(B) Graduated scales attached to the water glasses. (C) Tanks and platform scales for weighing water (or water meters calibrated in place). Wherever practicable the feed water should be weighed, especially for guarantee tests. The most satisfactory and reliable apparatus for this purpose consists of one or more tanks each placed on platform scales, these being elevated a sufficient distance above the floor to empty into a receiving tank placed below, the latter being connected to the feed pump. Where only one weighing tank is used the receiving tank should be of larger size than the weighing tank, to afford sufficient reserve supply to the pump while the upper tank is filling. If a single weighing tank is used it should preferably be of such capacity as to require emptying not oftener than every 5 minutes. If two or more are used the intervals between successive emptyings should not be less than 3 minutes. (D) Pressure gauges, thermometers, and draft gauges. (E) Calorimeters for determining the calorific value of fuel and the quality of steam. (F) Furnaces pyrometers. (G) Gas analyzing apparatus. 5. OPERATING CONDITIONS Determine what the operating conditions and method of firing should be to conform to the object in view, and see that they prevail throughout the trial, as nearly as possible. Where uniformity in the rate of evaporation is required, arrangement can be usually made to dispose of the steam so that this result can be attained. In a single boiler it may be accomplished by discharging steam through a waste pipe and regulating the amount by means of a valve. In a battery of boilers, in which only one is tested, the draft may be regulated on the remaining boilers to meet the varying demands for steam, leaving the test boiler to work under a steady rate of evaporation. 6. DURATION The duration of tests to determine the efficiency of a hand-fired boiler, should be 10 hours of continuous running, or such time as may be required to burn a total of 250 pounds of coal per square foot of grate. In the case of a boiler using a mechanical stoker, the duration, where practicable, should be at least 24 hours. If the stoker is of a type that permits the quantity and condition of the fuel bed at beginning and end of the test to be accurately estimated, the duration may be reduced to 10 hours, or such time as may be required to burn the above noted total of 250 pounds per square foot.