Page 45 - Kennemerland VOC ship, 1664 - Published Reports

K. MUCKELROY: WRECK SITES
in my opinion, constitute sufficient grounds for further intensive investigation, since adequate information on the operation of this factor on the site could be extracted from this initial sample. This approach evidently relies very heavily on the analysis of the initial sample of the site; a faulty conclusion at this stage would invalidate the whole exercise. In order that this analysis should fulfill the twin criteria of objectivity and repeatability, I
suggest that it must be a statistical analysis.
Statistics and maritime archaeology
aspiring towards the level of ‘inferential statistics’. In the programme outlined above, it is of prime importance that the analysis of the sample of the site should be undertaken rigorously, in such a way that any reader can understand how the conclusions presented were reached, and can thus assess their validity. The explicit use of generally accepted statistical procedures fulfills these criteria perfectly; no other approach is likely to be so satisfactory. There is a need for a more widespread recognition of the many potential applications for statistics in maritime archaeo- logy; a few of these are outlined in the final section of this paper, in the hope that there will be many more experiments in this field.
As mentioned above, the site on which I have been working, and which now serves as a testing ground for the potential of this systematic approach, is that of the Kennemer- land, a Dutch East indiaman wrecked in I664 (Price & Muckelroy, 1974). Within this site, a sample area of 9 0 m 2 of sea-bed has been excavated, in which 6-8000 artefacts were located. The areas concerned were chosen to represent the whole site, and included four particularly rich gulleys, along with an area of more open sea-floor (Price & Muckelroy,
‘Statistical methodology provides an array of
sharp and probing tools for etching and
defining the regularities concealed in data,
and an equally useful conceptual vocabulary
for model building. Because the tools are
sharp, their employment is always dangerous,
especially in a new field of material, but at
the present stage of archaeological indisci-
pline any trials in this direction are preferable
to the indolence of their critics. Statistical
methods are incisive analytical tools, they
bring to light information concealed within
the obscurity of noisy data’ (Clarke, 1968:
143). Whatever may be true of conventional
archaeology nowadays, there can be no doubt
that underwater archaeology is still in a state
of indiscipline ; furthermore, it seems reason-
able to suggest that the sensible application of
certain statistical methods may bring some
order into the subject. However, there is no
need for such an opinion to lead one im-
mediately into the stratosphere of advanced It is obviously important that the areas
statistical analysis; all the procedures used below have been around for several decades, and have proved themselves in many con- texts. Because the tools are sharp, and this is a new field of material for them, there are great virtues in showing restraint in their application.
analysed should have been thoroughly ex- cavated; we are confident that the methods we employed ensured that this was the case (Price & Muckelroy, 1974: 259-60). A more serious failing in this data is that the precision in recording was determined by the capabilities of a team of student divers, most of whom had no previous archaeological experience, rather than by the requirements of the system of analysis. This is the inevitable result of the fact that there are no precedents for this type
In the present paper, it is the power of
statistics to bring to light information out of
‘noisy’ data which is being exploited. Funda-
mentally, we are operating at the level of
‘descriptive statistics’, although in the extent of approach on an underwater site, the
to which we are using the results of our study on a limited sample to apply to a larger entity of which that sample is only a part, we are
programme outlined in this paper being developed only as a response to the data collected from this site. Nevertheless, the
1974: 259). These areas were not chosen in any systematic way; in the future, methods will have to be developed for defining what constitutes a fair sample of such a site, both in terms of the size of the sample, and its distribution across the whole site. The Kennemerland sample can be criticized for being too concentrated in one part of the site.
175