264 PRODUCTION OF RACES AND SPECIES IN LEPTINOTARSA. 



ined singly, they are unmistakably seen to be deviations of greater or less size 

 from the mode of the species. 



I have shown that these spots behave as units in the ontogeny and evolu- 

 tion of coloration. In these selection experiments, also, they behave as units. 

 If we combine large numbers of cases they follow Quetelet's law in the 

 polygon of distribution ; but this in no wise presents the real fact, namely, 

 that each heritable variation is a divergence from the species mode in some 

 individual line of descent. 1 The distribution, in the long run, according to 

 the law of error, of these diverging heritable variations is not without interest 

 or importance, indicating, as it does, that the causes productive of the trans- 

 missible variations also diverge from their standard, now more, now less, and 

 in different directions; but whether there is continuity or not can not be 

 determined until we know the causes. In any event, the distribution of the 

 causes of heritable variations according to the law of error results in the 

 long run in producing heritable variations in dccemlineata that are also dis- 

 tributed according to the law of error, so that the species is thereby kept upon 

 an even basis. 



The behavior of general color characters, such as melanism, albinism, 

 xanthism, and rufism has been studied, especially with the object of creating 

 races by the selection of variations through successive generations. Out of 

 the many sets of experiments tried, two will serve to illustrate the general 

 results obtained. I shall describe experiments for the production of albinism 

 and rufism. 



I attempted by selection to create an albinic race of dccemlineata in two 

 ways first, by selecting for breeding the most extreme albinic variations 

 found in nature, and, second, by creating extreme albinic conditions in experi- 

 ment and breeding from them. For the first set of experiments the selec- 

 tion was made from numbers of copulating pairs found in nature. The 

 selected pairs were kept in separate cages, as were their progeny, the only 

 lumping of material being in the statistical treatment of it. The great major- 

 ity of such pairs and their offspring were not of any interest. Out of 311 

 pairs selected and mated in the years 1 896-1904, only 26, or 8J per cent of the 

 total number of pairs tried, were found capable of transmitting their partic- 

 ular variations. In many of these pairs it was certain that only one of the 

 beetles had the character in transmissible form, so that in 311 pairs, or 622 



1 That is, between any given variation and the species mode the intermediate stages 

 are a greater or less distance back in the line of individual descent. Moreover, in the 

 lines of individual descent intermediate stages are always discernible, no matter how 

 suddenly the variation may seem to appear. In all of the cases that I have studied there 

 are no jumps or discontinuity of any kind in the heritable variations, but there always 

 exist transition stages, passed frequently in ontogeny, but nevertheless real and directly 

 observable, between the characters of parent and offspring. There is then no saltation 

 or "mutation," either large or small, in Lcptinotarsa, nor are there "immutable" unit 

 characters. 



