1896.] The Classification of Diatoms. 531 
raphe. The Cryptoraphidee are usually circular or angular and 
have nothing resembling a raphe. Upon the supposition that 
the raphe is an essential organ, and that it is present in one 
tribe, replaced by another structure in the second, and “ hid-. 
den” in the third, this might be a natural classification. But 
if the raphe is known to exist only in the first tribe and its 
existence in the others is wholly theoretical, it will hardly serve 
as a character on which to base a classification. It is true that 
the genera brought together by this system appear to bear 
more or less relation to each other, but if we knew as little 
about Phanerogams as we do about Diatoms, we should think 
that a division of them into Arboræ, Frutices, and Herbx placed 
related genera together, for it would be easy to see that Salix 
and Populus are related, and also that Solanum and Physalis 
are more or less closely allied. I venture to regard the Raph- 
idex, Pseudoraphidex, and Cryptoraphidex as having no greater 
naturalness than the divisions Arborx, Frutices, and Herbz ; 
and it is to be hoped that they will soon be consigned to the 
same botanical limbo in which the latter have long since found 
obscurity. 
It is true, however, that in the Raphidex, there seems to be a 
trace of naturalness in the system. The author begins with 
the bilaterally symmetrical forms, that is those in which the 
raphe is a median line, as for example, Navicula. Those with 
the raphe at one side of the center, as in Cymbella, he considers 
a modification of the first type by a curving of the frustule and 
thus bringing the raphe nearer the concave side. And in the 
third division the raphe has approached so near to the concave 
margin that it fuses with it, asin Amphora. If this is to be 
considered simply as a modification of a typical form, it means 
little. But if this modification shows the course of develop- 
ment from the Navicula form to the Amphora form, it means a 
great deal. In Navicula and Cymbella two auxospores are 
formed from two mother cells without conjugation, and in Am- 
phora two auxospores are formed from two mother cells by 
conjugation. It is probable that the method of reproduction 
found in the derived form is a development from that found in 
the primitive form. If then the Amphora form has developed 
from the Navicula form, there is reason to believe that the for- 
