86 Dent's new Compensation Balance for Chronometers. 
Another circumstance that tends to aggravate the error arising 
from the defect of compensation for the diminished tension of the 
balance-spring at high temperature, and the excess of compensa- 
tion for the increased tension at low ones,—is, the unfolding or 
straightening of the circular rim of the balance at reduced tem- 
peratures, and the contrary action at high ones. By this action 
of the rim, the compensating weights are made to describe por- 
tions of a spiral curve, whereby the variations in the central 
distance, due toa given change of temperature, are greater at 
the low than at the high temperature, which is the reverse of 
what is required in order to effect the compensation; and although 
such deviations from the required law of approach of the com- 
pensating weights may be rendered less apparent by increasing 
the weights, yet, in this case, other errors are introduced (which 
it will be needless here to allude to) that render this mode of 
proceeding inadmissible without much limitation. In the con- 
struction of the balance I shall here describe, it is not pretended, 
indeed, that the law of approach is mathematically what it ought 
to be, in order that the proper ratio may be obtained at all tem- 
peratures between the tension of the balance-spring and the 
inertia of the balance,—yet it may be safely affirmed that, in this 
construction, the variations in the central distance of the weights 
increase at the higher and diminish at the lower temperatures ; 
which is exactly the reverse of what has hitherto generally taken 
place in chronometers, and therefore will doubtless afford a much 
nearer approximation to the truth than heretofore attained. More- 
over, the correction of the error alluded to, will be a continuous 
correction ; an object of no little importance, and which is not 
effected in the contrivances lately put forth to remedy the defect 
by means of supplementary weights, which weights are brought 
into contact with the balance rim at amean temperature. In 
these contrivances by contact, although chronometers may be 
adjusted to equal rates at one of the extremes, and also at a mean 
temperature, yet between these limits, they are obviously subject 
to an error of the same nature as before, though of one half the 
amount only; and in the other half of the range of temperature, 
when the supplementary weights are brought into contact with 
the rim of the balance, the law of approach is the reverse of what 
it ought to be. Besides, the friction at the point of contact is 
highly objectionable in this mode of correction, and will not only 
