342 SECTION MECHANICS. 



chronometer to go right at the temperature of freezing, thirty-two 

 degrees fahr., we shall find that when we raise the temperature to- 

 100 degrees the chronometer with a gold spring will lose eight min. 

 four sec. a day, the chronometer with a steel spring six min. twenty- 

 five sec., the chronometer with a palladium spring two min. thirty- 

 one sec., and the chronometer with a glass spring forty seconds ouly. 



It is with the compensation for these errors that we have practically 

 to deal in chronometers, the loss of time due to the expansion of the 

 balance being slight in comparison. 



In the case of the glass spring chronometer, the smallness of this 

 error made it necessary to apply a very different to the usual form of 

 compensation balance : compound lamince formed of silver melted 

 upon platinum weighing about two grains, each are mounted upright 

 upon a disc of glass, and with any increase of temperature the tending 

 of these towards the axis of motion affords a sufficient compensation. 



Unfortunately the amount of compensation required for a steel 

 spring is so great that in correcting it we introduce a secondary error, 

 the nature of which I will try to explain to you. 



The loss of time due to the loss of elasticity of the spring varies 

 nearly in the same proportion as the temperature, and an ordinary 

 compensation balance advances or withdraws its weights to or from 

 the axis of motion practically in the same proportion. 



But the time in which the balance swings, varies not as the distance 

 of its weights from the axis of motion, but as the square of their 

 distance, and thus it requires a greater motion of the weights inwards,. 

 to produce the same amount of effect as a lesser motion outwards. 



Thus, in an ordinary compensation balance, if the weights were 

 adjusted to advance a sufficient distance to compensate the chrono- 

 meter for a rise in temperature of thirty degrees, a fall of the same 

 amount would make them retire so much too far as to make the 

 chronometer lose four seconds a day. And the best that could be done 

 under the circumstances would be to so adjust the chronometer as to 

 divide the error, and make it lose two seconds a day in the heat, and 

 two seconds a day in the cold. 



Many methods have been introduced for correcting this secondary 

 error ; some are based upon the principle of increasing the motion 

 of the weights when they move inward, and reducing it when they 



