Astronomy. 473 



where z denotes the zenith distance, p the position angle of 

 the red star from the comparison star, q the parallactic angle, 

 and A/3 the supposed color effect. 



The following pages give the measures corrected for refrac- 

 tion and aberration, the value of tan z cos (p— q) for each observed 

 distance, and the sums, differences, corrections for temporary- 

 scale value, and corrected differences, as in the series for paral- 

 lax determination. These last, by their variation from an 

 assumed mean difference, furnish the equations of condition 

 which follow, in which x and x ! represent the required corrections 

 to the assumed mean difference for each epoch, A/3 the quantity 

 to be determined, whose coefficient is the difference of the values 

 of tan z cos (jo — q) for the two distances, and the second mem- 

 ber the observed difference for each night minus the assumed 

 mean difference. The star Krtiger 1181 was specially selected 

 because it had a close neighboring star presumably of average 

 color, and as a test of the value of the method this latter also was 

 observed with respect to the same comparison stars in conjunc- 

 tion with the red star and symmetrically with it, so that the con- 

 ditions should be absolutely the same. For both stars, likewise, 

 the treatment and solution were carried out in precisely the same 

 manner 



If we arrange the results of the previous pages in the order of 

 the stars' redness as given by Krtiger, we have — 



Kriiger 985 Color 6'0 Aj3 = -0019" ± 0'019" Wt. 63-6 



1080 7-0 +0-005 ± 0-020 64'4 



1078 7-1 +0-009 ± 0-015 16*0 



1181 7-8 +0-014 ± 0-018 55-7 



1108 8-7 +0-046 ± 0-017 45-2 



W. B. 15 h , 745 .. -0-003 ± 0'02l 55-6 



An inspection of these results would seem to indicate that there 

 is a discernible color effect conforming with theory, the mean 

 light of the red star being apparently refracted less than that of 

 the comparison stars, except in the case of Krtiger 985, although 

 the amount is very small except for very red stars. This conclu- 

 sion is corroborated by a comparison of the results for Krtiger 

 1181 and W. B. 15 h , 745, investigated under exactly the same 

 conditions. The result for the latter coming out so nearly zero is 

 a most satisfactory confirmation of the accuracy of the work. 

 That the color effect is not greater rather tends to confirm Sir 

 David Gill's opinion that the heliometer observer's tendency is to 

 bring the similarly colored parts of the stars' spectra rather than 

 the brightest parts into coincidence. 



Now what influence this possible source of error, if real, may 

 have upon our parallax results can be easily shown. Scarcely any 

 of the measures were made at a zenith distance greater than 60° 

 and the average was about 50°. The following table gives the 

 value of the factor tan z cos (p—q) for each twenty degrees of 

 declination at 50° zenith distance, assuming p y the position angle 

 of the red star from the comparison star to be 90°, for which 



