244: WALLICH, ON TRICERATIUM. 
layers, I believe, with all deference to so deservedly high an 
authority, are, however, rarely more than two, and arise from 
the plates, during the commencement of division, in the frus- 
tules of this and many other genera, always consisting of two 
pieces, which, at first, entirely overlap each other; but, as 
the process advances, recede from each other, and whilst so 
receding, appear like three distinct parallel annuli, the centre 
being less diaphanous, and its markings more confused, in 
consequence of its being, in reality, the overlapping and 
double portion referred to. This appearance has led to much 
uncertainty and doubt in descriptions of the connecting mem- 
brane, inasmuch as, from its transparent structure, markings 
when they exist in the lower plates, are seen through those 
in the upper. In those genera in which the valves assume 
at times a great relative depth, we find not only that the 
connecting membrane is more largely developed, but that 
the valves are furnished with a constricted rim, to which the 
margin of the annular plate is attached, as if to afford a more 
powerful point of resistance from whence it can extend itself. 
In Amphitetras, and certain species of Triceratium and Bid- 
dulphia, the existence of marginal rows of puncta on the 
annulus, in close proximity to the markings on its surface 
generally, proves that the growth of each plate of the con- 
necting membrane takes place at the margin furthest off 
from the valve to which it is attached. Were it not so, the 
rows of marginal puncta would recede from the central mark- 
ings, an effect opposed to what in reality occurs. Growth 
thus takes place in both plates at once—the overlapping, to a 
greater or lesser extent, being dependent on the rate at which 
the new valves within happen to be developed. In the newly 
separated frustule, one end may constantly be seen imbedded 
in its own half of the connecting membrane, which, for a 
time, remains attached to it. The same structure exists, I 
believe, in nearly all the genera, although more readily discern- 
ible in some than in others, from the greater facilities they 
afford as regards size and figure. It may be thus seen in 
Mimantidium, Odontidum, Denticula, Eunotia, Grammatophora, 
Amphitetras, Biddulphia, Isthmia, Melosira, Coscinodiscus, 
Hydrosera, &c. I may observe, in passing, that the figures 
given in the ‘ Synopsis of British Diatomacee’ of Biddulphia, 
Amphitetras, and Isthmia, show the general aspect of the 
connecting membrane—but without any allusion, on the 
part of the author, to the striking mode of development now 
described. 
The arcuate bands are always arranged in the same direc- 
tion, that is, their concave or convex aspects always face 
