248 Holden.— Some Fossil Plants from Eastern Canada. 
3. The affinities of Tylodendron are uncertain ; it is not closely related 
to the Cordaitales, and there appear to be no sufficient grounds for relating 
it to any one living group rather than to another. 
New Brunswick. 
Just west of Martins Head, on the south shore of the Bay of Fundy, 
the grey sandstone characteristic of the Devonian gives way to cliffs of brown 
arenaceous rock, embedded in which there is a confusion of lignitic remains. 
Among them were found a number of tree trunks with woody cylinders 
much broken, but with the pith still present as a sandstone cast. Such 
casts were usually not more than 1 or 2 cm. in diameter, while others, from 
which all trace of organic material had disappeared, had sometimes a dia- 
meter of 5 or 6 cm. One, represented natural size in Fig. 11, was traced at 
least 100 cm. along the cliff without coming to any swellings such as are 
characteristic of the Tylodendron type of pith cast. In addition to these 
trunks there were some decorticated twigs, considerably flattened, but 
otherwise in a fair state of preservation. A few hundred feet further west 
this fossiliferous layer crops out again, containing bits of wood, but no twigs. 
During the past winter this lignite has been subjected to a microscopic 
examination. To prepare it for sectioning, it was first softened by a sojourn 
of several days on a paraffin bath, in a mixture of equal parts of 70 per cent, 
alcohol and 5 per cent, caustic soda. The alkali was then thoroughly washed 
out with alcohol, and the lignite transferred to a mixture of glycerineand 30 
per cent, alcohol. In this condition a preliminary examination may be made, 
the useless bits rejected, and those desired for further study treated with 
hydrofluoric acid, and then embedded in celloidin. Most of the wood was 
of no value from a structural standpoint, but occasionally well-preserved 
pieces were found. From these the ligneous structure could be ascertained 
without difficulty. 
Fig. 11 shows one of the casts. As may be seen, the entire surface is 
covered with rhombic scars. These scars are of variable length, but always 
lack the slit characteristic of similar structures in Tylodendi'on. It may be 
suggested that imperfect preservation might have obliterated this feature, 
but examination of a large number of such casts, most of which were 
beautifully preserved, proves such not to be the case. 
Thin sections of the small lignitic twigs referred to above show that the 
furrows of the pith cast represent instanding protoxylem strands. As would 
be inferred from the absence of a slit, the foliar trace passes out directly 
from the medulla, without the ‘ hump * characteristic of Tylodendron. The 
leaf-strand is single when it leaves the pith, but forks during its passage 
through the wood. A similar condition may be seen in the living Agathis. 
Fig. 24 shows in transverse section one of these traces in the process of 
splitting ; Fig. 23 shows the double trace immediately after the split. In 
