14 THE STUDY OF SPEECH CURVES. 



curve, or even that the pitch of the chief resonance tone is incorrectly given. 

 The vowel curves obtained by investigators using the phonautograph may 

 indicate correctly the pitch of one of the cavity tones in the vowel 

 recorded, although there is no proof of this fact. 



A phonautograph recorder could be trusted only if it could be proved 

 that it gives the proper curves. This might be done by turning its curves 

 back into speech or by comparing its curves with curves of known cor- 

 rectness. An apparatus that can be used for the former purpose will be 

 described at the end of Chapter III. The latter method might be practi- 

 cally applied Ijy using the phonautograph to record the tones of a series 

 of good tuning-forks or of a series of other sounds whose true curves are 

 known. Owing to the great convenience of the phonautograph it has a 

 future before it when the proper principles of construction are found and 

 followed, and when the tests are properly appUed. 



An improvement on the phonautograph is found in the instruments 

 that not only record the speech vibrations but also turn them into sound 

 again. 



The first of these instruments was the phonograph of Edison; later 

 the gramophone of Berliner was evolved. Most recently the telephono- 

 graph of Poulsen and the selenophotograph of Roumer have appeared, 

 but neither of them has yet been used for phonetic researches; the photo- 

 phonograph has not yet been made successful. 



In the following account I shall do no more than call attention to 

 some points that are usually known only to experts; it is unfortunately a 

 fact that workers with recording apparatus, as well as students and critics 

 of speech curves, are constantly led into failure and error by being unaware 

 of many essential steps of the technique. 



In the phonograph the sound wave passing down the receiving tube 

 or trumpet reaches the " recorder." The best form of recorder at present 

 is that shown in figure 1. A perfectly plane piece of thin glass is held 

 by a thin rim of wax. The light aluminum arm carries a sapphire cutting 

 tool, whose end is a plane making a certain angle with the axis. It is 

 preferable to have the sapphire bored hollow, so that its end is a flat ring 

 and not a plane surface. Many other details have to be regarded. The 

 construction of a good recorder is the work of an expert; out of a large 

 number made, only a few are found to give the best results. The "expert 

 recorders" are never found in commerce; the ordinary ones that can be 

 bought give very inferior results. 



The recorder is placed against a moving wax surface, usually a cyhnder, 

 just as a tool in a lathe against a brass bar; a continuous groove is made 

 in the wax. The vibrations of the diaphragm in response to the sound 



