.58 TECHNICAI, BULLETIN 7 



in this manner to measure directly the effect of broodiness on feciuidit\\ This 

 intense broody strain differs from the foundation birds more in tlie distribu- 

 tion of broody periods throughout the laying year thnn in the number of 

 broody periods. 



Complete records of broodiness are also maintained on every female of tlie 

 experimental flock to augment data collected in the broody ex})erin>ent. 



End to Be Attained 

 A flock of poultry breeding true for broodiness and non-broudiness. 



Scope of This Report. 



In this bulletin consideration is given to the actual relationsiiip between 

 pullet-year egg production and the broody trait as manifested during the first 

 laying year. Coefficients of correlation have been calculated as follows: 



Between broodiness and rate. 



December rate — Sections 1, 2, 3, 4, 5, \G, 17. 

 Winter rate— Sections 6, 7, 8, 9, 10, 18, 1!). 

 Annual rate— Sections 11, 12, 13, 14, 15, 20, 21. 



Betiveen tunes broody and len<ith of broody periods. 

 Section 22. 



Between zvinter rale ami annual rate. 

 Section 23. 



Betzceen zcinter rate and annua! eg(] yield. 

 Section 27. 



Betzveen annual rate and annual egg yield. 

 Section 28. 



Betzveen broodiness and egg yield. 



Winter production — Sections 24, 25, 26. 



Annual production— Sections 29, 30, 31, 32, 33, 34, 35. 



Coefficient of Courei.ation. 



The coefficient of correlation furnishes a concrete measure of tlie tendency 

 of two characteristics to move together, to move in opposite directions, or to 

 behave independently. In this particular study tlie characteristics studied 

 botli belong to the same individual fowl. Either a significant positive or 

 negative correlation coefficient is useful to the breeder as a guide, and the 

 magnitude of the coefficient shows him tiie relative amount of dependence 

 between the traits or characters considered. The value of a coefficient of 

 correlation from the biological standpoint depends upon its absolute magni- 

 tude and upon its relation to its probable error. A coefficient at least tiiree 

 times as great as its probalile error is generally considered significant, e\en 

 though its absolute magnitude is small. The deductions reported in this 

 bulletin are based on the above conception. King (1923), however, states 

 that the correlation coefficient should be more than six times its proIiaJile 

 error. He further states that a correlation coefficient of less than .30 indicates 

 a lack of marked correlation, that over .50 shows decided correlation. Furtlier- 

 more, the correlation coefficient with its regression coefficients may be used 

 for purposes of prediction. The value of a knowledge of the degree of correla- 



