334 The Mechanism of Evolution in Leptinotaesa 



than in the other lines. I knew from other portions of this investigation that 

 the position and relation of these spots is directly the product of a complex set of 

 body form-factors and that this complex set was capable of dissociation, and that 

 in some instances the factors went far astray as loosely attached members in 

 other portions of the system. There was no possible method of even guessing at 

 what the specific cause of the disturbance might be in this instance, if it were 

 that at all, and there were so many possibilities that any attempt to test out its 

 nature within the species L. multitceniata was hopeless or at least not hopeful. I 

 tried the tests with another species, L. oblongata, in which there is not present 

 the factors productive of the wide separation and divergence of the spots a and a'. 

 In this form they are always parallel and closely set to the median line. I had 

 some pure-line cultures of this species at that time and crossed the suspected 

 material with the L. oblongata lines, and the F-^ results showed at once that the 

 suspicion was well founded, as the four fraternities reared showed strong indi- 

 cations of the multilineata type of pattern ; and in the Fg results some forms 

 came out in the progeny that were in pronotal pattern multitceniata in every 

 respect, with the widely divergent a and a', as well as in other characters. The 

 result was clearly an indication that the " biotype strain " had something in it 

 that gave, in the new opportunity which cross-breeding gave it, the capacity to 

 form new associations and initiate diversity of characters, while in the first posi- 

 tion its location in the system, or lack of proper complements allowed only mani- 

 festation by the lack of stability found in the series of narrowly restricted strains. 



Thus far I had found only a possible source of contamination, but did not 

 know what the contaminator was, excepting that it visibly deranged the two 

 spots mentioned, a rather trivial indicator. It was hopeless to try to purify the 

 strain through the use of neo-Mendelian methods, as the characters were too 

 fine for certain recognition, and the fact of crossing would only introduce addi- 

 tional complications. The only resort left was the effort to purify the race 

 through the elimination in the breeding members of any indication of the con- 

 tamination as revealed in the spots a and a'. This I began early in 1909, paying 

 especial attention to the character of the spot system, with the result that at the 

 end of 1909, having put the stock through four generations, that the range in 

 the index had fallen to less than half of its former range, even though the indices 

 of the mated pairs had not been so rigorously coincident as in other matings. 

 The figures from these generations are shown in the table 34. 



After hibernation from the end of December 1909 to the first of April 1910, 

 the stock was put to breeding again, this time without any attention to either 

 pattern or index, mated at random for the first mating. The progeny showed 

 no change; and again mating as a group culture, the last generation reared 

 showed only the same narrow range in the index. This rather rough treatment 

 of the line gave some indication of its purity and fixity, as well as the freedom 

 from the formerly present factor or factors that had been productive of the wide 

 range in the index in the earlier generations, and there is no doubt that the proc- 

 ess of elimination carried on as I had begun it would have resulted in the for- 

 mation of a race that was free from this one impurity and would have been nar- 

 row-ranging in the index, as were some of the others. 



Not all of the lines gave results as easily as this one, or even were capable of 

 solution at all. Thus, the line shown in 10 g VII (BAB), table 33, was of the 

 same wide range in the index, but the same source of impurity was not present, 



