THE STRUCTURE AND AFFINITIES OF DIPLOLABIS ROMERI (SOLMS). 723 



trace. The central portion of the xylem bridge then unites across the centre of the 

 island with the horizontal bar, and thus two small islands are produced (Plate III. 

 figs. 39, 38, and 37, is., is., and Plate II. fig. 24) with a protoxylem group on the 

 inner side of each. The xylem on the outer edge of the two islands now thins out 

 and gradually disappears into the two very short arms (Plate III. figs. 36 and 35), 

 thus leaving the end of the trace open, with a prominence in the middle (fig. 34). 

 The tridentate appearance is still more clearly seen in fig. 33. The petiole now enters 

 the cortex of the stem. 



Immediately below the stage in the last figure, yet another bridge appears between 

 the two arms (Plate III. fig. 32, d). In some of the slides examined, this bridge seemed 

 to give rise to a root-trace, but it was always cut longitudinally, and could never be 

 followed very far out. Whatever the trace belonged to, petiole or root, the bridge 

 again connects the two arms, but it does not fuse with the central pillar of the 

 tridentate end of the petiole-trace (Plate III. fig. 32). The persistence of the 

 prominence causes the island to be reniform and it is concave outwards. 



Thereafter the reniform island becomes elliptical in shape (Plate III. fig. 30). 

 This has obviously been derived from the other by the disappearance of the xylem 

 separating the protoxylem groups. It is therefore permissible to conclude that the 

 pillar was originally derived from the division of a protoxylem group, and the sub- 

 sequent separation of the two parts by xylem-tissue. In the elliptical island only 

 two protoxylem groups occur, one derived from the last trace which joined the petiole, 

 and one from the fused petiole protoxylems. Thus the two protoxylem groups lie in 

 a line parallel to the long axis of the petiole-trace. Up to this point the protoxylems 

 at each end of the trace lay in lines perpendicular to that axis. The island of 

 parenchyma now completely disappears, and the petiole-trace takes the form repre- 

 sented in Plate II. fig. 22, where two groups of protoxylem elements prx. x and prx. 2 

 are found at each end of the trace. At the level of this figure the petiole-trace is 

 rather dumb-bell shaped, but, at a lower level, it becomes more rounded in outline, 

 PI. II. fig. 21. The double protoxylem groups are still, however, very distinct. These 

 double groups eventually fuse into one as in PI. II. fig. 20, prx. x and prx. 2 . A similar 

 succession can be traced in Plate I. figs. 11, 10, and 9. (The petiole- trace of fig. 11 

 is shown more highly magnified in PI. II. fig. 21.) In fig. 10 the trace is elliptical, 

 and has also two double protoxylem groups, one at each end. These double proto- 

 xylem groups are closer together than the corresponding groups in fig. 11. In fig. 9, 

 only one protoxylem group is seen at each end, and it is derived from the double group 

 of the petiole of fig. 10. 



In PI. II. fig. 20 the stem is also shown, and, in the section below, the petiole and 

 the stem are fused together. Plate II. fig. 19 represents the second section below 

 fig. 20, and there the petiole-trace is firmly fused to the stem-xylem. We have thus 

 followed the Diplolabis petiole of PL II. fig. 29 into the stem of Pi. I. fig. 1. 



The lowest trace which departs from the petiole is quite close to the point where 



