J. F. Tocher 
141 
by them of the importance of this scientific investigation, their cordial cooperation 
and most painstaking and laborious setting down of all the minute details 
required from each school, the survey would have been still in the limbo of 
fancy, to remain there until the census office should have the power to deal with 
the matter, along with the present ordinary details of this important statistical 
department. Only when the recording of measurable and non-measurable cha- 
racters comes to be included in the census, and is dealt with officially, will the 
importance of much voluntary pioneer work by the teaching profession be fully 
recognised. 
(4) The Problems to he discussed. 
Before proceeding to make a brief statement of the analytical methods employed 
and to follow with a general discussion of the resulting classified data, it seems 
desirable at this stage to enumerate the problems germane to the survey. 
(ft) The first problem clearly is : How are the children distributed with 
respect to the various colour classes, what is the proportion of children found in 
each class, and how does the general distribution among the classes compare with 
those of the continental countries already surveyed ? 
The answer to this problem is given (a) in Table XIII., where the general 
distribution and the percentages of the colour classes are given, aud (/3) in section 
(12), where the results are compared with those of continental countries. 
{h) The second problem deals with relative local differences in each colour 
class. Considering each colour class or category separately, one must ask, by how 
much does each locality in Scotland (division, county or district) differ from the 
remaining population ? In other words, is the distribution of colour uniform 
throughout Scotland, and if not by how much does the proportion for each class 
in each locality differ from the proportion which would occur on an even distribu- 
tion of the school population over the whole country ? This amount when found 
for each locality is termed the relative local difference and the complete solution 
of the problem is reached when significant relative local differences are determined, 
and separated from those relative local differences which are fair samples of the 
general population. This problem is dealt with under section (6). 
(c) The third problem is one bearing on the general resemblance of local 
populations to the general population. Here hair colour as a character is con- 
sidered as a whole in each locality, all the classes constituting the character being 
considered together. Similarly eye colour as a character is considered as a whole 
in each locality. The distribution in each locality of the classes constituting each 
character is compared with the corresponding general distribution of tlie classes 
for the same characters which is found for the whole country. Considering, in 
this manner, hair colour collectively or eye colour collectively, do or do not 
local populations resemble the general population ? If local populations do not 
resemble the general population how far do the actual local frequencies as a whole 
differ from the corresponding frequencies which would occur on an even distribu- 
