474 
BULLETIN OF THE BUREAU OF FISHERIES 
show as much as nine parts silica (as Si0 2 ) per million. Androscoggin water is, there- 
fore, almost as rich in silica as the spring water average just quoted (Clark’s exact 
figures are Si0 2 , 18.63 per cent of total solids, salinity 48.3 per million). According 
to one analysis the upper waters of the Merrimac are even richer in silica than this 
(Si0 2 , about 31 parts per million), but since the Merrimac flows for many miles 
through an alluvial valley in its lower course, and at the same time receives several 
important affluents from swampy areas, it probably reaches the sea with a much 
smaller percentage of silica in its water. 
If we can take the Androscoggin as fairly typical of the rivers of northern New 
England (including the St. John Biver, of which no analyses are available), which 
is justified by the nature of its watershed, it appears that, on the whole, the river 
water emptying into the Gulf of Maine is 7 to 8 times as rich in dissolved silica (Si0 2 ) 
as our analyses off Gloucester suggest as a fair average for the latter. A discrepancy 
of this sort obtains between the silica contents of river and sea water in temperate 
zones generally, and its effects are probably accentuated in the Gulf of Maine, 
just as the effect of land drainage is in reducing surface salinity by the concentration 
of the run-off from a large watershed into a comparatively small and topographic- 
ally circumscribed area of sea. It would therefore be reasonable to expect the waters 
of the Gulf of Maine to average high in silica when sufficient analyses are made 
to plot the distribution of silica in boreal seas generally. 
In addition to the silica brought down by the rivers in the dissolved state, 
probably much larger amounts are carried to the sea, suspended in the form of the 
finely divided clay which is derived from the disintegration of felspars, etc. Though 
most of this clay is precipitated to the bottom on mixture with the salt water, part 
of it is carried to great distances. Murray and Irvine (1892, p. 240) suspected from 
their cultural experiments “that the pelagic silicious organisms might, in part at 
least, obtain the silica for their frustules and skeletons” from this clayey matter. 
So far as I know these experiments have been neither confirmed nor refuted, 
nor is it clear whether they were sufficiently precise to eliminate other possible causes 
for the abundant growth of diatoms which ensued on the introduction of clay into 
the artificial culture solution. But we must reckon with the possibility that diatoms 
not only make use of the dissolved silica but also of the insoluble silicates, given 
vertical circulation strong enough to keep the latter in suspension in the water. 
A third possible source of silica is the slow solution of the rocks that form part 
of the coast line of the Gulf, and of its submarine boulders, sands, and clays. Silicious 
deposits of this sort have commonly been regarded as so nearly insoluble in sea 
water as to be negligible biologically; but as Clark (1916, p. 132) points out (geol- 
ogists generally recognize this), sea water does attack and in the end dissolve the 
most refractory silicates, even if very slowly. In fact, Joly (1901) found that sea 
water dissolves more silica (Si0 2 ) from felspar than does distilled water. 84 But 
there are two reasons for hesitancy in applying Joly’s generalizations to conditions 
as they occur in nature. First, I am unable to judge from his brief account whether 
his analyses took due account of the small amount of dissolved silica which we must 
» Earlier tests by Thoulet (1889) gave the opposite result. 
