DRS. GUY BARLOW AND H. B. KEENE ON THE ANALYSIS OF SOUND. 
137 
eventually replaced by springy brushes built up of long thin strips of phosphor-bronze, 
each strip being slit longitudinally into alternately two and three fingers. In this way 
a number of independent contacts was obtained, and much steadier conditions of running 
of the motor were secured. Even with the improved brush design occasional irregu¬ 
larities in speed still occur, and are usually traceable to the contact conditions on the 
commutator. This appears to be the outstanding difficulty in obtaining a constant 
speed of revolution with an electric motor. 
The speed of the motor was measured by a stroboscopic method.* For this purpose 
the shaft carries a cylinder the surface of which is divided into 21 rings. Each ring is 
marked out into 20, 21, 22 . . . 41 equally spaced black squares with white intervals, 
and for ease in identification every fifth ring is tinted red. The stroboscopic cylinder 
is viewed through double slits mounted on the prongs of a maintained fork (64/sec.) 
giving 128 views per second of the rotating patterns. The speed is determined by 
observing the number of the ring which appears to be stationary. When, as is generally 
the case, no ring is exactly stationary, then two consecutive rings are seen to rotate 
slowly in opposite directions with different speeds. By measuring the rate of progression 
of one of these the required frequency may be obtained with a degree of accuracy limited 
only by the constancy of the motor speed. In practice it is sufficient to interpolate by 
estimation, as this can be done without giving an error in the frequency of more than 
per cent. 
The laboratory experiments were made with a Broca galvanometer (10 ohms), and, 
when required, a transformer having a primary resistance of 4 ohms and secondary of 
90 ohms. At the reservoir a Broca galvanometer (100 ohms) was used, and also a 
transformer with resistances 60 and 110 ohms. The period of the galvanometer was 
in both cases adjusted to be 3 sec., and then made almost dead-beat. For the purpose 
of dealing with vibrations of great complexity, it would appear quite practicable to 
modify the present apparatus to give a photographic record of the “ spectrum.” 
Analysis of a Current. 
Before proceeding to analyse sound vibrations the following experiments were made 
to test the reliability of the method by applying it to analyse alternating currents of 
known characteristics. 
(1) Simple Harmonic Current. 
The current was generated in a small coil, wound in the form of a figure 8, by the 
motion through it of a U-shaped magnet (made from a piece of knitting-needle 4 cm. 
long) attached either to the prong of an electrically maintained fork, or in the case 
of the lower frequencies to an electrically maintained steel strip. 
* Rayleigh, ‘ Phil. Mag.,’ 1907. 
