228 H. GRAHAM CANNON 



survived tliu succoedmg ccdysis. The young in these cases 

 showed stain chiefly in the tips of tlie first and second antennae 

 and in the ' Haftorgan '. By the third instar all traces of the 

 stain had disappeared. 



These experiments show a similarity to those of Sitowski 

 (18) on Tineola hiselliela. This worker fed these 

 cateri)illars on food stained with Sudan red, and their fat 

 became stained red, giving them a red appearance. The eggs 

 laid were also stained red while the animals hatching from 

 them showed signs of a slight red coloration. 



They are also most probably similar to a certain experi- 

 ment of Agar (1) on Bimocephalus vetulus. In 

 Agar's experiment he fed the Daphnids on a food which pro- 

 duced in them a curious abnormality, which consisted in 

 a change from the normal, of the curvature of the valves 

 of the carapace. On removing the abnormal individuals to 

 normal conditions the aljnormality disappeared in a few 

 generations, and up to this point the result is analogous to 

 Sitowski's results and to the experiment recorded here. How- 

 ever, Agar states that not only did the abnormality disappear, 

 but in the third generation of the offspring there was a ' very- 

 decided reaction ' — the valves of the carapace not only 

 came back to their normal position but overshot the mark 

 and became more curved in the opposite direction. This is 

 stated to be due to the overproduction of an anti-body 

 antagonistic in its effects to the substance causing the abnor- 

 mality. I'his occurrence of a reaction w^as supported by 

 a table of ratios representing the transmission of the abnor- 

 mality, and about this table Agar says that, by itself, ' it 

 cannot be said to give unequivocal evidence, especially when 

 the high degree of inaccuracy in the original measurements is 

 considered ', but that this table bore a ' striking resemblance ' 

 to a second table representing the transmission of another 

 abnormality which was based on much more accurate measure- 

 ments and on a much greater number of individuals. But, 

 even supposing that this latter table accurately- represents the 

 course of the second experiment, the value of the resemblance 



