OP ARTS AND SCIENCES. 285 



At "Washington, and at most other first-class observatories, the first 

 method is employed. At Greenwich, both the first and the second 

 methods are in use, the value adopted for any day being the mean 

 result given by the two methods. At Harvard College Observatory, 

 the third method is employed in all differential observations. 



Each method has its advantages, and also its disadvantages. In the 

 use of the first method, either the latitude must be considered as known, 

 or else it must enter as an unknown quantity into the equations of con- 

 dition formed from the observations. Easy reference to a fixed point 

 would seem to be about the only advantage that can be claimed for 

 this metliod. Except for this tliere would seem to be no good reason 

 why we should measure a quantity which is not the quantity sought. 

 Polar distance is the co-oi'dinate to be directly measured, and the polar- 

 point correction is tlie correction needed. This can be obtained by 

 observing the polar distance of the Pole-star at both the upper and 

 lower culminations. This method has been exclusively followed for 

 the past five years in the series of observations undertaken by the 

 writer, for the determination of the absolute co-ordinates of about one 

 hundred stars between the first and fourth magnitudes. The defect 

 of this method consists in the requirement that the index error of the 

 circle must remain constant between two adjacent culminations. 



It is proposed to verify the constancy of this quantity in the fol- 

 lowing way. 



It has been found that the reversible level invented by Mr. John 

 Clark of the United States Coast Survey serves in the most admirable 

 way to define a fixed reference plane. The reading of the microscopes 

 of the Harvard College Meridian Circle for the indicated zero of the 

 level, which is attached to the cube of the telescope, has now been 

 continued without interruption for nearly two years. A provisional 

 discussion of the results shows that the reference plane thus indicated 

 remains nearly invariable, — certainly it is more steady than the 

 position of the mean of the microscopes upon the circular frame 

 upon which they are mounted with respect to the position of the circle 

 itself. 



It is proposed to mount a level of this form upon a horizontal table 

 attached to an arm having an angle with the axis of the earth nearly 

 equal to the polar distance of Polaris when the telescope is set at this 

 polar distance. If the lower end of this arm is mounted upon centres, 

 and if there is attached to the upper end a micrometer screw which is 

 tangent to the arc of revolution, it is obvious that, when the reading of 

 the screw for either culmination of Polaris is known, we can measure 



