FRICTIONAL LOSSES IN INTERNAL COMBUSTION ENGINES. 5Ol 
abandoned, and development took place in the use of an 
electro-magnetic shoe brake which attracted and attached 
itself to the rim of the flywheel. 
Owing to the brake B, Fig. 1, being necessarily heavy, 
it was found best to balance its weight by a counterpoise 
© on the end of a lever L, so that a spring balance placed 
directly over the brake would show only the tangential 
wheel-pull on the brake. Increasing or decreasing the 
exciting current round the brake windings increased or 
decreased the pull on the brake. 
Controlled by a spring balance in this manner, however, 
the brake was very irregular in its action, and values for 
**T’’ could not be obtained with a probable error of less 
than 16%. But advantage was taken of the electro-mag- 
netic brake to arrange an automatic control. Instead of 
counteracting the pull on the brake by a spring balance, a 
constant known weight W was hung on the other end of 
the lever, as well as the counterpoise O already there. 
This weight W being at the same distance from the fulcrum 
of the lever L, as was the brake, gave directly the pull on 
the brake, and to keep the lever floating steadily in a 
horizontal position an electrode EH was hung from the lever 
and dipped into water forming part of the electric circuit. 
Then, when the brake had a tendency to be pulled down, 
the other end of the lever would lift the electrode E so 
that less was immersed in the water; this weakened the 
excitation, and consequently the brake was allowed to rise 
again to the correct position. Another rheostat, operated 
by hand, was necessary in the circuit for preliminary 
adjustments, but during a run the automatic rheostat 
worked admirably. The brake pull by this means was able 
to be determined with an error of not more than one half 
per cent. 
