312 



THE AMERICAN NATURALIST [Vol. XLIV 



ably not absolutely insurmountable. In any event it would seem 

 that indirect methods of studying inheritance should be only a 

 sort of last resort when direct methods have failed or can not be 



The correlations are also found to be reduced in the nest as 

 compared with the "general" population. The amount of this 

 reduction is shown to be that which would be expected as a 

 statistical consequence of the observed reduction in variability 

 of the one population compared with the other. 



"Warren's (15) termite paper contains a wealth of interesting 

 biological data. It presents the constants calculated from a very 

 extensive series of measurements of the character head breadth 

 in the different castes of several species of South African ter- 

 mites. The bulk of the data are from one species, Termes vata- 

 lensis. A typical nest of this species contains the following 

 kinds of individuals: (1) A single king and a single queen, 

 these being the only sexually mature forms present; (2) soldiers 

 of two sizes, asexual; (3) workers of two sizes, asexual; (4) 

 winged males and females, not sexually mature; (5) young or 

 immature members of castes (2), (3) and (4). Some of the 

 more interesting results which come out of this work are as 

 follows : 



There is considerable seasonal variation in the individuals of 

 the same nest. The mean is smallest and the absolute varia- 

 bility greatest in material taken in November, and the mean is 

 greatest and the absolute variability least in March material. 

 Warren offers the following explanation of this result: 



This seasonal variation probably arises from two causes at least: 

 (1) the elimination of the physically unfit, (2) post- "adult" growth. 

 With reference to the first cause it may he noticed that it is very prob- 

 able that more individuals arrive at maturity from August to November, 

 that is during the first rains, than at other seasons, and therefore the 

 stunted adults will be more abundant during this period, with the 

 result that the mean would be lowered and the standard deviation 

 would be raised. By the time that March arrives the small and weakly 

 individuals of the nest are likely to have died, and consequently the 



age of the study 



