MECHANICS OF THE HUMAN VOICE 249 



The validity of a vowel theory can be tested by constructing an 

 apparatus to produce vowels on its principles. The Helmholtz 

 theory was tested by Helmholtz himself in his vowel apparatus. 

 This consisted of a series of tuning forks with pitch frequencies in 

 the relations of i: 2: 3, etc. By adjusting the resonators different 

 overtones could be reinforced. Helmholtz states that he could ob- 

 tain only a good o and u. The same results can be obtained almost 

 as well by simpler apparatus and in a way to refute the theory. 

 Any well made tuning fork of any moderately low pitch will pro- 

 duce a good u. Empty bottles often give good ti vowels through 

 several octaves of pitch. These facts made it unnecessary to con- 

 struct another apparatus on the Helmholtz theory. I therefore 

 began on the Hermann theory, which at the start I believed to be 

 correct. According to this theory the vocal cords emit a series of 

 puffs of air, each of which acts like a blow on the air of the vocal 

 cavities and arouses the tones of the cavities. A blow of the hand 

 on the open mouth will thus arouse the cavity tones. If we con- 

 ceive the blow to be repeated often enough —e. g., 100 or 200 times 

 a second — we should hear a vowel. To produce these puffs of air 

 I first used a siren comprising a disc with holes rotated before a 

 blast of air. I found that a resonator with hard walls (brass, wood, 

 etc.) would respond only when the puffs came at a definite pitch, 

 whereas resonators with soft walls (cotton soaked in water, gela- 

 tine, etc.) would respond to anj^ pitch. I could thus produce good 

 examples oi a, 0, and ti with a siren tone of any pitch. I then 

 tried a vox h7ima?ia organ pipe with the same results. These vow- 

 els were produced at a7iy pitch of tone, as required by the Hermann 

 theory. I utterly failed, however, to produce e, ce, or i. 



Certain puzzling phenomena in the speech curves seemed to indi- 

 cate that the action of the vocal cords differed for different vowels. 

 Such an unusual conclusion, though difficult to accept, might be 

 justified by considering (i) that the innervation of different parts of 

 the vocal cords may differ for different vowels, whereby the distri- 

 bution of the load, and consequently the mode of vibration, may differ 

 (it should be noted that the vocal cords, or, more properly, the 

 vocal bands, do not vibrate merely along the thickened edges, but 

 through the entire muscular mass); and (2) that the vocal cords as 

 soft bodies may be influenced in the character of their vibrations by 

 the size of the resonance cavities. The experiments in manufactur- 

 ing vowels were now renewed with tones from rubber membranes. 



17 



