CHAPTER XXVI 
HEREDITY IN PURE LINES 
H. H. NEwMAN 
In the last chapter we have seen how Galton formulated his law 
of filial regression, which means that average parents tend to produce 
average offspring, but that exceptional parents tend to produce off- 
spring less exceptional than themselves, but nevertheless more excep- 
tional than the average. . 
In studying this law the Danish botanist Johan&sen saw in it a 
possibility of racial improvement through the instrumentality of con- 
trolled selection. He thought that by continually selecting the most 
exceptional parents in each generation the degree of regression toward 
the average might be lessened until a pure non-regressing strain might 
be produced. 
In order to simplify the experiment and obviate the complexities 
inherent in intercrossing, he selected a self-fertilizing type, using the 
bean Phaseolus. Taking a set of nineteen beans from several plants, 
choosing the largest bean, the smallest bean, and seventeen inter- 
mediate types, he planted these, expecting to find that the bean pro- 
geny of the large bean would be mainly large, those of the small bean, 
small. In this he was disappointed, for the bean progeny of the large 
bean fluctuated about a mean, some being large, some small, but the 
majority average. Similarly the bean progeny of the smallest bean 
were of various types, mainly average. Each of these nineteen purelines 
had a mean of its own, and irrespective of which one was selected (the 
largest, the smallest, or the average) the mean and distribution of the 
progeny was practically the same. Thus it was concluded that within 
a pure line selection had no effect in modifying the character of size of 
beans. The reason for this rather unexpected result is not far to seek. 
Johangsen was selecting on the basis of somatic variations, the fluctuat- 
ing variations of Darwin, that are merely due to more or less favorable 
growth conditions and are not represented in the germ plasm, i.e., are 
not hereditary. Each germ cell in a pure line is supposed to have the 
same hereditary units, and, if that is so, selection would not be expected 
to effect any modification that would persist. 
376 
