268 PRODUCTION OF RACES AND SPECIES IN LEPTINOTARSA. 



of such races we really have two forces — a species selective tendency and a 

 local (or artificial in experiment) — acting against one another, with the 

 result that selective divergence to a certain limit is attained, but beyond that 

 the racial divergence is slow or entirely stopped. When the local or artificial 

 selection is removed the species selective tendency causes a regression to the 

 type of the species. It may be objected that my experiments do not cover a 

 sufficiently long series of generations to have accomplished the result 

 intended, and this may be true ; but selection is a powerful formative factor 

 and works rapidly up to a certain limit, and this has been abundantly proven 

 by plant and animal breeding for fifty years. Why should it not also be 

 able to establish a race on permanent footing with the same rapidity? We 

 know from the rearing of domestic animals and plants that constant selec- 

 tion is necessary to maintain the race. In as far as these color characters 

 are concerned, by artificial selection we can easily produce and maintain a 

 race, but we can not establish it as an independent one; we can create 

 isolated races from extreme variations, and by selection keep them isolated, 

 but we can not permanently establish them. On the whole, selection would 

 appear as a relatively impotent factor in the evolution of these color charac- 

 ters were it not for the fact that we usually try in our experiments to work 

 against the far more powerful natural selective tendencies of the parent 

 species. Two points we may note in passing, namely, the number of highly 

 divergent variations beyond the normal range of fluctuating variation pro- 

 duced in this series of experiments, and the increased percentage of indi- 

 viduals which show variations capable of being transmitted to the progeny. 



Sei/ECTion Experiments with Structurai, Characters. 



It is often asserted that color characters are not reliable for the study of 

 variation and evolution. The assertion is, I think, due entirely to ignorance 

 of the real nature of color characters. I have made a series of selection 

 experiments with structural characters — general, such as size and shape; 

 and special, such as glands, punctation, spines, etc. 



There exists in this species a very considerable variation in size between the 

 sexes, and also in the same sex, and this provides a good general structural 

 character for the experimental study of artificial selection and selective proc- 

 esses. Four main sets of experiments were tried, each aiming to create by 

 selection a race having certain characters. Four conditions were aimed at: 

 (a) To have both sexes large; {h) both sexes small; {c) females large and 

 males small; {d) and males large and females small. Only the first two sets 

 gave results of any value, the last two failing because of the difficulty of 

 breeding opposite extremes. Therefore I give here the data and conclusion 

 from the first two series. The general results of these experiments are given 

 in plate 30, the procedure and results being in general not unlike the experi- 

 m.ents with the color characters. 



