ence, due to the material and thickness of the plates used, but when passing from air to water this difference wholly disappeared. In passing from steam to water the rate of transmission increased very rapidly with increase in difference of temperature.

HEAT TRANSMITTED IN THERMAL UNITS FOR EACH SQUARE FOOT PER HOUR AND PER DEGREE DIFFERENCE OF TEMPERATURE.

48. Methods of Testing Radiators.-So far as the writer knows, no standard method has been adopted for use in the testing of radiators, and while numerous tests have been made by different engineers and experimenters, they are often not concordant either as to the method of testing or as to the results obtained. The results in the testing of radiators are greatly affected by small variations in temperature, by irregular air-currents, and by the amount of moisture contained originally in the steam. Obscure conditions of little apparent importance and often disregarded greatly influence the results. The heat emitted by the radiator is in all cases to be computed by taking the difference between that received and that discharged. This result is accurate, and easily obtained. This heat is utilized in warming the air and objects in the room, and to supply losses from various causes, which take place constantly, and is diffused so rapidly, and used in so many

ways, that it is practically impossible to measure it, although it is, of course, equal to that which passes through the radiator. The radiating surface is almost invariably heated either by steam or by hot water. In the case of a steam radiator the heat received may be determined, by ascertaining the number of pounds of dry steam condensed in a given time, multiplying this by the heat contained in one pound of steam, and deducting from this product the weight of condensed water, multiplied by its temperature. To make a test of this kind with. accuracy requires, first, a knowledge of the amount of moisture contained in the original steam; second, the pressure of the steam or its temperature; third, an arrangement for permitting

water of condensation to escape from the radiator without the loss of steam, and means of accurately weighing this water, and also of determining its temperature. The radiator can be located in any desired position in the room, on the floor, or slightly elevated therefrom. The temperature of the room during the test should be maintained as nearly constant as possible, and no test should be less than from 3 to 5 hours in length. The method adopted by Mr. George H. Barrus in making a radiator test is shown in Fig. 29. The one adopted by the author, in many respects similar, is shown in Fig. 30.

In some recent tests of steam radiators made at Sibley College* the author adopted the following plan of operation

* See Transactions vol. i., American Society Heating and Ventilating Engineers.

for measurement of the heat discharged and for operating the radiators:

First, the steam supplied to the radiator to be passed through a separator and a reducing-valve to remove entrained water and maintains a constant pressure during any given run. Second, the amount of moisture in the steam to be measured by a calorimeter, and corrections made to the result for the entrained water. Third, the pressure and temperature of the steam in the radiator to be measured by accurate gauges and thermometers. Fourth, the amount of heat passing through the radiator to be obtained by weighing the condensed

steam, measuring its temperature, and computing by this means the heat discharged.

Fifth, the air from the radiator to be effectually removed. Large errors are caused by leaving varying amounts of air in the radiator. The ordinary air-valve is often very unsatisfactory for this purpose; if used, it must be closely watched, or the results may be seriously affected.

The heat supplied was computed by knowing the weight, the percentage of moisture, and the heat contained in one pound of steam. Various methods were tried for drawing off the condensed water: in some tests a trap was used, but better results were obtained by employing a water-column with gauge. glass and drawing off the water of condensation by hand, at

such a rate as to maintain a constant level in the glass. To prevent loss by evaporation, this water needs to be received either into a vessel containing some cold water, or else into one with a tight cover, the latter being generally preferred.

Methods of Testing Indirect Steam Radiators.- For this case the general methods of testing should be the same as those previously described, and in addition the volume of air which passes over the radiator should be measured; also, its temperature before and after passing the radiator. For measuring the velocity of air the most accurate instrument at present known is the anemometer, which has been described and illustrated in Article 30, page 37. In measuring the velocity the anemometer should be moved successively to all parts in the section of the flue, and the average of these results should be used. The velocity in feet per minute multiplied by

the area of section in square feet should give the number of cubic feet. The number of cubic feet of air heated can also be computed, as explained in Article 30, page 40, by dividing the heat emitted by the radiator by the prod uct of specific heat of air and increase in temperature.

The heat which is absorbed by the air can be computed by multiplying that required to raise one cubic foot one degree, as given in Table VIII, by the total number of cubic feet warmed multiplied by the increase in temperature. Fig. 31 shows an arrangement adopted by the author in testing indirect radiators, the air-supply being measured by an anemometer not shown.

Testing Hot-water Radiators.-The amount of heat transmitted through the surfaces of a hot-water radiator can be determined in either of two ways: first, by maintaining circula

tion at about the usual rate, measuring the temperature of the water before entering and after leaving the radiator; also, measuring or weighing the water transmitted. The heat transmitted would be equal in every case to the product of the weight of water, multiplied by the loss of temperature. In making these tests the same precautions as to removing the air from the radiator must be adopted as in testing steam radiators.

These radiators can also be tested by filling with water at any desired temperature and noting the time required for the water to cool one or more degrees. In this case the iron which composes the radiator would cool the same amount, and a correction must be added. The easier way to correct for the metal composing a radiator is to consider the weight as that of the water increased by that of the iron multiplied by its specific heat. The specific heat of wrought iron is, practically, I divided by 9; that of cast iron, I divided by 8; hence for a castiron radiator the effect would be the same as though we had an additional amount of water equal to of the weight of the radiator.

In the practical operation of this test the water in the radiator must be kept thoroughly agitated by some sort of stirring device.

49. Measurement of Radiating Surface. The amount of radiating surface is usually expressed in square feet, and the total surface is that which is exposed to the air, and includes all irregularities, metallic ornaments, etc., of the surface.

Where the surface is smooth and rectangular or cylindrical it is easily measured, but where it is covered with irregular projections the measurement is a matter of some difficulty and uncertainty. The only practical method of measuring irregular surface seems to be that of dividing it up into small areas and measuring each one of these areas separately by using a thick sheet of paper or a bit of cord, and carefully pressing it into every portion of the surface. The sum of all the small areas is equivalent to the total area.

This method is at best only approximate, and even when exercising the utmost care different observers are likely to differ three or four per cent in their results. The writer has tried several other methods of measuring surface, but so far without