266 a>WENTY-FIRST REPORT. 



1918, and in the seedling stage two were classified as intermediate, one as 

 rogue-like, and one as aberrant. At maturity all were Rabbit-Ear rogues 

 except the one first classified as rogue-like, which was an intergrading form, 

 with type-like foliage and small rogue-like pods. Tlie reciprocal of this cross 

 gave four Rabbit-Ear rogues that were recognized as such at all stages of 

 growth. 



None of the plants so far described have boon grownx to the second hybrid 

 generation, but several crosses made in 191(i have been. The Fi plants of the 

 cross Rabbit-Ear rogue x type and reciprocal gave a total of 22 plants in the 

 Fa generation, which were all rogues at all stages in their development. The 

 results are given in Table 2. 



Three crosses of Rabbit-Ear rogue x Rabbit-Ear rogue gave in the first 

 and second generations only Rabbit-Ear rogues. The plants of both hybrid 

 generations were like the parent plants in all respects. (See Table 3). 



Considering again the results in the F^ generation of crosses between type 

 and rogue plants, it is seen that no matter which way the cross is made the 

 resulting seedlings vary in regard to the amount of resemblance to one or the 

 other parent. A summary of the classification of the seedlings given in Table 

 2 shows the intermediate to be the largest class, while the l-ogues are second 

 in number, and the aberi-ant plants least frequent. Plate X (figs a and b) 

 shows type and rogue seedlings. I'late XI (figs, a, b and c) shows the three 

 classes of hybrid seedlings. Because of the variation in stipule shape in the 

 young Fi plants it was thought that the best measui-e of the difference between 

 the hybrids and their parent forms could be obtained by getting the ratio of the 

 length to the width of stipule in the three forms, namely, pure types, pure 

 rogues and hybrids. To this end measurements of the stipules from the 4th to 

 the 9th nodes were made and the ratios computed. From the Fi hybrids 69 

 plants were measured, and from the parents 49 types and 43 Rabbit-P^ar 

 rogues. Where possible the type and rogue plants measured had been grown 

 from self-fertilized seed of the hybrid plants measured. Table 4 gives the 

 frequency distributions of the ratios so obtained expressed in percentages. 

 Due- to disease and mechanical injury, the same number of stipules could not 

 be measured at each node in any one category of plants. It was necessary, 

 therefore, to express the class frequencies at each node as percentages of tl»e 

 total number of stipules measiu'ed at that node before calculating the statis- 

 tical constants. A glance at the three distributions discloses the close simi- 

 laiity of the hybrid to the rogue parent. The similarity is shown in both the 

 lange of variation and in the distribution of the modal classes. The resem- 

 blance is further accentuated in the results given in Table 5 and Figure 3. In 

 Table 5 are shown the mean, standard deviation of the ratio, and the coeffi- 

 cient of variability as calculated for each node separately and as determined 

 for all the nodes of each kind as a whole. Figure 3 gives the graphic result of 

 plotting the mean ratio for each node in each of the three forms. The per- 



