thermometrical barometer for measuring altitudes. 185 
being thicker would occasion a greater variation. By altering 
the quantity of mercury in the basin I removed this cause of 
inaccuracy, and the instruments have agreed equally well in 
all parts between 30,68 on barometer and 28,23. The result 
of the comparison is, that a difference of i° Fahrenheit is occa- 
sioned by 0,589 on the barometer corrected. 30,603 corrected 
barometer = 213,367 thermometer, and 28,191 barometer 
= 209,263 thermometer. There will be variations from this 
general result when the difference below the mean heights is 
considerable ; but I did not attempt the observation of them, 
as my barometer is not provided with an adjustment of the 
mercury in the basin, and this thermometer is, I think, of 
too long a scale. 
By these trials being satisfied of the capability of the instru- 
ment, I have endeavoured to render it as portable as possible, 
for the farther purpose of measuring difference of altitudes, 
of which I found it very sensible ; and will describe what I 
may call, following Fahrenheit and Cavallo, a thermome- 
trical barometer, and may venture to recommend for use. 
PI. VII. Fig. 1. represents the thermometer. The bulb A, 
one inch diameter, is blown thick and strong on a tube of thick 
glass, the bore of which is not material, say inch. It is 
better to make the bulb on a separate thick tube, and to join 
the fine thread afterwards, than to attempt to blow it suffi- 
ciently large and strong through the fine thread itself. The 
thickness and strength of the bulb in every part is essential 
to its not yielding. Close above the bulb a swell B is made 
to contain, as near as may be , whatever mercury expands out of 
the bulb between the common temperature and that of boil- 
ing water. If this be too small, the mercury contracts into 
Bb2 
