42 Scott and Brebner. — On the Secondary 
round, but on the side remote from the branch-root it became 
very small in amount, and almost wholly parenchymatous. 
Nearly the whole of the secondary zone in this part was of 
cortical origin. 
So far we have established a definite relation between the 
secondary thickening and the insertions of the branch-roots. 
The observations which we made on D. Draco confirmed those 
on D. fragrans. In one root of the former we found, on 
examining radial sections, that no sooner had the secondary 
tissues begun to thin out in receding from a lateral root, than 
they began to widen again as the next lateral root was 
approached. In this case the whole thickening was pericyclic, 
the cortical stage of growth never being reached between the 
two rootlets. 
A strong presumption has been established that the se- 
condary increase actually starts from the insertion of the 
rootlets. In fact, the younger stages in which we have peri- 
cyclic thickening only, limited to the immediate neighbourhood 
of the rootlet, raise this presumption almost to a certainty. 
The usual case appears to be that the cambium forms in 
the pericycle at the insertion of the rootlet, and that the 
divisions spread gradually in all directions, but at first are 
limited to the layer in which they started. As secondary 
tissue is formed the continuity of the endodermis is broken, 
and at the place where it is interrupted the divisions are 
taken up by the neighbouring cortical cells. By the time 
that the cambial activity has extended to the more remote 
parts of the root, the pericycle in those parts will have 
usually become thick-walled and incapable of division. 
Hence at a certain distance from the rootlet only cortical 
cambium can as a rule arise. In one case we were so for- 
tunate as to observe, in radial section, the very first cambial 
divisions in the cortex. They took place immediately outside 
the place where the endodermis was ruptured, near the 
base of a rootlet. The dividing cells were thus in direct 
communication with the pericyclic cambium. When once 
started, the cortical divisions can spread up or down the 
