Page 56 - Kennemerland VOC ship, 1664 - Published Reports

NAUTLCAL ARCHAEOLOGY, 4.2
demonstrate objectively whether any factors of archaeological significance produced the patterns observed on the sea-bed. Our basic data is contained in Fig. 8, and its derived clustering diagram (Fig. 9B); Figs 5 and 9A will only be cited where they differ greatly from the other. The first step is to search for evidence reflecting the influence of re-sorting of these patterns subsequent to initial deposition. One consequence of this re- sorting could be to produce totally random distributions ; however, the strong clustering characteristics shown in Fig. 9 rule out that possibility at once. The other likely conse- quence of re-sorting is that the objects will have been rearranged according to the densities of their fabric, with the heavier objects tending to settle in hollows and similar traps, the lighter ones being more widely scattered. Inspection of Fig. 9 shows that this type of patterning seems to be present; both dia- grams display a well-defined cluster, only weakly linked to the rest of the assemblage, consisting of the heaviest classes, numbers V, VIII, and IX (lead shot, bronze fragments, and bronze nails). The central cluster is also fairly homogeneous in respect of fabric, being dominated by stoneware, glass, and p i p d a y . However, the presence of a metallic class, number 111 (pewter bottle tops), also in the central cluster challenges this hypo- thesis. Turning to the alternative interpreta- tion, we must consider whether these patterns can be seen to reflect, in any way, the associa- tion of items when they were on board ship. Inspection of the clusters produces the encouraging observation that the central cluster is concerned with the ship’s cargo, while the peripheral classes are concerned with its equipment, its provisions, and the possessions of its occupants. Classes I, 11, and 111 concern containers for the ship’s liquid cargo,whileclassTV(claypipes)islikelyto consist predominantly, although not ex- clusively, of cargo itself; witness the box packed with such pipes discovered on the near contemporary Vergulde Draeck (Green, I973 : 283). A further indication that this factor is at least partly operative in forming these clusters is the relatively low levels of correla- tion within the ‘metallic’ cluster, and the fact that the two bronze classes correlate more
highly with the lead shot than with each other. It is not possible to be more specific on this point because of uncertainty about the precise functions of the bronze classes on board the ship. In conclusion, it would seem fair to say that both factors appear to have been opera- tive, with that of ship-board association being, if anything, the predominant one. We are thus in a position to affirm that the distributions on this site do reflect factors of archaeological significance, and thus demand further study.
Although more detailed studies must await the complete excavation of this site, several interesting points can be raised at this stage. Once one has accepted that ship-board associations are an important factor determin- ing these distributions, then it should be possible to draw some further conclusions concerning these associations, particularly within the central cluster. The most striking feature here is undoubtedly the relatively low correlation between class IVb (clay pipe bowls) and the two other clay pipe sub- classes (coefficients of 130 and 142 in Fig. 8). It is difficult to account for this feature entirely, but one probable influence is the fact that in some places, particularly in Area A, the pipe bowls were more liable to destruction beyond recognition than in others. It may also be that our coefficients are slightly misleading in this context; it should be noted that the Simple Matching Coefficient pro- duced the very high figure of 0.8 for this relationship (Fig. 6). As noted above, this coefficient is more valuable when comparing abundant with sparse classes, as here. There are similar extenuating circumstances ex- plaining the relatively low figure of correlation between classes I1 and 111, as mentioned in the last paragraph. The highest coefficient within the cluster is that between classes I and 11
(159,both classes concerned with the storage of the ship’s liquid cargo. Evidence from the site shows that the Bellarmine flagons were used to carry mercury (Forster & Higgs, 1973: 298), and probably also preserved fruits (Price & Muckelroy, 1974: 266); we have no direct clues about what was in the glass bottles, but the ship was carrying a large quantity of liquor, which might well have been stored in them. The correlation between
186
~ ~ ~~