an improved Thermometer, jq 
tube turned up on the right fide, with fome mercury let down 
to the bottom, and the upper endclofely healed, as in the don* 
ble inftrument. Making a- ftandard mercurial thermometer, 
by which the fcale o£ the fpirit one was to be divided, I en- 
deavoured to- obtain as wide degrees as pofiible, that the motion 
of the mercury might thereby be rendered more confpicuous, 
and the height of it afcertained with greater prccifion. It is* 
true, the larger the degrees, the larger in fome meafure mu ft be. 
the bulb, and therefore the fluid contained in it not likely to be. 
fofoon affected by any change of heat or cold in the atmolphere 
as in a i maker. But as this thermometer was principally to 
be ufed immerfed in a large quantity of water,, gradually heat- 
ing or cooling,, little or no difadvantage could arile from 
making the bulb fomewhat larger than thofe- commonly made 
ufe of in the air. Not being able,, however,., to procure glafs 
tubes fo long as I had occasion for, whole inner diameters were 
perfecdy equal, 1 took the following method to adjuft the divi- 
fions on the Icale to the inequality of the tubes. Choofmg a 
tube of a length fuitable to my purpofe, with a proper bulb at 
the end, J put into it a fmall quantity of mercury * fufficient 
to form a column about one inch in length. Drawing then on 
a board the three lines aa, bh , cc, fig. 6. I placed the glafs. 
tube on the line aa, and while the mercury remained at reft at 
the end of the tube, near the bulb, I made two pencil marks, 
on the line aa, one at d, and the other at e , perfectly coin- 
* To put in. a fmall quantity of mercury, and meafure its length at different 
parts of the tube, as defcribed by Abbe nollet, vol. IV. p. 370. Lecons Pay- 
fique, is a very excellent method to difcover the error ; but in what manner readily 
to adjuft the fcale, fo as to avoid any inaccuracy from fuch inequality (which in. 
tubes of the length I had occafion for feerned to me unavoidable) was a matter 
concerning which I could meet with no information, 
ciding 
