REPORT OF EXECUTIVE COMMITTEE. 93 



eight of these objects are brighter than magnitude 4.5. These have 

 been photographed with the aid of a color screen. The instrument 

 emplo^-ed is the Sheepshanks equatorial of Cambridge Observatory — 

 a photo-visual Coude refractor of 12 inches aperture and 20 feet focal 

 length, at present set apart exclusively for this investigation. 



The photographic plates are made especially for this observatory 

 on plate glass. A reseaii is emploj'ed. The color screen consists 

 of a plate of plane-parallel glass, carrying a small rectangular patch 

 of yellow collodion film, and is placed directly in front of the sensi- 

 tive plate, so that the bright star is photographed through the film 

 at a reduction of brightness amounting to six magnitudes. The 

 plates obtained with this .screen are highly satisfactory, the defini- 

 tion, if anything, being better than on ordinarj- plates, and there is 

 no indication of sensible distortion. 



The first plate for measurement was taken November 18, 1903. 

 Up to June 28, 1904, 118 measurable plates, with four exposures on 

 each, have been obtained of 47 fields, 45 of which are of bright stars. 

 All photographs are taken within 30 minutes of the meridian. 



At present 84 plates of 34 fields have been measured, and 64 

 plates of 24 fields have been completely reduced down to the for- 

 mation of equations. The .v-coordinate is alone to be discussed, 

 thus halving the labor of measurement without sensible sacrifice of 

 accuracy. Two of the four images of a plate are measured in the 

 direct and the two others in reversed positions of the plate, thus 

 halving the labor of measurement without material sacrifice of accu- 

 racy. Eight sj^mmetrically disposed comparison stars on each plate 

 have usually been measured. With few exceptions these are in- 

 cluded between the eighth and tenth magnitudes. The use of a 

 number of comparison stars facilitates recognition of a sensible par- 

 allax for any one of them. One such case has alread)' appeared. 



No attempt is made to deduce the standard coordinates of the 

 plates from meridian observations, and the short methods of Turner 

 and Dyson are employed in the reductions. These methods are 

 justified on account of the care exercised to have the parallax star 

 very near the center of gravity of position for the comparison stars. 



The probable error of an jr-coordinate derived from a single plate 

 is about ±:o".o5, deduced from comparison of pairs of plates. 



In carrying on this work encouragement, criticism, and advice 

 have been received from Mr. A. R. Hinks, the chief assistant of the 

 Cambridge Observator}-, as well as from others, to all of whom 

 grateful thanks are due. 



