410 



G. S. WILSON 



In order to ascertain the error involved in the actual process of 

 counting, three experiments were performed on three separate 

 occasions. On the first occasion an emulsion of Bad. suipestifer 

 in 1 per cent phenoUsed saline was counted eight times in succes- 

 sion, and the results compared. They are shown in table 1, the 

 counts being in terms of the original emulsion. 



On the second occasion, using an emulsion of Bad. typhosum, 

 five counts were made, with a resulting percentage mean error 

 of 4.04 per cent. On the third occasion ten counts were per- 

 formed with a resulting percentage mean error of 4.95 per cent. 



TABLE 1 



From these three experiments it is seen that the mean error of 

 performing the total count under the conditions described may 

 be taken to be not greater than 5 per cent. It is interesting to 

 compare these results with those obtained by Glynn, Powell, 

 Rees and Cox. Working with staphylococci, streptococci, and 

 Bad. coli they found that in the case of an 0.02 mm., chamber 

 the average percentage deviation from the arithmetic mean was 

 3.1 per cent; with an 0.1 mm., chamber it was 5.4 per cent, and 

 with Wright's blood film method it was 34.1 per cent. With 

 regard to Klein's method, Gotschlich (1912) quotes the mean 

 error of the total count as 19 per cent. 



