138 A MANUAL OF MENDELISM 



was written. The accompanying diagram indicates 

 that Hermod is even a better sire than either Taurus IV 

 or Dan, for he gets no daughters below the 3'8 grade. 

 He was mated with better dams, of course, many of 

 which were Dan's daughters, but very few could have 

 been pure even for the factors for 3*6. The average of 

 the dams with which Hermod was mated was 3*80, 

 while the average of his daughters was 4'21. 



After Hermod came Skjold, but, as yet, only eight of 

 his daughters have long enough records to be relied 

 upon with confidence. Since these confirm the previous 

 results they may be set down, however. It will be seen 

 that Skjold is another sire whose daughters do not fall 

 below the 3*8 grade, although they do not rise to the 

 grade to which Hermod's daughters rose. 



Thus we have five sires, one of which threw daughters 

 which on the average were no better while four threw 

 daughters which were better than their dams. One 

 feature is common to all five cases, namely, that as 

 the parents are better the progeny are better ; and not 

 only so, but, as either parent rises in the scale of quality, 

 the chances of their producing progeny higher in the 

 scale increase while their chances of producing lower 

 progeny decrease. Thus we have a phenomenon similar 

 to that which was found occurring in Professor Nilsson- 

 Ehle's wheats, and it may be argued again that im- 

 provement is brought about by combining in certain 

 individuals the different factors for high quality found 

 in others and so expelling the factors for low quality. 

 Whether the factors are related to each other as domi- 

 nants and recessives or whether they produce inter- 

 mediates is not disclosed. Most probably there are 

 both kinds of pairs, but the fact that the four sires 



