g Art. 3 — K. Sotouio: 



where A unknown constant 



B yearly change of tlie micrometer vahie 



C temperature coefficient 



]) (hminishing factor 



t time 



d temperature 



a vakie of micrometer corresponding to the first 



assumption. 



1) excess of the micrometer vakie on the second 



assumption over that on the first assumption. 

 For the sake of convenience, I took the year 1910.0 and 7^0 for 

 the origin of the time and temperature respectivel3\ 



Assigning equal weights to all the equations of condition, I 

 treated them l^y the method of least squares. The solution of the 

 normal equations gave, among others, as the most probahle value 

 of the diminishing factor D, 



]) = 0.54 ± 0.17 mean error. 



This result shows that the regular displacement of the level 

 bubbles is due to two different causes : 



I A southward displacement of the bubbles due to some 



cause in the level, independent of the motion of 

 the telescope. This accounts for al)out half of the 

 total motion. 



II A regular northward depression of the telescope, 



corresponding to nearly half of the total displace- 

 ment of the bubbles. 

 Owing to these phenomena we should obtain for the micro- 

 meter too large a value from western elongation and too small a 

 value from eastern elongation on the hrst assumption; conversely, 

 too large a value from eastern elongation and too small a value 

 from western elongation on the second assumption. 



The first phenomenon may possibly be a disturbance due to the 

 heat of the observer and of the reading lamp. When observing lati- 

 tude by the Talcott-Horrebow method, the proximity of the observ- 

 er to the instrument is of short duration; so that the effect would 



