THERMOMETER FOR THE DETERMINATION OF RELATIVE HEIGHTS. 
123 
mine whether the instrumental graduations could be relied on ; and not having had 
an opportunity of comparing the instrument with any standard barometer in 
England, I was under the necessity of employing my Geneva observations for 
that purpose, comparing them with the corresponding indications of the barometer 
at the Geneva observatory, the height of which I estimated at seventy feet above 
the spot at which my observations were made. This difference of level will amount to 
^ =T28° Fahr. 
548 
to be added to the boiling-point as deduced from the height of the barometer. These 
corresponding observations were 
Geneva. 
Observed boiling-point. Reduced barometer. Calculated boiling-point. 
208 , 820 '72129 metre. 209'357 
209T15 '72345 metre. 209'500 
We have then 209'357 + T28 — 208-820 = '665 error, 
and 209'500+T28-209'H5= '513 error 
2 ) 1-178 
Mean error .... = — '589 
The non-accordance of the errors I am inclined to attribute to the latter pair of ob- 
servations not being simultaneous, mine being made at 6 a.m. and the barometric in- 
dication being deduced from that recorded at 9 a.m., which is the earliest. 
It was at once evident to me that the above error was not likely to afford a con- 
stant correction to be applied to all observations made at different points of the scale, 
because the lower point of graduation, being necessarily above the freezing-point, must 
have been determined by reference to a standard thermometer, and it was therefore 
not at all probable that it would differ from the truth by so large a quantity. An 
excellent opportunity of proving the truth of this conjecture offered itself at the Con- 
vent of the Great St. Bernard, where a register is kept of the variations of the baro- 
metric column. Arranging these as before, I obtain the results: — 
Great St. Bernard. 
Observed boiling-point. Reduced barometer. Calculated boiling-point. Error. 
197'640 '56538 metre. 197709 —'069 
197-680 -56630 metre. 197787 —TO 7 
There was a difference of level of ten or twelve feet between the two places of obser- 
vation, the first having been made in the salle-a-manger of the Convent, and the 
other in one of the dormitories on the next floor where the barometer is fixed. The 
correction for this change of level is 
