AS >. DR UH. SCOTT-LAUDER ON SOME 
each angle of the frustule; and (2) Bacteriastrwm (figs. 10 
and 11), circular in side view, with a varying number of 
awns springing radially from the circumference of the valve. 
The number of frustules in a filament may vary from one to 
one hundred or more, but, as a rule, it breaks up into 
shorter lengths of twenty or thirty frustules. A perfect 
unbroken filament is always. recognised by the terminal 
valves with the more robust awns of the parent frustule. 
In this family, as in the Diatomacex generally, it is often 
difficult to distinguish species from each other and from 
varieties of the same species. The variations of form merge 
so gradually into each other, that, without having seen the 
intermediate varieties, one is apt to put down as distinct 
species at each end of the scale what are, in reality, modifi- 
cations of one and the same plant, and so lead to unnecessary 
multiplication of nomenclature. 
Of the genus Chetoceros found in Hong-Kong harbour, I 
have identified twelve species, which, for the purposes of 
classification, I have divided into three groups, viz. :—(1) 
Those having spinous awns, (2) those having beaded awns, 
and (3) those having cellular awns, the members of each 
group generally having other characters in common. 
The genus Bacteriastrum (fig. 10), also very abundant in 
Chinese waters, can be divided into two fairly distinct 
species—B. varians, varying greatly in the number of awns to 
each frustule, and B. hyalinum, larger and more delicate in tex- 
ture, with a more constant number of awns, thirty to thirty-two. 
The frustules of which the filament is made up are connected 
together by the intimate apposition or fusing together, as it 
were, of the awns of contiguous cells for a certain distance 
from the margin of the cell; they then separate again into 
the two awns of which they were primarily composed, giving 
the appearance of a bifurcated awn. The awns of the 
terminal cells at each extremity of the filament are more or 
less curved, and are simple, having no other cell contiguous 
to assist in forming the compound bifurcating awn. 
This process of blending and bifureating again, which. 
takes place normally in Bacteriastrwm, also very rarely 
happens as an accidental abnormality, in Chetoceros. I have 
only met with one instance of it. 
