No. 639] INHERITANCE OF BLUE IN POULTRY 313 



As may be seen by inspection of Tal)li' l»nt one maK' 

 (296M) gave a group of offspring (."J) which wa< \ciy 

 close to expectation. The chances that as ureal a (h \ ia- 

 tion as this one would appear as a icsult of landnm 

 sampling are four to one. The mothers of this uioum 

 were the product of a cross which was the iccipidcal of 

 that which produced their sire. The otT>priii.u' .1!' I'lMiM 

 w^hen mated with females whicli wci-c the ])ro(hict of t!ir 

 same cross as himself (groiij) 2) gax'c a (h'vialion so ureal 

 that the chances against its ai)peariiig as a result of 

 random sampling are three to one. The chances of the 

 appearance of deviations as great as those &ho\vii by the 

 offspring groups of the other males were as follows : 86E 

 (group 1) one chance in a little less than five : (UiM ( uroup 

 4) one chance in about eight; 65E (group o) on,, chance 

 in approximately five ; 46E (group (i) one chance in ahout 

 two and a quarter; 258M (group 7) once in alumt iwelvi' 

 times with females from the sanu> cross as he. and once in 

 twenty-ti\-e when mateil with females fi'oin a recipi'oeal 

 cross (gron). S). 



Tt would l.e nnnsnal, ihongh not inipo-sihh'. to have so 

 ]nany comparatively wide .leviations fnmi expectation 

 simply as a result ..f ramloni saniplinu. 



If the genetic constitution of tin. F, 's was as has been 

 previously postulated, and these were in fact all chance 

 deviations, it would ]>e highly pro1)ahle that the luini)ing 

 of all the data given in Table 1\' would approximate the 

 c*alculated ratio fairly closely. 



The lumped data are given at the bottom of Table IV. 

 It will at once be seen that the goodness of tit as meas- 

 ured by P is poorer than the poorest constituent uroup, 

 and w.iuld be ])robabl(., on the basis of ran.hnn sampling, 

 once in about twenty^.i-ht lime.. It seems fairlv clear 

 that some .listurbing force was .^''''-nt ive. 



themselves are linkage and a differential viability of 

 classes, or it might be a combination of the two. 



