2 HEREDITY IN RABBITS, RATS, AND MICE. 



carried through 4 additional generations of selection. (See tables 

 5 to 8.) But this race has shown even poorer vitality than the plus- 

 selection race, so that in order to keep it alive practically no young 

 could be rejected as parents and consequently no further progress 

 has been made. The numbers of young recorded for the four addi- 

 tional generations have been 330, 130, 79, and 35 respectively, and 

 their mean grades -2.84, -2.89, -2.78, and -2.74. The race is 

 now practically stationary in grade, but seems likely soon to become 

 extinct despite our strongest efforts to keep it alive. Notwithstanding 

 the fact that the race is verging on extinction after long-continued 

 close breeding, the variability of the hooded character is still as great 

 as ever. The standard deviation ranges from 0.25 to 0.45 within 

 about the same limits as in the previous 17 generations from minus- 

 selected parents. In the plus-selection series the standard deviation 

 was also fully as high in the last generations as it had been in the 

 previous 10 generations. Only the initial 7 generations had shown 

 an appreciably higher variability. 



SELECTED RACES CROSSED WITH WILD. 



We may now inquire what happens to the races modified by selec- 

 tion in opposite directions, when they are crossed with an unselected, 

 non-hooded race, the wild race. This question was considered in 

 some detail, so far as the plus-selected race is concerned, in a previous 

 publication (1916), where it was shown that a cross of the plus-selected 

 race reduced the grade of the hooded character, undoing in a measure 

 the work of selection. Selected animals which mated with their like 

 should have produced young of about mean grade 3.75, actually pro- 

 duced hooded grandchildren, extracted in F2 from a wild cross, of 

 mean grade 3.17, a falling off in grade of over 0.50. A second cross 

 with wild showed no further falling off, but instead a movement in 

 the reverse direction to 3.34 (a movement probably not significant, 

 in the light of further experiments). A third cross with wild has 

 been made on a small scale; 19 hooded grandchildren extracted in F2 

 from this third cross have a mean grade of 3.04. (See table 9.) It 

 seems probable, therefore, in the case of the plus-selected hooded char- 

 acter, that the maximum efTect exerted by the residual heredity of the 

 wild race is to reduce the hooded character in grade by about three- 

 fourths of a grade. Selection in 10 previous generations had elevated 

 the grade of the hooded character by about If grades. A cross with 

 wild eliminated less than half of this change. The remaining change 

 must be ascribed to changes effected in the course of selection. 



The minus-selected hooded race has also been crossed with this 

 same wild race. Originally of grade -1, it had been altered by 15 

 generations of selection to the extent of about 1|- grades, to mean 

 grade -2.54. Females of generations 15 to 16 (table 10) and of 



