1 68 



NATURE 



\_y7me 13, 1878 



We have now obtained and analysed a very large number of 

 vowel curves, especially for the sounds o and u, and with your 

 permission will send a selection of these for publication after 

 the results of our investigation have been communicated to the 

 Royal Society here. These curves show what the voice effects 

 when singing at different pitches a vowel which remains of con- 

 stant quality according to the appreciation of the speaker. The 

 analysis of the curves gives the partial tones of which the vowel 

 sounds are composed, and it becomes a matter of considerable 

 interest to see how far these results confirm or contradict existing 

 theories. 



We therefore propose to give a short sketch of these theories, 

 hoping that if in error, we may be at once corrected. 



Prof. Willis showed (Cambridge P/ii/. Soc. Trans., vol. iii.) 

 that, by varying the length of a tube attached to a free reed, 

 he was able to produce the sensation of a change of vowel sound. 

 This sensation is not very definite, especially for the vowels i and 

 e ; but, when the length of the tube is changed rapidly, the ear 

 accepts the suggestion of a change from ti to o, a, e, and i. 

 Prof, Willis concluded that the vowel quality was given in each 

 case by the coexistence of the note proper to the tube with that 

 of the reed, whether the former was or was not harmonic to the 

 latter. One must bear in mind that Willis wrote before it was 

 recognised that all musical sounds are compounded of harmonic 

 partial tones, and also before the function of a resonator was 

 understood. This we may call the absolute single-tone theory. 

 Wheatstone pointed out that the tube used by Willis acted simply 

 as a resonator. 



F. C. Bonders {Danders' Archiv, vol. i.) observed that when 

 the mouth was formed to speak a given vowel, the cavity had a 

 certain definite pitch of resonance, or maximum resonance, which 

 he determined by observing the pitch of the whispered vowel. 

 Bonders and Helmholtz agree in considering that this charac- 

 teristic pitch is nearly constant in man, woman, and child, for a 

 given vowel. Bonders also observed that the vowels were, in 

 certain cases, accompanied by- a "geruisch" or "whish." 



Helmholtz attacked the subject in a different way. By means 

 of resonators he applied a qualitative analysis to the sounds 

 which came out of the mouth cavity when a vowel was spoken, and 

 pointed out that a vowel- sound at a particular pitch was charac- 

 terised, not by a single tone, but by many tones. In an early 

 paper, translated in the Phil, Mag. for i860, he describes not 

 only the analysis by resonators, but the synthesis by means of 

 tuning-forks, which is now a familiar experiment. In this paper 

 he appears inclined to believe that it is the relation between the 

 constituent tones which determines the vowel quality ; that, for 

 instance, any pair of simple tones one octave apart will, if 

 properly proportioned, always make an 0. This theory made 

 considerable way ; it is taught with small qualification as to 

 absolute pitch in Tyndall's lectures on sound, and elsewhere ; 

 it may be called the relative-pitch theory. 



Helmholtz himself does not seem ever to have formulated it, 

 for although the paper referred to distinctly suggests it, he 

 guards himself by saying that the relations observed can only 

 be considered as proved for the particular forks used, and, in 

 fact, his experiments were only made at two parts of the scale. 



In the " Tonempfindungen" the relative-pitch theory is entirely 

 abandoned, but it is not a little difficult to ascertain what Prof, 

 Helmholtz's latest theory is. This difficulty is indeed admitted 

 by his translator, Mr. A. J. ElHs. 



By Herr von Quanten, whose papers were published in Pog- 

 gendorffs Annalen for 1875, Prof. Helmholtz was understood 

 to mean that for each vowel one, and in some cases two, tones 

 of definite absolute pitch must be strongly present, these tones 

 being those which Prof. Helmholtz calls the characteristic tones 

 of the vowel. This would imply either that each vowel could 

 only be sung on a very few notes, or that the characteristic tones 

 were present as inharmonic partials. Neither of these conclu- 

 sions i being in accordance with fact, von Quanten concluded 

 that Helmholtz was wrong ; but Mr. Ellis, with justice, as we 

 think, points out that the true conclusion ought to have been 

 that Helmholtz could not possibly have meant to broach an 

 absurd theory. 



We confess that we were ourselves led to believe at first that 

 Helmholtz taught that each vowel contained strongly either its cha- 

 racteristic tone or some of the higher partials of that tone or tones 

 very near these, and that this was what gave it its distinctive cha- 

 racter. This was the theory which our first experiments seemed 

 definitely to contradict. We now believe, however, that this 

 is not the doctrine taught by Helmholtz. 



