296 DIATOMACE.E I STRIATELLE.E ; MELOSIRECE. 



cavity of the cell into imperfectly-separated chambers. In some 

 instances these annular septa are only formed during the production 

 of the valves in the act of self-division, and on each repetition of 

 such production, and thus are always definite in number ; whilst in 

 other cases the formation of the septa is continued after the pro- 

 duction of the valves, and is repeated an uncertain number of 

 times before the recurrence of a new valve-production, so that the 

 annuli are indefinite in number. In the curious Grammatophora 

 serpentina (Fig. 128) the septa have several undulations and 

 incurved ends so as to form serpentine curves, the number of 

 which seems to vary with the length of the frustule. The lateral 

 surfaces of the valves in Grammatophora are very finely striated ; 

 and some species, as G. subtilissima and G. marina are used as 

 Test objects (§ 132). The frustules in most of the genera of this 

 family separate into zig-zag chains, as in Diatoma ; but in a few 

 instances they cohere into a filament, and still more rarely they 

 are furnished with a stipes. — The small Family Terpsinoece is 

 separated by Mr. Ralfs from the Striatellese with which it is nearly 

 allied in general characters, because its septa (which in the latter 

 are longitudinal and divide the central portions into chambers) are 

 transverse and are confined to the lateral portions of the frustules, 

 which appear in the front view as in Biddulphiece (§ 230). The 

 typical form of this family is the Terpsinoe micsica, so named from the 

 resemblance which the markings of its costae bear to musical notes. 

 227. We next come to two Families in which the lateral surfaces 

 of the Frustules are circular, so that according to the flatness or 

 convexity of the valves and the breadth of the intervening hoop, 

 the frustules may have the form either of thin disks, short 

 cylinders, bi-convex lenses, oblate spheroids, or even of spheres. 

 Looking at the structure of the individual frustules, the line of 

 demarcation between these two families, Melosirece and Coscino- 

 discece, is by no means distinct ; the principal difference between 

 them being that the valves of the latter are commonly cellulated, 

 whilst those of the former are smooth. Another important dif- 

 ference, however, lies in this, that the frustules of the Coscino- 

 discece are always free, whilst those of the Melosirece remain 

 coherent into filaments, which often so strongly resemble those of 

 the simple Confervacece as to be readily distinguishable only by 

 the effect of heat. Of these last the most important Genus is 

 Melosira (Figs. 131, 132), long since characterized as a Plant by 

 the Swedish algologist Agardh, but ranked in the Animal kingdom 

 with- other Diatoms by Prof. Ehrenberg, who included it in his 

 genus Gallionella. Some of its species are marine, others fresh- 

 water ; one of the latter, the M. ochracea, seems to grow best in 

 boggy pools containing a ferruginous impregnation ; and it is 

 stated by Prof. Ehrenberg to take up from the water, and to 

 incorporate with its own substance, a considerable quantity of 

 iron. The filaments of Melosira very commonly fall-apart at the 



