87 
the same manner, that is to say by a growth from one 
form to another, so that on the simple tabular hexagon, 
growth takes place along the six sides, or from the corners 
arms may be thrown out, and such changes may, as my 
figures show, be repeated several times. But to explain 
the curved line found in the interior we must examine 
a little further. When a crystal begins to melt it loses the 
angular corners and gradually melts down, so that we can 
suppose a crystal, G, thawing down to take the shape of the 
meander line. This brings us to the explanation of these 
curved lines, which indicate in the first place a thawing, but 
as the arms extend beyond these lines it is clear that after 
this partial thawing freezing has begun again. 
The production of these lines thus shows that the crystals, 
after having been formed in a cold atmosphere, have passed 
into a warmer one, where part of the crystal has been 
removed by thawing; after this it has again passed into 
colder air, and has formed anew upon the partially melted 
crystal. 
It seems to me that these crystals bring down with them 
registers of the comparative temperature of regions in- 
accessible to the meteorologist, and that a systematic 
examination in favourable localities would add much to the 
knowledge of this science, as we may thus learn where 
counter currents exist. To take the crystal G, which fell on 
the 22nd of December, 1870, the meteorological observa- 
tions for Switzerland show that the prevailing wind east of 
Davos was north or north-east, with thermometer falling^ 
while to the west the wind was south and west. The con- 
clusion which I have arrived at from the lines under con- 
sideration is that above Davos a layer of this southerly wind 
intrudes into the colder northerly winds, and the observa- 
tions of other days do not militate against this theory ; but 
as I have been able to compare but few days, I should be 
glad to see a series of observations undertaken to obtain 
further information. 
