1916] on The Movements of the Earth's Pole 677 



where special observatories had been built for the purpose ; a new 

 observatory, established by the Russian Government at Tschardjui, 

 in Russian Asia ; and the existing observatory at Cincinnati. The 

 reason for selecting stations at the same latitude was that identical 

 sets of stars could be observed at each place, and thus any errors due 

 to defective knowledge of star places are similar for all. These 

 began work in 1899. Later, two stations in the Southern Hemi- 

 sphere, at latitude 81°*5 S. — Bayswater in Western Australia, and 

 d'OncatAYo in the Argentine — were added. 



The results were reduced and discussed by Prof. Albrecht at the 

 Geodetic Institute, Potsdam and published with a diagram showing 

 the actual polar movement as deduced from the mean of the observa- 

 tions at all the stations, from time to time. 



It was not long before these observations yielded a result of the 

 highest interest. The observatory which devoted itself most whole- 

 heartedly to the work and at which the observations were most 

 extensive and most precise is that in Japan. This was under the 

 able direction of Prof. Kimura. By a searching discussion of the 

 whole series of observations he showed that they became far more 

 consistent if a new term were introduced into the expression for the 

 latitude variation, this term having an annual period but being in- 

 dependent of longitude and having the same value for all the stations 

 at the same date. 



It will be readily seen that this term differs completely from those 

 we have been considering hitherto. It is not a shift of the earth's 

 axis or a movement of the pole of rotation, as it affects all places 

 along a parallel of latitude equally the pole evidently does not move, 

 but something which has an effect exactly the same as if the centre of 

 gravity of the earth were shifted a few feet up and down, northward 

 and southward, from its mean position. 



The great difficulty in elucidating the Kimura term lies in its 

 extremely small magnitude and in the consideration that there are so 

 many possible sources of error affecting observations of this class which 

 might have annual periodicities that their separation and evaluation 

 are extraordinarily complicated questions. This is not the place to 

 attempt any complete discussion but a mention of some of the lines 

 along which a solution has been sought may detain us for a few 

 minutes. 



The magnitude of the term at the latitude of 39° is about 

 6/lOOths of a second or arc, or 6 feet on the earth's surface. It has 

 the same value and phase for every station on the same parallel and 

 is zero on about March 9 and September 12, and maximum and 

 minimum on June 10 and December 10, i.e. about ten days before 

 the equinoxes and solstices respectively. It cannot be accounted for 

 as a real shift of the earth's centre of gravity. It is true that in the 

 alternate melting and accumulation of ice and snow at the two poles 

 we have a periodic factor at work which does do this, but the amount 