Indeed we fail to find in the "Tonempfindungen" any very 

 complete vowel theory, but we think that the following passage, 

 taken from Bonders' pamphlet "Be Physiologic der Spraak- 

 klanken, 1870," expresses very clearly a doctrine which is very 

 generally looked upon as that of Helmholtz. 



Bonders says (if our translation from the Butch be correct) : — 



"Vowels spoken loud are sounds of a determinate timbre 

 maintained unaltered, depending on the form of the mouth- 

 cavity and of the mouth -aperture, and, even without the accom- 

 panying * whish, ' characterised by strong comparatively low 

 upper tones not occurring in a definite order relatively to the 

 prime tone, but for each vowel of an approximately constant 

 pitch." 



We understand Bonders to believe that on whatever part of 

 the scale a vowel be spoken the pitch or pitches of maximum 

 resonance of the cavity are constant for a given vowel, and that 

 indeed the form itself is constant. This may be called the 

 constant cavity theory, and is taught by Mr. Ellis as the doctrine 

 of Helmholtz. 



We fail to find that Helmholtz himself has stated this doctrine 

 definitely in all its rigidity, although he accepts the results of 

 Bonders' experiments, and has himself confirmed and amplified 

 them. Almost every statement made by him concerning vowels 

 is limited to those which he could produce by forks forming an 

 harmonic series with Bl> for the prime. Any experiment de- 

 scribed by Helmholtz is of course to be relied on, and so far as 

 we have yet traversed the ground we find that the phonograph 

 gives results in accordance with his experiments as to these con- 

 stituents, but when we examine the one or two more general 

 statements made by Helmholtz we find room for doubt, both as 

 to his meaning and as to the truth or completeness of the con- 

 clusion. 



Thus at the end of Chapter V. Helmholtz says :— " Vowel 

 qualities of tone consequently are essentially distinguished from 

 the tones of most other musical instruments by the fact that the 

 loudness of their partial tones does not depend only on the 

 numerical order but chiefly upon the absolute pitch of those 

 partials; thus when I sing the vowel A to the note E[) the 

 reinforced tone b"^ is the twelfth partial tone of the com- 

 pound ; and when I sing the same vowel A to the note l/^ the 

 reinforced tone is still b"^, but is now the second partial." The 

 two words marked by us in italics have been introduced for the 

 first time in the fourth edition. 



This passage might very well be understood to mean that a 

 certain tone of perfectly or at least very approximately definite 

 absolute pitch must necessarily be present in a given vowel. 

 Further examination has, however, convinced us that Prof. 

 Helmholtz does not require the presence of any characteristic 

 tone or of any one of a group of characteristic tones in a vowel. 

 This is made obvious by Chapter VI. In this chapter, which 

 treats of artificial vowels, we find that in order to make an e 

 by tuning-forks Prof. Helmholtz employed B^ and b'\) as "being 

 adjacent to the deeper characteristic tone /'," which in fact lies 

 midway between them ; in the same way he employs /'", a"'[j, 

 and b'"^ for the same vowel, treating all these as adjacent to the 

 higher characteristic tone b'%. Thus the theory of Prof. Helm- 

 holtz is satisfied if tones lying anywhere within a whole octave 

 be present, provided the characteristic tone lie somewhere near 

 the middle of that octave. This is consistent with Bonders' 

 statement of the theory, provided " approximately constant 

 pitch " be allowed to signify anything within six semitones. 



We consider the following abstract as representing the doctrine 

 taught in the " Tonempfindungen " : — 



1 , For a given vowel there is a certain form of mouth cavity 

 which has a pitch (sometimes two pitches) of strongest resonance 

 — as b'\} for 0. 



2, If this vowel be spoken or sung on any subtone of this 

 pitch, the overtone corresponding to that pitch will be strongly 

 present, 



3, If the same vowel be pronounced at some other pitch then 

 these harmonic partials will be reinforced which lie within, say, 

 six semitones of the characteristic pitch. 



No opinion seems to be expressed on the following two 

 points : — 



I, Whether the mouth cavity for a given vowel remains 

 constant when the pitch of the vowel is altered. Mr, Ellis 

 understands Helmholtz to affirm this, which is apparently 

 Bonders' view, but we have failed to perceive any passage in 

 which this is definitely asserted, Helmholtz says the ca,vity for 

 a given vowel has a pitch of strongest resonance, but this is not 



