lO 



NATURE 



[Nov. 2, 1876 



two rods, RR, fastened upright upon the extremrties of the 

 loops. 



When the temperature rises, the flat bar bends upwards, 

 and tilts in the loops and the weights at their extremities. 

 But at the same time the loops open a little, and as they 

 are pointed towards the axis of motion, A A, advance the 

 compensation weights a little further in upon their own 

 account. Thus the action of the main bar is increased 

 in the heat. In the cold, the main bar bends downwards, 

 and tilts out the weights, but in this case the loops close, 

 and as they are now pointed away from the axis of 

 motion, by doing so, drzn^ back the weights a little. Thus 

 the action of the balance is reduced in the cold. The 

 secondary error is in this manner corrected. 



Balance-springs have this very important property, that 

 you are able to isochronise them, that is, so adjust them 

 that the balance shall perform long and short arcs of 

 vibration in the same time. The rule for doing this is 

 simple, though exceedingly difficult of execution upon 

 account of the minuteness of the operation ; if the chro- 

 nometer gains in the short vibrations, you shorten the 

 spring, and if it gains in the long vibrations, you lengthen 

 it. The best plan is to leave it gaining a trifle in the 

 short vibrations, for the reason I pointed out to you when 

 discussing the circular error in pendulums. A chrono- 

 meter gaining 3 or 4 seconds a day in the short vibrations 

 •will also be compensated against changes in atmospheric 

 pressure. 



The best watches are always timed in positions. There 

 is of course much more friction when the pivots are roll- 



ing upon their sides than when turning upon their ends ; 

 gaining in the short vibrations tends to correct variation 

 due to this. In general, timing in positions is a work 

 of the very greatest difficulty, and perfect accuracy can 

 only be obtained when the balance- spring weight of the 

 balance and mainspring are in exact adjustment. 



A very curious property sometimes exhibited by chro- 

 nomoters is to gain upon their rates, that is to say, if the 

 average daily rate of the chronometer were during the 

 •first month i second a day fast, during the second it 

 would be i^ seconds, and during the third month 2 

 seconds. 



This arises from the balance-spring having been left at 

 too high a temper after hardening. On the other hand, 

 if the temper were left too low the chronometer would 

 lose upon its rate. Glass springs exhibit the tendency to 

 gain upon their rates to a lemaikable degree. 



Watch Escapements. 



It would never do to have watch balances vibrating so 

 small an arc as clock pendulums ; the least we can do with 

 is 200°, that is, 100 times as much. If you will remember 

 the various clock escapements described, you will see 

 that not one of them is suitable to fulfil such conditions, 

 and something quite different had to be devised. 



The form most generally employed is Mudge's detached 

 lever, which is more m favour to-day than it ever was. 

 P P are the pallets (see Fig. 22) mounted along with a 

 lever, L L, upon a spindle, s. During the greater part of 



the swing of the balance, the lever and pallets are lying 

 against either of the banking pins,B B. There is a notch, 

 N, in the lever and a pin, i, to correspond upon the 

 disc, R, which moves around along with the balance in 

 the direction shown by the arrow. By-and-by the pin 

 upon the disc will catch the notch in the lever and 

 unlock the escape-wheel, a tooth of which is now being 

 held against the dead face, D, of the pallet ; the tooth will 

 immediately slide along the slknt, and deliver its im- 

 pulse, which will be transmitted to the balance through 

 the connection of the pin and disc R. The lever whilst 

 resting against either of the banking pins is held in posi- 

 tion by a little " draw " upon the dead faces of the pallets, 

 that is, they are slanted back^so that the pressure of 

 the wheel teeth thrusts them away from it. There is 

 also a safety disc, O, underneath the unlocking one, and a 

 safety tongue in the lever, which, in the event of the watch 

 getting a shake prevents its falling over to the opposite 

 side of the balance-spindle to that where the unlocking 

 pin is then situated. 



People are continually " improving " this escapement, 

 generally by making some slight alteration in the pins, 

 but the broad principle and details always remain the 

 same, as I have described them to you. 



Another form of escapement very much used in foreign 

 watches is the horizontal, invented by the same Graham 



already referred to ; but as it does not possess qualifica- 

 tions for accurate time measurement, I shall not describe it. 



Our last escapement is that universally employed in chro- 

 nometers (see Fig. 23) ; its original conception is apparently 

 due to Arnold, though it was modified and greatly improved 

 by Earnshaw. s S is the escape-wheel, which is now being 

 held by the detent, D. R^ R2 are two discs, the smaller 

 being situated in the plane of the detent, and the larger 

 in the plane of the escape-wheel ; both of them move 

 upon the same spindle with the balance. The balance is 

 now turning in the direction of the arrow; by-and-by the 

 finger, Pg, upon the smaller roller will come round and lift 

 away the detent, and the wheel will be free. The tooth, 

 T, will then drop upon the impulse pallet, P^, and deliver 

 impulse to the balance. Meanwhile, the finger, Pj, gets 

 clear of the detent, which it allows to fall just in time to 

 receive the succeeding tooth of the escape-wheel. 



The balance now passes on to the limit of its excursion, 

 and returns ; but in returning the finger does not interfere 

 with the detent for the detent, D, is actually too short to 

 reach it. Just now the finger really unlocked the detent by 

 means of the little spring, yy, which is fastened some dis- 

 tance down the detent, the little spring being supported 

 by the horn or extremity of the detent ; but when tfce 

 finger returns, it merely lifts out the spring, as there is 

 upon this side no horn or extremity to support it. 



