It remains, then, to consider the error of the meter. The custom is, in New York at least, not to set a meter that registers fast--that registers a greater volume of gas than actually passes through it. If it is found to be slow, however, and not more than three per cent., it is allowed to go out. As a result, the meter, when first placed, always favors the consumer, sometimes to the extent of recording only ninety-seven feet of gas for each one hundred feet actually passed. Owing to the aging of the mechanism and the drying out of the leathers, there exists a tendency to increase the registry for each cubic foot passed. In this way a slow meter may become a fast meter after a period of active service. From the meager data at my disposal, it would appear that every meter should be called in for a thorough overhauling and readjustment at periodic intervals of from three to five years. Assuming that there are several million gas meters in Greater New York alone, it is but natural to expect that out of this vast number, in spite of any reasonable care that may have been exercised in their adjustment originally, many will be found subsequently to be defective--some because of mechanical injury, some through sheer old age. Unfortunately, it is not possible as yet to obtain a convincingly large array of figures; but in the Borough of Brooklyn, where there are in service nearly a quarter of a million meters, and where complaints against them have been studiously encouraged by the authorities, one hundred and eighty-seven meters have been carefully tested. Here are the results: 21 correct { 3 more than 10 per cent. 114 fast, average 3 per cent { 42 between 3 and 10 per cent. (recording 103 cubic feet for { 69 less than 3 per cent. each 100 cubic feet actually { --- passed) { 114 { 0 more than 10 per cent. 52 slow, average 2¼ per cent { 13 between 3 and 10 per cent. (recording 97¾ cubic feet { 39 less than 3 per cent. for each 100 cubic feet { --- actually passed) { 52 --- 187