Analysis of Heterogeneity in Some Simplest Charactebs 235 



or at least was not to be detected by the same methods. It was tested with con- 

 trolled conditions to no purpose, and no clue was found to the range in the index 

 present, so that selection through 1909 and 1910 failed to change the array in 

 any respect. This may be interpreted in different ways, but the range was surely 

 not environmental, else it would have shown at least an indication to respond 

 to regulated conditions, which it did not at any time; or if smaller germinal 

 differences were present it must surely have been possible to have done at some 

 time the same here as in other lines, but more probable seems the hypothesis 

 that there was some " impurity " present, which my methods of detection did 

 not reach, and which by its presence in the system continued in spite of rigorous 

 methods to produce wide ranges in the index of this biot3rpic line. 



The results in this series were a decided surprise to me, and indicate clearly 

 the difficulties of the problem and the immense complexity thereof. As far as 

 the cause of the " fluctuations " was concerned, the outcome was not expected or 

 anticipated on the basis of the assertions made regarding these differences in 

 organisms, and the quite general independence of the index from environmental 

 influences is striking, as is the possibility of finer divisions and the presence of 

 impurities. It would be of interest to completely analyze the nature of the 

 entire fraternity resulting from any pair of parents, and I have often thought 

 that it would be worth while to attempt it, but I have never been able to obtain 

 the space or time free from other work to allow of the attempt being made. 



At the same time that the experiments with the muUilineata biotype series 

 were in progress, another part of the same species gave interesting confirmation 

 of the same principles, but, as might be expected, with diversity in the details. 

 In the course of the preliminary testing of this species I had separated out a 

 form " biotype " in which the index was low, on the average about 1.750, with a 

 range upwards to about 2.000 and on the lower side to about 1.550. The same 

 elimination of color and pattern differences were practiced, so that at the end of 

 1906 I had a weU-defined set of four races of this material, which were utilized 

 in some tests, that are companion to those already presented. 



The original stocks were uniform in aspect and in behavior in crosses and 

 in every respect appeared as homogeneous racial line cultures, and all showed 

 essentially the same " amount " of " fluctuations " in each fraternity, as shown 

 in figure 33B, in the pairs a, h, c, and d; and in the following generation progeny 

 of matings from these same fraternities showed the expected range in the " fluc- 

 tuations." These were random matings within the population, reared under the 

 conditions of the general breeding quarters, without any control of the condi- 

 tions. I then reared them through four generations under the same conditions 

 as in the muUilineaia series, with the same general result in the two lines carried 

 from pairs a and c. This difference was found in that in the line c the modal 

 condition remained at the same point on the scale, while the mode of the a 

 series was constantly shifting its position. Neither of the lines showed the 

 slightest indication of restriction in its range as a result of the controlled and 

 more uniform conditions of development, so that the environmental influences 

 were apparently not operative in this set of " fluctuations " more than in the 

 first one. 



The next step in the testing was the attempt to breed minor races from the 

 lines, as was done in the first series. With the line c, three sets of matings were 

 made from three points in the array of the fraternity, as shown in figure 33B. 



