Karl Pearson 
207 
If a population of a 1000 leucocytes, prepared in two different places and 
counted by those who prepared them, show practically the same range of variation, 
then it must be clear that if random samples of 50 or 100 were taken out of 
these 1000, we should reach the same distribution of possible means of bacilli 
for both of these populations and that accordingly the opsonic index as deter- 
Diagram I. (Fleming. ) 
-2-10 1 
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 
Bacilli per Leucocyte. 
mined by the individual as against himself would show just as wide variations 
as those demonstrated in recent papers in Biometrika* for Greenwood and 
White's material. There is only one flaw in this argument — possibly neither 
Fleming nor Greenwood and White would in an actual count of 50 or 100 
really take random samples from their 1000 counts — this is a point to which 
I shall return later on in this paper. 
(4) In order to confirm the result of the previous section, namely that 
whether the slide was or was not prepared in Sir Almroth Wright's Laboratory, 
it did not affect the variation in the resulting population, I took a slide prepared 
* See Vol. vii. pp. 505—531, and pp. 531—541. 
