THE FOSSIL OSMUNDACE^E. 767 



of the fully developed petiole are distributed in a manner characteristic of, and practi- 

 cally constant in, each species examined. At the same time, it varies sufficiently from 

 one species to another to render it suitable for the purpose of comparison. The diagrams 

 given in the figs. 6-12, PI. VI., indicate the arrangement of the sclerenchyma as 

 seen in a section taken about half-way up the stipular region of the petiole in the 

 various living species examined. On comparing them with the fossils, it is seen at once 

 that the leaf-base of Osmundites Dunlopi (fig. 3, PL VI.) comes very near that of 

 Todea barbara (fig. 7, PI. VI.); while Osmundites Gibbiana (fig. 4, PL VI.) (although 

 in some respects unique) approaches nearest to Osmunda regalis and 0. Javanica 

 (PL VI, figs. 6 and 8.). 



While referring to these diagrams, it should also be mentioned that the two 

 American species, Osmunda cinnamomea (PL VI., fig. 11) and 0. Claytoniana (PL VII., 

 fig. 12) distinguish themselves from all the rest by the fact that their sclerotic ring is 

 not homogeneous, but masses of specially dense and thick-walled elements occur at 

 certain well-defined points in it. 



In the stem itself the most important anatomical character that shows any consider- 

 able variation in the living species (apart from Osmunda cinnamomea) is the extent of 

 the interruption in the continuity of the xylem ring caused by the departure of the 

 leaf-traces. Osmunda regalis may be taken to represent one extreme of this variation y 

 in which the xylem ring is broken up into so many distinctly separate strands which 

 are quite free from one another, at any rate over part of their course (PL VI., fig. 1). 

 The other extreme is represented by Todea barbara and T. superba, in which the 

 strands are fused with each other and with the xylem of the leaf- traces, so as to form 

 continuous bands often of considerable extent (PL VI., fig. 2). As regards the two' 

 fossil species, Osmundites Dunlopi, with a practically continuous xylem ring, clearly 

 points in the direction of Todea barbara, while Osmundites Gibbiana points in the 

 direction of Osmunda regalis. So far as our own data permit us to judge, it appears 

 that among the living Osmundacese, Todea barbara shows most resemblance to 

 Osmundites Dunlopi, and Osmunda regalis to Osmundites Gibbiana. Nevertheless, 

 it does not follow that the two fossils actually belong to these two genera, and, until 

 their spore-bearing leaves are also known, it is advisable that both should remain in 

 the comprehensive fossil genus Osmundites. The question now to be considered is 

 whether our fossils are identical with any of the three other previously described 

 species of the genus. The first of these was found in the year 1847 by Pettko (2) in 

 Tertiary quartz in Hungary, and was described by him in 1850 under the name of 

 Tubicaulis. It was again briefly described by Unger (3) in 1854, who correctly named 

 it Osmundites schemnicensis. In 1870, Osmundites Dowkeri was obtained by Car- 

 ruthers (4) from the Lower Eocene of Heme Bay, and in 1902 Penhallow (5) described 

 Osmundites skidegatensis from the Lower Cretaceous in Canada. Seward and Ford (6) 

 also refer to a section of an unnamed Osmundites preserved in the collection of the 

 botanical department of the British Museum, and labelled " New Zealand." Through 



TRANS. ROY. SOC. EDIN., VOL. XLV. PART III. (NO. 27). 109 



