432 TRANSMISSION 



The breeder may choose either method, but having once 

 chosen, all his records must be made in the same way. In- 

 asmuch as the standard deviation has distinct mathematical 

 advantages over the simple average deviation, and inasmuch as 

 it is the one commonly employed in mathematical literature, it 

 is the one that will always be employed in this text. When, 

 therefore, a measure of variability is mentioned it will be the 

 standard deviation and not the average deviation that is meant. 1 



Meaning of standard deviation. The standard deviation is a 

 good measure of variability for the character in question and 

 the race involved. It therefore affords a reliable basis for com- 

 paring the variability of one race with that of another as to the 

 character under consideration, or of one character with that of 

 another either in the same or in different races. 



In ascertaining the standard deviation the mean, or average, 

 of the race was taken as the basis, and all deviation was reckoned 

 from that. It is manifest, however, that if the mode be taken 

 as a basis, and deviation reckoned from this, following the same 

 methods as for determining the standard deviation, a somewhat 

 different value will result. Whenever, as in most cases, the 

 mode does not coincide with the mean, this will represent the 

 deviation from the prevailing type, which is often of more prac- 

 tical importance to the breeder than the deviation just described, 

 especially to the breeder who proposes to deal with individuals 

 selected with reference not to the mean but to the prevailing 

 type. 



In the same manner the breeder may calculate the deviation 

 from his own ideal type or standard and in this way assess the 

 deviation, not from the present mean or type but from the one 

 he hopes to establish. In this way he is able to keep accurately 

 informed from year to year as to the progress he is making and 

 the degree of success that is following his selections. 



The writer has never seen either of these determinations 

 used, but they are especially valuable to the breeder who desires 

 to know how a certain variety deviates on the average from its 



1 In the section on " Probability Curve " (Appendix) it will be shown that the 

 standard deviation is on the whole preferable from the purely mathematical 

 standpoint. 



