THE FOSSIL OSMUNDACE^. 775 



stele possessing a true pith, surrounded by a stout and perfectly continuous ring of 

 xylem. The leaf-traces departed from this stele in a protostelic manner, i.e. without 

 leaving any depression or gap into which the external tissues could subside. Further, 

 it is possible to regard this stele as derived in turn from a still more primitive protostele 

 with a solid central mass of xylem. 



The evidence provided by the fossil Osmund acete, so far as it goes, may be taken as- 

 distinctly in favour of this point of view. For the requisite perfect continuity of the 

 xylem ring is almost, or, as we believe, actually, realised in Osmundites Dunlopi, one 

 of the oldest representatives of the order as yet recognised.* This theory is also in full 

 agreement with the ontogenetic evidence. For in several species it has been shown that 

 a continuous ring of xylem is maintained for some distance upwards in the stem of the young 

 plant {Todea hymenophylloides, Seward and Ford, I.e., p. 241 ; T. Frazeri, Chandler, 

 I.e., p. 398 ; Osmunda Claytoniana and 0. einnamomea, Faull, I.e., p. 387). Indeed, the 

 leaf-trace occasionally departs in a protostelic manner, even in the mature stem of 

 Todea barbara and T. hymenophylloides, as we have ourselves observed ; and, judging 

 from the description given by Seward and Ford (cf. fig. 29, I.e.), the same holds good 

 also in Todea superba. 



It should be noted here that both Jeffrey's view and our own are opposed to the 

 idea accepted by De Bary (12), that the vascular system of the Osmundacese is merely 

 a sympodium formed by the lower ends of the leaf-traces. On the contrary, as already 

 pointed out by Lachmann (13) and by Zenetti, the xylem strands undoubtedly con- 

 stitute a cauline network proper to the stem itself. 



Although protesting against the application of a theory of reduction to the order as 

 a whole, we do not, of course, reject, a priori, the possibility of its occurrence in any 

 one particular species. In fact, each case must be considered on its own merits, and it 

 is obvious that the structure exhibited by Osmundites skidegatensis will have an im- 

 portant bearing upon the discussion. This plant at first sight appears to provide some- 

 thing very like the dictyostelic ancestor postulated by Jeffrey's theory, and it certainly 

 establishes the fact that the Osmwidacese in the past have reached a far higher degree 

 in complexity than is represented by any of the living species. 



In face of this, the possibility that some of the existing species — Osmunda einna- 

 momea, for instance — are reduced can no longer be summarily rejected even by those 

 who are unwilling to apply a theory of reduction to the order as a whole. While ad- 

 mitting this, we venture at the same time to advance an alternative view. In the first 

 place, we would suggest that the type of dictyostely exhibited by Osmundites skidega- 

 tensis may have been attained by a " cladosiphonic " and not by a " phyllosiphonic " 

 method. That is to say, the internal phloem may have originated by the subsidence of 

 the external phloem into the pith through gaps in the stele produced by the branching 

 of the stem, and not through the gaps due to the departure of the leaf-traces — as 



* Since this paragraph was written we have obtained a fossil Osimmdaeeous stem which indisputably shows a 

 perfectly unbroken ring of xylem. 



TRANS. ROY. SOC. EDIN., VOL. XLV. PART III. ^NO. 27). 110 



