erations, was lost. The reason for this is, to us, not yet clear, 

 as in this new line the ground color of all wings is much 

 darker than in the old line of experiments, consequently there 

 is no want of dark pigment in the new development ; the only 

 reason we can see, so far, lies in the fact that the ocelli in the 

 butterfly of this new line of experiments are much smaller 

 than the ocelli of the butterflies of our old line, and many 

 butterfles of our old line developed first a larger .ocellus of 

 the forewing and only then developed the appendix. 



The fourth generation developed, in lot (a), 28 females 

 and 23 males; and lot (b) 23 females and 15 males; there were 

 in this generation about 407f- with purple colored ocelli; the 

 reason for the smaller percentage v;ith purple color seems to 

 be that the yellow color is a dominant one over the purple, and 

 as the more yellow color was fixed in the mother, it will 

 take some time before the purple color is steadily fixed. 



The fifth generation, in lot (a), developed 56 females and 

 65 males; and in lot (b) 42 females and 40 males; about 50%, 

 had in this generation, the new purple color. As by this cross- 

 ing method the size of the butterfles did not increase as in 

 the second generation, we gave it up for want of time and 

 space, and continued in a single line by inbreeding, without 

 crossing. As we had much trouble in the caterpillars with the 

 wilt disease, we raised the sixth generation of our caterpillars 

 in a cooler temperature to stamp out this disease. 



We made another experiment with this new line of but- 

 terfles by picking out from the second generation, bred in 

 humid air, the males and females which showed the most 

 purple color on all the wings, and obtained many eggs from 

 these extra selected purple ones; we raised the caterpillars 

 also in humid air. the temperature being on the averag_75°,_ a 

 little cooler than the above described to prevent the wilt dis- 

 ease ; and bred the chrysalides, as the others, in 90° tempera- 

 ture and humid air. We obtained in this line of experiment, 

 which formed the third generation, 81 females; 45 of these 

 had the large ocelli of the hindwings purple, and the fore- 

 wings had in manv a good purple color; and 80 males, 36 of 

 these had the ocelli purple and a few had also some purple 

 color on the forewing. In this lot they were all well-devel- 

 oped but none as large as those in the first generation; the 

 percentage of the purple colored examples was higher than in 

 the other' third generation which we had crossed in each gen- 

 eration ; the males, especially, showed the* purple color well, 

 as compared with the other third generation ; this higher per- 

 centage was undoubtedly due to the selection of the most 

 purple colored parents. As the descendants of the other Ime 

 did not increase in size by crossing the same in each genera- 

 tion, we crossed none in this line, and had for that reason a 

 much better chance to select the very best stock of the most 



33 



