216 Proceedings of Royal Society of Edinburgh. [sess. 
of auxospores h This remained a puzzle until when working over, 
and verifying observations in London, it occurred to me to test the 
siliceous character of the membranes of these young packets of 
Coscinodisci. Taking samples known to contain plenty of them 
among other diatoms, and “ cleaning ” them by burning and by 
treatment with nitric acid and chlorate of potash, it was found in 
every case that the packets totally disappeared, while the other 
diatoms remained. The conclusion, plainly warranted, then is, 
that the individual membranes of the young packet Coscinodisci 
are either not silicified or are incompletely so, and that they are 
therefore capable of superficial growth, however it may be with 
diatoms the walls of which are perfectly silicified. 
It appears, then, that these marine diatoms may reproduce them- 
selves, either by a rejuvenescence of the cell and the secretion of 
a new frustule within the parent ( Biddulphia , Coscinodiscus, and 
possibly Ditylum), which, escaping on the separation of the parent 
valves at the girdle, may grow, divide and multiply before fully 
attaining the characteristic external sculpturing and adornment of 
the parent ( Biddulphia ), or the number of the offspring may be 
increased by preliminary divisions of the protoplasm into two, 
four, eight, and sixteen ( Coscinodiscus ). 
Both last April and in August I observed in different species of 
Chcetoceros (in addition to the usual cell-division) a series of succes- 
sive divisions within individual cells which are manifestly prelimi- 
nary to the formation of reproductive bodies. In the ordinary 
multiplication of a Chcetoceros, the cell of course divides transversely 
to the axis of the chain of cells. In Ch. constridus (Plate III. 
fig. 3), when farther subdivisions of the cell-contents are destined 
to take place, the first division is parallel to the axis as shown in 
the upper cell. In the lower cell the second division, transverse to 
the axis, has taken place. In Ch. curvisetus (fig. 4) the same thing 
happens. In both cases a drop or cushion of oil separates the two 
halves at the first division. The second division (transverse) cuts 
right through the middle cell of fig. 4, and parts the resulting 
portions into two pairs, giving them the appearance of having first 
divided transversely, which however is not the case in these species. 
In Ch. borealis (fig. 1) the first division is always transverse. 
Though ordinarily all the cells of a chain are approximately at the 
