M. Greenwood and J. D. C. AVhite 
399 
Too much weight must not be assigned to this result, but the indication 
undoubtedly is that the Strangeways counts are more truly random samples, 
and therefore better measures of the phagocytic " population " than those ot 
Dr Fleming. 
While the evidence undoubtedly points to the conclusion that the difference 
between the F. and S. series is not wholly to be accounted for by the increased 
thickness of emulsion employed for the former counts, we arc not prepared to 
assert that the foregoing analysis is sufficient to render the conclusion certain, and 
we hope to deal with the matter on a subsequent occasion. For these reasons it 
would not be profitable to discuss the question as to how far differences in 
technique might operate in depriving a count of its character as a random 
sample. In justice to Dr Strangeways and his co-workers, however, we must point 
out that the indirect attacks which have been made* on their competency as 
observers of phagocytic distributions are entirely unsupported by our investigations. 
We see no reason whatever for refusing to assign at least equal degrees of 
importance to the S. data and to those of Dr Fleming. 
Gathering up our results, do we find that they affect one's estimate of the 
practical value of the opsonic test ? 
One thing is certain ; it is not proper to test the distribution either of means 
or opsonic indices on the basis of a normal population. The population is 
pronouncedly skew. In view of the extreme skewness of these phagocytic distri- 
butions, the mode would be a far better constant for the determination of the 
opsonic index than the mean ; the latter's physical significance has no longer the 
value attaching to it in the case of normal distributions. 
At one time we had a hopef that this skewness, so accentuated in Dr Strange- 
ways' counts of low mean value, might disappear with Dr Fleming's counts, which 
were rendered higher by the use of a thicker bacillary emulsion ; we surmised that 
the consequent increase in mean and modal values would be associated with an 
approach to symmetry which might justify the use of the normal curve in practical 
testing. We did indeed find in the latter curves a less pronounced degree of 
kurtosis, but the reason of this is a matter of doubt. In any case, even in the 
F. counts, the skewness is far too distinct to allow the hope of any Gaussian tests 
being yet valid. Moreover, we have reason to think that Dr Fleming has reached 
the limit of thickness in emulsion compatible with accurate counting (vid. sup. 
* As for example, that of Dr E. W. Allen, Vaccine Thevapij and the Op^onk Methud of Treatment, 
London, 1908, 2nd Edition, p. 40. Dr Allen's comparison of workers who find the opsonic method 
unreliable with those surgeons who are unable to perform certain delicate surgical operations seems to 
us altogether illegitimate. In this connection we must point out that a reference by Dr Allen (op. cit. 
p. 41) to a preliminary communication by one of us is liable to misconstruction. The statement that 
the "unavoidable error" involved in opsonic determination is not more than 10 per cent, is not a 
deduction from our preliminary note but rests entirely on the authority of Dr Allen and other patho- 
logists. 
t Greenwood, Practitioner, Vol. 80, 1908, p. 641. 
