Cfiruiiomelrical Ariattgemenls. 213 



this matter, we shall see how far the, trials to which the chro- 

 nometers now in the depot are submitted, accord with these ; 

 and we shall then be enabled to draw such inferences as the 

 comparison may warrant. 



Now, the rate of a chronometer is usually deduced from daily 

 comparisons made with a good clock adjusted commonly to 

 mean time ; I would, however, prefer immediate comparisons 

 with the transit clock, that being always considered the best in 

 an observatory, and unless thert are substantial reasons to the 

 contrary, I would unquestionably have tlie chronometer ad- 

 justed to sidereal time, by v/hich all trouble of reduction will 

 be spared, and the errors arising from the unsatisfactory na- 

 ture of solar transits entirely precluded. 



Should, however, necessity oblige us to have our chrono- 

 meter shew mean time, there is one circumstance which we 

 may avail ourselves of in comparing it with the sidereal clock. 

 The acceleration of sidereal on mean time is such, that the 

 former gains on the latter one second in about six minutes, now 

 the box chronometer usually beats half seconds, and conse- 

 quently synchronises with the clock's pendulum twice in this 

 time ; these instants are, with a little attention on the part of 

 the comparer, so easily ascertained, that, to make an error in 

 the comparison equivalent to one twentieth, or even one fiftieth 

 part of a second would be almost unpardonable. When pocket 

 chronometers are compared, the same principle may be acted 

 upon. These generally beat five times in two seconds, but, 

 owing to the seconds' circle being ill divided, and the seconds' 

 hand not accurately centred, equal accuracy cannot be ex- 

 pected. 



The instrument, whether large or small, is always included 

 in a box, whose base and top are connected together by four 

 sides, which we will denominate A, B, C, D. As sent home by 

 the maker, it will be undamped and horizontal ; in this state, 

 therefore, it should for a few days be compared with the clock, 

 and if its rate seem uniform, let it now, clamped in its gimbols, 

 be placed vertical, by making the box rest upon the side 4 ; on 

 the day following move it one quadrant, by placing the box upon 

 the side B ; on the next day, vary its position another qua- 

 drant, by resting its box upon the side C ; and on the subse- 

 quent day let it be passed through the remaining quadrant, by 

 making its box rest upon the sideD : it may then be restored to 

 its horizontal position, and if, on repeating these experiments, 

 examination should detect no material difference in the daily 

 rates, a great point will be gained. During these observations 

 the temperature should be kept as equable as possible ; the ther- 

 mometer, therefore, should be frequently appealed to, other- 

 wise the inferences deduced will be liable to some suspicion. 



Having gone thus far, it will be prudent to unclamp it, anfl 



