(55) 



operate in a constant direction for ratlier long periods, 3o. that 

 in the most accurate series of observations they appear in a less 

 degree. So it appears certain that systematic errors have been at 

 work, as indeed had already been found before (see above). 



How great however is the share of those systematic errors? In 

 this respect only the deviations of the various observatories compared 

 mutually for the same period can give us some indications and we 

 can use for this end the half-yearly means. If we look into these 

 more closely we often find considerable differences between observa- 

 tories situated close to each other, which must thus be due to 

 systematic error, but side by side with these we sometimes also 

 find criteria that real deviations from my formula may exist. 



To show the first I have derived for the 4 observatories which 

 have furnished observations during 6 successive half-years (Potsdam, 

 Strasburg, Carlsruhe and Lyons), by comparing the result of each 

 one with the mean of the 4, the mean value of the total error of 

 observation in the half-yearl\' mean of an observatory. The result 

 found is ± 0''.049, that is a considerable amount and which, 

 as it agrees exactly with the value of A'b obtained above, would 

 as far as it goes tend to show that in this case ^) systematic 

 errors of the observations alone may account for the deviations 

 from my formula. 



On the other hand a case pointing to a real deviation from my 

 formula is offered by the summer of 1895. Here follow for the 

 European observatories the deviations from Albrecht as well as 

 from me: 



If a real deviation from my formula is indicated by this, it would 

 prove that the combination of a constant motion of 431 days and a 

 constant yearly motion does still somewhat fall short of the reality. 

 That the latter may be different in the various years is a priori 

 not improbable. It must be left to further observations, which will 

 have to be kept still more carefully free from systematic error, to 

 furnish more certainty upon this point. 



^) The same criterion is not so easily applicable to other observatories and in other 

 years. 



