i9o6] FEMALE LINE IN POLAND SOWS. 247 



Date progeny was farrowed. Number of pigs in litter; boars raised, sows 

 raised; total raised. 



Volume and page reference. 



" The above litter would therefore appear on its card as follows : 



Gardner's Model, 122578. 



March 28. 1902. 7 — o 



26/282 



" After all the cards for a volume were written, two clerks compared 

 them with the originals and corrected errors in copying, one reading the 

 original records while the other corrected the copies. After all the cards 

 for a breed were written and verified they were sorted according to years ; 

 next the cards for each year were sorted according to size of litter farrowed ; 

 next the cards for each size of littler were arranged numerically according 

 to the numbers of dams and duplicates thrown out, and finally the cards 

 were counted and averages calculated. Cards were counted at least twice 

 by different persons, and the calculations were made twice by different per- 

 sons. In writing the cards for the ' American Poland China Record ' from 

 1898 to 1902 the page reference was not included, which we now know was 

 a mistake, as it is almost impossible to find a litter record without a page 

 reference. The cards for the ' American Poland China Record ' were arranged 

 alphabetically at first, and it was necessary to rearrange them numerically, 

 which eliminated a large number of duplicates." 



*' The reader will recognize the fact that the probability of error in this 

 work is large. Breeders are not always careful in reporting the number of 

 pigs farrowed, often relying on the memory. In spite of the painstaking 

 care of secretaries, errors in copying and typographical errors are sure to 

 occur, and in our own work errors in copying, indexing, and counting were 

 probable, although the cards were handled many times. At the same time, 

 although the probability of error is large, the probability of these errors 

 affecting the final results is very small on account of the large numbers used. 

 A mistake of one hundred litters for one of the later years of the ' American 

 Poland China Record ' would affect the average for that year slightly, but 

 would have no serious effect on the five-year average. The writer believes 

 that the factor of error in these calculations has been reduced below the 

 point of practical importance." 



The litters for 1902 of the " American Poland China Record " 

 were chosen for this additional work on inheritance. In addition 

 to the information as described above, the mother of the dam 

 was looked up and the size of litter in which she was far- 



PROC. AMER. PHIL. SOC, XLV. 183P, PRINTED JANUARY I4, I907. 



