474 



BELL SYSTEM TECHNICAL JOURNAL 



and from the rod is suspended a mirror m to indicate the amount of 

 twist. The air-pressure is adjusted to produce the maximum torcjue. 

 The outstanding defect of a receiver of this type is, that it imprints 

 its own characteristics upon the data. It will not respond efl'ecti\ely 

 to a wa\e-train not possessing a fretjuenry agreeing closi'ly with its 



h b 



^■i 



Fig. 6 — Radiometer and Radiometer \anes Used by Nichols and Tear. (Physical 



Review) 



own, or with some harmonic of its own; and has a tendency to ex- 

 aggerate the apparent proportion of such frequencies in a beam 

 which is a mixture of frequencies, as a damped wave-train is. In 

 general, the curve of receiver-reading versus mirror-displacement is 

 an unevenly wavy one which, when analyzed into components in 

 Fourier's manner, is found to contain at least two frequencies, one of 

 which is attributed to the receiver and the other to the radiation. On 

 the other hand, if a receiver having its natural frequency far away 

 from the expected periodicities of the waves is employed, it is found 

 too insensitive.^ Worse yet, if the wave-train pursuing the path 

 LiABM in Fig. .3 is a short heavih-damped one, while the natural 

 oscillations of the receiver are of comparativeh- low frequency and 

 slight decrement, the data will suggest that the wave-train is but 

 slightly damped and has the frequency of the receiver." 



'Thermal receivers having natural frequencies far below those of the incident 

 beams have been employed in studying wave-trains of much greater wavelengths 

 and much more intense than these. 



'This can l)e seen by considering an extreme case. Imagine that the wave is a 

 single infinitely thin pulse, while the natural oscillations of the receiver are quite 

 undani|x.'d. The pulse will l>e divided by the Boltzmann mirrors, so that two pulses 



