472 MR. W. C. WILLIAMSON ON THE MICROSCOPIC STRUCTURE OF THE 
minute kosmine tubes are seen arching downwards and inwards towards the centre 
of the plate, fig, 36 d. Similar tubes also descend from the convex upper surface of 
each plate, 36 e. From the top to the bottom of the segment are also seen numerous 
arched lines of growth, which run parallel to the upper surface ; these are best repre- 
sented in the segment, fig. 36 f, in which I have omitted the kosmine tubes, in order 
to exhibit the lamellae with more clearness. Running longitudinally through the 
centre of each plate are several narrow, depressed, semilunar canals, fig. 36 g, 
arranged at nearly equal distances from one another, and into each of which many 
of the kosmine tubes, coming both from above and from below, appear to open. 
Fig. 37 represents a horizontal section of two similar plates ; at 37 « we again see 
the orifices of the network of canals. Owing to the arched direction assumed by the 
kosmine tubes, this section divides them nearly at right angles at the central part of 
each plate, whilst it is almost parallel with their plane at each margin ; hence, at 37 b, 
these tubes are seen very distinctly, whilst at 37 c rows of minute dots alone mark 
the position of their divided orifices. The latter do not always run in lines exactly 
parallel to the plane of the marginal canals, but they exhibit a strong tendency to do 
so. Fig. 37 d shows the direction taken by one of the semilunar canals, fig. 36 g-. 
It is very easy to trace the process of growth in this interesting structure, in 
which each plate or segment represents a tooth turned inside out. The first deposi- 
tion of calcareous matter has been made in the form of a thin cylinder surrounding 
the central cavity, fig. 35 h. New lamellse, perforated with minute apertures, have 
been formed upon this basis ; the apertures in each lamella being arranged in exact 
juxtaposition with those of the contiguous lamellse. The minute tubes thus formed 
have at Jirst all opened at the external surface, but it will be readily seen, that, after 
the addition of many new lamellse, owing to their arched arrangement, the tubes of 
adjoining plates would meet at the line of junction, fig. 36 c, and thus each segment 
would contribute to block up those of its neighbour, preventing them from receiving 
their proper supply of nutritious fluids. To obviate this difficulty, however, the net- 
work of minute canals, fig. 36 a and c, has been left open between contiguous segments. 
The orifices of the kosmine tubes open into these canals, and maintain a free com- 
munication with the surface through the row of small orifices seen in each superficial 
sulcus, fig. 36 h. Thus the nutrition of the tissues in the interior of each segment 
continues to be provided for. The use of the longitudinal canals, figs. 36 g and 37 d, 
is uncertain. They have probably contained some form of pulpy matter. No trace 
of true bone is seen in the entire structure, which consists wholly of kosmine. In this 
it bears a resemblance to the rays and dorsal appendages of the placoid fish. The 
kosmine breaks with a translucent and shining fracture. 
It is difficult to conceive of any arrangement by means of which a stronger fabric 
could have been produced, than is exhibited in this cylindrical combination of long 
radiating plates. The true nature of the appendage itself, as well as that of the fish 
to which it belongs, is yet uncertain ; but if it has been a weapon of defence, like the 
