ASTRONOMY. 39 



face, the channels being half an inch apart. The back sur- 

 face of an ordinary rolled glass is turned to a plane and 

 moistened with oil, and placed on the iron back and fast- 

 ened round the edge to a projecting ring on this back with 

 marine glue. The air is then exhausted from the channels 

 in the cast-iron back through an orifice, and the o-lass and 

 iron are thus fastened together. The figuring is to be done 

 by a diamond point, which is mounted on a slide-rest, which 

 cuts at the same time both a concave surface on the glass for 

 the mirror, and a convex surface of the same radius on an- 

 other glass or metal which is to serve as a grinding tool. 

 The spherical surface to be thus obtained is to be turned into 

 a paraboloid by mechanical means not described. 



In Vol. III. of the Moscow Observations, M. Gromadski 

 gives the results of his discussion of the division errors of 

 the meridian circle of that observatory. It was made by 

 Repsold, aud a comparison of its division errors with those 

 of the two Repsold circles at Pulkova has led M. Gromadski 

 to the following important results : 



1. The precision of the copies of the original graduation 

 made by Repsold is far more exact than has been suspected 

 till now. 



2. The errors of the divisions of the original circle are ap- 

 plicable to all the copies, particularly when the abnormal er- 

 rors of a few divisions are determined independently. 



3. Since several of Repsold's circles are now in use, and 

 since others will doubtless be made, it is of importance that 

 the original circle should be carefully investigated. 



As an illustration of 2, Gromadski gives in three columns 

 the division errors of corresponding divisions of the two Pul- 

 kova circles and the Moscow circle. Considering the num- 

 bers in the three columns as comparable, the probable error 

 of a single division is 0.11". This is the measure of the 

 precision with which the original circle is copied. The prob- 

 able error of the mean of all is 0.06". From the table it 

 appears that the largest error is less than 1.2". 



ASTRONOMICAL BIBLIOGRAPHY, ETC. 



At a meeting held in London on October 2G, 1877, it was 

 resolved to found a society to be called the "Index Society/' 

 for the purpose 



