Diurnal Variation of Temperature. 119 
calculated temperatures, using only three variable terms of the 
formula, is less than the tenth of a degree. Usinglfive variable 
terms, it is less than four hundredths of a degree. 
n case the preceding month, April, had been selected for 
only six variable terms to be one minute /ater than if deduced 
from the complete series. 
Or if the next following month, June, had been taken, the 
formula with only five variable terms would have given the 
epoch of minimum not only within a single minute of that 
resulting from the full series, but also /ater than any of those 
resulting from the successive incorporation of the four following 
terms. 
_ When we examine the results obtained for Katharinenburg, 
the case is analogous, but even more favorable for illustrating 
the facts to which we call attention. Here the formula wit 
but five variable terms represents the observations with such 
completeness that the mean discordance amounts to only 0°02. 
With eight variable terms the discordance does not exceed one 
one-hundredth of a degree at any hour from 2 to 7 A. M.; and 
the same minute results for the epoch of minimum as when 
the complete formula is used. Indeed, with only six variable 
terms, the difference does not exceed five minutes. The value 
obtained graphically by Dr. Wild is five minutes later than 
that which results from the formula using the same data; and 
careful trials, employing the scale used by him, have convinced 
te that curves may be so drawn as to seem equally plausible, 
self; and it will be found that the errors, inevitably committed 
in the plotting and reading off, are far less than those resulting 
from the sketching of the curve. 
The extreme variability and uncertainty of these mean epochs 
of diurnal maxima and minima find an excellent illustration 
years from 1857 to 1862. [Temp. Verh., p. 54]. 
It has already been stated that in the trials for obtaining the 
daily mean, the time of maximum may, in the absence of any 
Am. Jour, if Pl Serres, Vou. XXIII, No. 134.—Frpruary, 1882. 
