120 KITTON, ON NEW GENERA AND. SPECIES. 
repeated, and only when reproduction by means of conjuga- 
tion takes place will the normal form be produced. Although 
diatoms multiply enormously by self-division, yet this has 
its limits—first, by the exhaustion of the spermatic force ; 
and, secondly, by the continued decrease in size of the frus- 
tule. It must be remembered that each successive frustule is 
formed within the parent, and the new-formed frustule does 
not increase in size, although it goes on secreting a thicker 
siliceous valve. I have some specimens of Surirella ovalis 
from Queen’s Park, Edinburgh, in which many of the valves 
have a central constriction; this arose from the abnormal 
state of the parent frustule. If this form had occurred but 
sparingly, and unmixed with the common form—one or two 
in a slide—it would probably have been described as a new 
species. Contour, that is to say, the outline of a valve, has 
very little generic or specific value. I have Triceratium favus 
from Sierra Leone of a semicircular form; <Amphitetras 
antediluviana from Joppa triangular. Triceratiwm orbicu- 
latum is sometimes so nearly circular that it might be taken 
for a Eupodiscus. Surirella fastuosa is frequently so truly 
orbicular that it may readily be mistaken for a Campylodiscus. 
Stictodiscus often seems to merge into a small Arachnodiscus. 
Pleurosigma rigidum occurs on the French coast and the 
Mediterranean without the sigmoid flexure, and the median 
line central, resembling in outline a Stauroneis. 
It would be occupying too much of your time if I were to 
enumerate all the species which gradually assume the appear- 
ance of other and supposed distinct species. I have, I trust, 
sufficiently proved the necessity of the greatest caution in 
publishing supposed new genera and species; above all, 
avoid doing so from sparse material or unique specimens ; 
rest assured that all recent diatoms will some time or other 
occur copiously, and then the range of variation will, to some 
extent, be seen. It would be far better to throw unique 
specimens into the fire than add an imperfectly described 
form to the already overburdened floras. It must also be 
remembered that a generic character taken from a single 
specimen can only belong to that specimen, and must be too 
limited to admit other specimens, varying however slightly, 
to be relegated to it. When a supposed new form occurs to 
an observer the better plan is to endeavour to discover the 
points of resemblance to already described forms than those 
where it departs from them, and by pursuing this plan more 
real benefit will be rendered to future observers than the 
addition of fifty new species. 
