FRICTIONAL LOSSES IN INTERNAL COMBUSTION ENGINES. 515 
during five seconds on each side of a mark was noted. One- 
tenth of this number was the revolutions per second at the 
time under consideration. 
If the test took less than, say, seven minutes, the follow- 
ing procedure was adopted. The test was divided up into 
about twenty equal intervals—in the case of the brake runs 
this meant every two seconds—and the total revolutions 
up to each division noted. The velocity at the middle of 
interval was taken as the result of dividing by the number 
of the seconds the number of revolutions which occurred 
during the interval. 
Very great accuracy in reading the records was required 
in the brake runs, and for the purpose of measuring the 
decimal parts of the revolutions use was made of the 
following simple device. A piece of unexposed but fixed 
photographic negative was marked by ten parallel equally- 
spaced lines. Placing this on the record in the manner 
shown in Fig. 3, it was easy to accurately read where the 
seconds mark divided the revolution space. In this way 
values of the angular velocity of the engine at certain 
intervals of time were obtained from the charts. These 
were shown graphically as a curve plotted to velocity and 
time co-ordinates. 
A typical curve is shown in Fig. 4, drawn from one of the 
National engine test when fully assembled. Bearing in 
mind that coefficient of friction is proportional to frictional 
resistance—pressure remaining constant—and that resist- 
ance or friction torque is proportional to the retardation, 
it is evident that the behaviour of the coefficient of the 
engine friction may be seen at a glance from the slope of 
the curve in Fig. 4. 
At high speeds the coefficient is comparatively great 
and decreases gradually as the speed falls, But when the 
speed becomes very low the coefficient rapidly increases 
