FRICTIONAL LOSSES IN INTERNAL COMBUSTION ENGINES. 503 
the elimination of the necessity for assuming, as above, 
that Ty remains constant at the same speed in the two 
tests—the unbraked and the braked. Ordinarily in tests 
on engines where so many different places exist in which 
friction occurs, T; was found to be far from constant, 
though every effort was made to run the tests under exactly 
the same conditions, and the resulting error in the deter- 
mination of I was too large to be neglected. 
Finally, the following scheme was arrived at. For the 
first part of a test a known brake torque was applied and 
a chronogram taken on the high-speed of the chronograph. 
After a sufficient interval the current was cut off the brake 
and for the remainder of the test the engine allowed to 
retard under its own friction alone. On drawing the 
corresponding velocity-time curve from this chart the curve 
was at first steep and then at the instant.at which the 
brake was removed the curve became suddenly much less 
steep (Fig. 5, Appendix). Tangents drawn to the curve 
immediately on each side of this change point gave the 
retardations m2, due to combined brake and friction torque, 
and m, due to friction torque alone, at obviously one and 
the same speed N at which the engine was revolving when 
the change was made, and equally certain is it that T; had 
not changed. The estimated error in the value of I obtained 
in this way is from one to two per cent. This error was 
greatly reduced by taking the mean of many values of I. 
6. Distribution of the Friction Losses.—Knowing ‘‘I’’ 
for any one engine the amount of power lost in friction in 
the various parts of the engine can be determined by a 
process of elimination of the parts of the machine, a separate 
retardation test being run after each successive part is 
removed from the engine. 
The general procedure was to run retardation tests when 
the engine was fully assembled at temperatures varying 
