152 THE QUANTITATIVE METHOD IN BIOLOGY 



prepared in this way, we determine exactly the weight of the 



precipitate in each 

 ; L mixture. The value 



of this weight de- 

 pends on two causes, 

 one of which (pipette 



A) is invariable, the 

 second one (pipette 



B) consisting of an 

 invariable cause + a 

 variable cause which 

 is chance. The varia- 

 tion (errors) of the 

 weight precip. is 

 therefore governed 

 by the rules of 

 chance, but here 

 the effects of chance 

 are of a peculiar 

 kind. 



Suppose that the 

 greatest errors (with 

 regard to the volume) 

 committed by using 

 pipette B have been 

 + 016 and -o-i6 

 c.c. Between these 

 extremes aU the 

 errors are distributed 

 according to a curve 

 similar to Fig. 18, 

 the negative errors 

 being equal in num- 

 ber to the positive 

 ones. 



Since the volume 

 of a is exactly 10 c.c. 

 in all the mixtures, 

 each negative error 

 results in a diminu- 

 tion of the weight 

 precip. For instance, 

 the greatest nega- 

 tive error, which 

 amounts to 0016 of 

 10 c.c, corresponds to 984 mgr. of precipitate instead of 1000, 

 etc. The positive errors, on the contrary, don't produce any 



Fig. 19. — ABCDEFGHIJK, curve of committed errors 

 (symmetrical). ABCDEL, curve of observed effects 

 (unilateral) 



