430 Journal of Agricultural Research voI.v.no. lo 



tion might prove a less valuable and reliable indicator. This may 

 possibly be so, though up to the present time no definite evidence on the 

 point has appeared. Another point which occurs to one is that possibly 

 a better measure of the winter cycle of productivity (this being the bio- 

 logical entity we attempt to measure by the record of production to March 

 i) might be obtained by using the egg production of a bird up to the time 

 when it has attained a definite age. Fowls are hatched at different dates, 

 while March i is a fixed point in time. Birds hatched at different times 

 will be of different ages at March i of their pullet year. Will the ^gg 

 production prior to the attainment of a definite age by a bird give a better 

 measure of her winter cycle than the production prior to a fixed date mth- 

 out regard to age, except so far as this is involved in having the birds all 

 hatched within a certain limited season ? It is the purpose of this paper 

 to present some data on this question. 



Specifically the material here presented has to do with the suggestion 

 that the egg production up to 300 days of age of the bird gives a better 

 measure of the winter cycle than does the production to March i , since an 

 age of 300 days will include the winter cycle, and will also allow for differ- 

 ences due to variation in date of hatching. Biometrically we can readily 

 test this question in two ways: On the one hand, we can determine the 

 correlation between the winter production as defined by the writer (to 

 March i) and the production to 300 days of age, on the other hand. If 

 this correlation is low, it will mean that one of the measures is probably 

 sensibly better than the other. If, on the other hand, the correlation is 

 very high, differing but little from perfect correlation, it will indicate, 

 so far as it goes, that there is little to choose between the two measures. 

 In the second place, we may examine the variabilities biometrically. 

 On theoretical grounds that measure of a character is best, other things 

 being equal, which exhibits the smallest relative variability. 



Evidence along these lines derived from extensive trap-nesting experi- 

 ments is presented in the following tables. The data cover three con- 

 secutive years. Two correlation tables are presented for each year: One 

 including the total flock of that year regardless of breed distinctions, the 

 other including only pure Barred Plymouth Rocks. The total flocks 

 were made up of various crossbred birds used in Mendelian experiments, 

 in addition to the pure Barred Plymouth Rocks. All birds included in 

 the tables are pullets — i. e., they were hatched in the spring of the year 

 indicated in the caption of the table. The computations were made 

 by Mr. John Rice Miner, staff computer of the Biological Laboratory. 

 See Tables I to VI. 



