678 Colonel Edmond H. Hills [May 19, 



is far too small. It was pointed out long ago by Van de Sande 

 Bakhuysen that to fit in with the observed value of this term the 

 apparent path of the centre of gravity must have an amphtude of 

 3 metres, which, if translated into terms of polar ice, would mean 

 that a cap of ice 1 kilometre thick and 244 square degrees in area 

 would have to form and disappear each year. This is obviously 

 quite impossible. There are certain possible errors in the accepted 

 values of the proper motions and parallaxes of the fixed stars which 

 might produce an apparent variation in the observed latitude of this 

 nature. As all parallaxes are based upon differential measures we 

 cannot with certainty say that such errors are impossible ; we can 

 only say that they appear to us very unlikely, and that, if they were 

 actually proved to exist our ideas of the stellar universe would be 

 profoundly modified. 



If there were a yearly term in the refraction which had the effect 

 of a periodic change in the apparent zenith we should get a corre- 

 sponding periodicity in the observations. If, for example, there 

 were a solar atmosphere, even of a quite tenuous nature, which 

 extended into space beyond the earth's orbit, we should get a 

 seasonal change due to the varying angular distance of the sun from 

 the zenith of the place of observation. An atmosphere which could 

 bend rays of light to the requisite amount, though undoubtedly 

 extremely rare, would, however, be dense enough to offer an amount 

 of resistance to a planet, or li fortiori to a comet, inconsistent with 

 observed facts. It is, however, quite possible that the changing 

 declination of the sun may curve or tilt the mean isobaric surfaces 

 in the upper atmosphere in such a way that the apparent zenith 

 moves north and south about its mean value and that it is to this 

 cause we owe the greater part, if not the whole, of the Kimura term. 

 Such a displacement of the isobars is highly probable and the phase 

 times of the latitude variation — nil at equinoxes, maximum northward 

 at summer solstice, and maximum southward at winter solstice — fits 

 in perfectly with this explanation. The observations made in the 

 southern hemisphere should form a crucial test. If this is the true 

 cause the apparent latitude of a southern observatory will be shifted 

 in the same direction as that of its northern counterpart, i.e. north- 

 ward in June and southward in December. We have only a short 

 series of observations from southern stations, but as far as they go 

 they appear to conform. There is thus fairly strong evidence in 

 favour of this explanation. 



It must not, however, be assumed that the matter is settled beyond 

 dispute. More observations are necessary and especially observa- 

 tions at widely different latitudes. The international stations are, 

 as to the northern ones, almost exactly on a parallel, and, as to the 

 southern ones, on a parallel differing only by 7|- degrees from the 

 northern. This uniformity, highly advantageous for securing a 

 precise record of the motion of the earth's pole, is disadvantageous 



