270 PRODUCTION OF RACES AND SPECIES IN LKPTINOTARSA. 



the greatest numerical strength and extensive distribution are provided with 

 numerous protective glands. Moreover, we know from the experiments that 

 the parents were normal, and that the selected parents possessed the varia- 

 tions in transmissible form. In the culture I acted as the insectivorous 

 enemy, eliminating all but those with the most glands, which were allowed to 

 breed. In this experiment were combined the conditions ideal for selective 

 modification, but the results in no wise differed from those obtained from 

 experiments with characters of no utility to the species. The similarity 

 between the results obtained in characters of utility and characters of no 

 utility is most interesting, and admits of only one conclusion — that as far as 

 the experiments with these beetles go local or artificial selection is a weak 

 factor, able to produce little or no real change in the species, and unable ta 

 create a permanent race by the preservation of extreme variations. 



Sei^ection Experiments with Physioi^ggicai, Characters. 



During the years 1899- 1904 various experim.ental pedigree cultures were 

 made in which physiological characters, such as habits and instinctive 

 responses, were chosen for experimentation. These characters were found ta 

 be only in the slightest degree influenced by selection, and the changes found 

 were too small to be of any real value or afford a basis for conclusions. The 

 failure to get positive results was not due to the non-heritable qualities of the 

 characters, but to the failure to get hold of sufficiently extreme variations to 

 form a basis for selective experiments. All of the variations used were so 

 near the mode that they afforded no real basis for work, and I was not able 

 by selection to increase such variations as were found. 



It has been shown in respect to these physiological characters that they 

 were exceedingly conservative. Thus, in the various habits associated with 

 reproduction or hibernation, only the modal individuals are able to go- 

 through the habit with success, and a variation of any moment is sufficient to 

 exclude the possessor thereof from participation in the propagation of the 

 race. As a matter of fact, variations of these characters are rare, and the 

 species are separated by wide divergences, far larger than those existing in 

 the case of morphological characters. 



When we bring together the results of the selection experiments upon 

 L. decemlineata several points of general interest are brought to light. First 

 and of fundamental importance is the fact that all of the heritable variations 

 are discontinuous in any given generation, and that continuity exists only in 

 individual lines of descent where discontinuity is impossible. The second 

 point of interest is the failure to carry, by selective processes, races beyond 

 the normal range of variation of the species. The third is the sudden devel- 

 opment of extreme variations differing in many points from the parents, and 

 lying beyond the range of fluctuating variations. These were found with 

 especial frequency in the series of experiments with general color tendencies. 



