No example was found of 83, say of one-third, of the 243 possible combinations. In one case no less than 24 sets fell under the same head; in another case 19 did so, and there were two cases in which 14, 11, and 10 severally did the same. Thus, out of 500 sets (see the five bottom lines in the last column of the above table) no less than 113 sets fell into four classes, each of which included from 10 to 24 entries.
The 24 sets whose Index-number is + M, + + + admit of being easily subdivided and rapidly sorted by an expert, into smaller groups, paying regard to considerable differences only, in the head-length and head-breadth. After doing this, two comparatively large groups remain, with five cases in each, which require further analysis. They are as follow, the height and eye-colour being added in each case, and brackets being so placed as to indicate measures that do not differ to a sufficient amount to be surely distinguished. No two sets are alike throughout, some difference of considerable magnitude always occurring to distinguish them. Nos. 2 and 3 come closest together, and are distinguished by eye-colour alone.
Five cases of Head-length 8·0, and Head-breadth 6·1.
1. { 72·4 38·0 4·8 { 71·2 { br. grey 2. { 72·6 { 37·0 { 4·7 { 71·4 { br. grey 3. { 72·7 { 36·7 { 4·7 { 71·4 blue 4. 73·9 36·4 5·0 70·7 brown 5. 75·3 37·9 4·8 73·4 blue
Five cases of Head-length 7·8, and Head-breadth 6·0.
6. 70·8 37·8 { 4·7 { 70·0 brown 7. { 71·9 36·2 { 4·7 { 69·3 blue 8. { 72·4 { 37·2 { 4·7 { 68·4 brown 9. 74·8 { 37·8 5·0 73·1 blue 10. 79·9 { 37·3 5·3 75·6 blue grey
This is satisfactory. It shows that each one of the 500 sets may be distinguished from all the others by means of only seven elements; for if it is possible so to subdivide twenty-four entries that come under one index-heading, we may assume that we could do so in the other cases where the entries were fewer. The other measures that I possess--strength of grasp and breathing capacity--are closely correlated with stature and bulk, while eyesight and reaction-time are uncorrelated, but the latter are hardly suited to test the further application of the Bertillon method.
It would appear, from these and other data, that a purely anthropometric classification, irrespective of bodily marks and photographs, would enable an expert to deal with registers of considerable size.