83 
There are a great many difficulties about such observa- 
tions, as, first, the condition of the snow is very variable, 
and the thermometer must to a certain extent be influenced 
by conduction. If it could be managed, the thermometer 
ought to be placed horizontally ; but practically this would 
be more difficult, as the condition of the snow in arranging 
it would be too much disturbed for constant observations. 
In order to test this, besides what I call the permanent 
thermometer, I placed one about an inch below the surface 
and another about a foot below; on February 22nd, for 
instance, at 11.30 a.m., the one 1 inch below showed 30*8 
F., the permanent one 23*7 F., and the one 1 foot below 
25 F. I of course found that the upper one was much 
more rapidly affected by the sun than the lower one, and 
further that a thermometer placed deeper than the perma- 
nent one registered higher. 
Freshly fallen snow is given in the text books as melting 
to one tenth of its bulk of water. I have not found quite 
as much with freshly fallen snow ; but this very soon varies, 
partly in consequence of pressure, but chiefly through re- 
crystallization, which takes place more rapidly as soon as 
the weather becomes a little warmer than usual. I have 
several times measured coarse-grained r eery stall ized snow, 
and found that it melts to one third of its bulk. This 
change of bulk often misleads people into thinking that a 
great deal has been melted, and two or three of our first 
rather warmer days caused the snow to sink down several 
inches without there being any melting. Nevertheless this 
change of condition is of course an important factor in the 
melting, as the quantity of snow which previously occupied 
a larger space comes now into contact with less air, and 
both the terrestrial heat and the solar can influence a greater 
weight of snow. 
A reference to the curves for January and March will 
show that the temperature in the snow only slowly follows 
