CHAP, iv.] APPLICATIONS OF THE LAWS OF HEAT. 373 



of expansions. Should any pendulum of this description be found 

 either to be over- or under- compensated, the defect can readily be 

 remedied by withdrawing or adding a little more mercury. The late 

 Mr. Dent used iron jars to contain the mercury, as he found great 

 difficulty in obtaining a perfectly true figure with glass ones. 



There are other kinds of compensation pendulums, but they are 

 all based upon the same principle, that is to say, on the unequal 

 expansion of the solids or liquids composing them, and on an arrange- 

 ment which in one way tends to raise, and in another to depress the 

 centre of oscillation, 



It is very important that a compensation pendulum should be so 

 constructed that its different parts may all take up any change of 

 temperature simultaneously. 



The necessity for this was well brought out, during the time that 

 the normal sidereal clock at Greenwich was under trial. That clock 

 had been originally fitted up with a heavy iron jar mercurial pendulum, 

 but it was found that the mercury being so little exposed in proportion 

 to its bulk, lagged behind the steel rod, both in expanding and con- 

 tracting, causing an error in the time of the clock by doing so. 

 The clock was in consequence supplied with a new form of zinc 

 and steel compensation. A central steel rod passes down to 

 below the pendulum bob, carrying a collet, or shoulder, at its 

 extremity. Upon this shoulder rests a zinc tube, which reaches 

 rather more than half way up the pendulum rod. To the top of this 

 tube is fastened another shoulder, and to this second shoulder is fixed 

 an outside steel tube inclosing the zinc one. To this outside steel 

 tube the bob (of lead) is fastened at the centre of the bob. The outside 

 steel tube is cut away, and the inside zinc one bored with holes, in 

 order to let in changes of temperature to the central steel rod. In 

 action this pendulum is very similar to the gridiron, but the 

 expansion of zinc being much greater than brass, much less of it is 

 needed. 



It is important that the bob should be fastened at its centre, 

 because owing to its bulk it would be sure to lag behind, and this can 

 only be got over by neutralising its expansion as far as possible. 



The compensation action necessary for watches and chronometers 

 is far greater than that required for clocks ; for, whereas a clock, with 

 a simple iron-rod pendulum, will lose three seconds a day for every 



