358 SCIENTIFIC RECORD FOR 1885. 



star. The screen employed reduced a<i to the magnitude 8^, and rendered 

 it impossible to distinguish n'l, but at the same time the measures of 

 distances from stars fainter than 7-5 magnitude could not be so well 

 made with a 4-inch heliometer. However, the 16 measures forming a 

 complete observation are always made as follows: Each distance is 

 measured twice for the sake of eliminating the constant error of the in- 

 dex, commencing with one of the stars, say o', then passing to /?, turn- 

 ing the tube 180° in its cradle, taking /? again and then a to complete 

 the series. The sum and difference of the distances of a2 Centauri from 

 the first pair of stars chosen by Mr. Gill have the following parallactic 

 factors : 



a 4- /5, 0-011 R cos (0— 96O'0), 



a - fi, 1-881 R cos (9—3470-3); 



El and indicating the radius vector and the sun's longitude ; the max- 

 ima and minima of the parallax for a — ^ occur upon March 7 and Sep- 

 tember 10. 



We may pass over all that refers to the various corrections, the errors 

 of the screw, refraction, the proper motion of a Centauri and aberra- 

 tion ; it will suffice to say that in adopting for the sum of the distances a 

 mean value derived from all the observations, the distances and their 

 differences were reduced to a normal scale, and the only question now is 

 whether these differences are not affected by systematic errors inherent 

 in the observer. It might be, for example, that the observer would in- 

 sensibly change his method of proceeding, thus introducing a term pro- 

 portional to the time, but the personal error might also depend upon the 

 hour angle if one observed to the east or to the west of the meridian, 

 and at a greater or less distance from it, and in passing from one posi- 

 tion to another a discontinuity might occur.* Mr. Gill used a reversing 

 prism for Mars in 1877 for the sake of eliminating this last cause of 

 error,t and he regrets not having continued its use, and avoiding thus, 

 as he says, so much labor. The observations commenced July 5, 1881, 

 and were continued regularly until April, 1882, when the existence of a 

 systematic error depending upon the hour angle became evident. After 

 this the conditions under which the observations were made were care- 

 fully noted, and it was possible to recall the conditions of the greater 

 part of the preceding observations. 



There are four series of observations for a Centauri — two by Mr. Gill, 

 and two by Mr. Elkin — with four pairs of comparison stars. One of the 

 first facts to be proved is the impossibility of treating the observations 

 of each series as a whole. Mr Gill's diagrams prove this, as do also 

 Mr. Elkin's numbers (giving his observations equal weights, Mr. Elkin 



* It is known that observations of the nadir and measures of the declination of a 

 star when between two threads are influenced by the position of the observer. 



t Dr. Seeliger has also recently recommended the use of such a prism {Bull. Astron., i, 

 p. 4^)0). We believe that one has been employed at the observatory of Leyden for a 

 long time. 



