THE ANTS OR PISMIRES 



appear in the head region and female characters in the more 

 posterior parts. Whiting has suggested that in ants the 

 mosaics are really intersexes or intercastes, and that the 

 second type of anomaly at least might be due to a change 

 in nursing behaviour during development. This is in fact 

 supported by Brian's observations. It cannot, therefore, be 

 claimed that the existence of caste-mosaics proves that the 

 castes have a genetic origin. To carry the argument further 

 it would be necessary to study the order in which the parts 

 of the ants' external skeleton are laid down, and to test 

 whether after such treatment as starvation the later developed 

 parts show any change in their caste character. 



No clear picture can as yet be given of the type of here- 

 ditary mechanism which could be operative in the more 

 polymorphic type of ant. Some scheme, like the rather 

 theoretical one already mentioned for stingless bees, might 

 be imagined. There are two special features about the social 

 life of ants which might make some such theory more tenable. 

 The possibility of the differential feeding of grubs whose 

 caste the worker ants had recognised at the very beginning 

 of development, has already been mentioned. It is further 

 possible that some of the egg eating which is common in ants, 

 especially in young colonies, may also be differential. Thus, 

 only worker or queen eggs might be allowed to survive at 

 certain seasons, though both types of egg might be produced 

 at all times. This would explain why the supposed here- 

 ditary mechanism, which would be expected to produce all 

 castes in fixed proportions, does not in practice do so, the 

 castes being produced rather according to the needs of the 

 colony. 



There is another process which operates, not only in the 

 ant colony but in any social animal. The unit whose 

 efficiency determines whether the species shall survive or 



k 145 



