M. Greenwood and J. D. C. White 
381 
measuring the entire adult population of a village will give as good a measure of 
racial characters as a sample of the same number of persons drawn indiscriminately 
from a whole population. 
Although we cannot satisfy ourselves that Di' Strangeways and his colleagues 
have demonstrated that the opsonic method is unsound, their paper marked, in 
our opinion, an important step in advance. They were, we think, the first students 
of the method to realise the necessity of counting large numbers of cells, in some 
cases extending their counts to 2000 cells, and to attempt to put the matter upon 
an adequate statistical or arithmetical basis. 
Should the present memoir be of any service to subsequent investigators, it 
will be owing to the kindness of Dr Strangeways, who has placed all his valuable 
material at our disposal. No useful purpose would be served by a further 
attempt to summarise the literature of opsonic determinations*. To do so 
would entail the recitation of absolutely irreconcilable conclusions, based, so far 
as we can judge, upon but slender evidence. 
The accounts to which we have more particularly referred, ably present the 
opinions of two schools of observers, and will enable the statistician " to see the 
cloud of doubt in which things are wrapped." We shall now describe the manner 
in which we have attempted to elucidate the points at issue. 
The practical man has a right to demand that a biometric problem should first 
be attacked in the most simple and direct manner. Only when simple methods 
are demonstrably insufficient are we justified in adopting a more elaborate and 
less direct plan. 
The most direct way to test the value of opsonic determinations is to compare 
the differences between the mean of a "control" phagocytic count and the 
respective means of patients' counts, made on the same day and with the use of 
the same emulsion, with the probable errors of such differences calculated on the 
assumption of a normal distribution of the observations. It could then be seen 
whether the indices corresponded to significant differences of mean values. 
Now, at the London Hospital, one or two control counts of 75 cells which have 
been in contact with the serum of the investigator, serve as a standard by which 
other counts, each of 75, made on the same day are measured. 
The actual figures having been preserved, numerous series were available for 
analysis, thanks to the kindness of our colleague, Dr Western. 
Assuming the normality of the distributions, results were obtained, a few of 
which are reproduced in Table I. 
At first sight these results suggest that variations of say + 20 per cent, in the 
mean value, or opsonic indices above 12 or below 0'8, are significant were it not 
for the existence of a striking peculiarity. 
* Eeference may be had to Manwaring and Euh, Journ. of Experimental Medicine, Vol. ix. 1907, 
p. 473. C. E. Simon, ihid. p. 487. C. F. Bolduan, Lomj Island Medical Journal, Vol. i. 1907, 
No. 10 and Medical Record, Jan. 4, 1908. 
