282 PROCEEDINGS OF THE AMERICAN ACADEMY 



appearing to withdraw from A after they had been set upon it. Ice 

 had filled the holes in which the points of the tripod rested, and 

 probably caused this motion. It may, however, have been due to 

 a real change in the refraction, or to a change in the instrument, 

 which had just been removed from a warmer place. In the subse- 

 quent sets the positions of A, B, and C remained constant near 91, 19, 

 and 90. 



24. Air unusually clear and steady. Wind high. Sun shining. 



25. Cloudy. 



26. Sun shining, wind high. 



36. (7 difficult on account of darkness ; settings discordant. 



37. Air clear, wind light. 

 46. Hazy. A difficult. 



53. Cloudy and a little hazy, but images good. 

 65. Images very bad. 



74. Broken clouds, sun shining on C and Observatory. 



75. Broken clouds, sun shining on A. 



86. Heavy thunder-storm to the southwest. 



87. A seen with difficulty owing to intervening rain. 

 107. A hidden in mist. 



Fifty-two sets were taken during the winter, and gave the mean 

 value of ^C= 1408", and BC= 1008". The mean height of the 

 barometer was 30.21 in., and of the thermometer 11°. Ninety-two 

 sets were taken in warm weather, and gave AC ^^ 1384", and BC :=■ 

 995". Barometer 29.78 in., thermometer 76°. 



The mean value of ^C for the whole series was 1392", and that of 

 BC ■=. 1000". The extreme range of the barometer was from 30.70 

 to 29.45 in., or 1.25 in. The thermometer varied from -|-91° to 

 — 16°, or 107°. The various values oi AC differ from their mean 

 of 1392" on the average by 16".8. But nearly all of the i-esiduals 

 of the winter observations are positive, those in summer negative. 

 If the residuals of each series are taken from the mean of that 

 series, the average value is reduced to 8".5. This is nearly the same 

 as 8."3, the mean of the residuals after correcting for temperature 

 and pressure. The probable error of a single setting is 2". 7, both 

 in the winter and summer observations. Since three pointings are 

 made on each object, the errors of observation are much less than 

 the deviations due to refraction. Moreover, some observations are 

 included in which one or more of the objects were seen with great 

 difficulty, on account of haze, unsteadiness of the air, or increasing 

 twilight. Probably the error due to phase caused by varying illu- 



