510 JUNCTION OF THE FRESH WATER OF RIVERS 
This anomaly is easily accounted for. Were the current of 
the tide confined entirely to the channel of the river, an ar- 
rangement of the waters, similar to that which existed in the 
first experiments, would have prevailed. But during the flow- 
ing of the tide, the sea-water soon occupies more than the 
channel of the river, and spreads itself in various streams 
among the hollows of the sand-banks. These streams reunite 
at different places with the principal current, and, in this man- 
ner, prevent the salt and fresh waters from gaining their natu- 
ral relative position. But as soon as these sand-banks are co- 
vered with water, the tide proceeds with regularity in its 
course, so that the different layers of water can then arrange 
themselves according to their specific gravities. 
A thousand grains of water obtained from the bottom, at the 
height of flood, yielded by evaporation twenty-three grains of 
salt, while the same quantity of water from the middle yielded 
only eighteen grains; and from the surface only seventeen 
grains. This was a difference of no less than six grains, and 
seemed to afford a decisive result. 
In order, however, to complete the series of observations, 
I examined the conditions of the currents at half-ebb. The 
same irregularities prevailed, as before observed at half-flood. 
A thousand grains of the water, from the bottom, yielded after 
evaporation eleven grains of salt ; from the middle, nine grains, 
and from the surface, twelve grains. At this time the densest 
water was at the surface, and the lightest occupied the middle. 
The cause of this was obvious. Extensive portions of the 
sand-banks had already been left dry by the receding tide, and 
various currents of water, disjoined from the main stream by 
the inequalities of these banks, were now re-uniting with it, 
through various channels, and disturbing the natural arrange- 
ment which had prevailed during the time of flood. 
Although 
