324 WHEELER— ANT LARV^. 



the other hand, it has often been suggested (<?. g., by von Buttel- 

 Reepen) that the three social subfamiHes, the stingless bees (Meli- 

 poninge), bumble-bees (Bombinse) and honey bees (Apinse) hax^e 

 developed from the solitary bees by another and more direct path, 

 for the Meliponinse, though living in populous societies, still bring 

 up their brood in essentially the same way as the solitary bees, i. e., 

 by sealing up the eggs in cells provisioned with honey-soaked pollen. 

 The Bombinae, however, keep opening the cells from time to time 

 and giving the larvae a little food at a time, and in the honey bee the 

 cells are left open till pupation and the larvae fed more continuously. 

 Numerous facts indicate that the Bombinae are the most primitive, 

 the Apinse the most specialized of existing social bees, and that the 

 Meliponinae, though closely resembling the solitary bees in the care 

 of the young, are nevertheless in other respects very highly spe- 

 cialized (vestigial sting, elaborate nest architecture, etc.). It is 

 therefore not improbable that these bees, after passing through a 

 stage more like that of the Bombinse, have reverted secondarily to 

 a more ancient method of caring for their brood. At any rate, the 

 Meliponinse have been so little studied, as compared with the Bom- 

 binae and Apinse, that they can be left out of the present discussion. 

 Sladen (191 2) has given us a good account of the queen bumble 

 bee feeding the larvae, but he says nothing about the salivary glands 

 of the latter. These are very large, as we know from the work of 

 Bordas (1894), but their development is perhaps fully accounted 

 for by the complete cocoon spun by the mature larva. Even in the 

 honey bee, which has been so thoroughly studied, I find no evidence 

 that the adult workers feed on larval secretions. In both cases, 

 however, it is impossible under natural conditions accurately to 

 observe the behavior of the larva while it is being fed. This 

 might, perhaps, be done if the bees could be induced to rear their 

 young in glass tubes made to resemble the cells.^ But even if it 

 should be found, on further investigation, that there is no indica- 

 tion of reciprocal feeding between the larval and adult Bombinse 

 and Apinse, we might still contend that these very highly specialized 



^ Dr. E. F. Phillips informs me that it would be possible to observe the 

 behavior of honey-bee larvae and their nurses in cells built against and partly 

 formed by the glass wall of an observation hive. 



