I 40 THE PLANT WORLD. 



ancestral impari])innatc leaf which i)rece(1ed the present paripin- 

 nate type. It must not l)e forgotten, however, that we do not 

 know the orio-in of the cassia leaf and that we are reasoning on 

 purely comparative data when we postulate that it was derived 

 from a typical imparipinnate leaf of the type which we now see 

 in so many o^enera of this family. 



Furthermore, the examination of a large series of specimens 

 showed that the origin of the extra leaflet from the terminal 

 process of the rhachis is open to question. Several leaves were 

 found in which the ultimate segment of the rhachis was much 

 reduced in length so that the two terminal pairs of leaflets were 

 inserted together as illustrated in figure 20B. while in one example 

 one of the leaflets of the ultimate pair was considerably smaller 

 than the other. The terminal process of the rhachis is rather 

 small in C. Saplicra and some related species, but it could be 

 made out easily at the end near the dorsal surface in the leaves 

 with two terminal pairs of leaflets approximated, and it could 

 usuall}' be detected in those in which there was apparently a 

 terminal leaflet. These facts indicate that the apparentl}' ter- 

 minal leaf does not originate by the development of a terminal 

 primordium. but that it is simply one of the ultimate pair, only 

 one member of which develops, possibly owing to the fact that 

 two pairs occasionally originate together at the end of the rhachis 

 instead of being separated by considerable distance. 



In the summer of 1905 many plants of cassia belonging to the 

 C. occidciitalis and C. Saplicra tyi)e were grown, and the same 

 peculiarities were noticed in occasional leaves. 



Early in October I had about ten plants of these cassias cut 

 and the leaves sorted and counted. The plants had grown 

 among many other forms as a border against the outer walls of 

 the greenhouses. They had in many cases been overtopped by 

 Iheir higher companions so that the leaves had fallen from the 

 ishaded, lower portions of the stems which were (|uite woody in 

 several of the specimens. Many of the leaves still on the plants 

 had lost some of their leaflets. All of these imperfect leaves 

 were discarded and the remainder. 2,740 in number, counted. 

 Of these, 2,669 ^vere of the normal i)aripinnate type, while 80 



