W. Batkson and Carolixk Pkli-k\v 21 



;i Intermediates. Leaves and stipules neai-ly as in type ; but 

 usually declining, especially in upper parts of plant, towards 

 the smaller size ■•md slia[)e characteristic of rogues. I'ods 

 dcfiniti'jy curved. 



/ 4. Hof/iies. Foliar parts (espeoi.ally at lowei- nodes) somewhat 

 larger than those of the ordinary rogues. Little marhling. 

 Pods curved. 



5. Rogues. Foliar parts small and pointed. Littli> or no maihling. 

 Pods curved. 



This classification is to some extent arbitrary. Owing to inter- 

 gradation, and especially to the fiict that the different parts of the same 

 plant are frequently not uniform, it is not possible to refer the indi- 

 viduals to the several classes with perfect certainty. This applies 

 especially to the distinctions between classes 1 and 2, and classes 4 

 and 5 respectively, and these minor subdivisions have not been systema- 

 tically recorded. 



Genetic Behaviour. 



The oft'spring of classes 4 and 5 are always rogues. As to the 

 interrelation of these two cla.sses it is remarkable that the gi-eat 

 majority of the offspring of class 4 belong to class 5. Nor have we 

 ever had a clear example of a plant which could be reckoned as 

 belonging to class 4 coming as the offspring of class 5. On no 

 occasion has a plant of class 4 bred true to that type, or even given 

 more than a small proportion of plants like itself In other words, the 

 larger-leaved rogues throw a great niajority of the ordinary small- 

 leaved rogues, and these always breed true. 



As to the genetic behaviour of the other classes the facts are not 

 wholly clear. Difficult}' of course arises from the impossibility of 

 referring the individuals to the various classes with perfect consistency, 

 and the genetic composition cannot be quite certainly determined 

 from an inspection of the somatic characters. Many families contain 

 occasional plants of class 2, namely types in general conformation, bul 

 having some slight degree of curvature in one or more pods. Plants of 

 this .sort have as a rule proved to be, genetically, types. 



Classified however according to the offspring which they produce the 

 whole series of plants (othor than the rogues, which, as stated above, 



