an ascending pipe, passing through this, conveys the oil up to the wick. Now, not only is the spring weaker as it expands with the descent of the piston, but the vertical height it has to raise the oil also increases: thus, if were not for the contrivance about to be described, we should have the brilliancy of the light continually lessening as the rate of supply of oil to the wick diminished. This difficulty the celebrated James Watt did not quite surmount when he turned his attention to this subject (Vide Life, pp. 462-465).

A straight wire, or rod, is placed concentrically within the ascending pipe, of a diameter but little less than it at its upper and thickest part, and long enough to enter the movable pipe when at its lowest position; the lower part is only a support. The oil in rising is compelled to pass through the narrow annular space between the interior of the moveable pipe attached to the piston and the rod (or moderateur); from this it results that it meets with a resistance which causes its upward movement to be very slow. Now as the moveable pipe and piston descend, the same length of the moderateur is not always engaged in the pipe: at first, when the spring is strongest, and the height that the oil has to rise is least, then also the annular passage is longest, and the resistance to the ascent of the oil greatest; and again, when the piston has descended, and consequently the spring is weaker, and the height the oil has to be raised greater, so also the length of this annular space is less, and the resistance to the ascent of the oil diminished in proportion as the ascending force itself is diminished. By a tentative process in each particular case-filing a portion into a flat surface-the needle is adjusted so as to give a uniform supply of oil, and make the lamp burn with equable light as long as the spring acts. This principle is evidently applicable to the discharge of water by simple modifications.

The "Module" adopted on the canal of Isabella II. is shown in Figs. 46, 47, 48, 49, taken from the work of Lieut. Scott Moncrieff. It consists of a float, M, and a plug, N, suspended from it, which works in a circular orifice in a plate set at the level of the bottom of the channel.

ELEVATION.

All being contained in a rectangular well of masonry 3.28 ft. by 3.94 and 4.16 feet deep, communicating with the main channel by a lateral opening having an iron grating in front, and covered by a locked iron trap-door, to prevent all tampering with it. The float M, which is formed of brass plate, is shown in plan in Fig. 47, and in elevavation, with plug attached, in the upper woodcut, Fig. 46. The surface of the water, and, therefore, the apparatus which rises and falls with it, being supposed at its highest level, and one meter in depth. It is also shown in Fig. 48, at a larger scale, in a vertical central section through AB in Fig. 47, and two of the three central supports are given, carrying a central disc, through which passes the screwed end of the rod that the plug is suspended from; a butterfly nut on the

Fig. 46.

PLAN.

Fig. 47.

top enables the whole to be adjusted. The plug and the plate, in which is the orifice for the outlet of the water, are of bronze to avoid rust.

C

D.

E

M

100

The water entering laterally from the canal passes down through the annular space between the plug and plate. From the form of the plug it is evident that this space increases as the level of the water is lowered, and if the area of the annular opening be inversely proportional to the square root of the "charge" or depth of the canal above the bronze plate, then we should have a constant discharge under all the variations in the level of the water flowing down the canal. In Fig. 49 we have represented four horizontal sections taken at the corresponding numbers on the right-hand side in Fig. 48; if the water surface were to descend until the horizontal dotted line at 2 reached the level of the orifice,

F

G

H.

0.20

Fig. 48.

then the area for the discharge would be the annular

Fig. 49.

space shown at the corresponding number in Fig. 49;

as the outer circle, representing the circular orifice, is constant, the area for discharge evidently increases as the plug descends.

The objection to this module is, that it involves a considerable loss of head, the level of the water in the culvert, flowing off to the irrigation channels, must be lower than that in the canal of supply by, at least, double the height of the plug, so that it would be inapplicable to the great plains of India, where every inch of level has to be economized. The diameter of the orifice in Fig. 48, which shows the float and plug at a larger scale, is 0.20 metre (= 7.874 ins.), and that of the plug at the same level, when in its highest position, is o.1653 metre (= 6.51 ins.), so that the area of the annular space in this case is 15.436 sq. ins.; and, as the coefficient of contraction, from experiment, was found to be 0.63, we have 0.63 x 15.436 = 9.72 square inches = m S, or 0.0675 square feet, which is the opening at 4 in Fig. 49, and the depth of the water over it, or the charge, is 1 metre (= 3.281 ft.). Had the water surface lowered until G, at 3 on the plug, coincided with the orifice at I, then the annular opening is that shown at 3, Fig. 49; at this point the diameter of the plug is 0.1554 ft.

Let D represent the diameter of the orifice in the bronze plate fixed in the bottom of the chamber,

And H the depth of water over it when the main channel is running full.

Let d, represent the diameter of the plug at the base,

and

d3, da, &c., the respective diameters at the several points so numbered, Fig. 48, between the base and vertex of the plug, when

h1, h,, &c., are the corresponding depths estimated from the lowered surface of the channel, these

two quantities being so related to each other that the increased area of the annular space may compensate for the diminished charge and give a constant discharge.

From experiments it was found that the coefficient of contraction in this case was 0.63. Let Q represent the unaltered quantity which it is desired to discharge at every different level of the water; then to compute d the diameter of the plug at the base when the charge is H, we have—

Hence, if we assign successive values to either h, or d1, we obtain the value of the other. If h, be given, we have

The following Table gives the dimensions, in metrès, of one of these modules from Lieut. Moncrieff's work. It may here be stated that an identical regulator for the flow of gas has long been in use in this country.