108 KENNELLY-TAYLOR— EXPLORATIONS OVER [April 22, 



resonant frequency, the amplitude of vibration, shown by the ex- 

 plorer, fell off very markedly. The curve of relative amplitude at 

 different frequencies is indicated in Fig. 9. It will be seen that 

 when exciting the diaphragm with vibrations remote from the 

 resonant frequency in either direction, the amplitude becomes so 

 small that the degree of precision which may be obtainable near 

 resonance is impossible to secure. The outline theory for this 

 resonance curve. Fig. 9, is given in Appendix II. It is shown that 

 if we multiply the successive ordinates by oj^ 271-11, the resulting 

 velocity-values correspond to vector chords on a certain velocity 

 circle. 



Fig. iB oi Appendix I. gives the graph of the explored vibration 

 amplitudes, at successive radial distances from the center of dia- 

 phragm No. I, for the frequency 896'--'. It will be seen that the 

 amplitude falls off smoothly from a maximum at or near the center 

 (r = o), to zero at the fiat-clamped edge (r = 2.62). The applica- 

 tion of Rayleigh's theory of free vibration to these curves is given 

 in Appendix I. In general, the agreement between the acoustically 

 forced amplitudes and theoretically computed free amplitudes was 

 satisfactory. 



At or near the resonant frequency, or natural frequency of a 

 diaphragm, especially when its damping coefficient is small, so that 

 the resonance is sharp, a small change either in impressed frequency, 

 or in the constants of the diaphragm due to change of temperature, 

 may have an appreciable influence upon the amplitude of vibration. 

 In other words, although the observed amplitudes are relatively 

 large, and the precision of measurement is seemingly high, yet the 

 system is in a virtually unstable condition. Consequently, although 

 there is no reason to suppose that the conditions at resonance differ 

 from those off resonance, nevertheless, when a reliable and repro- 

 ducible set of observations of ampHtude distribution is desired, it is 

 advisable to select a frequency not too close to resonance, or say of 

 about half the resonant amplitude. 



