40 Mr. de Luc on 
by which afterwards the firft produced is more or lefs com - 
perflated , is that operated on the width of the mejhes ; and it is 
becaufe the laft of thofe effedls is the only one that can affedt 
the length of the flips, that, in every change of msijlure , they 
move evenly , without any recoil. 
59* To that demonft ration of the exiftence of two oppofite 
effects of moifture in the threads , I (hall now only add an exam- 
ple of a fimilar phenomenon, in which the caufes alfo are 
vifibie. The compound frames, mentioned in § 24. are formed 
of two glafs rods 4 feet long, fixed together at the bottom and 
the top. A thin flip of brafs of determined length, fixed to 
one of the glafs rods towards the top, comes down from thence, 
paffes over a pulley at the bottom, and turns up for half an 
inch. To this end of the brafs flip is fixed the lower end of 
the hygrofcopic fubftance, the upper end of which is conne&ed 
with the index of the inftrument. It is by that means that, 
whatever be the changes of heat , provided they are flow, the 
lower end of the hygrofcopic fubftance remains fenfibly at the 
fame diftance from the axis. But if I take that inftrument out 
of water at a low temperature , and plunge it immediately into 
warmer water, the index inftantly moves as if the hygrofcopic 
fubftance had lengthened ; which is the effedl of the brafs flip 
dilating fooner than the glafs rods ; then the index recoils , and 
this is the effedl of a flower dilatation of the glafs . 
60. I have concentrated in thefe pages an account of twenty 
years afliduous labour in hygrometry , moftly occafioned by the 
anomalies of the hygrofcopic threads ; and the principal refults 
have been, fome determinations of the four principles that 
dire&ed 
