.1 Dyx.iMic.iL sruoy or run roiiF.L sorxos zis 



most laiii;uages. Now consider the soquence from this sound to 

 No. XIII at the i-nd of the diagram; these sounds include most of 

 those which are known to have two characteristic regions of resonance. 

 The main region of resonance now divides into two parts which 

 gradually recede from each other as we follow the diagram down- 

 waids. (Sound X (er) is difficult to fit into the diagram in an exact 

 fHisition, but it is evident that it beUmgs in the series of douMy-reso- 

 ii.mt vowels.) 



Contour lines (nearly vertical) have been drawn on the diagram 

 to indicate the progressive changes in regions of resonance. View- 

 ing the diagram as a whole it is important to consider not only the 

 location of the resonant ranges but al.so their extent, and their relative 

 separation from other resonant ranges in order to arrive at the essential 

 characteristics of the vowel sound. In other words the indi\idual 

 vowel characteristic depends not only on the absolute pitch but on 

 the relative pitches in case there is more than one region of resonance. 

 It is only in this way that we can explain what is a matter of uni- 

 versal experience in using the phonogr.i()h ; namely that moderate 

 variations from normal speed in recording and reproducing speech 

 leave the vowel sounds still intelligible. 



It is expected to deal in a later publication with the semi-vowel 

 sounds /, w?, «, m which seem to be related to the general diagram of 

 the vowel sounds, and on which a preliminary report has already 

 been made'. 



The more interesting features of the original records as such will 

 also be dealt with in a subseciueiit publication. 



APPENDIX 



Recording and Analysis of Vozcel Sounds 



Recording App.\r.\ti;s 



The apparatus usetl in recording consisted of a contlenser trans- 

 mitter, an amplifier, and an oscillograph, in which important modifica- 

 tions were made. The vibrator was given great stiffness and damp- 

 ing so that the frecjuency response of the vibrator was nearly uni- 

 form up to 5000 cycles. Instead of the usual 12 inch film, special 

 film 51 inches in length was used. This necessitated a much larger 

 film drum. Furthermore the desired length of the record was about 

 four times the circumference of the film drum, so the shutter was 

 arranged to stay open during four revolutions while the vibrator was 



' Phys. Rer. 2.?, 19i4, p. 309 — "Prt-Iiminary .Analysis of Four Semi-Vowel .Sounds." 



