ wares
Something must be said concerning the division of Roman pottery into ‘fine’
and ‘coarse’ categories. These terms are widely used in both specialist
reports and articles of a more general nature, without reference to a
standard definition. Ironically, one of the prime examples of this
malpractice is the Council for British Archaeology handbook (Webster, 1976),
which claimed to represent an ‘attempt to establish a consistent method of
describing Romano-British coarse pottery. . . as a move towards the
clarification of the terms in use for types of fabric, decoration and vessel’
(ibid., 3). The title and foreword (from which this quotation is taken) of
the volume imply that discussion is restricted to ‘coarse’ pottery; the
introduction states that ‘the local pottery of Roman Britain. . . consists
of a wide variety of fabrics both coarse and fine’ (ibid., 4). Fabrics
described include a wide range from the hand-made, calcite-gritted Huntcliff
ware to
|