. 
Diurnal Variation of Temperature. 107 
and by the seven independent equations thus available, theseven 
unknown quantities which they contain could be fully deter- 
mined ; so that the form of the daily curve would be absolutely 
known, excepting so far as it might be affected by epicyclical 
terms which complete their period in six hours or still shorter 
aliquot parts of a day. 
Unfortunately the data relative to the extremes are only 
roughly, even when at all, known; yet the condition that there 
shall be but one* contrary flexure in the curve, together with 
the possibility of inferring a very approximate value for M, 
from the observed temperatures at 7", 14%, and 21", provide the 
means of remedying these deficiencies to a considerable extent; 
while it may be justifiable to assume.an analogy between the 
form of the curve sought, and that of those already determined 
for Bahia Blanca, Buenos Ayres, and Cordoba, to an extent to 
diminish still further the probable errors of the determinations. 
n this way very approximate values of the constants of the 
first three terms may frequently be obtained ; and although no 
one would maintain that an absolute representation of the true 
daily curve could thus be secured, still we should certainly 
obtain a curve of which the deviation from the true one is 
very small, and which intersects it in six nearly equidistant 
ints, 
The first, and usually the most troublesome, process to be 
: * Notwithstanding indications of secondary maxima and minima have been found 
in observations made at sund places, and mentioned by Kamtz and Karsten, while 
they have been especialiy discussed by Hellmann, the reality of the phenomenon 
Gous ca 
62 8 pro he most conspic f their occurrence is at 
S tsbarg, in t eries of observations organized by D ld with 
such care that he thinks the error of a sing! termination can x 1 
atur- Verhiiltnisse, p. 2 n the mean of ars’ observations of 
a ries, minima, other than those which he accepts as true ones, occur a 
18 serie 8 the t 
about 345 in November, 213% in !)ecember, and 03" in January. Although the 
amounts of the abnormal fluctuations are petty in themselves, they are far from 
being unimportant relatively. Indeed for the month of December, for which the 
aximun mean temperature, as given by the observations, is 7°-36 at 13h 45", 
and the minimum 8°-28 at 21 42™, Wild adopts as the true minimum, 8°-03 
at gh or only thirty-three minutes earlier than the epoch which he himself 
d 
8 
old series show no indications of such a phenomenon. 
he real existence of any secondary maxima and minima is flatly denied by Wild, 
and in this we believe him to be correct. But in accounting for their apparent occur- 
Out, or generated by calculations wi : : 
Had he said “ erroneously inferred from inadequate data,” there might have 
h in his statement. But it ought to be needless to mention 
that errors can only be produced by the use of this general formula in the same 
So which they can be caused by the use of a book of logarithms; namely 
m 
