[ 3°8 ] 
will make the experiment lefs exadt : and therefore 
it would be much more convenient, if the fea-watcr 
could intirely be kept out ; which probably may be 
done by tying a bladder full of air to die neck of 
the ball C > which will contradt by the preflure of 
the water, without letting any of it get in. 
If this can be done, the inftrument may be filled 
with mercury and fpirits, juft like that at fig. i. But 
it would be more convenient to fill it with mercury 
only : in which cafe it may be made with as linall a 
tube and ball as common mercurial thermometers : 
or it may be filled with fpirits only. The inftru- 
ment will thereby become much lefs bulky ; which 
will compenfate the want of the common icale of 
heat. 
It is better to put but little mercury into the ball- 
at top, for fear of its getting into the capillary tube 
by the motion of the inftrument. 
The thermometer for finding the greateft cold, if 
applied to this purpofe, muft alfo be left open at top. 
There is another inconvenience to be avoided ; which 
is, that the mercury in the ball A, by the tofting of 
the inftrument, might fometimes get into the fhorter 
leg of the fyphon ; which would fpoil the experi- 
ment. To prevent fuch an accident, the raoft con- 
venient conftrudfion, which occurs to me, is that of 
fig. 4. which differs from fig. 3. in having the ball A 
omitted ; fo that the mercury running out of the 
fhorter leg will fall to the bottom of the cylinder, 
and will not be fo liable to get back into the tube by 
motion. The cylinder is made to ftand not quite 
parallel to the legs of the fyphon, that the mercury 
contained therein may more eafily be brought to* 
touch 
3 
