2 Rutherfurd Photographic Measures. 



but it has been included to secure evidence as to possible variation of 

 .scale value with position angle. It is perhaps necessary to remark 

 that I have preferred to base the study of parallax upon measures 

 of distance only. The original plan did not include the pair c, d; 

 but the observations of the pair c, 6, having produced a discordant 

 result, the pair c, d, was also computed. From this discordance it 

 has been possible to obtain a value for the parallax of e Cassiopeise, 

 a result not anticipated when the reduction of the observations was 

 begun. It is for this reason that the star c appears twice in table 

 ] I. The method of computation is the one commonly employed. 

 The two stars of each pair were selected so as to differ approxi- 

 mately 180° in position angle with respect to ^ Cassiopeiae. The 

 scale value was then determined for each pair, on each plate, so as 

 to make the sum of the distances from ,a constant. The difference 

 of the same distances was then taken as the quantity from whose 

 variation the parallax should appear. This method gives the excess 

 of the parallax of the principal star over the mean of the parallaxes 

 of the two comparison stars.* 



Every observation of distance contained in the Rutherfurd 

 observation books has been used, the treatment of the observational 

 data being as follows : First, means were taken of the separate 

 pointings of the microscope, each measure of distance depending 

 upon ten independent pointings on fi Cassiopeise, and ten on the 

 comparison star. The distances thus obtained from the two sepa- 

 rate impressions were combined into a single mean depending alto- 

 gether on 40 pointings, and this mean was then considered as one 

 complete measure. The distances thus obtained are expressed in 

 divisions of the glass scale of the measuring micrometer, one such 

 division being approximately equal to 28". or. The same unit of 

 measure has generally been employed throughout all the subse- 

 quent calculations. The following corrections were then applied: — 



1. Correction for division errors. These were taken from the table 

 of corrections determined by RoGERS.f 



* Tins is of course not strictly true unless the two comparison stars are 

 equidistant from the principal star, — a condition which should always be 

 approximately satisfied. Nor is it possible to deduce the parallax of the 

 principal star with respect to each comparison star separately, since the 

 parallaxes of both comparison stars will always influence the result through 

 the scale value determination. 



f Ann. N. Y. Acad, Sci., vol. vi, p. 250. 



