OF THE DIATOMACEJE. 757 



The word "transverse" is, for the sake of bre\'ity, omitted before strise in 

 the definitions, but, unless the contraiy be expressed, it must always be 

 understood. 



A\Tien the frustules are lunate or curved, the convex margin is called the 

 dorsimi and the opposite the venter. 



We have not mentioned the Sporangia in the generic and specific descrip- 

 tions, because the examples recorded are too few, and that condition is too 

 seldom met ^vith to be practically useful. With respect to the general 

 history of the Diatomaceae, the importance of Mr. Thwaites's discoveries can 

 scarcely be overrated (see p. 61). We consider it, however, desirable to point 

 out, that whilst the similarity of their conjugating process to that of the 

 Desmidicae affords a powerful argument in support of the vegetable natiu-e of 

 the Diatomacea), the widely different characters of their sporangia, not merely 

 in form but in subsequent changes, furnish UTesistible evidence of the pro- 

 priety of separating the Desmidiese from the Diatomaceae. The resemblance 

 of the reproductive bodies in the latter to the parent frustules, and their con- 

 tinuous growth and increase by self- division, is so unlilie what we find in 

 the sporangia of the Desmidieae and Conjugatae, as to appear more like an 

 *' alternation of generations " than examples of true sporangia. 



The fii'st attempt at a scientific arrangement of the Diatomaceae was by 

 C. A. Agardh in the ' Conspectus Criticus Diatomacearum.' He distributed 

 them into three families — CjTabeUeae, Styllarieae, and Fragilarieae, according 

 to the form of their frustules. He considered that lq each family the frus- 

 tides might be free, stipitate, united into a filament, or enclosed in a frond. 

 This system was greatly extended and improved by Professor Kiitzing ; and, as 

 we believe his arrangement (p. 101) is the best and most natural yet pro- 

 posed, we have used it in this work, admitting, however, some judicious 

 alterations proposed by Meneghini and others. It is true we do not meet 

 with examples of the four conditions in each family ; but they may fairly be 

 anticipated to occur, and their absence regarded as lacunae likely to be filled 

 up by futiu'e discoveries. We have thus brought together nearly allied genera ; 

 for it is often difiicult to distinguish a Eunotia from a Himantidium, a Tri- 

 ceratium from an Amphitetras, a Cymbella from a detached Cocconema, and 

 an escaped frustule of CoUetonema from a Navicula. The arrangements of 

 Ehrenberg and Smith we regard as far inferior, — separating, as they do, such 

 nearly allied forms. Indeed the fame of those eminent observers must 

 depend on their intimate knowledge of genera and species, and on their 

 definitions being superior to those of their predecessors, and not on their 

 primary divisions. We feel persuaded that, but for his lamented death, 

 Professor Smith would have been led by increased acquaintance with the 

 Diatomaceae to modify his views in that respect in a future edition of his 

 valuable and beautiful work on the British Diatomaceae. 



ANALYSIS OF THE FAMILIES OF DIATOMACE^. 



A. 



{Yalves with central nodule and median longitudinal line B. 



,, with umbilicus or pseudo-nodule and radiant lines or 

 ceUules 12 



„ without a central nodule 2 



2 r Frustules in side view lunate or arcuate 3 



'[ ., ,, with symmetrical margins 6 



3 I Valves dissimilar Striatelle.e. 



■\ „ similar 4 



A f Valves cellulose, without transverse striae ANGULiFERiE. 



'1 „ not cellulose 5 



