114 THE STUDY OF SPEECH CURVES. 



The results of this investigation — as far as they are appU cable here — 

 may be summarized as follows: 



A spherical brass resonator blown by a series of smooth puffs responds 

 loudly when the period of the puffs is the same as that of the resonator. A 

 very soft resonator, for example, a water resonator, responds practically 

 alike to all frequencies of the puffs within a large range. This response, 

 however, appears partly as an increase in the loudness of the puff tone, 

 but mainly as an addition of a peculiar vowel tinge. The effect becomes 

 less prominent as the softness of the resonator diminishes. 



The resonance effect depends also on the form of the puff. It is well 

 known that to a perfectly smooth puff, as from a tuning-fork (a. sin pt), 

 a spherical resonator with hard walls will respond only when its period 

 coincides with that of the puffs. As the puff becomes sharper the required 

 coincidence changes from a point to a range. To moderately sharp puffs 

 a resonator with hard walls will respond not only when the puff period 

 is the same as its own but also when it is nearly the same — the effect 

 diminishing with the difference between the periods. For very sharp puffs 

 my experience is that hard walls respond in two ways, with a loud tone 

 when the periods correspond and with a modification of tone at all times. 



The principles involved in these experiments can be directly applied 

 in a vowel theory. Thus the loudness of the tones with sharp puffs explains 

 the loudness of the glottal tone although the simple harmonic analysis 

 indicates a weak sinusoid of that period. The prominence of the cavity 

 vibration in the curve over the glottal vibration is also explained. 



With funds from the Smithsonian Institution* I made an attempt to 

 construct a musical instrument to produce the vowels. Any attempt to 

 build up the vowels by adding sinusoids (tuning-fork tones), whether 

 harmonic or inharmonic, can not succeed, because the curves of the vowels — 

 even sung vowels — are of an utterly different character. Helmholtz seems 

 to have satisfactorily produced [u], [o], and [a] in this way, but there is 

 much doubt whether his reported success with [e], [i], etc., was a close one. 

 No one else ever seems to have succeeded with the same method. Experi- 

 ments were begun on the basis of the WilUs-Hermann theory that a vowel is 

 produced by the action of a series of puffs on a resonating cavity. For 

 the source of tone, striking reeds of metal, wood, etc., in vox humana 

 pipes were used; for the cavities gelatin and water resonators singly 

 doubly, and triply, also gelatin copies of the mouth, also a skull fitted 

 with gelatin instead of flesh and with exchangeable gelatin tongues; flesh 



♦Scripture, Report on the Construction of a Vowel Organ (Hodgkins Fund), Smithsonian Miscel- 

 laneous Collections, 1905, xlvii (3), 360. 



