102 KNOTT AND TANAKàDÂTE 



times of transit. By applying a clock error reckoned from the pre- 

 ceding set of ol)servations, Ave obtained an approximate hour angle ; 

 and from this with assumed latitude it was easv calculatinn- the as- 

 tronomical azimuth of I?olaris. The finding ot{ this azimuth was much 

 facilitated l)y the use of a table of Polaris azimuths, previously pre- 

 pared, and tabulated AA'ith latitudes and hour angles as arguments. 

 On applying the level correction, we have the azimuth to a first 

 approximation. This value was used to set the instrument in the 

 approximate meridian. 



From the list of stars given in the Berlin Yahrlmch, four stars 

 were now selected — if possil)le, two near the zenith and two near the 

 horizon, and lying in pairs on each side of tlie zenith. Each star's 

 transit was observed across three of the micrometer wùres. Immediately 

 after the last transit was taken the star was bisected by the horizontal 

 wire and the zenith distance read otf on the vertical circle. The 

 striding level was read twice, once with the telescope pointing north, 

 and once with it pointing south. The level in connection with the 

 vertical circle was also observed, the telescope being then set to zero 

 reading. 



The azimuth was now read, and the Y's turned through 180°. 

 Another set of four stars was chosen and their transits observed in 

 the same way. 



On a fine night, from 1}, to 2 hours' observations sufficed for 

 carrvino- out the complete transits of the eio-lit stars : and from these 

 the collimation, azimuth, and clock errors could be calculated. A 

 similar set at the sajuo sj)ot the next night fnrin'shed tlie means of 

 calculating the clock rate. 



On one occasi(^n, at Mivazaki (Xo. GO) the clock rate was olitain- 

 ed by making two sets of ol)ser\ations not (jiiite 8 liours apart — the 

 one set in the eveninii'. the other set in the moniin"'. 



