MICHIGAN ACADEMY OlT SCIENCE. 205 



been Griulus, seed of which was first ohtaiiietl from the J. I'.. Kict; Seed Coni- 

 pany of Cambridge, New York. 



OCCUURENCE OF MtJTATIONS. 



Kubbit-Ear rogues ai'e linown to arise dc novo from type plants or from 

 various intergrading forms. To illustrate the numerical relations that exist 

 between the type and the Rabbit-Ear rogue in families of Gradus in which 

 they occur, the results of the past season (1918) are given as typical. From 

 150 type plants selected in 1917 were raised over 2,500 plants in 1918. Of the 

 150 families grown oidy one produced any Kabbit-Ear rogues. In this family 

 of 19 plants (Record No. Sag. (iO) three were typical Rabbit-Ear rogues 

 and the other 16 were type plants. Twenty-eight typical plants of another 

 strain, and belonging to the F3 generation from a plant (G 37) selected in 

 1910, gave in 1918 a total of 312 F, plants. Two of the 28 families produced 

 I'ogues in the ratio of 14 type to 1 Rabbit-Ear, and 21 type to 1 Rabbit-Ear, 

 respectively. Another Fj family descended from the same parent, G 37, 

 contained in 1918, 12 type plants, and one plant with type-like foliage and 

 small rogue-like pods. In all, adding together the results from all Gradus 

 families in which Rabbit-Ear rogues occurred, we get a total of 5 rogues to 

 51 type plants, or 9.6 per cent. 



BKHAVIOR OF RABBIT-EAR ROGUES ON CROSSING WITHIN THE GRADUS STRAIN. 



In 1917, forty-eight crosses were made between Rabbit-Ear rogue and type 

 plants. In 1918 from these crosses were raised to maturity 148 plants. An 

 attempt was made to classify the hybrids when they had developed to the 5th 

 node. As a great amount of variability existed in the shape of the stipules it 

 was decided to group the seedlings arbitrarily as type-like, intermediate 

 (stipules intermediate between the two parent forms), and rogue-like. Many 

 seedlings were found injured by fungus attacks, and such plants were lumped 

 together in a group as aberrant. It was noted that Fj plants from one cross 

 might be fairly uniform in appearance, or the individuals might vary consid- 

 eral>ly, or the reciprocal cros.ses were unlike in the .seedling stage. At maturity 

 it was found that all the Fj plants, with one exception, were like the Rabbit- 

 Ear parent. A few hybrids of the cross (Gradus type x Ral)bit-Ear, and rccip 

 rocal), which have been carried into the F. and F3 generations, were Rabbit- 

 Ear rogues at all stages in development. The same results were found true 

 for several crosses of type and Rabbit-Ear rogue plants in other varieties, 

 from which F^ and F3 generations have grown. Table 1 shows the results 

 of classification of the progenies in the seedling stage, and later, at maturity. 

 As mentioned before, when classified at maturity the hybrids were, with one 

 exception, indistinguishable from the Rabbit-Ear parent, not only in the 

 character of the stipules but in all the characters that differentiate the rogue 

 from the type. The one exception occurred in the protected cross (G. 19-1-3, 

 R. E. rog. X G 13-1-6-1 type). Four plants were grown from this cross in 



