108 BELL SYSTEM TECHNICAL JOURNAL 



second, the reflected waves reaching the microphone at 12 feet from a 

 small source contribute much more to the microphone output than 

 does the direct sound. To illustrate the effect of these reflections, the 

 curve {b) in Fig. 105 has been plotted. It is based on the data of Fig. 

 10 and Fig. 11, and assumes that the transmitter is acted upon by 

 progressive plane waves arriving with equal intensity from all directions 

 in space. Their phases are taken to have random distribution. At 

 any one frequency the response of the transmitter is then proportional 

 to 



41 



U 



lA{d)J ■ sin e-dd, 



where A{^) is the azimuth factor taken from Fig. 11. The result is 

 seen to be intermediate between the pressure and the normal field 

 calibrations, for frequencies up to about 8,000 c.p.s. Under these 

 circumstances it is immaterial which way the diaphragm faces, but 

 this holds only for sustained sounds. For sounds of short duration, 

 the peak amplitudes in the microphone output often are of particular 

 interest. They will be more nearly given by that single field curve 

 corresponding to the azimuth with respect to the sound source in which 

 the transmitter happens to be. 



The above discussion of directional effects is simplified by the fact 

 that the No. 394-Type Transmitter is symmetrical about any dia- 

 phragm diameter. Hence a single parameter — azimuth angle — is 

 sufficient. The amplifier mounting cases usually employed destroy 

 that symmetry. The directional effect becomes much more compli- 

 cated since it involves two parameters, e.g. two direction cosines of the 

 diaphragm axis. It has been suggested ^^ that this complication can 

 be done away with by placing the transmitter and its amplifier case in 

 a rigid hollow sphere, only the transmitter front being exposed. If 

 the front contour of the instrument be designed closely to conform to 

 the rest of the sphere, and if the diaphragm subtend a sufficiently 

 small angle at the center of the sphere, the directional effect can be 

 computed. ^^ 



The simplest directional properties, i.e. uniform response for all 

 directions of incidence, require a transmitter whose linear dimensions 

 are small (say < }iX) relative to the shortest sound wave-length to be 

 picked up. For a frequency range extending to 10,000 c.p.s., this 

 means a transmitter less than 0.85 cm. in diameter. In general such 

 restriction on the permissible size adds to the difficulties of construction 

 and operation of the instrument. It is not intended to imply that non- 



