be of lower temperature at the outlet of the meter, subtract the product from the gasholder reading. Ex. :

86. A simpler, but less accurate, method of correcting for temperature, is as follows:

RULE.-With gas issuing from the meter at a higher temperature than from the holder, subtract the per-centage difference in volume from the per-centage of error as per Table, and when the gas issues from the meter at a lower temperature, add the per-centage difference to the percentage error of the Table.

Ex.:

87. The illustrations which have been given show the importance of due regard to the maintenance of equality in temperatures in meter testing, or for somewhat careful observations as regards differences. Sufficient care, up to the present time, has certainly not been generally exercised, and many a correct meter has, from this cause, been made to appear inaccurate. The Author has had many meters which had been tested by other persons, and pronounced by them to be in error, but which careful testing under his hands proved to be absolutely correct.

THE TEST-METER.

88. Place the test-meter on a tolerably level and firm stand or board, as in fig. 20, as near to the meter to be tested as may be convenient. Open the cocks, A, A, B, B', of all the gauges. Insert a funnel into the charge-socket

on the top of the test-meter, and if the meter has three gauges pour in water until it reaches very nearly to the marked water-lines, then by means of the screws in the feet adjust to a perfect level and add more water until in each gauge the lowest part

of the meniscus or cup formed by the water coincides with the water indicator. If the meter is fitted with spirit levels (26) and has only one gauge it may be levelled first and then be charged with water, but the spirit levels should be examined after so charging with water and a readjustment made if necessary; should too much water have been put in, the excess must be drawn out through the cock in the lower part of the meter case. Insert the thermometer in its socket (25), and the meter is ready for use.

89. If the meter is to be used at or below its nominal rate of working (120 cubic feet per hour, or 2 cubic feet per minute for a 20-light, and 600 cubic feet per hour, or 10 cubic feet per minute for a 100-light), the cocks A, A, B, B', of all the gauges must be open, and the cock C, on the tube connecting the back gauge to the inlet, must be closed, but if the meter is to be worked faster than its normal speed, close the cock A' of the back gauge and open C, and while the meter is working at the speed desired (38), add more water until the level in the back gauge coincides with the "working level." This addition of water can easily be made through a seal or syphon-funnel, screwed into the charge socket of the meter; or, better, the meter may be made with an elbow joint near its base to receive an upright pipe with a funnel on the top, after the manner which is usual with station meters. (W fig. 20).

90. A little practice is needful to make an adjustment of the level while the meter is working, owing to the oscillation of the water, but there is no practical difficulty, and a 100-light may be so adjusted in a very few minutes. The true line is fixed upon by observing that the range of oscillation is equal on each side, that is, above and below the working line.

*Take great care not to have A' and C of the back gauge open together, or gas will pass from the inlet into the case, and thence unmeasured from the outlet of the meter.

91. As a rule, it will be found most convenient to adjust the water-level when the meter is not in action, and then to work it near to its normal speed. As is shown in paragraph 32, et seq., a considerable variation in the rate of speed is needful to induce an error of one per cent. Some experiments with a 20-light test-meter, showed that when it was worked at half speed (60 feet per hour) it registered 4th per cent. more, and when worked at double speed (240 feet per hour) one-fifth per cent. less, gas than it delivered. The difference in a 100-light would no doubt be nearly or quite half per cent. each way.

TESTING OPERATIONS.

92. There are three conditions under which a meter may be presented for verification by the aid of a test-meter.

1. Fixed in position for use, complete with service pipes, but having no gas delivered into it.

2. Fixed and in actual use.

3. Unfixed and standing in a yard or shop.

93. Dealing with the first condition, and assuming that it is a station-meter which is to be tested, the following description of the diagram, fig. 20, and the accompanying remarks will render the arrangement intelligible. A represents the back of the test-meter, B the back of the station-meter to be tested, C is the inlet pipe of the later with a valve at D. The outlet pipe is E, having a valve at F, while G is another valve to serve as a byepass. All these valves are supposed to be, and must be, closed when the test is made. H is a pipe connecting the inlet pipe, below the valve D, to the inlet of the test-meter, but the gas may, of course, be obtained from any other convenient source of supply. K is a pipe connecting the outlet of the test-meter to the inlet of the station-meter at a point above the valve D. The pipes H, K, may be of metal or vulcanized rubber, the latter