TH) 



THE AMERICAN NATURALIST 



[Vol. XLIX 



do not show the full strength of the correlation, remove at once 

 all question concerning the relatively permanent differences in 

 productiveness of the individual hills. 



Consider next an illustration from hybridization of measur- 

 able characters. 



Goodspeed and Clauson 10 have given the mean values of meas- 

 urements of the flowers of individual plants of Nicotiana hybrids 

 cultivated in 1912 and of corollas of the same plants cut back 

 and flowered in 1913. The correlations between the mean di- 

 mensions for the two years I find to be : 



N. Tdbacum var. macrophylia $ X N. sylvestris <$ 

 F x plants, 2V=21. 



For spread of corolla, r = .044 ± .147. 

 For length of corolla, r = . 169 ±.143. 

 Hybrid produced by crossing F 1 of the hybrid N. Tdbacum 

 " Maryland $ by X. Tahacum q. with X. sylvestris, N = 19. 

 For spread of corolla, r = . 560 ±.106. 

 For length of corolla, r = .788 ± .059. 



These correlations show at once the high degree of uniformity 

 of the Fj of the first as compared with that of the second series. 

 In all four cases the signs of the coefficients are positive, but 

 those of the first class arc insignificant in comparison with their 

 probable errors. In both cases length of corolla is more closely 

 correlated than breadth. Possibly this is due to errors of sam- 

 pling only, or to greater difficulty in obtaining an exact measure 

 of the spread of the limb. It may, however, indicate that some 

 characters are more sharply and permanently differentiated 

 from individual to individual than others. 



That the latter may sometimes be the case is clearly shown by 

 unpublished data of my own for the ligneous perennials Sfi/ph- 

 ylea trifolia and Hibiscus Syriacus. 11 



io Goodspeed, J. H., and R. E. Clauson, • • Factors Influencing Flower Size 

 in Nicotiana with Special Eeferenee to Questions of Inheritance," Amer. 

 Jour. Bot., t, 232-274, 1915. 



of ovaries of shrubs well established in the Missouri Botanical Garden. In 

 such work the number of individuals can never for practical reasons be very 

 large, if a fairly large number of countings be made for each shrub. Fur- 

 thermore much of the work which one does may be lost by some accident 

 which precludes the securing of countings from each individual every year, 

 omitted entirely. P e e * d both of a P air of > ears * must be 



