180 On Criminal Anthropometry 
Head Length (in cms.) 
Class 
18-8 and under 
18-9 to 19-1 
19-2 to 19-3 
19-4 and over 
Totals 
13-4 and under 
453 
222 
135 
229 
1039 
13-5 to 13-G ... 
147 
99 
63 
152 
461 
13-7 to 13-8 ... 
112 
99 
74 
176 
461 
13'9 and over 
146 
177 
156 
560 
1039 
Totals 
858 
597 
428 
1117 
3000 
From this the following 9-fold table is constructed : 
Head Length. 
Class 
18-8 and under 
18-9 to 19-3 
19-4 and over 
Totals 
13-4 and under 
453 
357 
229 
1039 
13-5 to 13-8 ... 
259 
335 
328 
922 
13-9 and over 
146 
333 
560 
1039 
Totals 
858 
1025 
1117 
3000 
and finally a 4 -fold table to find coefficient of correlation : 
Head Length. 
Class 
19-1 and under 
19-2 and over 
Totals 
13-6 and under 
921 
579 
1500 
JS'7 and over 
534 
966 
1500 
Totals 
1455 
1545 
3000 
(It may be remarked that this latter table would also give the means, but we shall 
use the 9-fold table for this purpose.) 
It seemed at the outset that the best results would be obtained by choosing 
the limits of divisions so that the totals of each column and of each row in the 
4-fold and 9-fold tables should be as nearly as possible equal to a half and a third 
respectively of the whole 3000 ; but this principle was found in some cases to 
be inapplicable*. 
(4) Before making use of these short tables we will first consider the results 
obtained in the case of our 3000 criminals from three frequency correlation tables 
(Tables I. II. and IIL) made up in the usual way. The methods of preparing the 
tables and calculating the results, with their probable errors, are very fully explained 
* In fixing the limits much help was obtained from Dr Garson's results. 
