Diurnal Variation of Temperature. 109 
to a relatively small error in the final result. Where no spe- 
cial means of forming an estimate can be made useful, it may 
be well to begin with the supposition that the highest tempera- 
ture occurs at 2 Pp. M. daily. 
If we disregard the third variable term in the equation (1) to 
(4), we have now a means of determining, first, approximate val- 
ues for the constants a, A, b, B; and from these a correspond- 
ing curve, which must necessarily intersect the true mean daily 
curve in the three points given by observation. But, unless the 
assumed values of both M and H have been very near the 
trath, this very circumstance will cause the general form of the 
computed curve to be markedly discordant from that of the 
true one, already approximately known; and very frequently 
it will be found that double curvatures will present themselves 
as the only geometrical mode of satisfying the condition that 
the areas above and below the medial line must be equal. 
A second computation is now to be made with a slight 
e 
convenient to change the assumption by about the same amount 
as before, to permit an easy use of the rule of proportion. It 
will be found that the increase or diminution of the values of 
the resultant constants depends upon the value assumed for M; 
and a comparison of the corresponding curves will usually per- 
mit its true value to be fixed within extremely narrow limits. 
revised, although any change in this would be due to the infla- 
ence of terms of the second order. 
