ATMOSPHERIC REERACTIONS. 169 
arc, perhaps, steadily elevated. In most cases, the 
refracted portion of the distant ice is closely con¬ 
nected with the ice of the horizon, from whence 
it takes its rise; and when it assumes the 
columnar form, it presents the appearance of 
a vast amphitheatre, which is so disposed, that 
every observer, whatever may be his position, 
' imagines himself to be in the centre of it. But 
in some instances, and these not unfrequent, 
the stratum of refracted ice is completely de¬ 
tached from the horizon, and appears to form 
a white horizontal streak in the lower part of the 
atmosphere. And occasionally, multiplied images 
of the ice, as well as other objects, occur, forming 
a parallel vertical series. According to the laws 
of optics, were the vapour universally distributed, 
and of regular density, there might be an eleva¬ 
tion of the distant objects, which would be equal 
and uniform ; but there coidd be no extraordinary 
rearing of some, with the distortion of others, such 
as I have had occasion to describe : these pheno¬ 
mena, therefore, must be owing to unequal and 
changeable density; and some of them, such as 
the repetition of images, to alternations of parallel 
strataof different densities, in the medium through 
which the refracted objects are seen * 
* Dr Wollaston, with his usual ingenuity and preci¬ 
sion, suggests an explanation of the ordinary phenomena of 
