June 19. 1916 Correlated Characters in Maize Breeding 447 



of segregation in the first generation, it should be recalled that neither 

 of these characters, which belong to the Waxy Chinese variety, is univer- 

 sally present in the plants of that variety, and the parent plant may 

 have been heterozygous. There is also a less pronounced indication in the 

 second-generation plants that one-sideness is Mendelian in its inheritance. 



CORRELATIONS 



Eleven of the characters most definitely contrasted in the parents were 

 selected and the correlation coefficients between all possible combina- 

 tions were calculated for both the first and second-generation plants. 

 The results are shown in Table IV. The correlations are so stated that a 

 positive, or plus, correlation indicates a correlation between the char- 

 acters derived from the same parent; in other words, a coherence. For 

 example, the Waxy Chinese variety has a large number of tassel branches 

 and no tuberculate hairs, while the Esperanza variety has a small number 

 of tassel branches and well-developed tuberculate hairs. In expressing 

 the relation between these two characters, when a large number of tassel 

 branches is found associated with short tuberculate hairs, the correlation 

 is recorded as positive. 



Since ears i and 2 were reciprocals and no significant differences were 

 found between their progenies, the observed values were used directly 

 in calculating the coefficients of correlation. Where the mean progeny 

 of ear 3 differed from the mean of the combined progenies of ears i and 2 

 with respect to any character, all measurements in the progeny of ear i 

 were multiplied by the percentage difference between the means before 

 combining the populations in a correlation table. 



The combined progenies of the three first-generation ears numbered 

 183 individuals. Complete notes could not be taken on all the plants, 

 so that the number of individuals entering into the different correlation 

 tables was reduced to from 125 to 150. Assuming all correlations that 

 are more than 3.5 times the probable error to be worthy of consideration, 

 an examination of Table III shows that 20 of the 55 character pairs 

 fall into this class. ^ With three exceptions the coefficient for the charac- 

 ter pairs of this group is 0.2 or larger. Of these 20 character pairs that 

 may be held to show definite correlations in the second generation, 17 

 are positive — that is, in the nature of coherences — and 3 are negative. 

 All but 5 of the 20 are, however, open to the suspicion of being physio- 

 logical correlations, since they do not differ materially from the corre- 

 lations shown for the same characters in the first generation. 



The 5 character pairs that show most evidence of genetic correlation 

 are given in Table V, Even here there are no very striking differences 

 between the coefficients of the first and second generations, and it is by 

 no means impossible that even here the differences may be due to chance. 



' These coe£5cients are printed in bold-face type in Table IV. 



