NOTES (1) Herschel placed the pair in his "Class II," which indicates that he estimated the distance as between 4" and 8". Otto Struve considers that at this time the distance must have been less than 4", which seems the more probable. No use has been made of this measure in the final adjustment. (2) Excessively difficult. The angle was estimated roughly as being in the direc tion of the principal star, of which the position angle is 107°. The entire unreliability of this measure was first pointed out by Burnham in 1894. (3) No trace of duplicity. (14) This is merely a rough estimate. Knott used a 73 inch refractor. (51) "Nearly invisible.'' (53) and (54) Made with the 12-inch. I have given half the theoretical weight to numbers (5), (38) and (43). (57) Was not used in the computation; these observations were made after the work was completed. These observations were corrected for precession, and then plotted as above described, and the elements of the true orbit were de rived from them. These elements were the following: For the purpose of effecting a least square solution, twelve normal places were next formed from the observations of the preceding table, as follows: From these there resulted twelve equations between the six unknown quantities, the residuals in angle only being employed. These equations were weighted and solved for the corrections to the elements, the results being as follows: The residuals from these elements were not wholly satisfactory, especially between the years 1853 and 1879, when they steadily maintained the positive sign. For the purposes of a further improvement, therefore, the original observations were next grouped into the following thirty-three normal places : These measures were corrected for precession, and to the resulting 33 equations there were assigned two series of weights, the first depending only on the number of nights, and the second being arbitrarily assigned. Only the residuals in angle were employed, so that there resulted 33 equations between the six unknowns. The final values obtained from this solution led to the following elements: The value of a was obtained from a series of equations of the form The weighted mean was taken for the value of a. The following table shows the agreement of the observed positions with the positions computed from the final elements. These residuals are perhaps as small as can be expected with a star of this character: 0.00 O C2 57.21 55.22 -1.91 3.33 3.25 + 2.11 2.88 2.26 1895.89 84.17 1" A very vague estimate 2"No trace of duplicity." 3 This is a rough estimate merely; Knott used a 7-inch. + .56 0.03 + 0.04 2 Leavenw'th 6 Hall 3 Tarrant -0.39 4-1 Schiaparelli -0.62 + 0.16 Ο.ΟΙ +0.06 0.02 + 0.26 0.06 0.02 2 Schiaparelli 5/ Hall 4 "Nearly invisible." 5 12-inch. IH. Struve -0.37 2.95 2.56 + 0.12 0.02 -0.02 0.89 2.67 2.65 2.48 0.05 B. |