386 PROCEEDINGS OF THE AMERICAN ACADEMY. 



the distance of the reflector from the pipe, in Curve 1 of Figure 5. 

 This curve shows the stationary wave system set up hy the interfer- 

 ence of the direct and the reflected waves. The distances between al- 

 ternate nodes and alternate loops of the curve give the following values 

 of the wave-length : 



49.7, 49., 45.8, 51, 46.5 ; Average, 48.4. 



The velocity of sound at the temperature of the room, 18°, was 34200 

 cm. per second, whence the period 



34200 ^^^^ 



^ = l8T-=''^^' 



while the actual value of the pitch of the pipe FJ 4 is 705 vibrations 

 per second. This agreement is evidently better than is to be expected 

 from the method, on account of the uncertainty of locating the nodes 

 and loops of the curve. 



It is seen, however, that the points of the stationary wave lie well on 

 the curve. A repetition of the observation on a succeeding day gave 

 substantial agreement with Curve 1. It is to be observed that the 

 first maximum, with the reflector in the neighborhood of 23.5 cm. from 

 the pipe, is weaker than the second and third maxima. This is prob- 

 ably caused by the fact that the wind-chest on which the pipe was 

 mounted intercepted the reflected wave more strongly when the re- 

 flector was close up than when it was more distant from the pipe. 



The horizontal dotted line through the curve at 3.30 gives the mag- 

 nitude of the current when the reflector was removed. It is seen that 

 the peaks of the curve above the line of no reflector are much greater 

 than the neighboring depressions of the curve below the line. This 

 distortion was found to be chiefly due to the current-voltage character- 

 istic of the rectifier, and is eliminated by the calibration of the recti- 

 fier with an alternating voltage, and by plotting the stationary wave 

 in terms of alternating voltage instead of galvanometer current. 



In making the substitution of voltage for current it would be in- 

 structive to impress the known alternating voltage on the primary of 

 Figure 2, and take the corresponding throws of the galvanometer in 

 the secondary. We should then be able to know the voltage generated 

 by the telephone when we know the galvanometer current. However, 

 on account of the influence of the transformer, this could be properly 

 done only with an alternating voltage of the same frequency as the 

 sound, in this case 705 cycles. A generator for this frequency was not 



