DISCUSSION : FORESTS, RESERVOIRS, AND STREAM FLOW 359 
of similar reservoirs would be required for protecting effectually the Mr. Harts. 
basin of the Garonne. The capacity, however, of the reservoirs which 
would be of value in reducing the height of floods, and could be con- 
structed in the upper valley of the Garonne, would amount to only 
one-sixth of that required for protecting Toulouse. The investiga- 
tions led to similar conclusions in the case of the lower Garonne, and 
of the other principal river basins of France. 
“The distribution of rain and advantages of situation would oblige 
the reservoirs to be placed as much as possible in the hilly portion of 
the basins, where, however, owing to the declivity, the reservoirs would 
be smaller, and would be liable to be rapidly filled up with detritus. 
It might also sometimes happen that a flood would be produced by a 
heavy rainfall in the lower part of a river valley, as was the case in the 
great flood of the Garonne in 1875, and then the reservoirs in the hills 
would be useless. 
“The action, moreover, of reservoirs on the river further down, 
with an increasing discharge, becomes rapidly less Thus a lowering 
of the Rhone at Lyons of 34 feet, by means of reservoirs above, would 
amount to only 14 foot below the confluence of the Saéne, and to only 
8 inches below the confluence of the Isére at Valence. In like manner 
a retention of 180000000 cubic .yards on the Garonne at Toulouse 
would only act like a retention of 65 000 000 at the mouth of the Tarn, 
of only 52 000 000 at Agen, and of only 23 000 000 at Tonneins, and its 
effect would become insignificant at Langon. This shows how large 
the reservoirs in the upper valleys would have to be made to produce 
an adequate effect in the lower portions of the rivers. Reservoirs on 
tributaries, by retarding a flood, would modify the flow in the main 
river in a variety of ways, which could hardly be determined before- 
hand, and might be injurious by causing a coincidence of floods. 
“Often continuous floods are produced by. the coming down of 
several high floods at short intervals apart. Thus the inundation of 
the valley of the Garonne in 1856 was caused by five successive floods, 
keeping the river in flood for more than two months. Under such cir- 
cumstances the reservoirs would be filled by the first flood, and would 
not be available for lowering the subsequent floods. The reservoir 
embankments might be endangered by the flow of water over them, and 
a failure of an embankment might entail more disasters than a very 
high flood. 
“Tt has been proposed that flood reservoirs should be also utilized 
for supplying canals, for irrigation, and for water power, thus serving 
a double purpose. Flood reservoirs, however, to be useful against floods, 
must be kept empty throughout the whole season when floods may 
occur. Storage reservoirs have small outlets, and a flood discharge 
would have to flow over the embankment, or over a wide overfall weir. 
Such an arrangement might be adopted in a rocky valley, and where 
the discharge of a river is small; whilst the erection of high embank- 
ments across the lower valleys, where the discharge is large and the 
ground liable to be eroded by the fall of the water, would be dangerous. 
“The system of reservoirs for providing against floods must accord- 
ingly be entirely abandoned, not only on account of the excessive ex- 
pense and difficulty of obtaining adequate capacity, but also because a 
carefully designed system of reservoirs along a river valley, which 
