216 B, A. Gould— Use of the Sine-formula for the 
vation, inasmuch as the effect of a given error in the time dif- 
fers at different seasons. Yet, in spite of all this, these insig- 
nificant minutiz seem to be thought worthy of application to 
data obtained by the free hand of a draughtsman, rounding o 
errors in the observations, while the representation and generali- 
zation of the results by a formula accompanied by an exact 
exhibit of its discordances from the several observations, is 
deemed inadmissible. 
If the epochs of daily minimum, as deduced from monthly 
means of hourly observations by the two methods, differ (as at 
Katharinenburg, for example) by amounts varying between 
+57 and —71 minutes, I cannot agree with the distinguished 
author in attributing a large share of such discordances to 
errors occasioned by the use of Bessel’s formula; nor even in 
preferring the results of the graphical method. It seems much 
more probable that the phenomenon ought to be attributed toa 
very different class of influences. It is at least certain that 
what the draughtsman obtains without gauge or measure in 
equating out a series of data, the formulas will afford together 
with accurate indications of the probable errors. Yet if in the 
accurate interpolation is needed, any number of observations 
may be represented with absolute precision by the formula. 
Yet recourse to a computation in which twelve variable terms 
are used to represent twenty-four hourly observations is very 
nearly equivalent to a declaration that the computer believes no 
periodic law to exist. 
That entirely similar discordances present themselves be- 
tween the times of maxima and minima, as obtained by the 
two methods, for other stations than the one cited can cause no 
surprise. The only surprising fact in the case is that the 
inference should have been drawn that the employment of a cy- 
clical formula is inadmissible even when it represents the entire 
series of observations within the limits of reasonable error. _ 
Still retaining the hypothesis already mentioned regarding 
the character of the service rendered by the sine-formula, @ 
