[ 2°4 3 
the divifions and center of the index, pointing 
thereto, not being coincident, the index cannot move 
over fpaces geometrically proportionable to one another 
in all fituations of the fcale ; yet, the whole fcale 
not exceeding 30 degrees of a circle, it will be found 
on computation, that the error can never be fo great 
2s Tso-'^ part of the fcale, or i degree of the hygro- 
meter; which in this inftrument being confidered as 
an indivifeable, the mechanical error will not be 
fenfible. 
Choice and Preparation of the Cord, 
The cord here made ufe of is of flax, and be- 
twixt -rV*" and of an inch in diameter ; which 
can readily be afcertained by meafuring a number of 
turns made round a pencil or fmall flick. It is a 
fort of cord ufed in London for making nets,^ and 
is of that particular kind called by net makers faxen 
three threads laid. I do not imagine that the fabrick 
of the cord is of the moft material confequencc ;• 
but yet I fuppofe, when cords can be had of fimilar 
fabrick, and nearly of the fame fize, that fome fmall 
fources of variations will be avoided. In general I- 
look upon it that cords, the more they are twifted, 
the more they vary by different degrees of moiflure, 
and the lefs we are certain of their abfolute length-; 
therefore thofe moderately twifted, I fuppofe, are 
Fikely to anfwer beft. 
A competent quantity of this cord was boiled in 
one pound avoirdupoifc of water, in which was put 
two pennyweights troy of common fait ; the whole 
was reduced by boiling to 6 | avoirdiipoife, which 
3 was 
