37 8 Mr. cavendish’s Account of 
two fets of divifions on fuch thermometers as are in- 
tended for trying the heat of liquors; one of which 
fhould be ufed when the tube is immerfed almoft to the 
top of the column of quickfilver; and the other, when 
not much more than the ball is immerfed ; in which laid 
cafe the obferver fhould be careful, that the tube fhould 
be as little heated by the fteam of the liquor as poffible. 
It mud be obferved, however, that the heat of the liquor 
may be eifimated with much more accuracy by the firft 
fet of divifions, with the help of the correction, than it 
can by the fecond fet, as the latter method is juft only in 
one particular heat of the atmofphere, namely, that to 
which the divifions are adapted; but, if they are adapted 
to the mean heat of the climate for which the thermo- 
meter is intended, the error can never be very great, 
and, when the liquor is much hotter or colder than the 
air of that climate ever is, will be much lefs than if the 
firft fet of divifions were ufed without any correction ; but, 
when the liquor is within the limits of the heat of the 
atmofphere, greater accuracy will lometimes be obtained 
by ufing the firft fet of divifions than the fecond, for 
which reafon the latter fet fhould not be continued within 
tliofe limits. I would willingly have given rules for the 
conftruCtion of this fecond fet of divifions, but am obliged 
to omit it, as it cannot be done properly without firft de- 
termining, by experiment, how much the quickfilver in 
the tube is heated by immerfing the ball in hot liquors. 
In a fpirit thermometer, the error proceeding from the 
fluid in the tube being not of the fame heat as that in 
the 
