W. E. Macdonell 
179 
on probation, and as their advancement in the Service depends on the accuracy 
with which they measure and describe the prisoners, they may be expected to do 
so with the utmost care. Indeed, I understand from Dr Garson that these 
" practice " forms may be accepted with as much confidence as the official forms 
for habitual criminals. 
It is with these non-habitual criminals that we are concerned. Obviously, the 
larger the number of individuals we include, the better will be the results; on the 
other hand, the number must not be extremely large, otherwise the investigation 
becomes too laborious. Accordingly a group of 3000 was decided on as likely to 
give good results, while still keeping the work within manageable compass. Three 
thousand metric forms were thereupon obtained from Scotland Yard, containing 
particulars of 3000 male prisoners undergoing their sentences in the chief prisons 
of England and Wales. The majority of the prisoners were English and Welsh, 
many were Irish, and only a few Scotch ; no foreigners or youths under 21 were 
included. The forms were drawn at random from the mass on the office shelves ; 
we are therefore dealing with a random sampling. 
The metric description recorded on the forms includes {inter alia) certain 
physical measurements, viz., those (^f the Head Length, Head Breadth, Face 
Breadth, Left Middle Finger, Left Cubit, Left Foot, and Height, of each individual. 
These measurements are taken, in the case of height, to the nearest ^ of an inch, 
in the case of the other characters, to the nearest millimetre ; thus. Head Length 
of 19"2 centimetres includes all head lengths from 19"15 cm. to 19'25cm. ; 
Height of 5 ft. 5^ in. includes all heights from 5 ft. Sj^g in. to 5 ft. 5yg in. These 
characters form the subject of the present memoir, and will be studied from the 
anthropometrical point of view in the earlier part of the paper ; the problem of 
identification, with which Scotland Yard is concerned, will be discussed in the 
latter part. 
(3) Methods employed. In dealing with this large mass of observations, I 
propose to apply the methods explained in Professor Karl Pearson's memoirs in the 
Philosophical T^-ansactions, Vol. 195, A., pp. 1 — 47, 79 — 150, in order to calculate 
the Means, Standard Deviations, and Coefficients of Correlation, with their probable 
errors, for the above seven characters. These methods were devised by Professor 
Pearson to deal with characters not quantitatively nieasureable, but they are used 
here with the object of saving much of the labour involved in making these calcu- 
lations in the ordinary way from the usual elaborate correlation tables. It will be 
seen later on that the saving of time and labour is very considerable. 
To obtain the coefficients of correlation Professor Pearson shows {loc. cit. p. 2) 
that a 4-fold table is required ; to find standard deviations and means, and probable 
errors and error correlations of all the quantities involved, a 9-fold table is required 
(loc. cit. pp. 82, 84). If a 16-fold table is formed, it will give the required 4-fold 
and 9-fold tables, as is shown in the following example, which is a frequency corre- 
lation table for the Head Length and Face Breadth of the 3000 criminals. 
15—2 
