332 PROCEEDINGS OF THE AMERICAN ACADEMY 



its coming to rest relatively to the last recorded swing, it was easy 

 to calculate the number of revolutions per second corresponding to a 

 given deflection. 



Taking different distances along the bar, as well as different inclina- 

 tions of the two commutators, over a thousand determinations were 

 made in the course of a few days. The result was in general a con- 

 firmation of the previous and more exact experiments. 



In many cases an actual reversal of the current took place during 

 the period of stopping of the machine ; but of course when the incli- 

 nation of the commutators was such that a reversal would indicate a 

 retardation of more than the maximum found before, no such reversal 

 took place. 



These experiments serve, moreover, nicely to illustrate the fact 

 upon which all thorough understanding of the above phenomena is 

 based, namely, that magnetism requires a fixed time to reach a fixed 

 value, and a greater time, proportionally, for greater distances from 

 the magnetizing coil ; so that, with even the moderate velocities em- 

 ployed (20 to 30 reversals per second), it has not time enough to 

 reach its maximum even within the primary itself between two suc- 

 cessive reversals ; and for still shorter intervals of time reaches a 

 proportionally less value. 



This is proved by the fact that, when the commutators are adjusted 

 to best advantage, the deflection of the galvanometer is independent 

 of the speed of the commutator within certain limits, as 5 to 12 revo- 

 lutions, or 10 to 24 reversals per second, after which it falls off a 

 little ; gradually diminishing more and more nearly in proportion to 

 the increase in velocity, until, at 140 reversals, it is reduced to a 

 quarter or a thirtieth of its maximum value, the diminution being most 

 marked for distances moderately remote. For, if each wave had time 

 to reach its maximum, the deflection, instead of diminishing, would in- 

 crease in direct proportion to this velocity, as is indeed the case when 

 the velocity is only a few turns per second. The actual results are con- 

 tained in the Table of Galvanometer Deflectious, which is appended. 



It is evident that, by dividing the deflections contained in this table 

 by the corresponding number of revolutions of the commutator, we 

 obtain a series of numbers proportional to the actual height of each 

 wave at the end of the short intervals of time during which the cur- 

 rent acted. In the table below, the first column shows the duration 

 of each current in terms of an arbitrary unit ; the second column 

 shows the corresponding height of the wave at a distance of two 

 inches from the primary, expressed in percentages of that height 



