SELECTION EXPERIMENTS. 269 



I started in these experiments with 10 pairs of modal beetles that had hiber- 

 nated, and bred them, getting a second generation, normal in every respect, 

 and from these selected 10 modal pairs and obtained a third normal genera- 

 tion of 1,288 beetles. From this third generation, out of no pairs 2 were 

 found that had variations capable of transmission, giving two divergent 

 groups of beetles in the fourth generation. One of these consisted of beetles 

 larger on the average than the parents and the other beetles smaller than the 

 parents. From the fifth to the tenth generations I practiced constant rigorous 

 selection, and in the eighth there was a relaxation in the selection of part of 

 the large beetles, with the general result shown in plate 30. It is possible 

 to increase and decrease the size, but the mode of any one generation could not 

 be brought to the limits of the variability of the species. This result, with the 

 general structural character, size, differs in no way from that obtained with 

 color characters, nor from the common experience of breeders of domestic 

 animals and plants. The breeder of large horses has not yet by selection 

 been able to create dray horses as large as elephants, nor is it possible to 

 double the size of these beetles by the same process. The change produced 

 in size was, if selection was removed, as shown in plate 30, immediately oblit- 

 erated through regression to the standard of the race. 



An interesting set of experiments with the scent glands was made. I have 

 shown how these glands act as protective structures, and how distasteful their 

 secretions are to insectivorous birds, reptiles, and amphibia. They are there- 

 fore highly useful to the beetles, serving as active protection, hence of real 

 utility, and they ought by selection to be capable of increase in number and 

 in quantity of secretion, a result that could not be otherwise than a decided 

 advantage to the race, since it is shown that the species most numerous, 

 widely distributed, and immune from attack are those with the largest number 

 of these glands. 



All that the experiments with the glands show is that, given a variation 

 capable of being inherited, with it as a basis we can by selection create a race 

 that possesses as its modal condition very nearly the extreme variations in the 

 characters chosen for experiment ; but the race persists only as long as the 

 stimulus of selection is continued, and I do not see how this process could lead 

 to the establishment of independent races or species. 



While the selection experiments with structural characters show that the 

 same results are obtained as with color characters, they are strong evidence 

 against the assumed ease with which useful characters can be augmented and 

 preserved by selection. They, however, show clearly that in these beetles 

 even a most useful protective device is not easily modified in a permanent 

 manner by artificial selective processes. In this series of cultures we are not 

 dealing with uncertain and unknown quantities. We- know the direct and 

 important use to which these glands are put by the beetles ; we know that in 

 the genus some beetles possess more, others less ; and that those species with 



