718 MR W. T. GORDON ON 



the single bow-shaped strand thus formed then joins on to the )— (-shaped petiole-strand, 

 bridging over the space between the two arms on one side of the petiole-trace and 

 enclosing an island of parenchyma between the entering pinna-traces and the two arms. 

 The island then becomes smaller in size and the xylem-bar divides in the centre, one 

 half disappearing gradually into each arm. The diminishing size of the arms in pro- 

 portion to the horizontal bar is very marked at this stage (PI. II. fig. 26). Another 

 pair of traces of reduced pinnae enter at a still lower level, and again a bridge is formed, 

 but the island now enclosed is very small indeed compared with that enclosed by the 

 pair of pinna-traces immediately above. The petiole is still free from the cortex of the 

 stem at this point. Continuing downwards, this second bridge is seen to bend in 

 towards the inner margin of the island (PI. III. fig. 40), which thus becomes reniform 

 in shape, and ultimately the middle part of the xylem bridge fuses with the end of the 

 horizontal bar of the H -shaped petiole-trace, cutting the island of parenchyma into 

 two. It is just about this point that the petiole enters the cortex of the stem. The 

 protoxylem groups are situated one in each parenchymatous island. The xylem on the 

 outside of these islands now disappears into the arms on each side just as in the case of 

 the first-mentioned reduced pinna-traces, but, in disappearing, it leaves the end of the 

 trace tridentate (PI. III. fig. 34). The three arms are very short : two of them 

 correspond to the two arms of the trace higher up, and the third — the central arm — is 

 produced by the union of part of the entering reduced pinna-traces with the base of the 

 island of parenchyma which they enclose. 



At this stage another bridge of xylem joins into the petiole -strand and connects the 

 ends of the two outer arms of the tridentate end ; thus a reniform island is again 

 produced. In this case, however, the two extremities of the island are on the inner 

 side, i.e. point towards the middle of the petiole-trace, whereas, in the former reniform 

 island, the two extremities pointed out from the trace. At least one protoxylem group 

 enters the trace with this new strand, and so there appear to be three protoxylem groups 

 bordering the island, one at each extremity and the third in a median position at the 

 apex of the island (PI. III. fig. 31, prx. t , prx. 2 , prx. 3 ). Further down, the two inner 

 protoxylems fuse, and thus the tongue-like central arm of the trace disappears. In this 

 way an elliptical island is substituted in place of the reniform one. The trace mean- 

 while has become rather shorter and more dumb-bell shaped. The island of parenchyma 

 soon dies out, and we are left with a double protoxylem group sunk in the xylem at 

 each end of the dumb-bell shaped trace (PI. II. fig. 21). The trace now becomes shorter 

 and more rounded, while the double protoxylem groups unite in pairs. This takes place 

 at each end of the trace, and so we have now one sunk protoxylem group at each end 

 of an elliptical xylem. Immediately below this level the trace joins on to the stem 

 without any gap being formed in the outer xylem zone of the latter. The petiole-trace 

 gradually disappears into the stem, but the petiolar protoxylems can be followed down 

 into it for a considerable distance before they disappear (PI. II. figs. 21-16, and PI. I. 

 figs. 15 and 14). 



