FRICTIONAL LOSSES IN INTERNAL COMBUSTION ENGINES. 505 
Further, two tests, in general, have to be made to 
determine the friction loss in any one part as, for instance, 
the valve gear. One test is made with the shaft and valve 
gear in place, and the other after the valve gear has been 
removed. The valve gear friction will be the difference 
between the friction losses found from the two tests, pro- 
vided that the losses occurring in those parts that are 
present in both tests have remained altered. In order that 
this last condition be attained as nearly as possible, special 
care was taken to keep the lubrication of the engine con- 
stant. Separate tests made with varying rates of lubrica- 
tion showed clearly how greatly the friction varied with 
the quantity of lubricant supplied to the different bearings. 
Finally, the fact that the measurement of the friction 
loss of one part depends on the difference of two other 
measurements is in itself a possible source of error. 
So much for the limitations of the retardation method as 
applied to measuring the distribution of friction in engines. 
In its favour there are several very striking points. From 
the fact of its being a retardation test, values of friction 
at any required speed may be obtained from one test. 
The necessary apparatus is simple, easily applied and 
capable of performing highly accurate measurements for 
this class of work. It depends for none of its results on 
instruments needing previous calibration. There are no 
uncertain conditions to be allowed for, as there are for 
instance, in the case of a belt drive from a dynamometer. 
The engine, when run under the conditions ordinarily 
present in retardation tests, is left entirely to itself. An 
important advantage is the existence of a time-velocity 
curve drawn for each test. For the alteration in the slope 
of the curve shows at a glance how the coefficient of fric- 
tion varies with the speed. 
