296 THE THEORY OF THE GENE 



must, under these conditions, do so by changing the gene 

 itself. 



The beans produced by each plant are somewhat vari- 

 able in size, and when arranged according to sizes they 

 give the normal curve of probability. All the beans from 

 any one plant and all of the descendents of this plant 

 have the same distribution (Fig. 150), no matter whether 

 large beans are continually selected, or small beans are 

 picked out in each generation. The offspring always give 

 the same groups of beans. 



Johannsen detected nine races of beans in those he 

 examined. He interpreted his results to mean that the 

 differences in size of the beans from a given plant are 

 due to its environment in the widest sense. It was pos- 

 sible to demonstrate this with material in which the mem- 

 bers of each pair of genes were identical when selection 

 began. Selection is shown to have no effect in changing 

 the genes themselves. 



When sexually reproducing animals or plants are 

 selected that are not homozygous at the start, the imme- 

 diate outcome is different. There are numerous experi- 

 ments showing what happens, such as Cuenot's results 

 with spotted mice, or MacDowell's results with ear-length 

 in rabbits, or East and Hayes' experiments with corn. 

 Any of these might serve as an example of what takes 

 place under selection. One example will suffice. 



Castle studied the effects of selection of the color- 

 pattern of a race of hooded rats (Fig. 151). Starting with 

 the offspring of commercial animals, he selected in one 

 direction those rats that had the broadest stripes, and 

 in the other direction the rats that had the narrowest 

 stripes, keeping these two lines apart. In the course of a 

 few generations the two populations became measurably 

 different — in one the dorsal stripe was broader, on the 

 average, than in the original group of rats ; in the other, 



