Page 53 - Kennemerland VOC ship, 1664 - Published Reports

K. MUCKELROY: WRECK SITES
I 11 riI 1Va IVb IVC V VI VIII IX XI
F1 23.90
F 2 9.23
F3 21.48
D 5.54
E 9.23
B 5.20
A1 1.o 0.54
15.42 12.94 29.85
8.96 12.44 9.95
0.49
1.49 4.48
1.49 1.49
0.49 0.49
17.36 14.46 26.86
7.85 12.40 11.57
0.41
1.24 3.72
1.24 1.24
0.83
0.41 0.41
3.23 10.75 70.43
5.38 245 1.07
0.53 5.38 0.53 0.53
0.25
7.14
4.41 64.29 4.17
0.25
10.71
3.13 1.47 2.08
2.08
3.13 1.47 5.20 24.02 5.20 1.72
0.25 14.29
2.20 3.57 10.42 10.29
14.59 6.86 3.92
14.59 8.58 2.08 16.91 17.46
A2 1.o
A3 0.34
A4 2.19
A5
A6 3.86
A7 2.85
A8 3.86
A9 4.38
A10 1.68
A1I 3.02 A12 0.68 A13
A14 0.84
A15 0.34
A16 050
1.90 0.54 3.13 0.14 0.54 0.95 0.82 0.54 0.82 0.82
0.41 0.14 1.09
30.20 3.13
21.90
25.76 31.81 4.55 6.06 10.60 7.58
1.52 6.06
1.52
1.52 1.52
1.52
26.83 21.95 12.20
2.44 12.20 19.51
4.88
11.0
1.34 1.52 0.12 0.06
067
0.12
0.79 17.15 3.71 20.26
8.27 12.65 2.19 4.50
13.69 12.96
17.40 8 .o 9.80 18.23
Figure 7. Matrix of percentages of 11 classes of artefact in each four square metre area.
I 11 111 IVa IVb IVc V V1 VIlI IX XI
I+
11 155 -t
111 127 1Va 143 IVb 130 IVc 146
V 48 VI 88 VllI 54 1x 41 XI 82
131 f
153 121 +
153 149 130 +
161 129 188 21 18 16 85 67 108 30 28 26 25 22 18 75 47 97
142 f
10 16 +
64 102 8t
15 25 125 16 -k
20 19 125 12 80 +
60 92 41 145 35 14 i
Figure 8. Similarity matrix, Method 2 (Robinson/Brainerd).
presence/absence matrix cannot adequately deal with classes of artefacts which are highly concentrated in a small number of squares. However, this comparison does confirm that the Coefficient of Jaccard produces a more trustworthy result than the Simple Matching Coefficient, which supports the previous suggestion that this will always be the case on sites where a number of classes are relatively sparse, or highly clustered in a few areas. With respect to the question of choosing between the two principal approaches, my recom- mendation is that, while the Robinson/
Brainerd method gives a more discriminating result, Method 1, using the Coefficient of Jaccard, can give a useful view of the data in circumstances where either the quality of the data, or the facilities available to the operator, preclude the use of Method 2. One can also draw conclusions from these tests regarding the level of accuracy in recording required in the excavation of the initial sample of a site. Since all the possible techniques of analysis make use of unit areas (squares), a system using a permanent framework on the sea-bed would appear to be most suitable, although
183