290 THE DIVISION OF LEUCOCYTES 



The table and its summary supply further details. 

 As was expected, the reduction is not very large, but 

 the striking point is that, in addition to the total 

 cancer averages being below the normal, a subdivision 

 into such groups as male and female demonstrates 

 that the reduction in cancer is again present in both 

 groups. 



Every individual case of cancer in the category 

 does not, by any means, have a reduction in the 

 average number of granules contained in the cells, and 

 it will be seen that many of the individual controls 

 showed a reduction; but when one comes to deal 

 with comparatively large numbers, the reduction in 

 carcinoma is demonstrated. It must be remembered 

 that in everybody there is a great variation in the 

 actual number of granules contained in individual cells; 

 and when sampling say five cells from a person, one 

 may by chance hit upon five larger or five smaller 

 cells. Obviously, therefore, it is only by the observa- 

 tion of many cells from large numbers of persons that 

 one can reduce to a minimum the "error of random 

 sampling." We think, however, that the enumeration 

 of the granules contained in 235 cells, from 22 cancer 

 patients and 47 controls, diminishes this error to such 

 an extent that the results are fairly trustworthy. At 

 the same time, it must be remembered that in experi- 

 mentation of this nature the error of random sampling 

 can never be altogether eliminated, and therefore the 

 reliability of the averages depends entirely on the 

 extent of this error among the cells which have been 

 photographed. 



