ix HEREDITY 159 



these angles, and whether a wide L s is necessarily linked 

 with a narrow Lt as in the Egyptian parent. The result 

 of dissecting our curves in this way is very definite ; there 

 is no such linking. Thus, taking the 26 plants which had 

 the narrowest Lt in an F 2 of 180 individuals, we find that 

 while 24 had very wide Ls, there were two with a very 

 narrow L s, far removed from the rest. Conversely, taking 

 out the 26 plants with the widest L t, we find from one to 

 five possessing very wide L s, while the rest had narrow L s. 

 Thus the two angles may definitely be said to be inherited 

 factorially, and the origination of new leaf-forms in F 2 

 need no longer be surprising. 



Flower form. The characters studied in the flower 

 were the length of the petal, style, and column, measured 

 from the apex of the ovary cavity, the length of the 

 filaments, and the width and length of the involucral 

 bracts. The form of the corolla has not been examined ; 

 it ranges from an open cup to a long narrow one. 



The story of the inheritance of all the first four 

 characters is much the same. In the first cross studied 

 we found that by correcting for fluctuation, taking the 

 petal length as the standard because it has been the same 

 in both parents, our graphs for the F 2 composition assumed 

 the 3 : 1 form. 13 ' 15 Long was dominant over short, and all 

 the extracted shorts tested bred true up to F G . Here, in 

 spite of the JEgypto-American crossing, the characters 

 seemed to be under the control of a single pair of factors. 

 In this way we bred out the " short-style flower " 

 mentioned in connection with natural crossing (Frontis- 

 piece). 



In the King x Charara cross we found indications of 

 segregation in petal length also, and it was in this con- 

 nection that we developed the correlation-diagram method 

 of seeking for segregation (Figs. 53 and 60). The method 

 is essentially the same as the use of the slide-rule, but it 

 enables us to deal with two variables, without correction. 



