1 04 Mr. Cavendishes Obfervations 
dire 6I1011 of the arch, we have none of any having been feen 
in places much diftant from each other in the contrary dire&ion ; 
and, fecondly, that moft of them have pafled near the zenith, 
whereas otherwife they ought frequently to appear in other 
lituations ; for if they appeared near the zenith to an obferver 
in one latitude, they fhould appear in a very different fituation 
in a latitude much different from that. 
I wifh it to be underftood, however, that I do not offer this 
as a theory of which I am convinced ; but only as an hypo- 
thecs which has fome probability in it, in hopes that by encou- 
raging people to attend to thefe arches, it may in time appear 
whether it is true or not. If it fhould hereafter be found, that 
thefe arches are never feen at places much diftant from each 
other in a diredtion perpendicular to the arch, it would amount 
almoft to a proof of the truth of the hypothefis ; but if they 
ever are feen at the fame time at fuch places, it would fhew 
that the hypothefis is not true. 
Suppofing the hypothefis to be well founded, the height 
above determined will anfwer to the middle part of the band, 
provided the breadth of it was fmall in refpedl of its diftance 
from the earth, but otherwife will be confiderably below the 
middle. If the breadth of the band was equal to th^ diftance 
of its lower edge from the earth, the height of the lower edge 
would be three- fourths of that above found ; and if the breadth 
was many times greater, would be half of it. 
In the common aurora borealis, an arch is frequently feen low 
down in the northern part of the lky 5 forming part of a fmall 
circle. What this is owing to, I cannot pretend to fry ; but it is 
likely that it proceeds from ftreams of light which appear more 
condenled when feen in that direction than in any other, and eon- 
fequently that the ftreams which form the arch to an obferver 
m 
