BINAURAL LOCATION OF COMPLEX SOUNDS 35 



been inconsistent, and in some cases thc>' have not e\en corresponded 

 to any actual source whatever. Under these circumstances, if an 

 image is formeci at all, some purely ps\'chological factors must enter 

 in. For pure tones it has been found possible to explain much of the 

 experimental data obtained under circumstances such as this by 

 assuming that the observer subconsciously judges one or more of the 

 characteristics to be in error and applies such corrections as will 

 make all of the data correspond to an actual source. As a criterion 

 for determining which characteristics will be altered, it is assumed 

 that, in general, those are chosen which require the smallest changes. 



Let us now consider what characteristics are available for locating 

 sounds of different kinds. A pure tone from a source at rest with 

 respect to the observer has at any point only two physical character- 

 istics which are subject to change with the position of the source. 

 They are its amplitude and phase. Corresponding to each position 

 of the source there is a particular amplitude and phase at each of the 

 two ears so that a total of four properties — the loudness of the sound, 

 the average phase, the difference in amplitude (which mayconven- 

 iently be expressed as a ratio) and the difference in phase at the two 

 ears— are available for determining the position of the source. It is 

 inconceivable that the average phase can have anything to do with 

 the location of the sound since it may be changed at will without 

 altering the position of the source. The same remark applies to the 

 loudness of the sound except in those instances where the observer 

 is familiar with the source to such an extent as to know how loud 

 it may be expected to be. Hence, if we restrict ourselves to the 

 cases in which prejudicial information of this sort does not exist, 

 we find that the observer has only two quantities from which he may 

 deduce the position of the source. We should therefore expect that 

 these two quantities would make it possible to locate the tone with 

 respect to two coordinates only. This is found to be in general 

 agreement with experiment, for most observers locate all sources of 

 pure tones in the same horizontal plane with their heads and determine 

 only the distance and angular departure from the median plane. If 

 the source is more than a few yards away the intensity ratio and phase 

 difference change very slowly with distance so that in this case even 

 the sense of distance is not keen and a feeling of certainty exists with 

 respect to the direction only. 



In many experiments the tones at the two ears have been varied 

 arbitrarily so as to give combinations having equal phases and un- 

 equal intensities or vice versa — combinations which cannot arise 

 from actual physical sources in the absence of distortion. Under 



