igO LECTURE XVII. 



machinery, and points out the time elapsed without an error of the hundredth 

 part of a second. 



The alternate motion of a balance, thrown backwards and forwards by the 

 successive actions of a wheel impelling its pallets, is also capable of produc- 

 ing a degree of uniformity in the motion of the wheel; for the force operat- 

 ing on the pallet is consumed in destroying a velocity in one direction, and 

 in generating a velocity in the contrary direction ; and the space in which it 

 acts being nearly the same in all cases, the velocity generated will also be 

 nearly the same at all times, as long as the force remains the same. The ad- 

 dition of a balance to a clock was made soon after the year 1400, for Ar- 

 nault, who died in 1465, describes a planisphere, constructed by his master De 

 Fondeur, which had a balance with a scapement like that of a common watch, 

 but without a spring. Such a balance vibrates much more slowly than a ba- 

 lance provided with a spring; if the balance spring of a common watch be re- 

 moved, the hands will pass over the space of about twenty eight minutes in 

 an hour. 



It i| said that before the pendulum was used, a balance wheel was some- 

 times suspended in a horizontal position by a thread passing through its axis, 

 which coiled round it, and caused it to rise and fall as it oscillated^backwards 

 and forwards. This mode of regulation differed but little in principle from 

 the modern pendulums, but it was more complicated and less accurate. 

 Huygens, in somewhat later times, constructed a clock with a revolving 

 weight, wliich rose higher, and increased the resistance, whenever an aug- 

 mentation of the force increased the velocity; and he caused the thread, 

 which supported the weight, to bend round a curve of such a form as to pre- 

 serve the equality of the revolutions. 



A chronometer maybe constructed on this principle for measuring small por- 

 tions of time, which appears to be capable of greater accuracy than Mr. 

 Whitehurst's apparatus, and by means of which an interval of a thousandth 

 part of a second may possibly be rendered sensible. If two revolving pendu- 

 lums be connected with a vertical axis, in such a manner, as to move two 

 weights backwards and forwards accordingly as they fly off to a greater or 

 smaller distance, the weights sliding, during their revolution, on a fixed sur- 



