TIME KEEPEES. 429 



•r two, or more, then fet it going, and if it be a good time- 

 keeper fo as not to be atFe6^ed by the weather it will go fome 

 feconds per day fafter than it did when it was let down, but 

 jt will again lofe its quicknefs in a gradual manner gaining lefs 

 aqd lefs till it comes to its former rate. Therefore finding 

 that ifochronal fprings would not do— and likewife having 

 made fprins;^s of fuch fliape as would render long and fliort vi- 

 brations equ?il in time— conftantly lofe the longer the watch 

 went, I then made them of fuch Oiape as to gaip in the (hort 

 vibrations about five or fix feconds per day more than the 

 long ones, this quantity could only be found by long ejcpe- 

 tience, and the way I proved this was to (ry the rate of the This gradual ef- 



watch wilh the balance vibrating about 4 of circle, then tried ^^^""^" * '°^* 

 *> 3 ^ ' on the rate 



its rate vibrating one circle and a |, and if the (hort vibra- which may be 

 tions SCO flower than the Ion? ones that watch will lofe on its compenfated by 



/-/-111 g'ving greater 

 rate, and it they are equal, it will likewife lofe, but that only fpeed to the 



from relaxation, and if it gains in the fhort vibrations more ^o"^"^ \''>",-^ 



', ,. /- /- , . /• , • 11 - 1 tions in the t^f ft 



than five or fix leconds m twenty-fpur hours it will in the conftruftion. 

 long run gain on its rate, but if not more than that quantity^ 

 ^nd the tinjejieeper is perfect in heat and cold and every other 

 part, the above properties will render it deferving of the 

 name of a perfed timekeeper, and this i? a principal caufe of 

 my timekeepers excelling all others, and this the principal 

 caufe of (bme of my timekeepers going better than others, 

 though made by me, the I'prings of them being made to ac- 

 cord more exaflly to the above proportions ; and this is the 

 caufe which has enabled me to foretel what my timekeepers 

 would do, which Dr. Ma-^kel; ne, ■ Mr. Crosley, and others 

 pan teflify. The above elfecl is produced as follows. I find 

 the common relaxation of bal9nce fprings to be about five or 

 fix feconds per day on their rates; in thecourfe of a year, there-, 

 fore if the fliort vibrations are made by the Ihape of the 

 fpring to go about that quantity fafler than the long ones, and 

 as the fpring relaxes in going by tjme fo the watch accumu- 

 lates in dirt and thickening of the oil which {hortens the vi- 

 brations, the fliort ones then being quicker, compenlates for 

 the evil of relaxation of the balance fpring. From this it is 

 plain, that the caufes of error in timekeepers are not unde- 

 fined and vague in their nature, which has been fuppofed ; 

 for when it is certain that all caufes of error may be over 

 ^pn[}pe|ifat^d wQ cannot defpair of finding the medium, and 



which 



