1897-98.] J. Boeke on Derivation of Curves of Vowel Sounds. 93 
trumpet — the length of which ought, obviously, to be constant — - 
by means of the drum, Q, I found that the number of divisions 
through which the drum, Q, had to be turned for each period was 
by no means constant, but varied in the most capricious manner. 
As I soon found out that this result was the consequence of the 
teeth of the wheel, A, and its pinion not always being in close 
contact, I tried to mend this by making the teeth cling constantly 
together by means of a weight, turning the axis of the cylinder in 
a direction opposite to that in which it was turned by means of the 
drum, Q. This arrangement, indeed, made the readings of the 
drum, Q, reliable. But only where there were a small number of 
underwaves did it prove practicable to measure the abscissm as well 
as the ordinates of the curves. In most cases it would have been 
too fatiguing for the eye to read alternately the ocular micrometer 
and the divisions of the drum, Q. 
II. For this reason, in most cases a second mode of procedure 
was employed to determine at the same time the abscissae as well 
as the ordinates of the vowel curves under examination. 
This consisted in measuring out the transverse diameter of the 
impressions on equidistant spots, which proceeding would obviously 
lead equally well to the exact form of the curve. I tried to do 
this by means of a clockwork, working on the axis of the mandrill 
on which the phonograph cylinder was placed. This clockwork 
was provided with a stay pressed upwards by a spring. While 
looking through the microscope a gentle pressure of the finger was 
sufficient to make the clockwork run through ^-gVo a revolution. 
But in this case also it was found that the apparatus did not do 
what it should. It was, however, a long time before I discovered 
that the movement of the clockwork was not at all uniform, so 
that the abscissae of the curve which I derived from my measure- 
ments were not equal to each other, as I had supposed them to be. 
Since I came to this conviction only gradually, after having 
tried, without success, to regulate by different means the movement 
of the clockwork, many of the numerous measurements which I 
have made by means of this clockwork lack the precision and 
reliability which vowel curves ought to have. Neither by a 
stronger spring in the clockwork, nor by the application of a weight 
pulling the cylinder either in the same direction as the clockwork 
