OF ARTS AND SCIENCES.' 269 



convenience in identifying the objects observed. The pitch of the 

 screw was about 0.1 cm., and its head was divided into hundredths. 

 The value of one division in seconds of arc was found by placing the 

 instrument upon the cube of the telescope of the Meridian Circle of the 

 Harvard College Observatory. A series of readings was then taken, 

 setting the screw in various positions, and inclining the cube until the 

 bubble showed that the small telescope was horizontal. The position 

 of the bubble was then read, and the inclination determined by the 

 large circle of the Meridian Circle. The value of one division was 

 thus found, on June 7, 1877, to be 13". 95. As the instrument is used 

 only for small altitudes, the angles are practically measured by an 

 accurate tangent screw on a circle about 38 cm. in diameter. 



Instrument C is larger than B, and the cross-hairs in the latter are 

 replaced by a filar-micrometer. It therefore combines the advantages 

 of A and B. Its telescope has an aperture of 4 cm., a focal length of 

 45 cm., and a magnifying power of sixteen diameters. 



Instrument D closely resembles B. Light is secured by the use of 

 an inverting eyepiece. The aperture of the telescope is 2.5 cm., its 

 focal length 21 cm., and its magnifying power ten. The value of one 

 division of the screw was found as described above, and gave the result 

 13".95, or the same as B. 



An important improvement was devised by Mr. J. R. Edmands, and 

 introduced in this instrument, in March, 1882. The level was attached 

 to the bar carrying the wyes, instead of to the telescope. As the in- 

 strument was originally used, the observations were rendered differen- 

 tial, the quantity measured being the difference in apparent altitude of 

 the various objects. The only quantity liable to vary in using the in- 

 strument is the angle between the true level line and the axis of the 

 instrument when the bubble of the level is the centre. This angle may 

 be divided into two parts : first, the collimation error due to the devia- 

 tion of the axis from the line passing through the centre of the pivots 

 by which the telescope is supported in its wyes ; and, secondly, the 

 level error, by which this last line deviates from the horizontal when 

 the bubble is in the centre. Each of these constants may be determined 

 by the device proposed by Mr. Edmands. The level is attached to the 

 bar carrying the wyes, instead of to the telescope. If now the telescope 

 is turned around its axis 180°, the mean of the readings in the two 

 positions will eliminate the collimation. The value of this constant is 

 equal to one half the difference in these readings. The level error is 

 similarly eliminated or determined by turning the telescope end for 

 end around a vertical axis. Rotating the telescope around a horizontal 



