size, and which offered a large resistance with a corresponding loss
of pressure; on the line there were also 23 siphons and 42 stop
valves.

These trials were repeated several times to secure accuracy, and the
speed of the air was brought to 49 ft. a second. The results obtained
in one of these trials may be taken as an example. The main between
the Rue St. Fargeau and the Fontaine au Roi, on which there are no
collecting reservoirs, but three siphons and eight stop valves, gave,
with an average speed of 21 ft. 3 in., a loss in pressure of 0.05
atmosphere for each kilometer of main.

From these experiments it would appear that, assuming a speed of 21
ft. per second, a loss in pressure of one atmosphere would correspond
to a distance of 20 kilometers; that is to say, a central station
could extend its mains on all sides with a radius of 20 kilometers,
and the motors at the ends of the lines would receive the air at a
pressure 15 lb. less than at the central station. Professor Riedler
states that as an actually measured result, the velocity of the air
through the mains of the St. Fargeau system is 19 ft. 8 in. per
second, and that the loss in pressure per kilometer is 0.07
atmosphere. From this it follows that including the resistances due to
the four reservoirs, and other obstructions actually existing, an
allowance of one atmosphere loss on a 14 kilometer radius is ample. By
increasing the initial pressure of the air, much better results can be
obtained, and future attention in practice should be devoted to this
point. The amount of work required to compress air does not increase
in the same ratio as the pressure, and for this reason considerable
economy can be effected at the first stage, and the loss in the mains
will be reduced.