DESMIDS AND DIATOMS. 511 
not an accurate one, many linear forms being marine, and 
a few discoids fresh water, but the distinction is a con- 
venient one, and sufficiently characteristic to apply to the 
majority of cases. . 
Among the forms whose general outline is linear, we 
have those which are straight and free (such as Witschia, 
Pl. 13, fig. 8), those whose “frustules,” as the separate 
individuals are termed, are adherent by the base, and 
which produce fan-like clusters, or even star-shaped ag- 
gregations (as in Synedra, fig. 33; Asterionella, fig. 17; 
and Licemophora, fig. 20); those with the individuals 
adherent by the sides, and producing lengthened fila- 
ments (fig. 18), which if one end of the cell be smaller 
than the other, will give rise to spirals (as in Meridion, 
fig. 12), or again lying free with others of its kind in a 
sort of gelatinous envelope or cushion (as in Mastogloia, 
fig. 19, Encyonema, etc.), or adhering by alternate cor- 
hers and producing zigzag chains (as in Diatoma, fig. 16; 
Grammatophora, fig. 30; Tabellaria, fig. 13) ; or, lastly, not 
included in either of the above divisions, but still having 
a certain general resemblance to the typical straight line 
(as Cocconema, Gomphonema, etc.). These are by no 
means natural or scientific divisions, but aid in fixing in 
the memory their characteristic shapes. Between each 
other and the second great class of circular or discoid 
forms, there is every variety of gradation, the one insen- 
sibly passing into the other. The latter, or discoid forms, 
which, as I have said, are mostly but not exclusively 
marine, are .also characterized by even greater variety of 
form and outline, often exceeding in beauty of conception 
and in delicacy of execution the most elaborate works of 
human ingenuity. As a general rule the surface of their 
valves is more or less broken into numerous dots, depres- 
