G28 
LIEUT.-GENERAL R. STRACHEY OX HARMONIC ANALYSIS OF 
again at about 60°, or 4 a.m. But a sudden change takes place in passing from 
October to November, which might be attributed either to the time of maximum 
rapidly advancing, that is, occurring later, or to its sudden recession, or becoming 
earlier ; and, in whichever way it is brought about, one of the recurring epochs of 
maximum is established, from November to January, at about 10° after midnight. 
There is a like sudden change between January and February in the opposite direc¬ 
tion, which again brings the maximum to about 60° from midnight, or 4 a.m. It will 
be seen that in February and November the absolute amplitude of this component is 
very small, and probably these sudden changes are coincident with the component 
becoming zero. (See Plate 23.) 
Remembering that the fourth component of the diurnal curve includes a series of four 
undulations 90° apart, the explanation of what has just been stated, is probably to be 
found in a rapid change of conditions, under which the position of the first or earliest of 
these undulations recedes, until its place is taken by what was the second ; so that as the 
maximum of the first undulation gradually becomes earlier, and at length occurs at 
0°, or midnight, the maximum of the second undulation is approaching 90°, or 6 a.m., 
from which position it raj)idly further recedes between January and February, in the 
last of which months it is found at 60°, or 4 a.m. The converse of such a process 
would explain the sudden change between October and November. On this hypo¬ 
thesis the numerical value of in the months of November, December, and January, 
as given in the Tables, should be increased by 90° to render them properly comparable 
with the values in the other months. This would change the mean value for the year 
from 35° to 57°. 
In proportion as the diurnal curve of temperature tends to greater simplicity, the 
magnitude of the component of the first order exceeds that of the others, and, in the 
months of May, June, and July, in which the amplitude of the first component is 
more than ten times that of any of the others, the temperature during the day, 
between the hours of 8 a.m. and 8 p.m., hardly deviates from a curve of sines. The 
mean temperature of the day coincides almost exactly with the temperature of the 
last-named hours, and the excess of the diurnal maximum over the mean hardly 
ditfers from the value of P^, which, as before shown, is directly dependent on the 
Sun’s meridional zenith distance. (Plate 21.) 
The periodical variations in the amplitudes and epochs of the second, third, and 
fourth components, wdiich are indications of the departures of the hourly temperatures 
from the simple curve of sines represented by the first component, are, without doubt, 
connected with the vaiying length of the day and other infiuences dependent on the 
time of year, among which the direction of the prevailing winds, the greater or less 
transparency of the atmosphere, and the amount of vapour will have places. All of 
these, however, will be connected in a more or less direct manner with the Sun’s 
position and surrounding terrestrial conditions, though an expression for this connec¬ 
tion may be difficult to discover or define. 
