ON TIME MEASURERS. 343 



move outward ; others are arranged to move an auxiliary weight 

 towards the axis of motion, both for an increase and decrease of 

 temperature ; others check the motion of the main weight away from 

 the axis of motion ; and others assist it only when it is going towards 

 it. Illustrations of all these different forms are here exhibited, but 

 not to occupy your time we propose only to describe one of them. 



Here is a bar formed of brass, melted upon steel, the brass being 

 underneath. Do not consider the staples at present, but imagine for 

 the present that the weights by which they are crowned are mounted 

 upon little pillars straight from the main bar. 



Now let us see what happens with any increase of temperature : the 

 brass expands, so does the steel, but the brass expands most, and the 

 only method by which the two can accomodate matters, is that the bar 

 shall form a curve, the brass being outside. In other words the bar 

 bends upwards and tilts the compensation weights in towards the axis 

 of motion. The reverse action takes place in the cold. 



Now see, the weights move inwards and outwards equally for equal 

 rises and falls of the temperature. Now let us consider the action of 

 our staples. They themselves are compound pieces, and when the 

 weights are pointed towards the axis of motion in the heat, they open 

 and advance their progress a little ; and in the cold, when the weights 

 are pointed away from the axis of motion, they close and retard it, 

 producing in this way the necessary correction for the secondary error. 



But although balance springs as compared with gravity are at this 

 very great disadvantage with regard to changes of temperature, they 

 have one exceedingly important peculiarity which their rival has not : 

 that is to say, that with a spring of suitable form and strength, by 

 slightly altering the length, or for very small amounts by arching it, 

 you are able almost completely to isochronise the long and short arcs 

 of oscillation of the balance. You can make the chronometer keep 

 the same rate when its balance is swinging 1 8oS from rest as when it 

 is swinging 22 5 9 . The direction you alter the spring in is this : if 

 the chronometer gains during the short arcs you shorten the spring, 

 and if during the long you lengthen it. 



The property of the spring is exceedingly convenient, because by 

 leaving the spring rather fast in the short vibrations you can correct 

 the chronometer for any occasional irregularity owing to an increase 



