W. F. Harvey and A. M'^Kendrick 
GO 
what it is defined to be, the ratio of two averages, and it is with the value of the 
particular average used, as a measure of the phagocytic distribution, that we are 
chiefly concerned. Whether, with varying concentrations of emulsion, longer or 
shorter incubation, the ratio of the averages of normal and test serum distri- 
butions remains the same is quite a separate question and involves the discussion 
of the comparability of indices estimated under different conditions and at different 
times. But if we were merely to limit ourselves to an opinion as to whether a 
given serum is subopsonic, normal or hyperopsonic, we must still know whether it 
is sufficient for this purpose to compare averages based upon a limited count and 
what that count ought to be. 
We distinguish therefore two questions — in reality they merge into one 
another: — (1) the question as to whether the average of the pliagocytic count as 
ordinarily carried out affords a characteristic which is practically sufficient to 
differentiate blood sera ; (2) whether, if the average of a test serum is in every 
case expressed by a separate determination in terms of the average of a normal 
serum taken under the same conditions, we sliall actually get a number which 
is not affected by peculiarities of the observer or other sources of variation. 
Question (1) in the first place requires an answer either in the affirmative or the 
negative, and in the second place necessitates a statement of the approximate 
limits within which that differentiation is justifiable. Now no one who has 
worked much with opsonic indices should have much hesitation in saying that 
they will serve to differentiate a highly specific serum from a normal one. The 
results which we obtain in the case of the glanders bacillus, staphylococci and 
streptococci, the Micrococcus melitensis, and the organisms of typhoid and Malta 
fevers are all so striking as to leave no shadow of doubt upon this simple point. 
It becomes more difficult, however, to dogmatise on the minimum limits within 
which we may accept difference of index as equivalent to significant differentiation. 
This is a point on which mathematician and observer might well work hand in 
hand to reach a conclusion. It will be surmised by workers on the subject that a 
different answer might have to be given to this question in the case of different 
bacteria. Such differences — as far as the specific bacterium alone is concerned — 
may well be connected with the capability of the bacillus to act as an antigen or 
its capability of being opsonised. The susceptibility of one of the reacting sub- 
stances (the opsonisable substance) to the action of the other (the opsonin) is, I 
suspect, one of the elements which enter into the differentiation of phagocytosing 
leucocytes as regards the number of bacteria ingested by them. If the organism 
whether through age, quality of culture medium, virulence or some such cause is 
hard to opsonise, then weakly opsonic sera will not give that differentiation from a 
normal serum which they would do were the antigen somewhat more susceptible 
to the influence of the antibody. Dr Greenwood in an admirable paper on this 
subject has given us some idea of the limits of reliability for differentiation of the 
index in the case of the tubercle bacillus for counts of 25 leucocytes. This 
reliability lay outside the range 0"85 to 1"3. It still remains to examine in similar 
