( m ) 



maximum, to which I add those for the whole of the period 1890 — 96. 



If we derive the length of the period from those couples of par- 

 tial results, lying three periods apart, we shall find: 



from u- 431-1.3 

 r, y 430.3 



The surprising agreement with the results obtained from great 

 intervals of time had of course to be regarded as partly accidental. 

 Now in order to investigate more closely what accuracy might be 

 arrived at, I fell back on the original values for the coordinates x 

 and y as they have been derived by Albrkcht. In my preceding 

 paper 1 gave on page (53) 12 a comparison of tliese values with those 

 computed by my formula. In entirely the same way I now made 

 comparisons with formulae in which 423 and 428 days were suc- 

 cessively assumed for the length of the 14-monthly period, but which 

 agreed for the rest, mean epoch included, (which mean epoch coin- 

 cides approximately with 1893.0) with those employed for the former 

 comparison. These lengths of the period were taken from Chandler's 

 two formulae. 



I shall not here communicate these comparisons themselves, l)ut 

 shall give only the sums of the squares of the deviations and the 

 mean values of the latter, repeating also the values formerly fouivl 

 with the length of the period 431'' : 



