The Resin-Canals in the Guayule. 167 



cal canals. At the level at which the earliest canals appear, namely, im- 

 mediately above the level of the cotyledonary node, the difficulty is not 

 as great as higher up. Here the endodermis is evidently involved, and it 

 seems conclusive that the whole of the canal structure is derived from it, 

 though the cell lineage is not as evident even in a young condition as it is 

 at higher levels in Parthenium incanum. This at any rate accords with 

 previous observations, 1 and is without any doubt the case in those parts 

 of the stem where the endodermis is regular enough to display its morpho- 

 logical relations. I therefore conclude that, were it possible to follow the 

 development of the structure, it would be found, even in the higher parts 

 of the stem in Parthenium argentatum, where the endodermis is quite ill- 

 defined, to have originated in this. 



The course of development is as follows: A tangential division takes 

 place in one, or it may be two or three neighboring endodermal cells. In 

 the cell destined to give rise to the canal a radial 2 division crosses the first 

 wall so as to form four cells, realizing the "division crucial" of van Tieg- 

 hem. Periclinal divisions, however, take place, cutting off special secre- 

 tory cells, four in number, from a tier of supporting cells, while these suffer 

 a still further subdivision. Two pairs arising from the inner two cells of 

 the original four are cut off, and are, so to speak, discarded from the canal 

 structure, as occurs also in the primary root-canals. Only the outer cells 

 of the outer original two become divided, so that fourteen cells in all arise, 

 of which four are the original secretory cells, six are the supporting 

 cells, and four, or perhaps six, excluded this in the mariola, Parthenium 

 incanum (plate 37, fig. 8). 



The canals of guayule (plate 37, figs. 1-5) bear sufficient resemblance 

 to those of the mariola, so that it would be unsafe to deny their entirely 

 endodermal origin. Secondary changes, by which the number of secre- 

 tory cells as well as that of the supporting cells is multiplied, need not 

 be described, as they consist only of repeated radial divisions and some- 

 times of tangential ones in the secreting cells. 



These canals suffer more or less displacement (plate 37, fig. 3) accord- 

 ing to circumstances, often sufficient to mask their origin. For this reason 

 they have been alluded to as cortical by Ross (1908) and by Fron and 

 Francois (1901), without raising the question as to their origin. This is, 

 perhaps, the reason that, although Col (1903) asserts the endodermal 

 origin of the cortical canals in the Tubuliflorae, his drawings sometimes 

 fail to show clearly this derivation, as, e.g., in Aster astivalis. 



In Anthemis mixta and Lasthenia glabrata, however, the origin is 

 clearly shown, and it seems that the canals are less elaborately organized 

 than in Parthenium and suffer less displacement. My own effort has been 

 to show conclusively the origin of these canals, with the result that the 

 work, in part of Vuillemin, of van Tieghem, and of Col, is supported. 



1 Van Tieghem (1884) insists correctly upon the endodermal origin of the 

 primary canals, but I am unable to recognize the distinction between canals 



hordes" and "non-horde's," though, correlated with the more definite character 

 of the endodermis in the roots, the canals are here more regular and somewhat 

 simpler in their structure (but certainly not "non-horde's") than in the stem. 



2 With respect to the stem. 



