THE SPEECH OF MONKEYS. 1 79 



chords may be detected, and every part of the 

 sound compared to every other part. 



Having thus augmented the quantity of sound 

 by increasing alike the sound-unit and interval, 

 it can be recorded on another cylinder and mul- 

 tiplied again as long as the vibrations can pro- 

 duce sound. From the constant relation of parts 

 and their uniform augmentation under this treat- 

 ment, it has suggested to my mind the idea that 

 all sounds have definite geometrical outlines, and 

 as we change the magnitude without changing 

 the form of the sound, I shall describe this con- 

 stancy of form by the term contour. 



In a few instances I have been able, by reduc- 

 ing the record of certain sounds from a high 

 pitch to a lower one, to imitate the sound thus 

 reduced with my own vocal organs ; then by re- 

 storing this record of my voice to its normal speed, 

 I have obtained almost a perfect imitation of the 

 sound. This effect, however, does not always 

 follow, and in many instances my best imitations 

 have not developed the original at all. But this 

 presents a new problem in acoustics. I must 

 here take occasion to say that the difference of 

 pitch, quality, etc., in sounds does not appear to 

 me to depend alone upon the length of the sound- 



