188 



THE AMERICAN NATURALIST [Vol. XLIX 



to normals in pairs, gave 778 normals and 691 confluents, show- 

 ing that they too were heterozygous. 2 The remaining six were 

 bred together in pairs and gave 261 normal and 431 confluent 

 progeny, or a ratio of approximately 1 : 2 instead of the expected 

 1 :3. According to expectation one third of the 431 confluent 

 offspring in this generation should be homozygous, and random 

 matings in pairs (confluent by confluent), should give in five 

 cases out of nine only confluent progeny. Sixteen such matings 

 have been made, none of which gave this result ; instead each gave 

 approximately one normal to two confluent, just as did the $\ 

 heterozygotes. Normal brothers and sisters of confluent in both 

 generations bred en masse gave only normals, showing that none 

 of them was heterozygous for confluent. From these data we 

 conclude that the homozygous confluent flies are not viable, and 

 that the 1:2 ratio is due to the total absence of this class. To 

 our knowledge such a condition as this has been previously re- 

 corded in only three cases: the "aurea" Antirrhinum of Baur, 

 the yellow mouse of Cuenot, Castle, etc., and the dwarf wheat of 

 Vilmorin. Baur's ease differs somewhat from the others and 

 from ours in that the homozygous mutant class appears, but 

 soon dies (due to the absence of chlorophyll). 



With regard to the origin of mutations the present cases are 

 instructive in showing that they may appear without the use 

 of artificial chemical or physical agents, and without hybridiza- 

 tion. No radium, X-rays or any chemicals whatever have been 

 applied to these cultures, except ether, and that only for anes- 

 thesia of the adult flies in each generation. The stock of D. 

 tripunctata from which axillary arose was obtained wild, and 

 had been inbred for six or seven generations; that of the other 

 species, from which confluent arose, is all descended from one 

 pair of wild flies, almost certainly brother and sister, and had 

 been inbred for about twelve generations when the mutant ap- 

 peared. In neither case had flies from two localities been crossed ; 

 both stocks were pure and inbred. The only agent that could 

 possibly fall under suspicion as a causative one, then, is ether, 

 but this was used uniformly throughout the experiments, and 

 since only two mutations appeared among many thousands of 

 flies, there is no reason for attributing them to the specific 



2 The offspring per pair were respectively: 242 : 194, 93 : 73, 97 : 106, 



