112 THE SHORTER SCIENTIFIC PAPERS 



both for the number of leaves and the height of the plant, 

 and it may be noted that the coefficient of variation has 

 dropped from 19.5 5 to 9.40 in the one case and from 17.3 5 

 to 13.60 in the other case. Thus variability as measured 

 statistically has decreased. Those who have advocated 

 an increased variability as the result of hybridization are 

 correct when comparison is made of the F2 generation 

 with the Fi generation or with a single parental genera- 

 tion. They are not correct, however, in making a general 

 statement that cross breeding increases variability since 

 the variability of the group composed of both parental 

 types must be considered and upon so doing, it may 

 normally be found that there has actually been a decrease 

 in variability. 



The possibility exists however that the variability will 

 appear to have been increased when forms having the 

 same phenotype but different genotypes are bred together. 

 Such a condition may be illustrated by the two strains of 

 white sweet peas crossed by Bateson which produced 

 purple flowers in the first (Fi) generation, and purple, 

 pink, mixed and white flowers in the second (F2) genera- 

 tion. New combinations had arisen^ but only as an ex- 

 pression of that which already existed in the phenotypes, 

 for there is no evidence of an increase in unit characters 

 nor was there an actual increase in variability. 



There are only three papers of a statistical nature in 

 which it has seriously been asserted that cross-bred forms 

 or conjugating forms produced greater variability than 

 resulted in close-bred forms or non-conjugating forms. 



The first is that of Castle, Carpenter, Clark, Mast and 

 Barrows ('06) based on a series of observations as to the 

 effect of cross breeding and close breeding on the varia- 

 bility and fertility of the small fruit fly Drosophila 

 ampelophila Loew. In conclusion it was stated that 

 "inbreeding did not affect the variability in the number 

 of teeth of the sex comb of the male, nor the variability 

 in size," the first opinion resulting from the value of the 

 coefficient of variation in the number of tibial spines, the 

 second from the standard deviation in the length of the 

 tibia. In the former case the data certainly do not 



