388 



THE AMERICAN NATURALIST [Vol. LIII 



From the data presented in Table II it appears that, on 

 the average, the birds that manifested the higher per- 

 centages of increase in the weight of the September eggs 

 were characterized by higher annual production. Those 

 showing a weight-increase of 10 per cent, or more gave a 

 mean annual production of 143 eggs, while those birds 

 characterized by a decrease in mean egg-weight showed 

 an annual production of only 108 eggs. When the flock 

 was divided into two groups according as the egg-weight 

 had increased by more than 6 per cent, or less, the high- 

 percentage group gave a production of 141 eggs as op- 

 posed to 111 eggs laid by the low- percentage group. Thus, 

 dividing the flock on the basis of a 6 per cent, increase in 

 the mean weight of all the September eggs, gave a group 

 of 12 hens (out of 33) which showed a mean production 

 17.5 per cent, higher than the flock average (120), about 

 27 per cent, higher than the mean production of the low- 

 percentage group, and 30 per cent, higher than the mean 

 production of the small group of eight hens which mani- 

 fested a decrease in mean egg-weight at the period con- 

 sidered. 



It will hardly be necessary to call the attention of the 

 reader to the circumstance that this method of demon- 

 strating the correlation involved in the frequency distri- 

 bution of these two variables (increase in egg-weight and 

 numerical production) is, by its very nature, such as to 

 constitute a practical application of the means involved. 



The correlations between weight-increase and produc- 

 tion, considered in the foregoing paragraphs, were so 

 obvious that the question arose as to whether satisfactory 

 correlations could not be demonstrated between these two 

 variables under conditions in which a smaller amount of 

 statistical data was employed. For instance, if the 

 method should prove of value to poultrymen in affording 

 a means for the detection of the higher producers of the 

 flock, it would be desirable to reduce the machinery of 

 computation to the lowest point consistent with valid 

 results. ' It thus appeared pertinent to inquire whether 

 computations basQd upon the weight of only ten eggs, laid 

 as closely as possible to the periods of the absolute vernal 



