Karl Pearson 
205 
it, especially having regard to the other statements in his paper, that this 
functional error may be due primarily to two sources : 
(a) the technique of preparing the slide for counting, (b) the actual method 
adopted in counting. We might hope to consider these facts in the following 
manner : 
(i) The slide might be prepared by one of Sir Almroth Wright's own 
workers. 
(a) The actual counting might be undertaken by one of his staff. 
(/?) The actual counting might be undertaken by somebody else. 
(ii) The slide might be prepared by one of the critics suggested by Sir 
Almroth as having " enormous functional errors." 
(a) The actual counting might be undertaken by the same critic. 
(J3) The actual counting might be undertaken by somebody else, e.g. an 
assistant of Sir Almroth. 
It is clear that if we could have the results obtained in these four different 
ways, we should — with the assistance of a mathematical statistician — be able to 
throw much light on the problem of whether the variations of sampling or the 
variations of functioning are the more important factor in opsonic index variation. 
Now unfortunately I am not in possession of data covering all the four cases 
cited above. But it is possible, from published work, to compare the results 
obtained when : 
(i) (a) A slide was prepared and the bacilli counted by Dr Fleming of 
Sir Almroth Wright's own laboratory*. 
(/3) A slide was prepared in Sir Almroth Wright's own Laboratory, but 
counted by Dr Strange ways f. 
(ii) A slide was prepared elsewhere and the count made by the operator, 
Dr White +. 
The whole material dealt with in these cases was for the tubercle bacillus. 
(3) The first point I asked myself was this : Does the actual population 
on the slide differ substantially according to whether the slide has been prepared 
in Sir Almroth's Laboratory or not ? I took therefore the data for Fleming's 
* Biometriha, vi. p. 384, Slide T. A. i. Our results here and later differ slightly from Dr Greenwood's 
(loc. cit. pp. 385 — 6, and Graph xiv. ) because, (i) we have taken the first thousand (not the whole 1100) 
of Fleming's returns and (ii) Dr Greenwood used Dr Fleming's means, which we have recalculated 
as we found them not without error. 
f This is the only interpretation I can put on the footnotes to Strangeways' original paper (Bulletin 
of the Committee for the Study of Special Diseases, Vol. i. p. 133) and to Greenwood's paper, A Statistical 
View of the Opsonic Index (Proceedings of the Royal Society of Medicine, March 1909, p. 7 of Offprint). 
I am subject to correction, but I do not understand why there should be any secretiveness as to the 
origin of these slides. 
% Biometriha, Vol. vn. p. 505. 
