Page 46 - Kennemerland VOC ship, 1664 - Published Reports

NAUTICAL ARCHAEOLOGY, 4.2
precision achieved was reasonably well suited to the operations subsequently performed on it, as can be seen below; we estimate that we achieved a general accuracy of plus-or- minus half a metre (Price & Muckelroy, 1974: 258).
Statistical procedures
Since the information concerns only a sample area of the site, we can learn little from the absolute locations of every artefact within the South Mouth; that can only be studied once the whole site has been ex- cavated. For present purposes, when we are concerned with determining whether any significant information is contained in these sea-bed distributions, rather than the full extent of that information, more progress can be made by analysing the relationships between the distributions of the different types of find on the site. The archaeological discussions, presented in the next section, are therefore orientated towards identifying the significance of distributions which are very closely related, and those which are mutually exclusive. In this section, attention is focused on the statistical methods of quantifying these interrelationships. The mechanics of the analysis are given in some detail, firstly, so that the reader can reach an objective assessment of the strengths of the various
methods presented, secondly, so that he can criticise the validity of the archaeological conclusions reached, and thirdly, to serve as a guide for anyone .wishing to undertake the same analysis on different data.
The first stage involves transforming the material into a form in which it is susceptible to mathematical treatment. During the course of the excavation the finds were catalogued, either singly or in groups, on the basis of discrete find spots, whose positions were recorded by triangulation from a fixed base-line. Thus the first task is to use this master catalogue to produce sub-catalogues of the different classes of artefact, the classes themselves being selected so as to represent all types of material present, and thus to yield the most information. All the classes are defined so that their status on board the
ship (cargo, personal possessions, cquip- ment, etc.) can be identified. In every case, the unit of measurement is the smallest identifiable entity, on the grounds that any other approach involves unjustifiable assump- tions concerning relative importance. In other words, there is no a priori weighting, a procedure which, in this context, is generally regarded as undesirable (Sneath & Sokal, 1973: 109-1 3). The classes isolated for analysis are:
176
I .
Stoneware potsherds, all apparently from Bellarmine flagons; the unit was a sherd.
IT. Green bottle glass, all apparently from square-based glass bottles; the unit was a piece of glass.
111. Pewter bottle tops, from the bottles of‘ class 11; the unit was a piece of pewter identifiable as coming from a top.
IV. Clay pipes, subdivided into : (a) Pieces of pipe stem.
(b) Pieces of pipe bowl.
(c) All fragments of clay pipes.
V. Lead shot, larger than 0-001ni in diameter; i.e. not ‘scatter shot’.
VI. Personal possessions, as defined in Price & Muckelroy (1974: 264-6).
VII. Items associated with the ship’s arma- ment.
VIII. Fragments of bronze sheeting.
IX. Bronze nails.
X. Fragments of lead, of all shapes.
X1. Bones.
XIl. Pieces of rope.
It is estimated that, in toto, these classes account for over three-quarters of all the finds made on this site. Table I summarizes the incidence of each class in each of the areas excavated. It shows wide variations between areas within each class, but, since the areas are so different in size, direct comparisons at this stage are evidently not possible. The table makes it clear, however, that further analysis of certain classes is not justified; classes VII and XI1 (armaments and rope) are too small to be significant, and are limited to one area, while class X (lead fragments) is also too small, as well as not being functionally homogeneous.
The next step is designed to eliminate the