ag Le Roy’3 Memvois on the best Method 
too frequently found it so. Therefore I shall ailvance nos 
thing of which I am not assured by facts, as the Commis- 
sioners may verify: 
Part IT; 
Examination of the causes whieh make watches vary. 
Article I. 
Of the spring in general, and of the alterations which may 
happen in the force of the spiral spring. 
The first question which presents itself to be cleared up 
in treating of the theory of watches, and which, however, 
appears to be made for the first time, is this: Is the spring 
in itself, abstracting from the effects of heat, a constant 
power, on which we may establish a principle of petfect re- 
gularity, or is it not ? 
These axioms of the philosophers, that there is no perfect 
spring in nature; that it does not admit of any precision, 
&c., would appear at first to announce in the spring a 
power not very proper to give the required accuracy in a 
marine clock. But, on the other hand, many philosophers 
and artists think that the spring is a constant powcr, when 
it is but little contracted. 
To have more exact notions on this subject, I have con- 
struicted an instrument, fig. 1. Plate I.; 1 call it the elate- 
tometer, ‘This is in some degree only a long spring rr, 
stretched by a weight p, which, according as the force of 
this spring increases or diminishes, ascends or descends, the 
distance that it moves being rendered a hundred times more 
sensible by means of a long index //, whose weight on the 
spring is insensible ; this weight being counterbalanced by 
an opposite branch J/, which makes it almost in equili- 
brinm. 
By nftans of this instrument we find that, the spring loses 
a considerable part of its force in the first month of its ten- 
sion; that then the loss is much less; that at last it be- 
comes almost insensible, unless the spring receives a con- 
siderable degree of heat ; for then the index falls several de- 
grees; and when the thermometer returns to the degree 
where 
