Omitting the subject of the pumps, and passing on at once to the discharge main, the author may first point out that the distinction between the ascending and descending mains of the system is of no importance, for two reasons: first, that nothing prevents the motors being supplied direct from the first alone; and second, that the one is not always distinct from the other. In fact, the reservoir may be connected by a single branch pipe with the system which goes from the pumps to the motors; it may even be placed at the extreme end of this system beyond the motors, provided always that the supply pipe is taken into it at the bottom. The same formula may be adopted for the loss of initial pressure in water pipes as for compressed air pipes, viz., ; h being the difference of level between the two ends of the portion of conduit of length, L, and the sign + or - being used according as the conduit rises or falls. The specific weight, δ, is constant, and the quotients, p/δ and p/δ, represent the heights, z and z, to which the water could rise above the pipes, in vertical tubes branching from it, at the beginning and end of the transit. The values assigned to the coefficient b in France, are those determined by D'Arcy. For new cast-iron pipes he gives b - 0.0002535 + 1/D 0.000000647; and recommends that this value should be doubled, to allow for the rust and incrustation which more or less form inside the pipes during use. The determination of this coefficient has been made from experiments where the pressure has not exceeded four atmospheres; within these limits the value of the coefficient, as is generally admitted, is independent of the pressure. The experiments made by M. Barret, on the pressure pipes of the accumulator at the Marseilles docks, seem to indicate that the loss of pressure would be greater for high pressures, everything else being equal. This pipe, having a diameter of 0.127 m. (5 in.), was subjected to an initial pressure of 52 atmospheres. The author gives below the results obtained for a straight length 320 m. (1050 ft) long; and has placed beside them the results which D'Arcy's formula would give.

``` Loss of head, in meters or ft. respectively

per 100 meters or ft. run of pipes.

+-----------------^-------------------+

| |

Calculated loss.

+-----------^-----------+

| |

Velocity of flow Actual loss

per second. observed. Old pipes. New pipes.

Meters. Feet. Met. or Ft. Met. or Ft. Met. or Ft.

0.25 0.82 1.5 0.12 0.06

0.50 1.64 2.5 0.48 0.24

0.75 2.46 3.7 1.08 0.54

1.00 3.28 5.5 1.92 0.96

1.25 4.10 6.1 3.00 1.50

1.50 4.92 7.3 4.32 2.16

1.75 5.74 8.0 5.88 2.94

2.00 6.56 10.2 7.68 3.84

2.25 7.38 11.7 9.72 4.86

2.50 8.20 14.0 12.00 6.00 ```

Moreover, these results would appear to indicate a different law from that which is expressed by the formula b u2, as is easy to see by representing them graphically. It would be very desirable that fresh experiments should be made on water pipes at high pressure, and of various diameters. Of machines worked by water pressure the author proposes to refer only to two which appear to him in every respect the most practical and advantageous. One is the piston machine of M. Albert Schmid, engineer at Zurich. The cylinder is oscillating, and the distribution is effected, without an eccentric, by the relative motion of two spherical surfaces fitted one against the other, and having the axis of oscillation for a common axis. The convex surface, which is movable and forms part of the cylinder, serves as a port face, and has two ports in it communicating with the two ends of the cylinder. The concave surface, which is fixed and plays the part of a slide valve, contains three openings, the two outer ones serving to admit the pressure water, and the middle one to discharge the water after it has exerted its pressure. The piston has no packing. Its surface of contact has two circumferential grooves, which produce a sort of water packing acting by adhesion. A small air chamber is connected with the inlet pipe, and serves to deaden the shocks. This engine is often made with two cylinders, having their cranks at right angles.

The other engine, which is much less used, is a turbine on Girard's system, with a horizontal axis and partial admission, exactly resembling in miniature those which work in the hydraulic factory of St. Maur, near Paris. The water is introduced by means of a distributer, which is fitted in the interior of the turbine chamber, and occupies a certain portion of its circumference. This turbine has a lower efficiency than Schmid's machine, and is less suitable for high pressures; but it possesses this advantage over it, that by regulating the amount of opening of the distributer, and consequently the quantity of water admitted, the force can be altered without altering the velocity of rotation. As it admits of great speeds, it could be usefully employed direct, without the interposition of spur wheels or belts for driving magneto-electric machines employed for the production of light, for electrotyping, etc.

In compressed air machines the losses of pressure due to incomplete expansion, cooling, and waste spaces, play an important part. In water pressure machines loss does not occur from these causes, on account of the incompressibility of the liquid, but the frictions of the parts are the principal causes of loss of power. It would be advisable to ascertain whether, as regards this point, high or low pressures are the most advantageous. Theoretical considerations would lead the author to imagine that for a piston machine low pressures are preferable. In conclusion, the following table gives the efficiencies of a Girard turbine, constructed by Messrs. Escher Wyss & Co., of Zurich, and of a Schmid machine, as measured by Professor Fliegnor, in 1871: