THE SOi'XnS OF SPEECH 593 



tu'iut- "the solution for transient oscillations of the system is reduced 

 to forinul.ie which are functionally the same as those for steatly state 

 oscillations" (reference 10; see also reference 11). But before leaving 

 this tliscussion of speech characteristics it should he noted that the 

 essence of the matter lies not so much in reconciling the two theories, 

 of the vowel sounds as in ascertaining what motions really take place 

 in the oral cavities, anti in the air near the vocal cords. Though the 

 process of harnjonic analysis is to be applied to the records of the 

 vowel sounds, we must recognize its limitations, and not necessarily 

 infer steady state conditions. Indeed the most casual inspection 

 of the records shows a certain lack of periodicity in the phenomena 

 recorded; and it is hardly to be expected that all the phenomena can 

 be satisfactorily summed up on the basis of the harmnnic theory. 



II 



Till-: RixoRDiNT. App,\ratus'= 



In providing means for accurately recording .sound waves, use has 

 been made of three devices recently developed in this Laboratory and 

 we believe that by properly connecting these together we have obtained 

 a recording instrument which is superior in accurac>- and power to any 

 heretofore used. These three devices were each nearly free from dis- 

 tortion, and such residual distortions as could not be eliminated were 

 so controlled that they practically offset one anoltuT n\cr a wide range 

 of frequencies. 



The first element in the recording set is the condenser transmitter, 

 which has been thoroughly investigated by Wente (refs. 8b, 8c, 8f); 

 its frequency characteristics, in both amplitude and phase are shown 

 in Fig. 2. The particular transmitter u.sed was of recent design and 

 had been carefully standardized and calibrated especially for this 

 work. 



The condenser transmitter was connected to the input terminals 

 of a seven-stage amplifier as shown in the large diagram of Fig. 5 

 which gives the details of the electrical circuit, including tlio third 



"J. R. Carson: Phys. Rev. X, 1917, p. 217, "On a (jencral Expansion ThcarL-m 

 for the Transient Oscillations of a Connected .System." 



" T. C. Fry, Phys. Rev. XI\', V)V), p. 1 17. "The Solution of Circuit Problems." 

 '= Thanks are due to Messrs. C. F. Sacia and C. J. Beck for the skill and care 

 with which they assembled and calibrated the recording apparatus, and made the 

 complete set of records. The writer is also under obligation to Mr. Sacia for aid in 

 choosing the sounds to he recorded, an<l systematizing the collection; Mr. Sacia 

 also developed and applied the photomechanical method of analyzing records, the 

 results of which arc given in Figs. 13 and 14 of this paper. 



