36 THE STUDY OF SPEECH CURVES. 



of two notes. The ostensible trill is therefore not a proper one, but merely 

 a tremolo or quavering on a high note. The false nature of this trill is 

 not apparent even to trained musical ears on first hearing the record, 

 but by making the gramophone disc turn slowly the pitch of the tone 

 is lowered and the tremolo becomes evident. 



It would be interesting to investigate the nature of the mechanism 

 of the tremolo and the reasons why it is substituted for a trill ; some records 

 and observations on these points will be reported in a future publication. 

 Here I will only say that in mj^ opinion the tremolo is often not produced 

 by movements of the muscles of expiration, but by variations in the 

 action of the larynx. The effect of any such variations in the adjustment 

 of the larjaix for a given tone — the pitch of the tone remaining unchanged — ■ 

 would be to modify the timbre of the tone in addition to the alterations 

 of intensity; this would show itself in differences in the form of the waves 

 in the loud and weak portions of the curve. This is apparent at once 

 on close inspection of the curve given; the two portions have two typical 

 forms, with gradual passage from one to the other. How this fluctuation 

 in intensity is produced by the larynx is a matter for future investigation. 

 Before leaving this curve we note that in some of the weaker portions 

 the waves fall into groups of twos or threes or even fours; an explanation 

 is suggested in the discussion of the curves of laughter in the chapter on 

 Qualitative Analysis. 



The next specimen on the plate is the curve of the first naif of a piano 

 chord with one note more prominent than the others. The curve shows 

 great complexity, due to the summation of the different waves for the 

 separate notes and to the fact that these die away at different rates. A 

 detailed study of the curve requires mathematical methods; it will be 

 given on a future occasion. 



The curves of laughter give the last two of a set of laughs which may 

 be indicated by "Ah, hah, hah, hah." These will be discussed in detail 

 in the chapter on Qualitative Analysis. 



The curves on this plate were made with a single lever. The time 

 equation is lmm.=0.0004s. ; the magnification is 150. The curves on plates 

 VIII and IX were made with the compound lever. The time equation for 

 plate VIII is 1mm. = 0.0007s. ; that for plate ix, lmm.= 0.0016s. The magni- 

 fication in both cases is 150. 



The phonograph tracing apparatus (figure 42) was constructed by a 

 grant from the Elizabeth Thompson Science Fund.* It is adapted solely 

 for tracing celluloid cylinders of the Lioret type. 



* A New Machine for Tracing Speech Curves, Amer. Jour. Sci., 1903, xv, 147. 



