171 
it cannot be expected to lead to an accurate determination of the 
quantity A and has to be superseded by a method which is 
free from these objections. 
A second method, which has been proposed (9, is to follow 
the way adopted in determining periodically varying astro- 
nomical and physical quantities by calculating the epoch at 
which the same phase occurs for two different series and 
interpolating a whole number of cireumyolutions between the 
two epochs, starting from a known approximate value of the 
period of revolution. This method, although unobjectionable 
when physical and astronomical quantities are to be determined , 
where, owing to the method of observation, the error com- 
mitted is small with respect to the quantity to be calculated, 
still necessarily leads to uncertain results when meteorological 
data are concerned, where individual observation of the in- 
vestigated phenomenon, in the proper sense of the word , does 
not take place, but, on the contrary the errors (in casu 
the adventitious disturbances) are large with respect to the 
quantities which are sought for. 
Proof will be given from the following results that, for an 
exact determination of the epoch a very large bulk of observa- 
lions has to be taken into account; when a limited series 
is employed the error committed in calculating the epoch 
may be so large as to attain almost any value between zero and 
Эт and the resulting value for the period of rotation deduced 
in this way bears rather a conventional than a real character. 
As il is, I do not think that a series of observations is anywhere in 
existence long enough to suffer division into two.groups from each 
of which the epoch may be determined with sufficient accuracy. 
Perhaps the series of magnetic observations made at the Paw- 
lowsk observatory may be found to be the only one which satis- 
fies this peremptory condition for the application of this 
method 
ee A 12 
(1) Sitzb. der Kais. Acad. Wien. Bnd. XCVI. Nov. 1887. 
