As these tests represent widely varying conditions it is clear that there is a trustworthy connection between the conditions of refrigeration and the power required for it in any case.1 The algebraic expression of these values is R = 1 + 4.15 T1-T2/T.

The relation between temperature range and power requirements for a given quantity of heat having been determined, we need only Fig. 241. The coefficient of T1-T2/T1 based on these results becomes 3.7 instead of 4.15, corresponding to a smaller indicated power than shown in the earlier results. The difference is not great for the conditions of blast refrigeration and makes power estimates based on the subsequent curves in this quotation somewhat high and correspondingly safe. Tests on both wet and dry compression were given in the original paper but now the dry is so much superior that the wet compression tests have been dropped and allusions to them omitted.

1 Although this seems such an obvious means of determining the actual power requirements in any case, when the general conditions are known, it is entirely new, and has never been used by refrigerating engineers, so far as known to me.

In the original paper reference was made to the work of Professor Denton in New York and Professor Schroter in Munich on tests of the power consumption of refrigerating machines, but a much more extensive series of tests have since been conducted by the York Manufacturing Company, under the direction of Mr. Thomas Shipley, its general manager, and the results of these tests have been collected and plotted in a paper on " The Power Required for Refrigeration" (Metallurgical and Chemical Engineering, December, 1912) and the diagram giving these results is substituted as to know in addition the quantity of heat to be removed to determine the actual horse-power required.

The quantity of blast used being commonly measured in thousands of cubic feet per minute, the quantity of heat required to be removed per 1000 cu. ft. of blast is obviously the most convenient basis, but we are at once confronted with the difficulty that, as we change the temperature of the air, we change the volume of a given weight as well, 1000 cu. ft. at 70° F., becoming 900 cu. ft. at 21° F.; so that in considering the quantity of water-vapor contained in the blast as affected by refrigeration we not only diminish it by the fact that a cubic foot of space will contain less moisture at 21° F. than at 70° F., for instance, but that there are only nine-tenths as many cubic feet for a given weight of blast.

Fig. 241. Chart showing ration of actual to theoretical horse-power required for refrigeration with different temperature ranges.

To overcome this difficulty the standard temperature at which blast is measured is taken at 70° F., at which 1000 cubic feet of air weigh exactly 75 lbs., and all calculations are based on 75 lbs. of air, correction being made for variation in volume. For the convenience of those desiring to make calculations on blast drying, a diagram (Fig. 242) is given herewith, which gives various relations of this standard quantity of air, based on its temperature.

The lower half of the diagram contains three curves whose abscissae are temperatures, and their ordinates quantities of heat above 0° F.

The straight line I shows the sensible heat, above that at 0° F., of 75 lbs. of air at any given temperature.

The upper curve gives the quantity of heat present above 0° F., in the water-vapor which will saturate 75 lbs. of air at different temperatures.

The latter quantity is not that given in steam tables as the "total heat of steam," but represents as nearly as may be the heat which would require to be removed by refrigeration in order to reduce the vapor present in the air to the quantity sufficient for its saturation at 0° F.

This is made up as follows, for temperatures up to the freezing point:

1. The sensible heat of the ice above 0° F. (the specific heat taken as 0.50).

2. The latent heat of freezing.

3. The heat of vaporization.

These are all to be taken for the given temperature and the quantity of moisture actually present in 75 lbs. of air at that temperature.

Above the freezing point the conditions are altered; for part of the water, as soon as condensed, will not wait to be cooled down to the temperature of the refrigerating coils, but will drip off them soon after its deposition. As a fair estimate, therefore, one-half of the sensible heat (above 32° F.) for all the vapor in excess of that present at freezing is added to the other items.

This is shown on the diagram as curve II.

The lower curve III is derived by adding the ordinates of the first two and represents, therefore, all the heat requiring to be removed to reduce the air from saturation at a higher temperature to saturation at 0°.

The small quantities of heat A Q, of which these larger quantities are made up, are each removed at a different temperature, and therefore correspond to a different T1, in formula,(I); hence these total quantities cannot be applied directly in that formula, but the latter must be put into the form HP = R(778/33,000) ∆ (T1-T2/T2) and integrated in order to give numerical results. This would be so tedious a process as to be useless if done analytically, but may be done conveniently by the application of the entropy diagram invented by Professor J. Willard Gibbs of Yale University many years since, and brought to the attention of the engineering world at large some fifteen years ago by Mr. Macfarlane Gray of England.