430 KRETZ. 



The second category, projection errors, have been ehminated 

 entirely in the case of the Repsold machine in use for the pres- 

 ent research by an improved guiding way with which it was 

 equipped in 1896. 



As regards the deviation of the cyhnder from straightness, 

 an investigation made under Professor Jacoby'sMirection shows 

 that no appreciable error is introduced thereby, the greatest 

 range of variation not exceeding o".04. I have therefore ne- 

 glected this correction. 



In Table V are recorded the final corrected coordinates ob- 

 tained from the measures of Table III. The process of com- 

 puting them is very simple : To the number of the line add 

 the mean of the two corresponding y^nis corrected for runs and 

 division errors. In case two lines are used, substitute for the 

 line of above the mean of the lines. The result is the mean 

 position of the star with respect to the scale for each of the four 

 positions of the plate. Then calculate the measured coordi- 

 nates, as previously explained, by comparison of these four 

 quantities with the corresponding quantities for the central star, 

 obtained in the same manner. Apply thereto the rotation cor- 

 rections, having care for the sign, and the result will be the quan- 

 tities set down in Table V. No further explanation of the terms 

 there used is necessary ; it may be mentioned, however, that, 

 as before, unity in the fourth place of decimals corresponds to 

 about o".oo5. 



1 " Permanence of the Rutherfurd Photographic Plates," Annals New York 

 Academy of Sciences, Vol. IX, p. 207. 



(90) 



