of 9.142 kilometers; to the whole system of mains, 16.5 kilometers; to the northern mains running from St. Fargeau to the Rue de Belleville, 1.4 kilometers; and from St. Fargeau to the Rue des Pyrenees, 6.5 kilometers.  It will be seen from the figures given in the table that the actual loss is small, and it is stated that this is due chiefly to the elastic joint employed throughout the system, excepting in the Rue de Belleville, where rigid couplings are used, and continual trouble is experienced from loss by leakage.  In all cases the losses given are the maximum, which only occur under the most unfavorable conditions.

It was found, during the first, second, and fourth tests, that considerable leakage occurred between the St. Fargeau central station and the Rue de Belleville.  During the trials two and four, an uncertain amount of loss occurred from the consumption of air required to work the pneumatic clocks, and also motors in the circuit, that could not be stopped.  The tests two and four include all losses in the service pipes, as well as the mains.

The production of compressed air at the central station is assumed at 30,000 cubic feet per hour (atmospheric pressure), and in all cases the loss in the mains is taken as a percentage of the total production.

The losses due to resistance in the mains were also examined with great care, over independent sections, as well as through the complete reseau.  During the early part of these trials, an unusual and excessive loss was recorded, the cause of which could not be at first ascertained.  At intervals along these mains are placed a number of water reservoirs which receive the water injected into the mains; in addition to these the direct flow of the air is interrupted by numerous siphons, the stop valves to branches, etc.  Investigation showed that the presence of these reservoirs created considerable resistance on account of an increased and subsequently reduced section.  The exact loss from this cause was, therefore, carefully measured, as well as the losses existing in the mains not so interrupted.  The results show that the loss by expansion at one reservoir, when the speed of the air flow was 23 ft. per second, was equal to 0.15 atmosphere; with a speed of 29 ft. 6 in. per second, it amounted to 0.2 atmosphere.

Therefore, the presence of five such reservoirs would cause a loss in pressure equal to one atmosphere.  This very undesirable arrangement is not repeated in the new system, the sumphs being connected in such a way as not to modify the section of the tube, nor consequently the pressure of the air.  The presence of the siphons and stop valves did not seem to affect the pressure to any measurable extent.  The following table contains a list of the more important mains tested, and it may be mentioned that the resistance, due to the reservoirs, was at first partially included.  The trials were carried out while the mains were not being drawn upon by subscribers.