1865.] Notes on Central Asia. 127 
sheltered ravines or defiles which are hardly reached by the rays of 
the sun. Had I fallen into these errors in my determination the 
results-would have been to lower instead of to raise the height of the 
snow-line, as compared to its true limits. But these sources of error 
were fully anticipated and averted; my observations were made at 
points where regular layers of eternal snow occurred, and moreover on 
mountain-ridges and not in hollow depressions, in some of which 
I really did find eternal snows in some cases several hundred feet 
below the limit of 11,000. 
With regard to the other point, I must observe that the method of 
determining heights by the temperature of boiling water, is certainly 
one which is far from being perfect ; and leads only to approximate 
results ; but the inaccuracy of these results becomes more inappreci- 
able, the greater the height which is being measured. For inconsi- 
derable elevations this method of measurement cannot be adopted. I 
may, however, observe that the other method, namely that of commer- 
cial determination, can scarcely be expected to give more accurate re- 
sults when the conditions are unfavourable, as for instance on a journey 
through an extremely wild and dangerous region, where the traveller 
is obliged to form his own track, and stands every moment in danger 
of an attack; under such circumstances all simultaneous observations 
of the barometer, at the base and summit of mountains, or a series of 
observations at any one point, are quite out of the question. Experi- 
ence has also shewn me the complete impossibility of keeping the 
barometers (I had two with me) from breaking, in a country so 
mountainous as that I traversed, where, on each expedition, the pack 
horses and camels stumbled repeatedly, and were occasionally dashed 
to pieces by falling over precipices. Hence travellers (Humboldt 
amongst the rest on his famous journey in the Andes and the 
Cordilleras) have invariably had recourse to the method of determining 
heights by the temperature of boiling water. The results obtained in 
this manner are regarded by science merely as approximations, until 
they are superseded by more accurate data, obtained when the region 
is more accessible to scientific exploration. 
Although incomplete, these results are nevertheless of undoubted 
value to science, as the magnitude of probable errors even under such 
an imperfect method, cannot exceed certain limits. 

