lo Reynolds and Moorbv, Equivalent of Heat. 



time to adjust, that is, they are limited to the fluctuations 

 which occur in a single revolution of the engines. Fluc- 

 tuations of this kind occur in all reciprocating engines 

 from causes which are well-known and hence admit of 

 estimation, and by such estimation the limits of errors 

 from this cause have been found to be such that at 300 

 revolutions a minute the relative error would be less than 

 one forty-thousandth part. 



Besides these fundamental errors, the effect of the 

 friction of the automatic gear {^Fig. 4) had to be considered. 

 In designing the brakes, it was hoped that the slight 

 tremour to which they would be subject during their 

 motion, would so far relax this friction that the gear- 

 ing would adjust without pressure on the support, and 

 this proved to be the case. The question of the dash-pot 

 was also a matter of consideration; but so long as the 

 viscosity of the oil is constant, and the passage for the oil 

 to flow past the piston of constant dimensions, the time- 

 mean of the resistance is simply proportional to the 

 distance traversed divided by the time, so that, as the 

 piston does not move through O'l ft. in an hour, this also 

 became insensible. 



The only other source of error in taking the product 

 of the angle turned through and the moment of the 

 load as measuring the work done, is possible end-play 

 of the wheel in the case. This is very slight ; it cannot 

 exceed ^4 of an inch, while a side pressure of less than 

 Solbs. will always stop it ; so that this again became 

 negligible. 



The exact determination of the moment of the load 

 which involves the exact balance of the brake, and the 

 measurement of the length of the lever and the weight of 

 the load was,of course, necessary for the exact measurement 

 of the work. 



