HARMONY IN MUSIC. 83 



the more accustomed we are to the compound tones of which 

 they form a part. This is especially the case with the human 

 voice, and many skilful observers have consequently failed to 

 discover them there. 



The preceding theory was wonderfully corroborated by leading 

 to a method by which not only I myself, but other persons, 

 were enabled to hear the upper partial tones of the human voice. 



No particularly fine musical ear is required for this purpose, 

 as was formerly supposed, but only proper means for directing 

 the attention of the observer. 



Let a powerful male voice sing the note e k to the 



vowel o in ore, close to a good piano. Then lightly touch on the 

 piano the note b' fe 1r^f' =: ~ in. the next octave above, and listen 



attentively to the sound of the piano as it dies away. If this 

 b' fe is a real upper partial in the compound tone uttered by 

 the singer, the sound of the piano will apparently not die away 

 at all, but the corresponding upper partial of the voice will be 

 heard as if the note of the piano continued. 1 By properly 

 varying the experiment, it will be found possible to distinguish 

 the vowels from one another by their upper partial tones. 



The investigation is rendered much easier by Urming the ear 

 with small globes of glass or metal, as in Fig 12. The larger 

 opening a is directed to the source of sound, and the smaller 

 funnel-shaped end is applied to the drum of the ear. The in- 

 closed mass of air, which is almost entirely separated from 

 that without, has its own proper tone or key-note, which will be 

 heard, for example on blowing across the edge of the opening a. 

 If then this proper tone of the globe is excited in the external 

 air, either as a fundamental or upper partial tone, the included 

 mass of air is brought into violent sympathetic vibration, and 



1 In repeating this experiment the observer must remember that the e ft of 

 the piano is not a true twelfth below the b'h. Hence the singer should first be 

 given b'h from the piano, which he will naturally sing as b h, an octave lower, 

 and then take a true fifth below it. A skilful singer will thus hit the true 

 twelfth and produce the required upper partial b'b. On the other hand, if he 

 sings efe from the piano, his upper partial b'V. will probably beat with that of 

 the piano. TK. 



a 2 



