D (Fig. 1), attracting the armature, and drawing the disk forward sufficiently for the roller at I to pass over the center of one of the pins, and so drop in between that and the next pin, thus completing the motion, and holding the disk steadily opposite the figure.  This action takes place on any meter completing a unit of measurement of (say) 1,000 cubic feet, at which point the contact makers touch.  But suppose one meter should be moving very slowly, and so retaining contact for some time, while other meters were working rapidly; the armature at D would then be held up to the magnet by the prolonged contact maintained by the slow moving meter, and so prevent the quick working meters from actuating it; and they would therefore pass the contact points without recording.  A meter might also stop dead at the point of contact on shutting off the air, and so hold up the armature; thus preventing others from acting.  To obviate this, we apply the disengaging apparatus shown at L (Fig. 4).  The contact maker works on the center, m, having an armature on its opposite end.  On contact being made, at the same time that the magnet, D, is operated, the one at L is also operated, attracting the armature, and throwing over the end of the contact maker, l, on to the non-conducting side of the pin on the disk.  Thus the whole movement is rendered practically instantaneous, and the magnet at D is set at liberty for the next operation.  A resistance can be interposed at L, if necessary, to regulate the period of the operation.  The whole of the meters work the common dial shown in Fig. 1, on which the gross results only are recorded; and this is all we want to know in this way.  The action is so rapid, owing to the use of the magnetic disengaging gear, that the chances of two or more meters making contact at the same moment are rendered extremely small.  Should such a thing happen, it would not matter, as it is only approximate results that we require in this case; and the error, if any, would add to the apparent amount of leakage, and so be on the right side.  Of course, the record of each consumer’s meter would be taken by the inspector at the end of every quarter, in order to make out the bill; and the totals thus obtained would be checked by the gross results indicated by the main dial.  In this way, by a comparison of these results, a coefficient would soon be arrived at, by which the daily recorded results could be corrected to an extremely accurate measurement.  At the end of the working day, the engineer has merely to take down from the dial in his office the total record of air measured to the consumers, also the output of air from the compressors, which he ascertains by means of a continuous counter on the engines, and the difference between the two will represent the loss.  If the loss is trifling, he will pass it over; if serious, he will send out his inspectors to trace it.  Thus there could be no long continued leakage, misuse, or robbery of the air, without the company becoming aware of the fact, and so being enabled to take measures to stop or prevent it.  The foregoing are absolutely essential adjuncts to any scheme of public motive power supply by compressed air, without which we should be working in the dark, and could never be sure whether the company were losing or making money.  With them, we know where we are and what we are doing.