528 On the Distribution of the Means of Samples 
TABLE XV. 
Comparison of Odds against Certain Index Deviations as Calculated 
on Different Assumptions for Savqyles of 50. 
Index 
Using True 
Mean 
Experimental Curve 
without Keplacement 
Experimental Curve 
with Keplacement 
•7 
88-3 to 1 
21-3 to 1 
20-8 to 1 
•8 
15-2 to 1 
■ 6-16 to 1 
5-88 to 1 
■9 
3-38 to 1 
2-26 to 1 
2-23 to 1 
1-1 
3-63 to 1 
2-19 to 1 
2-16 to 1 
12-0 to 1 
4-45 to 1 
4-57 to 1 
1-3 
39-3 to 1 
8-81 to 1 
9-99 to 1 
Without in the least wishing to underestimate the shortcomings of both our 
material and method, it appears to us that Table XII can fairly be used as a 
rough practical test of the accuracy of the opsonic method in the case of the 
Tubercle Bacillus and that it confirms the opinions expressed in our former paper. 
A theoretical discussion of the index problem in general is reserved for another 
paper and more suitable material. It is now proper to consider very briefly some 
practical points. 
In the first place, it has been asserted that the Tubercle Bacillus is an 
unfavourable organism to select for the purpose of studying the accuracy of the 
opsonic technic. While we cannot assert that no more favourable example could 
have been chosen, we are of opinion that the Tubercle Bacillus is certainly not 
the least satisfactory of the organisms commonly dealt with. For insjiance, Tubercle 
Bacilli act, or are acted on, almost entirely as single individuals, while Staphylococci 
or Streptococci occur in groups or chains respectively. These groups or chains are 
counted as so many individuals which, from the point of view of phagocytosis, may 
or may not be correct. The opsonic power required to attract a group of cocci to 
the leucocyte may not be greater than that associated with the ingestion of a 
single coccus. Under such conditions we have the difficulties as to the correct 
treatment of doublets and triplets— to which allusion has been made by Harvey 
and McKendrick — in an acute form. 
Next as to the practical accuracy of the method for diagnostic purposes or in 
the control of dosage. The results here communicated, which are in agreement 
with our previous observations and also, mutatis mutandis, with those of Harvey 
and McKendrick, seem to demonstrate that single observations falling within the 
limits of 20 % ^''^ either side of the mean cannot, in the case of Tubercle Bacilli, 
be regarded as strongly suggestive of abnormality. If it is possible, as will often 
be the case when it is a question of diagnosis alone, to repeat the observation 
several times under comparable conditions, the opsonic method may be expected 
to yield results of a certain value. When, however, the opsonic index is used to 
