THROW-TESTING MACHINE FOR REVERSALS OF MEAN STRESS. 
275 
(1) To determine the extent of the variation due to the first cause, the curves of 
displacement for the oscillating parts were carefully drawn to a large scale, and the 
harmonics of the motion found by the usual graphical method. The harmonics were, 
however, very small, and as the kinetic energy of the reciprocating parts is only 
about yg-th of the total kinetic energy of the rotating parts, it is evident that the 
fluctuation of energy of the parts could only introduce infinitesimal fluctuations in 
angular velocity. 
(2) The fluctuations of velocity of the engine in a cycle, loaded as it was by a 
heavy fly-wheel and rope pulley, and connected with a long line of shafting having a 
large number of heavy pulleys attached, are also negligibly small. Thus the only 
real difficulty in eliminating errors in the measurement of co was found in overcoming 
the secular variations of velocity. 
(3) In the first series of experiments, the machine was driven by a Crossley’s 
oil engine of three horse-power, but the fluctuations of velocity were not small 
enough, even when the engine was working with full load, for this mode of driving 
to be considered satisfactory. Although the author was not content with the 
results obtained under these conditions, yet, for the sake of comparison, the results 
of one set of 20 tests are given in this paper. 
The machine was finally driven by the low-pressure engine of the triple expansion 
experimental engines. These engines are described in a paper on ‘ The Mechanical 
Equivalent of Heat,’ by Professor Osborne Reynolds and W. H. Moorby, 1898. # 
The secular changes of velocity were again found to be great, and it was only after a 
great number of trials that the following method (suggested by Professor Reynolds) 
was hit upon to reduce them to a minimum, it being the only method suitable for 
this work. 
The boiler was worked at 120 lbs. pressure, and the steam was throttled so as to 
reach the engine at 5 lbs. per square inch. In this way, small variations of boiler 
pressure were rendered less effective in causing variations of velocity. 
The engines were run so as to give out approximately 20 horse-power, and drove 
by means of a rope a long line of shafting from which the power was taken to the 
counter shaft of the machine by means of a 2 \-inch belt. The surplus work was 
dissipated in a hydraulic brake, also described in the paper just referred to. The 
brake was not loaded in the ordinary way, but was allowed to bed against an upright 
or dead-stop behind the brake ; a fairly constant flow of water was supplied to the 
brake, and the resistance offered by the brake was varied by regulating the quantity 
of water passing out of it. 
A speed indicator, similar to the one attached to the testing machine, and 
previously described (see p. 272), was driven directly from the engine, and the 
heights of the water columns in the two indicators were constantly watched by 
* ‘ Phil. Trans.’ 
2 N 2 
