ANIMAL NATURE OF DIATOME^. 475 



alone the order Stomaticse. The two preceding families 

 {Licmophorece and Striatelleae) are comprised in the order 

 Astomaticse. These two orders constitute the tribe of 

 Diatomese vittatse. 



Any one examining these beings with dihgence will 

 entirely convince himself that the distinction of the two 

 orders is altogether insufficient. No Tabellaria has a 

 central perforation in the secondary surface, at all to be 

 compared with that of the Diatomese, constituting the 

 order Stomaticse in the preceding tribe. Nor have I 

 succeeded, moreover, in discerning the other four 

 perforations, described by Ehrenberg as existing in the 

 middle of the terminal surfaces. As to the four genera 

 comprised in this family, since the sole character by 

 which they were united together, and which was brought 

 to establish the principal distinction between the Fra- 

 gillariese and Striatellese, has been proved to be erroneous, 

 it will remain to be inquired whether there be other dis- 

 tinctive characters. I firmly believe that Tabellariese 

 and Striatellese ought to constitute one family, since the 

 diaphragms, which Ehrenberg considers characteristic 

 of the second exclusively, are not wanting in the first. 

 The only doubt that I could rationally entertain would 

 arise in respect to the genus Terpsinoe. The figure 

 of the lateral surfaces, the total form of the frustules, 

 the arrangement of the vittse, the direction of the appa- 

 rent transverse canal, finally, the similarity to the other 

 allied genera in external aspect, would induce us to regard 

 the frustules as connected to the head of each other, 

 rather than laterally ; as we have already stated, to give 

 a clearer idea of the subject. But then, how can we recon- 

 cile the transverse deduplication with all the other instances 

 in which, on the contrary, it is longitudinal? Eor if we 

 would attribute a greater degree of generality to this law 

 than to any other ; setting out from the direction of the 

 deduplication, always parallel to the secondary surfaces, to 

 decide upon the correspondence of all the other parts 

 with each other, we must then withdraw the title of 



