756 PROFESSOR FLEEMING JENKIN AND J. A. EWING ON THE 
The figures given for the amplitude of the partial tones are the actual 
amplitude calculated in y3,ths of an inch, from measurements of the tran- 
scribed curves shown above. They are therefore not merely relative, but 
afford an indication of the loudness of each utterance. ‘This indication is, 
however, only a very rough one, for the amplitudes depend on many other 
conditions than the loudness of the tones; in particular, they are greatly 
affected by the closeness of the mouth to the vibrating disc, and no special care 
was taken to keep that distance constant. 
Even in the absence of slight and unavoidable irregularities in the curves, it 
would have been almost impossible to have made the measurements with such 
accuracy as to determine with certainty the values in the unit place for the 
above numbers. The last figure of each group is not to be depended upon, 
and when an amplitude, such as 2 or 3, is given for any tone, that tone may 
very possibly have been entirely absent in the round ring, as it may very 
possibly have been present to double the extent indicated by the figures. 
Table II. gives the analyses of the examples in fig. 2 of 6 sung by voice 5. 
The most cursory examination of the above curves and tables show that at 
different pitches the vowel sound 6 is by no means composed of the same 
relative constituents, but, that as the pitch alters, changes take place which are 
fairly gradual and consistent throughout. Above g, indeed, the results might 
be described with little reference to absolute pitch. In that region of the 
scale the sound 6 consists almost wholly of two partial tones—the first and the 
second, the proportion between which, however, depends partly on the pitch 
and partly on the quality of the voice. On the highest notes reached the pro- 
portion of prime to second with each voice is greater than it is a few tones 
lower down, but the proportion varies greatly with different voices, even at the 
same pitch. Below g the third partial begins to be prominent; the second, 
however, still remains very strong, and the prime moderately so. This state of 
things continue on ¢ and d. Then the fourth partial appears, and is very 
strong on B, B?, A, and even G (No. 16); the prime, however, is now very 
weak ; the second partial continues pretty strong, except on G, and the third is 
very strong. Thus we have, on going down the scale, the general result that 
one after another of the upper partial appears in succession, while first the 
prime and then the second partial become much weakened; there being 
always at least two partials strongly reinforced, and generally more than two. 
Before, however, proceeding to examine these figures in detail, and to 
point out their bearing on existing theories of vowel sounds, we shall give a 
number of additional examples of 6 sung by other voices. Fig. 3 gives the 
curves for a set of o’s, sung by voice 3, and fig. 4 a similar set by voice 4. 
Table III. and Table IV. contain the results of the analyses of the curves 
in figs. 3 and 4 respectively. 
