No. 591] SHORTER ARTICLES AND DISCUSSION 



181 



tions had occurred, this material would show greater variability 

 than the ancestral race before inbreeding began. This, I take 

 it, is the point which Walton has in mind when he asserts that 

 inbreeding has a tendency to increase variability. But this is 

 very different from the condition to be expected in any single 

 line considered separately, as in one of our inbred lines of Droso- 

 phila. Such a line should be less variable than the population 

 from which it arose, provided that population contained any 

 genotypic variations whatever! 



The question is decidedly worthy of consideration, which Wal- 

 ton's paper suggests, is evolution more rapid in a self -fertiliz- 

 ing or habitually close fertilizing population on one hand or in 

 a habitually cross fertilizing population on the other hand. 

 The importance of the question is not lessened by the fact that 

 Walton has brought into the discussion material wholly irrele- 

 vant, including his own observations on the zygospores of Spiro- 

 gyra and the observations of Jennings on Paramecium and Bar- 

 rows on Drosophila. But the work of Hayes on the variability of 

 pure races of tobacco and of their hybrids, which Walton cites, 

 does bear directly on this question. By combining the observa- 

 tions on the parent races into one mass of data and treating 

 this statistically, Walton has shown that self-fertilizing lines 

 mixed together would form a population more variable as re- 

 gards number of leaves and height of plant than the popula- 

 tion produced by cross-breeding of these same lines. Hayes's 

 observations verify Foeke's law already cited, that the variabil- 

 ity of F 1 does not exceed that of the more variable parent race, 

 but that F 2 shows increased variability. Theoretically F 2 should 

 show the maximum variability. Walton's figures indicate 

 clearly that this maximum variability under cross-breeding is 

 less than the variability of a mixture of the two inbred races 

 and consequently that continuous self fertilization within a 

 mixed population will produce a more variable population than 

 will result from continuous cross fertilization. This is an im- 

 portant generalization which demonstrably will hold good in 

 all cases in which "intermediate" or "blending" inheritance 

 occurs. It would not hold good for cases in which completely 

 dominant and quantitatively invariable Mendelian factors are 

 concerned, but it is doubtful whether such eases occur, as I have 

 elsewhere tried to show. It is the great variability of self fer- 

 tilizing populations and the stability of variations arising under 

 self fertilization (since no variations will be "swamped by cross- 



