182t) 



primifive means. To obtain a comparison-scale a portion of a photo- 

 graph of Coma Berenices was cut out, containing side by side 

 exposures of 12, 15, 19, 24, 30, 38, 48, HO, 76, 95, and 120 seconds, 

 which means 11 images of every star, increasing 0"\ 2 in magni- 

 Inde. Hy pressing this plate to the back of phite A or /i, (ilm 

 against film, and compaiing with an ocular enlarging 5 times, each 

 star on .1 or B could be inserted by means of eye-estimate between 

 the terms of the scale. The numbers of the scale-values represent 

 the ap|)roximate magnitudes of stars that would have the same 

 images on plate B. 



By means of this .scale in a number of regularly distributed 

 sfpiares the magnitude of all the stars distinctly visible on plate B 

 was estimated and the like on A for all clearly \isible and measur- 

 able images. Thus can be found the diflferences in magnitude between 

 the vai'ious exposures, expressed in the |novisional scale. To express 

 the unity of this provisional scale in the absolute scale of magni- 

 tudes, two strips, North and South, were measured on either plate 

 C in such a way as to leave each strip on the one plate entirely 

 covered by the Halbgitter, and on the other quite free. By deducing 

 from this the difference in magnitude of the images with and with- 

 out the Halbgitter in the provisional scale and comparing it with 

 the known absorption-coefficient of the Gitter, one can find the 

 reduction to absolute scale. By means of a few stars of known 

 magnitude the absolute magnitude can then be deduced. 



The execution and reduction of the measurements showed that 

 in case of the more brilliant stars with large images there existed 

 •systematic differences, that rendered a further use of them undesi- 

 rable. With the fainter stars of the scale other erroi-s presented 

 themselves. The smaller images showed as somewhat irregular spots, 

 and neither did these always ditïer 0"\2 in magnitude. This may 

 be caused partly by local differences of sensitiveness and a not 

 wholly regular spreading of the silver-grains, which influence the 

 look of these small faint spots, partly in the accidental coinciding of 

 scale-images with images of other invisible stars. It proved necessary 

 therefore, to ascertain separately the magnitude of all images of the 

 scale that were often used. This was done by estimating them 

 between the images on a polar plate, likewise following each other 

 with a theoretical interval of 0'",2 ; as each scale-image was inserted 

 in various polar-star series, the errors of these series passed into 

 the magnitudes of the scale to only a very slight extent. Thus for 

 the magnitude of the faintest (0) up to the brightest image (10) of 

 the stars 7v, s and /■ we found : 



