334 GENETICS IN RELATION TO AGRICULTURE 
entirely regular or continuous. The saltations or jumps revealed by 
Surface’s analysis were directly consequent upon lump elimination of 
a number of mediocre lines. These results, therefore, are in entire har- 
mony with the known nature of allogamous populations. This conclu- 
sion is further corroborated by the recent report of Reitz and Smith on 
the statistical study of indirect effects of selection for high and low pro- 
tein and oil. These authors state: 
“Tt is found that four distinct types of corn as regards length, circumfer- 
ence, weight of ears, and number of rows of kernels on ears are so well estab- 
lished that we may assign orders of values to the means of these characters that 
persist with but a few exceptions in such changes of environment as have been 
experienced in 11 years of planting, from 1905 to 1915. 
“While a few slight but progressive changes have been noted, the selections 
for chemical composition from 1905 to 1915 have not changed decidedly the dif- 
ferences in mean values of these characters. In fact, we are unable to assert with 
any high degree of probability that the strains differ more or less with respect to 
these characters during the second half of the period 1905 to 1915 than during the 
first half.” 
The italics are ours. It is of especial significance that careful biometrical 
study has failed to reveal any progressive change as a result of continued 
selection in these strains of corn. For the results of these experiments 
have been cited as evidence par excellence by Castle in support of his 
hypothesis of factor variability. 
The ear-to-row method has been modified in various ways but it 
still forms the basis of most systems of commercial corn breeding. A 
popular feature of systematic corn improvement is the use of score cards 
in judging. Aspecial development of the score card method of selection 
is the use of selection index numbers as advocated by Pearl and Surface. 
In this plan arbitrary values are assigned to various characters of the 
corn ear, for example, such as absolute size of the ear, average percentage 
depth of the grains, etc. The idea is to combine in a single numerical 
expression the values of a series of variable characters with regard to 
all of which the breeder wishes to practice selection at the same time. 
The index numbers of .different varieties are not directly comparable 
but for a given variety they may be useful as an fadjunct of the score 
card method. However, their use requires more attention to details 
and hence greater expense than most breeders can afford to give. Their 
use in plant breeding will probably be limited to experiment stations 
(see Chapter XX XI). 
The danger of continued ear-to-row selection or ‘‘narrow breeding”’ 
within a variety was pointed out in 1909 by Collins, who emphasized 
the importance of “broad breeding’’ in such crops as exhibit loss of 
vigor when closely inbred. About the same time Williams inaugurated 
Digitized by Microsoft® 
