480 ON IMPRESSIONS OF COLD 
weather being very unsteady, I had to watch the intervals of an 
opening sky. Covering the mouth of the pitcher with a pane 
of glass, the pyroscope would sometimes indicate one or two 
degrees of heat ; but, on removing that screen, it always mark- 
ed impressions of cold amounting perhaps to five or six de- 
grees, as often as the clouds parted, and left a blue space near 
the zenith. The continual darting, by day and night, of cold pul- 
sations from the sky was thus ascertained; and nothing seemed 
wanting but to improve the construction of the instrument, and ~ 
render it commodious and portable. I resolved to inclose both 
balls within the same cavity ; the lateral ball being gilt, and of 
the same colour as that of the annexed reflector;.and the naked 
sentient ball being placed in the focus. By this disposition, 
the influence of light was almost completely destroyed, its ac- 
tion being made equal on both the balls. But, to protect those 
balls from the disturbing effect of wind, it was requisite, that 
the cup containing them should be deep and rather nar- 
row. If the cold pulses to be measured were darted only in a 
vertical direction, the parabolic conoid would certainly be the 
proper form of the reflector. I was convinced, however, from 
some imperfect trials, that such impressions were likewise sent 
down at different angles of obliquity. Supposing their inten- 
sity to be equal in all directions, an hemispherical reflector would 
answer best. But the focus, instead of a single point, would then 
have branched into the diverging arcs of the caustic curve, and 
the sentient ball of the pyroscope would have required half the 
dimension of the reflector itself, (see fig. 11. P]. XI.) The main 
cbject, however, was to measure the impressions received from 
the upper portion of the atmosphere, that part near the horizon 
being generally obstructed by clouds or vapour. I therefore a- 
dopted an intermediate plan, and selected for the reflector a trun- 
cated oblong spheroid, cut through the upper focus by a plane per- 
pendicular 
