THERMOMETER FOR THE DETERMINATION OF RELATIVE HEIGHTS. 131 
Though the discrepancies in these results, when compared with the previously de- 
termined altitudes, appear very considerable, they are probably in most cases not 
more so than would have resulted from the employment of the usual form of baro- 
meter. I am led to this conclusion from the inspection of De Luc’s series of baro- 
metric measurements, wherein the determinations of the height of the same station 
at different times frequently differ from one another 250 feet, and, in one case, as 
much as 360 feet. The greatest error in the above table, viz. that at Aosta*, 400 
feet, certainly exceeds either of these, but there is, I think, good reason to suppose 
that the observed height at Aosta is not so erroneous as the comparison with De 
Saussure’s result makes it appear to be, when we remark that the height of the 
Great St. Bernard determined from it is one of the most accurate of the series, and 
also that the observations made the next day at the Great St. Bernard and at Aosta 
give the same difference of level, within twenty feet. 
The only other instance in which I was enabled to repeat an observation after the 
ascent of any considerable height occurred at Chamounix, and in it a tolerably close 
accordance in the observations before and after the ascent of the Montanvert is also 
to be remarked, though the height derived from the intermediate observation differs 
very considerably even from that given by Professor Forbes, which is however equally 
in excess above that by De Saussure. 
Since the calculation of the above results I have had an opportunity of comparing 
the instrument with a standard barometer, and have been surprised to find the in- 
strumental correction, under the pressure 30 inches, very different from that ob- 
tained above from the Geneva observations. I cannot assign any reason for sup- 
posing an error in making these observations; indeed, every consideration induces 
me to believe that no such error could have occurred, but at the same time I find it 
very difficult to account for this change in the correction, unless it be partly attribu- 
table to a difference in construction of the two barometers and partly to a differ- 
ence in the degree of purity of the water, or to a change in the instrument itself; 
all of which suppositions are highly improbable. To whichever of the above causes 
this remarkable change is referred, I do not think it should be considered as vitiating 
the results derived from the recorded observations, since they have been connected 
with those made with the barometers by means of which the correction made use of 
has been determined ; and moreover the time elapsed between the first and last re- 
corded observations was only one month, and the sources from which the water was 
derived were in all cases so far similar as apparently to ensure its chemical identity 
with respect to boiling temperature. 
The utility of an instrument to be applied to purposes which involve the necessity 
* The circumstances of danger under which the observation on the Mole was made, make me think that an 
erroneous reading is not unlikely to have occurred there, and I accordingly exclude it from consideration in 
these remarks. 
