100 
POUL TRY SCIENCE 
cuss the relative magnitude of the correlations for the individual 
months in greater detail since constants for other breeds are be¬ 
ing made ready for comparison, and all may be considered at the 
same time. 
We mav in passing note from table 3 that the coefficients for 
November to February inclusive, are larger in the first year than 
in the second year, while those for March to September are 
larger in the second year. A number of these differences may 
be considered statistically significant in comparison with their 
Diagram /. Magnitudes of correlations between the egg record of the 
individual months and the egg record of the whole year (solid dots) and of 
the correlations between the egg record of the individual months and the 
total egg production of the remaining eleven months of the year (circles). 
The shaded area indicates the magnitude of the difference between these 
values. 
probable errors. The difference for October is a slight exception 
to the other autumn months. 
The average values of the 12 coefficients are 
For first year,_54 1 5 
For second year_-5 2 9 2 
These show that the coefficients are, roughly speaking, equally 
large in the two years. Prediction of annual production from 
the record of any month should, therefore, be possible with a 
considerable degree of accuracy, providing that regression be 
linear. 
Graphic tests for the linearity of the regression equation relat- 
