Vol. V, No 10.] The Rivers of Bengal. - 401 
[N.S.] 
of the stream, and to the conditions under which the water 
flows off the catchment. What flood banks can do, and do do, 
is to alter the distribution of the flood water. They do this 
in two ways: Ist, in the case of a river which flows in a 
hollow, such as the Ganges above its deltaic portion, they 
reduce the reservoir space for the spread of the inundation 
in the valley or basin of the river; 2nd, in the case of 
streams of a deltaic character which flow on ridges, they 
part of the stream. That this has occurred to some extent 
along the Bur Gandak is probable, but here, as generally, 
accurate information is wanting. 
aptain Hirst has also drawn attention to the embank- 
ment which carries the Tirhut State Railway along the 
northern bank of the Ganges to the west of the Kosi. This 
embankment to a small extent limits the reservoir space of the 
Ganges, but its action in this respect is inconsiderable, and it 
Ganges and that caused by the overflow of the streams from 
the north, which turn eastwards and join the Kosi near its 
mouth. It does not appear to be probable that it will cause 
any sensible alteration in the flood level of the Ganges, and 
do not share Captain Hirst’s apprehension that it may injure 
the navigable channel of that river. 
With respect to the Kosi itself, I think that there will be 
a general agreement by those who have had any experience of 
tiver embankments, with Captain Hirst’s conclusion that the 
time has not yet come for any attempt to embank or to train 
rs) j 
“the Bulan, which is the central and the largest of about a 
‘dozen streams which intersect the district from north to 
