No. 613] EXPERIMENTS CONCERNING EVOLUTION 35 



of change in dominance during development.) (2) The 

 same thing happens in certain F 2 crosses, reversing com- 

 pletely during larval life the original ratio of lights and 

 darks. (3) The different races involved are characterized 

 by a difference in the speed of differentiation, as shown in 

 the actual curves. This velocity is also caused by genetic 

 factors. Where these recombine with the pigmentation 

 factors, the entire situation of the F 2 curve is shifted 

 (without changing the 3 : 1 ratio), showing that the visible 

 effect of the pigmentation factors is bound to a certain 

 velocity of differentiation. (4) The shifting of the type 

 of pigmentation from light to dark during larval life of 

 certain races or hybrids is a process which progresses 

 constantly with time. This is seen when isolated indi- 

 viduals are studied which belong to races that differ in 

 regard to the number of moults and exhibit the shifting 

 simultaneously. There are races where all the male cater- 

 pillars have four moults and the females either four or 

 five ; other races where the males have four, the females 

 five; others where both sexes have five moults ; and in the 

 last case even a sixth moult occasionally occurs. In these 

 cases we see that every new moult produces a further 

 shift to the dark side of the curve, showing that the class 

 of pigmentation to which a full-grown caterpillar belongs 

 is in this case a function of the time of differentiation. 

 The same fact can be demonstrated in a shifting race by 

 prolonging the time between two moults by starvation 

 (which succeeds only to a certain extent). In experi- 

 ments of this sort it has been possible to get the shifted 

 type of pigmentation, characteristic of the fourth stage r 

 in some individuals in the third stage. Table VI also 

 contains a few random data on the first point. (6) In 

 shifting hybrid cultures there appear comparatively 

 often mosaic-caterpillars, showing different classes of 

 marking on right and left sides. The distance between 

 these two different classes is approximately kept up when 

 shifting occurs during development. The following ex- 

 ample demonstrates this fact: 



