18 
have been no errors either in calculating or in transcribing the figures, then 
it is evident that the figures given above denote accurately the mean 
statures of the two groups of individuals who were measured; but, to what 
extent are we entitled to suppose that they apply equally to those members 
of the bands who were not available for measurement or who refused to be 
measured? The answer to this will depend upon whether random samples 
of the populations were taken; whether the numbers examined were ample; 
and whether the populations were unmixed. Since these Indians were 
measured in the order in which they came to receive their treaty money, 
and since subsequently every tent and habitation was visited in order that 
no one should be overlooked, it is clear that the individuals were not selected 
but were taken at random. Was a sufficient number of adult males in 
each band examined? As it should be the aim to obtain at least 25 indi- 
viduals, it may be said that in the case of Fond-du-lac and Fitzgerald the 
numbers are reasonable; in the case of Chipewyan they are meagre. It 
cannot be claimed for any band that it is totally pure and unmixed Indian, 
but the mixtures essentially consist in having varying amounts of white 
blood, and though pure Indians and those who are tinged with wdiite blood 
are being grouped together, there is no suggestion of including pure Indians 
and pure white persons. 
It has been ascertained that the probable error of the mean stature of 
the Fond-du-lac men is ±0-78 and of the Chipewyan men +1-27. (This 
is set out in table XI.) How ‘ ‘Probable Errors” are calculated need not 
concern us; we may accept them and may proceed in the following manner 
to employ them. 
Stature and probable error of : 
Chipewyans at Chipewyan 167 1 ± 1-27 
“ “ Fond-du-lac 164-7 ± 0*78 
2-4 cm. 
The difference in the statures is obviously 2-4 cm. 
If the probable errors ±1-27 and ±0-78 be squared, the results — as 
any book of tables will show — are 1*6129 and 0*6084. These when added 
together come to 2*2213, the square root of which is 1*490. If this result 
(1*490) which is known as the probable error of the difference of these 
statures (P.E. diff.) be divided into the difference between the statures 
(2*4 cm.) the answer is 1*6; that is to say the difference between the 
statures is 1 *6 times greater than the probable error of that difference, or in 
other words, the ratio between them is as 1 ■ 6 is to 1 * 0. On consulting an 
appropriate table of odds (See page 20), it will be seen that this ratio, if 
translated into terms of odds, will read thus: the chances or odds are 
2*57 to 1*0 in favour of our finding a difference of at least 2*4 cm. in the 
mean stature of the Chipewyans at Fond-du-lac and at Chipewyan, had we 
been able to measure infinitely larger numbers than circumstances per- 
mitted. 
Or if might be expressed thus: were 3*57 such groups of Chipewyan 
Indians to be measured their mean statures would be found to differ by 
2*4 cm. or more in 2*57 of the groups, whereas in one out of the 3*57 the 
difference would be less than 2*4 cm. It is on this basis then that com- 
