ANIMAL AND PLANT BREEDING 313 



practised by breeding only from hens which laid at least 160 eggs in 

 their first year and from cocks whose mothers had laid at least 200 eggs 

 in their first year. Egg-production sank steadily. From 1908 onwards, 

 the selected hens were those which had already had high-producing 

 daughters, while the cocks were sons of hens which had had high- 

 producing daughters. By this system of using breeding stock whose 

 usefulness had already been tested, egg production in the selected flock 

 went up steadily, at least till 1920. 



The great difficulty in the use of progeny performance tests is, of 

 course, the time they take; from the time of the male's maturity, a 

 sufficient period must elapse for several of his offspring to be born and 

 grow old enough for their characteristics to be tested. The test there- 

 fore requires that a considerable number of males shall be kept for 

 rather a long period while they are being tested and their breeding life 

 will be correspondingly shortened. The testing of dairy bulls would not 

 be complete till they were about nine years old, by which time at least 

 three-quarters of them would have been eliminated by death or for 

 reasons of selection. The progeny test, therefore, although much more 

 accurate than the pedigree test, does not in practice supplant it. Efforts 

 have been made to work out, from an animal's pedigree, an index 

 based not on the actual characters of the ancestors but on their progeny 

 performances. These indices, like all those based on pedigrees, can 

 only have a statistical justification, since they can take no cognizance 

 of segregation of important factors. But for characters such as milk 

 production, which are obviously influenced by many genes of more or 

 less equal importance, the basis for a statistical treatment is present and 

 the indices may be useful when large numbers of animals are bred 

 from, though unreliable in particular cases. 



3. Selection of Both Parents 



Seleaion, either by phenotypic characters, pedigree or progeny 

 performance tests, may be applied to both parents. This is common 

 practice in animal breeding, where matings are usually controlled in 

 any case, but is less common in plants where controlled mating requires 

 the rather laborious process of artificial pollination. A classical example 

 of this type of selection is the investigation by Castle^ of the hooded 

 pattern in rats, in which two lines, one of increasing and one of de- 

 creasing pigmentation, were built up. 



Castle originally suggested that the selection had actually altered the 



^ Castle 1916. 



