24 Art. 3-K. Sotome: 



will not respond immediately, because of the defect discussed above. 

 When the disturbance has exceeded a certain limit, the level begins 

 to indicate it to a certain degree. The thermal effect on the level 

 would appear later, as the levels are more distant and better pro- 

 tected. After a fairly long time, the resulting effect would be the 

 observed systematic shift of the level l)ubbles, in Avhich the regular 

 tilting of the ground partakes to some extent. 



Now the observation of latitude variation is based on the 

 cyclical system of star groups, consisting of pairs of stars selected 

 for the Talcott observations. This method is the so-called chain 

 method. As the result of this procedure, we ol)tained the polar 

 motion and Kimura's z term. The closing sum is also a product of 

 the chain method. The z term and closing sum form the principal 

 enigmas of present-day practical astronomy. 



Now for the first subject numerous causes have been proposed, 

 among which we may mention the following, — 



1. Yearly atmospheric refraction. 



2. Yearly cosmic refraction. 



3. Improper value of the parallax and proper motion of the 



oliserved stars. 



4. Ditto of nutation and aberration. 



5. Actual change in the earth's centre of gravity. 



6. Eesult of computation. 



7. Latitude variation of short period. 



All these hypotheses may be in a greater or less degree probable, 

 and at the same time no one of them has yet such a firm I)asis of 

 proof, as to secure our universal assent. 



The same may be said of the closing sum, for which the fol- 

 lowing hypotheses may be mentioned, — 



1. Erroneous value of aberration constant. 



2. Diurnal atmospheric refraction. 



3. Latitude variation of short period. 



The first explanation has been universally accepted. But the 

 usually adopted value 20. "47 is quite consistent with the recently 

 determined value of the solar parallax, and does not permit so 



