48 
SIR J. F. W. HERSCHEL ON THE CHEMICAL ACTION OF THE 
light circular card (to keep it flat) by slight touches of soft cement. This appendage 
was brought as close as possible to the dial-plate, avoiding contact, and was of 
course carried round uniformly once in twenty-four hours by the rotation of the axis. 
Over this a shade or cover was adapted, having an aperture in the form of a very 
narrow sector of the circle, to admit the light. If the paper be as sensitive as it ought, 
the smallest chink suffices, an amplitude of a degree in angular subtense at the centre 
being more than enough. The edges of the aperture should be so bevilled, or other- 
wise defended by screens, as to admit no direct ray of the sun when north of the 
equator, nor any other light than what emanates from that definite circumpolar re- 
gion of the sky to which it may be considered desirable to limit the observation, such 
as, for example, a circle whose semidiameter shall equal the latitude of the place in 
extratropical countries ; unless we prefer to register the actual illumination of the 
sky at the pole, in which case a tubular screen pointed to the pole must be adapted 
to the whole apparatus. Under these circumstances, we shall find at the end of each 
day the changes of illumination registered on the sensitive paper by a fan -shaped al- 
ternation of light and shade, radiating from the centre over so much of the area of 
the circular disc of paper as corresponds to daylight, or to such twilight as the paper 
is sensitive enough to be discoloured by. Of course it will be recollected in reading 
this description that the pole is the only point in the hemisphere whose illumination 
is independent of the sun’s hour-angle, and which can therefore properly be taken as 
likely to afford an average of the clear or cloudy state of the sky. 
121. For registering the direct effect of solar radiation it is necessary, 1st, that the 
paper should at every moment be presented at right angles, or nearly so, to the inci- 
dent beam. The greatest inclination of a sunbeam, however, not exceeding 23^ de- 
grees to the equator, a paper wound on a cylinder having its axis pointed to the pole 
will always fulfil this condition near enough, the difference between the cosine of that 
angle and radius being only eight per cent. ; and the error being, moreover, one sus- 
ceptible of allowance by a trifling calculation, or by a table, being the same for the 
same day in every year. Moreover, as the sun travels round in its diurnal course, 
and is required to throw a limited beam in succession on every part of the paper, it 
is clear that the cylinder on which the paper is fastened must either remain fixed in 
darkness, while a cover having an aperture to admit the ray revolves at the same rate 
as the sun, or else the cylinder must be caused to revolve in the same direction as the 
sun, but with double the velocity; this last condition, however, would induce need- 
less complication into the wheel work of the mechanism. The simplest mechanism I 
have been able to devise, which meets all the conditions, is as follows. 
122. To the lower end of the steel rod which penetrates the axis of the hour-index 
of the watch described in Art. 119, must be firmly adapted a hollow cylinder of thin 
and light material, as very thin sheet-brass, closed at that end which is attached to 
the steel rod, but open at the other, and having its axis coincident with the prolon- 
gation of the rod. In one side of this cylinder, viz. on that which corresponds to 
