THEORIES IN PRACTICE 349 



that "about one-third" of the second-generation 

 hybrids would revert toward one parent, while 

 another third would revert toward the other 

 parent, the remainder being intermediate in 

 character, and in this corresponding to the first- 

 generation hybrid that was their parent. 



This implies a fair understanding of the com- 

 bination of characters of the two parent species 

 in the first-generation hybrid, and the segrega- 

 tion and recombination of these characters in the 

 second-generation hybrid. It will be noted also 

 that the distribution of these characters in the 

 second generation (as predicted on the basis of 

 my observation of earlier seasons) was essentially 

 that which has come to be familiar everywhere 

 within recent years as the typical distribution of 

 characters among second-generation hybrids in 

 what is now known as Mendelian heredity. 



To be sure, the figures given are only approxi- 

 mate, nor have I in any of my experiments 

 endeavored to keep accurate account of the pre- 

 cise numbers, the large scale on which I operate 

 making this scarcely practicable — but the close 

 approximation of the rough estimate that I 

 made to the precise figures that have been deter- 

 mined by more recent investigations, sufficiently 

 attests the accuracy of the observations on which 

 the estimate was based. • 



