March i6, 1922] 



NA TURE 



141 



^ m 



Nature of Vowel Sounds. 



The following observations as to the nature of the 

 wel sounds of a single voice (my own) — details of 

 ich form the subject of a separate communication 

 the journal of the International Phonetic Associa- 



may be of interest. 

 The sounds were observed by ear, first, for the 

 "spered vowels, and afterwards with a larynx note 

 iperim posed. In the whispered series it was found 

 t each of the separate vowels was characterised by 

 resonant notes, an upper component ranging 

 m about Jd" (6o8~) to e"" (2579-), and a lower 

 ponent ranging from Jd' (304-) to ^a" (912-). 

 _ he upper components are produced in a manner 

 nnalogous to that of whistled notes, and their pitch 

 mainly controlled by the distance of the tongue 

 m the palate and teeth. The same note may be 

 'uced with almost any degree of opening of the 

 uth ; about one octave of the scale can be pro- 

 ced — through the nose — with the mouth closed 

 altogether. These notes, for convenience, are referred 

 to as " whistle notes." 



The lower components appear to depend, Uke the 

 pitch of a Helmholtz resonator, largely on the area 

 of the mouth opening — they are referred to as re- 

 sonator notes. 



The tw^o series are independent of each other, so 

 that, for example, an ascending scale of resonator 

 notes and a descending scale of whistle notes may be 

 produced (whispered) simultaneously. 



The characteristic whistle and resonator notes for 

 each vowel sound are not absolutely fixed (for an 

 individual voice), but may vary in some cases over 

 as much as 5 semitones without loss of the vowel 

 characteristic. 



The ranges of " neighbouring " vowels often overlap, 

 so that two different vowels may have the same 

 whistle or resonator note, but in such cases the other 

 component will be substantially different. 



In one or two cases, such as " ii " (eat) and " i " (it) 

 the ranges of both components overlap, and the differ- 

 ence between these vowels may be produced in some 

 cases mainly by difference of stress. 



When a larj^nx note is added — as in singing or 

 talking — the pitch of the resonator note does not 

 appear to be affected at all by variations of pitch of 

 the larynx note. 



The whistle notes generally are not affected, so 

 long as the pitch of the whistle note in question differs 

 sufficiently widely— say by 2 to 3 octaves— from that 

 of the larynx note. 



As the pitch of the larynx note is further raised 

 towards that of the whistle note, the latter tends to 

 adjust itself or " draw " towards the nearest harmonic 

 of the larynx note which lies within it's characteristic 

 range for the vowel in question. 



Thus, if a chromatic scale be sung to a given vowel 

 sound, the resonator note will remain constant, but 

 when the note sung approaches within say 2 octaves 

 of the whistle note, this latter may be heard to alter- 

 nate between or jump from one to another of 3 or 4 

 neighbouring semitones at each change of pitch of 

 the larynx note. 



This last phenomenon has, I find, been already ob- 

 served by Mr. Perrett. 



From these observations it would appear to be 

 possible to make an exclusively acoustic classification 

 of the vowel sounds depending on the range of their 

 whistle and resonator notes respectively. 



TT ^T ^- XT o R. A. S. Paget. 



East India House, 74 Strand, London, W.C.2, 

 March 3, 1922. 



NO. 2733. ^OL. 109] 



Protective Colloids— A Pretty Lecture Experiment. 



As the result of a large number of experiments 

 carried out in the Chemistry Department of this 

 School by Messrs. Vallance, Dennett, Trobridge, Ham- 

 mond, and Tidmus in conjunction with the writer, it 

 appears to be a general law that protective colloids 

 or organic emulsoids tend to retard the velocities of 

 such reactions, whether chemical or physical, as in- 

 volve a change of state in one or more of the com- 

 ponents. 



Thus it is found that the rates of solution of metals 

 in acids, of corrosion in neutral media, of solution 

 and precipitation of salts, of replacement of one metal 

 by another, as, for example, in the famiUar lead-tree 

 experiments, etc., are all retarded by protective 

 colloids. In many cases the rate of retardation con- 

 forms to the requirements of the adsorption law. 

 Details of these experiments will be published in due 

 course elsewhere. 



A very pretty lecture experiment illustrating this 

 retardation is afforded by the precipitation of mercuric 

 iodide on addition of the chloride to potassium iodide. 

 If this is effected in fairly dilute aqueous solution, 

 the unstable yellow form is first precipitated and 

 rapidly turns from orange to red as it becomes con- 

 verted to the more stable variety. 



If, however, the reaction is carried out in the 

 presence of gelatin, say one per cent., the Uquid first 

 turns momentarily yellow, due to the formation of 

 colloidal mercuric iodide, then becomes turbid, and 

 a beautiful canary colour develops, which remains 

 practically unchanged for half an hour or more, 

 according to circumstances. Only very slowly does 

 it change to the red polymorph. The protective 

 colloid retards the growth of the yellow particles. 

 Sunlight accelerates the change markedly. With the 

 aid of the ultramicroscope (iVth inch oil immersion) 

 these changes may be seen beautifully. Drops of 

 gelatin and dilute potassium iodide are mixed under 

 the coverglass and the ultramicroscope focussed as 

 usual. A drop of mercuric chloride solution is 

 brought to the edge of the coverglass and is drawn 

 under by capillary action. The field of the ultra- 

 microscope becomes swept with a stream of luminous 

 particles moving with dazzling velocity — the Brownian 

 movement of the colloidal mercuric iodide. The 

 velocity slows down as the particles increase in size, 

 until the colloid range has been passed, and in a few 

 minutes a fine precipitate is obtained evincing scarcely 

 any movement. J. Newton Friend. 



The Municipal Technical School, Birmingham, 

 February 27, 1922. 



A Problem in Economics. 



Many economic applications of meteorology depend 

 upon the use of forecasts in deciding whether or 

 not to incur expense by taking precautions against 

 some particular phenomenon which would cause 

 damage. A good example is provided by forecasts 

 of ground temperature in deciding whether to pay 

 men to spread sacking over newly-laid concrete road 

 surfaces which would be injured by frost. In the 

 simplest form of such problems the three possible fines 

 of action are (i) to take precautions only on occasions 

 when the phenomenon is forecasted, (2) to take pre- 

 cautions on all occasions, (3) to take no precautions 

 at all. It is of interest to examine the circumstances 

 under which (i) is the most economical Une of action. 



Let a be the cost of precautions against an event 

 whose probability is P and which will cause damage 

 b if it occurs in the absence of precautions. Suppose 

 the forecast to take the form of a plain " Yes " or 

 " No," and let p be the probabiHty that an occurrence 

 of the event will be preceded by a forecast of " Yes." 



N 2 



