200 Le Roy’s Memoir on the best Method 
main always of the same length, which I have proved, Arti- 
cle III. Part V. to be absolutely necessary: likewise, when 
I endeavoured to compensate the effect of heat and cold by 
pieces of brass and steel, riveted together as Mr. Harrison’s, 
T endeavoured not to change the Jength of the spiral, but to 
make a considerable part of the circumference approach or 
recede by this means from the centre of the balance. For 
this purpose I used a balance (fig. 4 and 5. Plate I.) com- 
posed of two semicircles, each formed of a piece of brass 
and steel, united as in Mr. Harrison’s thermometer picce. 
The effect sufficiently answered my intentions : ] ever 
observed, by means of the index i, which was moved by 
similar pieces ¢/ (fig. 2. and 3. of the same plate), that by 
heat and cold the motion of these pieces would follow very 
exactly the motion of the thermometer; there resulted from 
this a compensation for heat and coid, whose effect might be 
increased or diminished at pleasure by putting a greater or 
less mass at the extremities of these semicircles ; but by this 
construction the balance does not appear to me to haye suf- 
ficient solidity: besides, in the different degrees of heat and 
cold, it would be difficult for it to preserve its equality of 
weight in all points of the circumference, Lastly, having 
exposed the machine, provided with this regulator, to a heat 
of 35 degrees * of the thermometer, after having replaced 
it in a common temperature, J saw that it would advance 
about six seconds in 24 hours; this appeared to me a con- 
sequence of what we have observed to happen in pieces of 
metals heated’and kept in a state of constraint; for, here, the 
piece of brass dilating more than that of steel, neither one 
nor the other is ina state of freedom ; besides which, it is 
almost impossible for them to be nayeteil together, and ad- 
justed in such a way that their figure does not arise from 
their mutual constraint ; inconveniences to which our ther- 
mometers are by no means subject, _ 
Before finishing this article I shal] make a remark which 
may Le of some utility to those who would construct similar 
thermometers ; which is, that it is necessary, before using 
the watch to which they are applied, that it be placed in the 
* 1104 of Fahrenheit.’ 
greatest 
