506 Trans. Acad. Sci. of St. Louis. 
intervals. Only stars of the Berliner Jahrbuch and the 
American Ephemeris were used and the average number 
observed each night is 20. As at Columbia, a complete deter- 
mination of the clock error and instrumental constants was 
made both before and after exchange of signals, and by 
reversal of the instrument and judicious selection of the stars 
observed, the azimuth and collimation constants were well 
determined and their effect eliminated as far as possible from 
the clock 47. 
The time-piece used at St. Louis was sidereal clock Howard 
No, 214 of the Observatory of Washington University. 
Reduction of the Observations at Columbia and St. Louis. 
The observations at Columbia and St. Louis were both 
reduced by myself. With a single exception the right ascen- 
sions of the stars were taken when possible from the apparent 
places of the Berliner Jahrbuch. The right ascensions of 
those stars of the American Ephemeris which are not con- 
tained in the B. J. were taken from the apparent places 
of the American Ephemeris and reduced to the B. J. system 
by means of the corrections given for that purpose in the 
B. J. of 1884. In case of a few B. J. stars whose apparent 
places are not given, the reduction from mean to apparent 
place was computed with the Independent Star Numbers of 
the American Ephemeris. In case of the star 61 Cygni the 
place given in the Am. Eph. was used in preference to that 
given in the B. J. and the correction for reduction to the 
B. J. system was applied. 
Two complete reductions were made of the observations at 
Columbia and also of those made at St. Louis. The first or 
preliminary reduction was made by a method of my own pub- 
lished in No. 238 of the Astronomical Journal. Following 
this method, the corrections for level, diurnal aberration and 
for an approximate value of the collimation were applied to 
the observed times. The observed times thus corrected were 
subtracted from the right ascensions of the stars, giving as 
remainder a quantity M which is the clock 47 uncorrected for 
azimuth and also uncorrected for whatever error there may be 
in the assumed collimation. We then have for each star 
circle west an equation of the form 47,=— a, 4+ in which 
