Page 76 - Kennemerland VOC ship, 1664 - Published Reports

NAUTICAL ARCHAEOLOGY, 6.1
ffollundiu(16), Scilly Isles (Cowan et al, the environmental characteristics which deter- 1975) mine to which class a site is to belong. The next Lustdruger (7), Shetland Isles (Ste‘nuit, section is concerned with describing the tech- 1974) nique used for discriminating the relative Sunto Christ0 de Gzstello (lo), Cornwall importance of the various characteristics, after which there is a discussion of the main conclu-
(McBride et al, 1975)
In none of these cases is the destruction of material as extensive as established theories would have led us to believe. In some cases, furthermore, the sea-bed distributions are evidently of some archaeological Significance, in reflecting the process of wrecking. It should be noted that the distinction between this class and the previous one is essentially the presence or absence of organic remains, and that some sites have been placed in this class simply on the basis of silence in the relevant sources; this assumption may not have been justified in every instance.
Class 5. Finally, there are those sites which really have been smashed up by natural forces, and on which only heavy metal and stone objects survive, often in a heavily abraded condition. The most interesting point about this class is its size; it appears to be nothing like as large as Class 4, although this is partly because such unproductive sites have not attracted much archaeological attention to date. Included in the class are:
Adeluur (4), Barra (Martin, 1972)
Gun Rocks, Northumberland
Low Lee Ledges (13), Cornwall
Mewstone Ledge (14), Devon (Middlewood,
1975)
Penlee Cannon Site (12), Cornwall (Bax &
Farrell, 1975) Wendela(17), Shetland Isles
As with the artefactual collections from these sites, which are not representative of the range of items on board the vessels involved, the distributions of the material over the sea-bed are usually of limited significance. However, as potential sources of a certain amount of infor- mation which is otherwise totally unobtainable, they deserve more serious attention than they have on the whole received hitherto, with the notable exceptions, of course, of the Adeluar and Penlee sites.
Having thus established the archaeological categories involved, the next stage is to identify
50
sions suggested by this exercise.
3. Site attribute correlation
In one sense, there are an almost limitless number of factors which could be considered when seeking to understand the marine environ- ment of any particular site. However, the field of enquiry can be restricted by considering only those factors whch parallel studies in other marine sciences have found to be of impor- tance; in particular, reference has been made to work in marine biology (Hiscock, 1974) and coastal geomorphology (King, 1972). Unfor- tunately, there are no studies in the latter discipline giving any direct answers to our problems; on the contrary, some workers in submarine geomorphology are looking t o maritime archaeology t o provide them with new data on sea-bed processes. Consideration has thus concentrated on fetch, sea horizon, storm winds, tidal currents, depth, underwater slope and topography, the nature of the sea-bed, and any covering of marine plants. In trying to quantify all these factors, one is faced with a wide range of scales, ranging from the qualita- tive to the numeric, which are in turn to be compared with site archaeological character- istics, ordered in five discrete classes. In order to allow statistical correlation between them all, they have all been converted into ranked values, as displayed in Fig. 2. The Kendall Rank Correlation Coefficient 7 (tau)has been used to quantify these relationships (Siegel, 1956: 213-23), comparing each attribute across 20 wreck sites, as indicated in Fig. 1. This is the maximum number for which satisfactory information can be obtained concerning both archaeological and environmental characteris- tics. The calculations have been performed on the University of St Andrews IBM 360/44 computer, using the RAX statistical system. In order to try and ensure that all the correlation coefficients are positive, each scale has been ranked in increasing levels of ‘severity’ (e.g. increasing fetch, stronger tidal streams, decreas- ing depth). This ranking procedure allows the