32
THE STUDY OF SPEECH CURVES.
portion off is changed each time to avoid getting the portions reversed.
The portions are then pasted on pasteboard sheets of the required size.
For reproduction in the form of plates for printing, the curve has
to be made somewhat whiter than is otherwise necessary. A small wax
weight is added to the outer part of the recording point, so that the glass
knob rests a little more heavily on the smoked paper. The danger lies
in the bending of the lever due to the greater resistance of the smoke.
A test of amplitudes is made before and after adding the wax; if the altera¬
tion falls within the limit of error (say 1 per cent) the wax is retained.
The only available processes of reproduction seem to be copper
etching and zinc etching. The former gives excellent results, but is
expensive. The latter is successful only when the engraver employs the
best zinc. The mount is photographed by the engraver. Ordinarily the
negative is then stripped from the glass; since the gelatine stretches and
introduces errors, this process is avoided by mounting all the curves back¬
wards (from right to left) on the pasteboard and instructing the engraver
not to strip the glass plate. In the print the curves appear to read from
left to right. With the usual construction of the apparatus used for
taking records the phase of condensation of the original sound registers
with an outward (centrifugal, away from the center) deviation of the
sound groove. With the tracing point as shown in figure 21 this phase
gives a deviation upward on the page. The curve as seen in the figure
reads backward. When the strips are mounted backward, the blocks
produced from them without stripping have the phase of condensation
also upward.
Illustrations of the work done by the gramophone tracing apparatus
are given in the accompanying figures and plates. The time equation
for figures 28 to 40 is 1mm. = 0.0004s. The curves in figures 28 and 29
were made by the apparatus with single lever as shown in figure 18;
the pressure of the glass point against the paper was very light
and the line—reproduced by zinc etching without retouching—appears
like a series of dots. For the curve in figure 30 the glass point was
weighted and the zinc etching—with no retouching—appears clearer.
The curve in figure 31 was produced in the same way, but was retouched;
the line is stronger, but is at a decided disadvantage on account of its
raggedness and also because its width renders it difficult to measure.
The curve for the yodel (figure 28) is a short portion out of a single note; it
shows strong, smooth waves, each of which corresponds to one vibration of
the voice. The short portion in figure 29 shows a softer yodel combined
with piano vibrations. The trombone vibrations (figure 30) show consider¬
able fluctuations in intensity even in the short time represented by the piece