of) Le Roy’s Memoir on the best Method 
rollers, turning freely in two small frames ce, cc, (fig. 6, 
Plate III.) the one. for the lower pivot adapted to the lower 
part of the large frame; the other to the upper, for the pivot 
or trunnion é, oak 7, Plate I.) at some distance from which 
is attached the wire of suspension. All this is arranged with 
the necessary precautions, so that the wire and the axis of 
the balance may form always the same vertical line. 
This balance thus suspended makes vibrations of about 
20” duration each, by means “of the elasticity of the sus- 
pending wire. Two spiral springs ss, 5s, (fig. 6, Plate III. ) 
similar to those which serve as a mover in common watches, 
adjusted at the bottom of the balance arbor, by means of 
their ferrules, as the spiral in common watches, and in a 
centre of equilibrium absolutely zdle (as M. Daniel Bernoulli 
recommends in the researches above cited), act so that these 
vibrations are each made in about half a second. 
By this construction I avoid those defects of watches re- 
marked Articles III and V. of the preceding part ; for, the 
balance being freely sustained by the suspension wire, the 
friction which it would occasion by its weight, the very 
rapid wear which would result from it, &c., are absolutely 
suppressed, and-by means of the rollers, whose properties 
are well known, those which are produced by the effort of 
the regulating spring, hy shocks, and the Jateral motions 
of the balance, by the effect of the escapement, &c., 
are reduced to the Jeast quantity ; whence it happens, that 
instead of only preserving its oscillatory motion for about a 
minute; as the balance of the watch in the experiment re- 
Jated, (Article V. Part II.) the regulator keeps going: here 
more than half an hour; the two springs contribute also to 
this ; their efforts on the pivots being opposite, are recipro- 
cally destroyed. 
Article IIT. 
New method by which the most perfect isochronism is, given 
to the vibrations of the balance. 
It appears so simple, when we are occupied with the 
theory of watches, to try, first, whether the different lengths 
of springs produce no changes in the proportion which exists, 
etween 
ae 
a 
