KErOET ON THE DIATOMACE/E. 113 



one another as those belonging to this group are, cannot be rigorously applied, since, for 

 example, in Stictodiscus buryanus, Grev., 1 the radiating appearance is still retained by the 

 valve, although some of the radial folds are found to be interrupted. 



Again, the radiate Stictodisci may be divided into two groups according as the 

 central area is or is not distinguished by a simple or double corona of fine points or 

 granules, and these groups may be indicated as coronate or non-coronate respectively. 



It must further be remarked that among triangular frustules having the very greatest 

 affinities to Triceratium, some have been observed which, with the exception of their 

 non-discoidal form, exactly fulfil all the conditions of structure recpiired by the definition 

 of Stictodisciis as given by Greville. Thus among the triangular specimens brought home 

 by the ChalleDger, one had its surface folded in a radiating manner, and also showed the 

 central Stictodiscoid corona of points, so that notwithstanding its non-discoidal form it 

 must still be regarded as a true Stictodisciis. On this ground, therefore, the generic 

 definition as given by Greville should be modified, and may be stated in the following 

 words : — " Frustula simplicia, discoidalia, vel polygonalia, per lineas radiantes divisa in 

 areas plicatas, lineis centrum non attingentibus ; areis plerumque conspicuis punctis vel 

 granulis instructis." 



We cannot, indeed, believe that the same species can assume sometimes one form 

 and sometimes another, or that from the same Stictodiscus sometimes discoid and some- 

 times triangular or polygonal forms arise — -before admitting such an anomaly it would 

 be necessary to have the experimental results of artificial cultivation of some given species 

 to go upon — yet from their structural characteristics I must conclude that a Stictodiscus 

 may exist a3 a discoid or triangular or even polygonal body. The two last groups of 

 forms may be indicated by adding to the specific name the words forma triangularis or 

 forma polygonalis respectively. 



Having arrived at the conclusion stated above, it seems a fitting occasion to review 

 the long series of species of Triceratium to determine whether any of these forms might 

 not more properly be relegated to our present genus. 



Greville in 1861 2 described under the name of Triceratium harrisonianum a magni- 

 ficent triangular form adorned with rows of pearl-like puncta, which formed a large 

 circumscribing belt. This frustule, moreover, was provided with sparsely disposed pearl- 

 like granules in the central space, in which there also occurred a conspicuous network of 

 large elongated radiating cellules from which lines passed outwards between the rows of 

 granules to the margin. But on examining the figure the cellules in question are areas 

 or compartments which are bounded by depressed lines that pass between the rows of 

 granules, and not by lines of relief as might readily be inferred from the bold manner in 

 which they are represented in Greville's figure. The fact that Greville had not found 

 examples belonging to any genus which had sometimes a discoidal, sometimes a triangular 



1 Micr. Journ., n. s., voL i. p. 40, pL iv. fig. 1. - Micr. Journ., n. s., vol. i. p. 76, pi. ix. fig. 9. 



(bot. chall. exp. — part iv. — 1886.) D 15 



