42 
maximum has advanced into July. The principal minimum 
is still in December, but the secondary minimum occurs in 
May instead of June, though, like the first maximum, it has 
almost disappeared. 
It seems therefore, from a comparison of these five curves, 
that the general curve in diagram A, which is laid down 
from the monthly means of all the observations, may be 
regarded as compounded of two primary curves, — one 
having two well marked maxima in April and September, 
and two minima in December and J une, and the other having 
only one maximum in J uly and one minimum in December. 
From curves 3 and 4 it appears that the influence of the 
second primary curve upon the general features of the first, 
does not become apparent while the number of selected 
days is limited to five per month. The first primary curve 
therefore represents the monthly changes in the calorific 
intensity of the sun’s direct rays on cloudless or nearly 
cloudless days, and it leads us to this remarkable conclusion 
that the heating power of direct sun-light on clear days in 
the latitude of the British Islands, is greater in the months 
of April and September than in the month of June, when 
the sun attains his greatest meridian altitude. 
The second primary curve represents the intensities when 
the solar rays are more or less intercepted and dispersed by 
clouds and haze, and it approaches in form the annual curve 
of temperature; but the fiist curve, of which curve 3, 
diagram C, may be taken to be a fair representation, is 
unlike that of any other thermometric element. It has, 
however, a remarkably close resemblance to the curve 
representing the monthly changes of one of the magnetic 
elements, namely, that of the monthly means of the diurnal 
ranges of the magnetic needle. In the volume of the 
Greenwich Observations for 1859, the Astronomer Royal 
has given a table showing the monthly means of the diurnal 
ranges of the magnetometer at Greenwich from ten years’ 
