are made by the splitting in half radially of each of these sixteen bundles. 
The thickenings are thus all on the inner cell wall of the peristomial layer 
as is required by Philibert’s hypothesis. He states that in the basal mem- 
brane of this Tortula the layer inside the peristomial layer consists of just 
twice as many rows of cells as appear in the region from which our fig- 
ure is taken, thus making the homology with Dicranuni and Funaria more 
complete. 
Returning now to the relations between the arthrodont and nematodont 
types, M. Philibert considers Encalypta a primitive form connecting the two 
types and giving rise to all the varied kinds of arthrodont peristomes. In 
this connection I am very forcibly struck with the leaf resemblances between 
Georgia, Weber a, Encalypta and the Tortulaceae , which last are evidently 
next of kin to Encalypta on the arthrodont side. Moss students will remem- 
ber that so eminent an authority as Braith waite places Weber a sessilis next 
the Tortulaceae because of its leaf structure. 
The leaves of Buxbaumia are so reduced as to be of little value in show- 
ing relationships, but the peristome shows undoubted affinity to that of some 
species of Encalypta. My studies on the peristome have convinced me that 
the following arrangement of families would more truly represent the order 
of evolution than the one I have previously followed : Georgiaceae, Poly- 
trichaceae , Buxbaumiaceae, Encalyptaceae , Tortulaceae , Ephemeraceae , 
Grimmiaceae , Dicranaceae , Fissidentaceae. 
In Encalypta we find a curious combination of peristome characters. 
The peculiar “extinguisher-like” calyptra and the leaf characters make it 
certain that no mistake is made in putting all the species into one genus, yet 
within the limits of this genus we have almost all degrees of completeness of 
the peristome from none at all in E. co7mnutata to simple in E. ciliata , and 
highly developed and double in E. procera. The relationship to the Nema- 
todont peristome is shown in E. longicolla and E. brevicolla, espe- 
cially the former. Comparing the figures of the peristome of Georgia 
pellucida and those of the peristome of E. longicolla (Plate IV.) a most 
striking superficial resemblance appears, and this seems to increase rather 
than decrease upon closer investigation. The structure of the peristome of 
Georgia has previously been carefully explained, the only new points 
brought out are the thickening of the cell walls of the outer rows of cells. 
The peristome of Encalypta longicolla consists of sixteen lanceolate teeth, 
each composed of a bundle of jointed, red, papillose filaments. On the outer 
face of each tooth one can count four or five of these filaments, cohering at 
their joints but more or less free in the intervals between the joints, so that the 
tooth appears somewhat perforate. Sometimes these filaments are divided 
irregularly into two groups, separate at base but united at the top. 
These filaments contain cellular cavities as in Georgia , two in the middle 
and thickest portion of the tooth but usually only one on the edges. These 
cell cavities are enclosed by walls composed of two layers of plates, as are the 
teeth of the Arthrodonteae. “To pass to the normal type of the Arthro- 
donteae nothing is necessary but a reduction in the number of elements of 
