bles him to regulate the inclination to correspond with the
altitude of the luminary. The heater is composed of rolled plate iron
0.017 inch thick, and provided with bead and bottom formed of
non-conducting materials. By means of a screw-plug passing through the
bottom and entering the face of the hub the heater may be applied and
removed in the course of five minutes, an important fact, as will be seen
hereafter. It is scarcely necessary to state that the proportion of the
ends of the conical heater should correspond with the perimeters of the
reflector, hence the diameter of the upper end, at the intersection of the
polygonal plane, should be to that of the lower end as 8 to 6, in order
that every part may be acted upon by reflected rays of equal density. This
condition being fulfilled, the temperature communicated will be perfectly
uniform. A short tube passes through the upper head of the heater, through
which a thermometer is inserted for measuring the internal temperature.
The stem being somewhat less than the bore of the tube, a small opening is
formed by which the necessary equilibrium of pressure will be established
with the external atmosphere. It should be mentioned that the indications
of the thermometer during the experiment have been remarkably prompt, the
bulb being subjected to the joint influence of radiation and convection.

The foregoing particulars, it will be found, furnish all necessary data
for determining with absolute precision the _diffusion_ of rays acting on
the central vessel of the solar pyrometer. But the determination of
temperature which uninterrupted solar radiation is capable of transmitting
to the polygonal reflector calls for a correct knowledge of atmospheric
absorption. Besides, an accurate estimate of the loss of radiant heat
attending the reflection of the rays by the mirrors is indispensable. Let
us consider these points separately.

[Illustration: _Fig._ 2.]

_Atmospheric Absorption._--The principal object of conducting the
investigation during the summer solstice has been the facilities afforded
for determining atmospheric absorption, the sun's zenith distance at noon
being only 17 deg. 12' at New York. The retardation of the sun's rays in
passing through a clear atmosphere obviously depends on the depth
penetrated; hence--neglecting the curvature of the atmospheric limit--the
retardation will be as the secants of the zenith distances. Accordingly,
an observation of the tempe