G ALTON AND STATISTICAL STUDY OF INHERITANCE 1 77 



from ears with many rows, and that the greater the number of 

 rows on the ear from which the seed is taken, the smaller is the 

 number of ears produced with a small number of rows. It is also 

 plain that, as the number of rows on the ear from which the seed 

 was taken increases, the number of ears produced with a large 

 number of rows increases, and that we have in each case a very 

 considerable number of ears which equal their parents and a few 

 which excel them, even when the parent seeds are far beyond the 

 maximum for all ordinary corn. Fritz Miiller says he has never 

 under ordinary conditions, except in three instances, found an ear 

 with more than i8 rows, and Darwin puts the maximum at 20 

 rows ; yet we have among the children of seed from a 22-rowed 

 ear no less than 4.8 per cent, or 18 ears out of 373 with 20 

 rows, and one ear out of 373 with 26 rows, and it will also be 

 seen that the number of children which equalled their parents 

 increases in each case in each successive generation. 



Thus the seed planted in 1867 from an i8-rowed ear produced 

 12.6 per cent of i8-rowed children. The i8-rowed ear planted 

 in 1868 from an i8-rowed parent produced 18.2 per cent of 18- 

 rowed children, and the i8-rowed seed planted in 1869 from 18- 

 rowed parents and grandparents produced 18.6 per cent of 

 i8-rowed children. The series is 12.6 per cent, 18.2 per cent, 

 and 18.6 per cent. The rapid change which took place in the 

 "type" after only three years of selection is well shown by the 

 following table, which gives the dominant number of ears at each 

 sowing and also the percentage of ears which had this number : — 



The minimum for the third generation is equal to the mean 

 for the first; the mean for the third generation, 16 rows, is very 

 near the maximum for ordinary corn, and the maximum for the 

 third generation is far beyond the maximum for the grandparents, 

 and much beyond the maximum for the parents. 



No one can dispute the well-known fact that this sort of pedi- 

 gree selection for a single point quickly grows less and less effec- 



