Cc. A. BARBER. 27 
in its cells, also takes part in the formation of the sucker (Plates 
IV and V). Asin Santalum, the sucker of Cansjera consists 
partly of nuclear and partly of cortical cells: that of Olaz is 
purely nuclear. 
The further development ofthe sucker in Cansjerais extreme- 
ly limited as compared with that in Santalum and Olax. Having 
reached the woody cylinder, and being prevented from further 
progress in this direction, it spreads along the cambial line and 
thus becomes bilobed. But the lobing in Cansjera is very slight 
and sometimes altogether absent. In not a few cases the sucker 
remains plug-like or columnar (Plate IX, figs. 1 and 5). This 
is in marked contrast with the long-continued growth of the 
sucker lobes inthe other haustoria. Besides this, the sucker 
does not greatly increase in thickness, and, as a consequence, the 
wings of host’s bark thrust aside by its entry are comparatively 
small. It may be noted, in passing, that the haustoria of 
Ximenia americana, although not yet fully examined, appear 
to present the opposite extreme, as well in the growth of the 
sucker lobes as in that of the cortical folds. Both of these 
grow right round the opposing root until they completely 
envelope it. One result of the meagre development of the 
sucker in Cansjera is that the whole haustorium, from the 
start almost, is much broader above than below, a character 
which becomes intensified when secondary thickening takes 
place in the vascular system. 
On the other hand, the sucker is not active only at its end 
where it touches the woody cylinder of the host’s root, but also 
at is sides where it is in contact with the cortex and bark. A 
thick layer of yellow substance lines the inner sides of each wing, 
and the cells of the sucker extend as papille into this layer just 
as they do at the edges of the lobes. At some later period the 
active cells at the sides are usually cut off from the main body 
of the sucker by a collapsed layer, and when this happens, they 
rapidly become permanent and are ultimately crushed (Plates 
X, figs. 1 & 2,a, and XI, fig. 1). Asimilar state of things has 
been occasionally met with in Santalum. 
