OP ARTS AND SCIENCES. 



169 



each pair exists ; which ought not to be, but is often. The weights to 

 each pair being now roughly assigned, the observations should be 

 treated by least squares (if they can be improved in this way), con- 

 sidering the latitude and the value of one micrometer revolution as the 

 unknown quantities. 



A weak point in the zenith telescope is the connection of the level 

 readings with the actual position of the vertical axis ; arising from 

 the fact that the level has to be much handled, and to be tilted in 

 observing. It might be well, therefore, to employ the delicate level 

 only for the reading off, and have a separate rougher one for the 

 setting circle ; placing the former in direct connection with the vertical 

 axis. The instrument ought to be so constructed, that the two delicate 

 levels used for time and latitude respectively could replace one another ; 

 saving one spare level, or else diminishing the chance of loss from 

 their breakage. Some of the earliest as well as of the latest meridian 

 and equal altitude instruments are reversed by a machine, instead of 

 being turned around a vertical axis. I think this is an improvement in 

 solidity, if not in rapidity of observing. 



I will give, as an example of a method of discussing latitude obser- 

 vations, the latitude of Colorado Springs as observed by Dr. Kampf. 

 (See Lieut. Wheeler's Report on Cheyenne and Coloi'ado Springs, 

 pp. 70fF.) The stars are taken either from my catalogue of 981 stars, 

 or computed by myself on similar principles : the quantities /i(J) are here 

 added from a completer discussion of the declinations than given in the 

 Report. 



PAIRS ONCE OBSERVED. 



All these observations except those of pair 4 were taken on one day, 

 August 2d, on which a constant difference of about — 0".7 from the final 



