M. Arago on Double Slavs. SS 



the possibility of errors such as two or three seconds in the dif- 

 ferences of the measurements of the angular elevations of the 

 same star. And, it is freely granted that, with first-rate instru- 

 ments and long experience, such errors may be avoided in the 

 mean of a great number of observations ; in such, for example, 

 as the planetary diameters. But the observations of parallaxes 

 are a very different thing. 



Let it be first remarked, that these observations require in- 

 struments of very considerable dimensions. Without this con- 

 dition, a second will not be visible upon the graduated scale. It 

 must be added, that the earth occupies six whole movihs in pass- 

 ing from one point of the terrestrial orbit to the point diame- 

 trically opposite ; — that, if the angular elevation of a star has 

 been measured in the first station in winter, it can only be mea- 

 sured in summer in the second ; — that if the whole apparatus is 

 not kept in exactly the same state during the six months, it will 

 be impossible to compare the observations; — and, finally, it 

 seems most difficult to avoid the slight bendings and smaller al- 

 terations which this instrument, so large, so massive, and com- 

 posed of so many separate parts, which, in the two epochs, must 

 be in thermometrical conditions entirely dissimilar, must neces- 

 sarily undergo, &c. &c. But in spite of all these obstacles, what 

 with consummate skill on the part of the artists, and care and 

 patience on the part of astronomers, we can now answer for a dif- 

 ference of the angular elevations of the same star, observed at 

 the distance of six months, within nearly two or three seconds. 



This space, seen through the focus of the great telescopes of 

 our graduated circles, does not equal the thickness of the thread 

 of the spider ! Can we, after this, be astonished that there was 

 little prospect of surpassing this limit of precision by the usual 

 modes of procedure ? 



We have, however, to add, that, in certain circumstances, 

 which we proceed to point out, the double stars would enable us 

 to value the change of the angular elevation, not only by three 

 whole seconds, but even to the accuracy of a tenth of a second ; 

 that is to say, thirty times more accurately than has hitherto 

 been done. 



And now is the time to revert to a remark which was put in 



