146 Atmospheric Origin of the Aurora, §c. 
Below is a passage from the paper above referred to, and relates 
to its different classes of facts, propositions and speculations. It 
may show that the author haa not confounded their different de- 
grees of evidence. 
‘The present article not Leica: been saateinierd nih vie 
ence to any comprehensive theory, presents some miscellaneous 
facts, which are, thrown into the common stock for the use of 
others. Even among the relevant facts, there are, undoubtedly, - 
interesting relations yet to be traced. 2d. The article contains 
some generalizations, whose results, whilst they may suggest to 
others a more correct theory, cannot be thereby invalidated. 3d. * 
There are inferences of another class which may be modified, — 
but probably not overthrown by the progress ‘of discovery. For 
example, that the aurora is an electrical phenomenon ; that itis 
intimately connected with the elements of clouds, and with these 
elements only when they are generated in air intensely cold as 
well as nearly saturated ; and that cirrus clouds of a certain class 
are intimately connected with auroral action, and that both these 
phenomena, and also coronze, do, for some reason or other, require 
a cold adequate to the shyetallization of aqueous vapor, are propo- 
sitions which will not lose all their interest nor any of their truth, 
even if the discovery should be made that the elements of clouds 
= Bocceeme sal globular or vesicular, and that the vapor is not yet 
stallized at the time of the phenomenon. It may be neces- 
= to. remark, that-we have not intimated that all snow is no 
tallized. On this-subject crude notions have prevailed. 4 
is to the views which belong to a more hypothetical class, t 
author will cheerfully renounce them when a more plausible he 
ory shall appear, as they are designed to facilitate, not to limit, 
investigation. ‘This theory may contain much that is novel, valu- 
able and true, without being in the highest sense ¢he truth.’ 
ip 
¥ 
The individual facts on which the generalizations are founded, 
cannot be here repeated. Of the second class, or the generaliza- 
tions, are the ee 
lf cores gian : 
Which may be regarded as approvimately aaa generally true, 
in relation to mean results, though not universally, or in relation. 
to each = instance. 
~ 
