218 CERTAIN LAWS OF VARIATION. 



1.9 per cent., were put into one group, and a mean 

 taken of their " variability " numbers. The larvae 

 which, owing to a more favourable environment, were 

 larger than the normal, and which differed from it by 

 from + 2 to + 4.9 per cent., were similarly grouped 

 together. Still other groups were formed of those dif- 

 fering by from '+ 5 to + 9.9 per cent, and by + 10 

 per cent, and more. Like groups were made of larvae 

 which adverse environmental conditions had rendered 

 smaller than the normal. The following are the mean 

 results : 



VARIATION IN BODT LENGTH OP NUMBER OF SERIES VARIABILITY 



LARVAE FROM NORMAL. MEASUREMENTS. 



-f- 10 per cent, and over, . . '. . 26 89 . 6 



5 to -f 9. 9 per cent 30 968 



-j- 4. 9 per cent., ... . 51 96.4 



1.9 per cent., . . . .170 96.8 



2 to 4.9 percent., . . . . 63 100.8 

 - 5 to - 9.9 per cent., .... 43 104.4 



10 per cent, and over, .... 29 113.2 



Here we see that, with one slight exception, the varia- 

 bility becomes steadily greater and greater as the en- 

 vironment becomes more and more unfavourable, the 

 extreme value in one direction being 26.3 per cent, 

 greater than that in the other. This experimental re- 

 sult, bearing out as it does our theoretical conclusion, 

 seems to justify one in assuming the existence of a defi- 

 nite law of variability. This may be worded as fol- 

 lows: " An organism varies least when it is best adapted 

 to its surroundings, so that the less it is adapted, the 

 more variable does it become." Extended research 

 alone can teach us how general may be the application 

 of this law, and afford adequate numerical estimates as 



