620 



A. S. PARKES 



same manner, this precaution should remove the first source 

 of error. 



Secondly, bias particularly easily arises in such work and 

 may almost unconsciously detract from the accuracy. A con- 

 scious attempt to discount this bias may lead to the opposite 

 extreme and cause an equal inaccuracy. Also, the work is 

 very trying and strain seriously disturbs the measurements. 

 In general, however, the personal factor was discounted, as far 

 as possible, by only working for very short periods -^A a time. 



Text-fig. 1. 



V »,5 S 55 -b -66 t -IS 6 es •' 



Frequency polygon of head lengths of human spermatozoa. 



Thirdly, the unavoidable margin of error must be considered. 

 Fortunately, however, this is a calculable quantity, and the 

 following tests were made. In the first case the same sperma- 

 tozoon head was measured under three different magnifications, 

 the co-efficients of which were known, and the results com- 

 pared. The following table sums up the results : 



Ob. and Oc. Size in fj.. 



Bausch and Lomb ^" and Zeiss 18 . iOd 



Koristka yV" and Bausch and Lomb 10 4- 10 



KoristkayV" and 'Zeiss 18 . . . 4-14 



