[ 32 ] 84 
2 feet deep: (Ik? gate wiiere t'ue feeder discharged into the summit 
level is 3 feet; 4, and the water 7 inches deep. The slope of this 
feeder was 5i inches per mile. The quantity of water received into 
the feeders of Languedoc canal, is 72,000,000 metres^ of which 
37,256,000 is discharged into the canal, and 35,344,000 is lost by 
filtration and evaporation. Without attempting either to account 
for this loss, or for the fact that feeders, of comparatively small di- 
mensions, ajjpear to lose as much as canals, the conjecture may be 
hazarded tiiat the shallowness of the water permits its temperature 
to be raised — its velocity increases exposure to the air, and the more 
so if the wind is against the current: tlie absence of hydrostatic pres- 
sui-e on the banks, as the volume diminishes, allows them to become 
more dry and absorbent. These concurrent causes may account for 
the disappearance of this large proportion, and may suggest, besides 
the expedient of covering them from the sun, others by which they 
may be adapted to our climate. 
Tiic construction of dams of great elevation, though by no means 
impracticable, are not only attended with expense and difficulty, but 
with some uncertainty. That of the St. Ferreol reservoir of Lan- 
guedoc, is 110 feet in height; in masonry, parallel walls of great 
thickness, filled between with earth. But it is stated, as the result of 
experience in dams of masonry, that they are found too often to re- 
quire repairs. In Scotland, preference is given to building a puddle 
wall in the centre of an embankment. There is one of this kind, 90 
feet high, at Glencorreburn, near Edinburg. In building them, the 
course of the stream must be diverted. Puddle is any tenacious 
earth, compacted under water, by which the air is excluded from it — ■ 
the particles of the mass are afterwards kept in contact by ti»e weight 
of the atmosphere. 
The uncertainty of success arises principally from the hydrostatic 
pressure, in any deej) artificial water, which sometimes occasions 
small leaks secretly to spring out, perhaps through fissures of the 
rock, if on lime stone; or, by the porosity of the earth, discharging 
much water by small and imperceptible openings. Reservoirs are, 
of course, experimental, in paoportion to their magnitude; but most 
likely to succeed where the ground has been occasionally or annually 
flowed. 
We have shown that the loss by evaporation more than rain, may, 
in summer, be one foot. Du Cros, (a writer on these subjects,) states 
the absorption on canals to be one and an half times the evaporation. 
In this ratio, we must allow for a reservoir, were it of the same 
depth only; according to the tables for the summer season, 31.45 
inches, plus 15.77, but as a di'ougbt for sixty days would be so ex- 
traordinary as to preclude the navigation of the western rivers — we 
may safely assume the results of iJalton's experiments which were 
12 inches for sixty days, and allow the filtration to be 2 feet. This 
addition has been therefore made on the elevations of the dams. 
It has been shown, that, by these means and methods, there may be 
water enough for the longest, but lowest, line. 
