A.—­That will not be difficult.  In low pressure engines the pressure in the boiler may be taken at 5 lbs. above the atmospheric pressure, or 20 lbs. altogether; and as high pressure steam is merely low pressure steam compressed into a smaller compass, the size of the feed pump in relation to the size of the cylinder must obviously vary in the direct proportion of the pressure; and if it be 1/240th of the capacity of the cylinder when the total pressure of the steam is 20 lbs., it must be 1/120th of the capacity of the cylinder when the pressure is 40 lbs. per square inch, or 25 lbs. per square inch above the atmospheric pressure.  This law of variation is expressed by the following rule:—­multiply the capacity of the cylinder in cubic inches by the total pressure of the steam in lbs. per square inch, or the pressure per square inch on the safety valve plus 15, and divide the product by 4,800; the quotient is the capacity of the feed pump in cubic inches, when the feed pump is single acting and the engine double acting.  If the feed pump be double acting, or the engine single acting, the capacity of the pump must just be one half of what is given by this rule.

337. Q.—­But should not some addition be made to the size of pump thus obtained if the pump works at a high rate of speed?

A.—­No; this rule makes allowance for defective action.  All pumps lift much less water than is due to the size of their barrels and the number of their strokes.  Moderately good pumps lose 50 per cent. of their theoretical effect, and bad pumps 80 per cent.

338. Q.—­To what is this loss of effect to be chiefly ascribed?

A.—­Mainly to the inertia of the water, which, if the pump piston be drawn up very rapidly, cannot follow it with sufficient rapidity; so that there may be a vacant space between the piston and the water; and at the return stroke the momentum of the water in the pipe expends itself in giving a reverse motion to the column of water approaching the pump.  Messrs. Kirchweger and Prusman, of Hanover, have investigated this subject by applying a revolving cock at the end of a pipe leading from an elevated cistern containing water, and the water escaped at every revolution of the cock in the same manner as if a pump were drawing it.  With a column of water of 17 feet, they found that at 80 revolutions of the cock per minute, the water delivered per minute by the cock was 9.45 gallons; but with 140 revolutions of the cock per minute, the water delivered per minute by the cock was only 5.42 gallons.  They subsequently applied an air vessel to the pipe beside the cock, when the discharge rose to 12.9 gallons per minute with 80 revolutions, and 18.28 gallons with 140 revolutions.  Air vessels should therefore be applied to the suction side of fast moving pumps, and this is now done with good results.

339. Q.—­What are the usual dimensions of the cold water pump of land engines?