ACOUSTICS AND GRAVITATION. 59 



apparatus, for the distance and rate at which waves in the plate travel in two 

 diametral directions at right angles to each other will not probably be quite the 

 same. The hard-rubber casing, for instance, expands and contracts with tem- 

 perature much in excess of the iron plate. It is not probable that in such a 

 case there would be concentric and rigorously symmetric compensation. 



From this point of view, oscillations of the plate about a vertical axis are to 

 be associated with the lines, oscillations about a horizontal axis with the 

 shadow waves or again with gaps , if either are present . Hence , if the telephone 

 is rotated 90 on its axis (normal to the plate) the relations should be ex- 

 changed. It is in this way that figure 82 was obtained from figure 81. 



Waves which open out or develop from the cuspidal to the sinuous form 

 might possibly be called shock-waves. They result from the successive 

 magnetic shocks which the plate experiences. Their average period can not be 

 estimated even as high as a thousandth of a second. They are to be associated 

 with the free vibrations of the telephone plate, in the interval (T = O.I second , 

 o.o i second) between the inductions. 



53. Effect of temperature. Miscellaneous observations. A result with a 

 bearing on temperature effects was obtained incidentally. The telephone 

 having been adjusted for the phenomena, figures 81 and 82 were left at rest 

 with the window of the room open on a cold winter day. After adjusting the 

 interferometer for the temperature change, very large fringes appeared, which 

 when the telephone sounded with moderate loudness (20,000 to 10,000 ohms in 

 circuit) remained flawless fringe bands throughout. In other words, there was 

 no suggestion of motion of the plate of any kind. Not until all but a few 

 thousand ohms were withdrawn from the circuit and the telephone spoke 

 harshly, was a wave-form apparent. Sharply circumscribed patches of jagged 

 waves entered the field of the vibrating telescope on the right and left, crossed 

 at the center, and moved out of the field at the opposite sides. Apart from 

 these interruptions, recurring regularly every few seconds, the fringes remained 

 straight and in place, giving no evidence whatever of motion in the plate. The 

 same phenomenon, only more marked, was obtained incidentally on the follow- 

 ing day, in a cold room. In place of the patches, there were mere V-shaped 

 fringe-band incisions, moving in pairs, in opposite directions and meeting at 

 the center. These again were wholly obliterated when 10,000 or 20,000 ohms 

 were put in circuit. On warming the room, however, they did not essentially 

 change. In fact, the phenomenon was not controllable and the quiet fringes 

 passed in the lapse of time erratically into very different forms. 



Among these, the discontinuous or faulted fringes, figures 85 and 86, as 

 they may be called from the appearance given to the fringe band, are the most 

 interesting. In this case the phenomena of figures 8 1 and 82 , with their waves, 

 etc., occur as before; but the even horizontal band separates into sections w, 

 w', respectively, high and low or the reverse, and terminating in very sheer 

 lines of fault /, /', f", f'" . The sections w' may be crossed by vertical lines s, 

 or these may be absent. Finally, the sections w and w' are impulsively in 



