|Share page||Visit Us On FB|
THE PHONOGRAPHIC METHOD 47
The notes I made as to the changes of rate of the phonograph during the work of examining these songs are in detail as follows: —
BATTERY 5 (150 AMPERES)
July 25. I began the notation of the songs, expecting to assure myself of the constancy of the phonograph by the method I had employed in studying the Chinese melodies before mentioned ; viz., by returning every few minutes to some conspicuous note at the beginning of the record, to see whether I should be tempted to give it a different notation from that already chosen. I carried out this plan without making any notes of the results during this day's work, which consisted in the notation of Snake Songs Nos. 1 and 2.
July 26. I wrote down Snake Song No. 3, and made the following note of the running of the instrument. " To write this out took over an hour, and yet in going back to the initial note I could detect no change from the written pitch."
The ratio of the numbers of revolutions made by the cylinder in a given time, in two reproductions of tone from the same inscription, is that of the vibration numbers of the two tones produced. We may express this by saying that the ratio of any two numbers of revolutions per minute, executed by the cylinder at different times, gives the interval by which the pitch of the phonograph differs in the two cases.
The counting of the revolutions of the cylinder is conveniently effected by holding the finger lightly against a screw-head that projects from about the middle of the axis. The number of touches received in this way, during the time that elapses before the second hand of a watch passes again beyond the spot in its dial where we count the first one, exceeds by one the number of complete revolutions made by the cylinder in a minute. Although the point from which I began to count was always one of the ten second marks on the dial, I took the precaution to wait until it seemed to me that a touch was closely synchronous with the tick made by the watch (marking each .2") as it crossed such a mark. I found myself in practice estimating the fraction of a revolution often seeming to take place between the last touch and the tick of the watch that next took the hand over the initial ten second mark, either as a half revolution (e. g. 170.5) or as less than a whole revolution (e.g. 171 — ) or as not quite a half (e. g. 170 + ). Even with these efforts at exactness, the probable error of such estimates of the speed of the instrument might easily be half a revolution or more. Assuming it at half a revolution, the error may be in each case either by excess or by default. But it is only when it happens