402 Journal of the Asiatic Society of Bengal. {November, 1909. 
“south, runs along the summit of two great plains which 
‘slope from it downwards towards the two marginal rivers. 
‘* There is a counterslope inland for a few miles from the Kosi, 
‘* but little or none from the Kumla.’’ 
The levels taken showed that the lowest low water level 
of the Kosi was only 5 to 6 feet below the surface of the 
Since 1876 many further changes have occurred and the 
stream is unstable in the highest. degree. The Kosi when in 
flood carries immense quantities of silt, consisting largely of 
1 j 
altogether the channel in which it was at that time flowing, and 
to work back to some previous course much to the east, which 
would endanger the Civil Station of Purnea. 
n January 1895 I was deputed to examine the channels 
as most remarkable to find large areas of country 
on which the Sal trees andjlarge Cotton trees had been killed 
r 
the homes of tigers. The impression is given that the river 
must be engaged in cutting out for itself a channel through 
some former deposit of silt’ within the hills, and appearances 
give strong support to the view taken by Captain Hirst that 
the Kosi, as a stream of its present volume, is of very recent 
date, and that very many years must elapse before we can 
hope for even a moderately stable channel in the plains. At 
- the'same time it does not seem to be probable that the stream 
will make any sudden change from the general direction of its 
resent course. I venture to quot my report 
of 1895 :—*« When we consider that the spill water is flowing 
_, {2 @ much shallower channel than the main stream, and that 
