correction K= 7-05 sec. (as also in series D). The value of this azimuth constant 

 has been calculated from local time sights of rr Argus on Aug. l.'tth, \vliich lias a probable 

 error for the determination of 2-0 sees. 



The actual error introduced by an uncertainty in the value of the azimuth constant, 

 though appreciable, is not much greater than the probable error of the sights themselves. 

 From Table XXXII we have the following values for ill :- 



losing. 

 July 12th .. 27-21 sec. 0-25: mean value of A - -8-12 ser.. rates 0-14 1 



14th .. 27-48 0-20 -7-98 



15th .. 27-88 -7-08 0-49 



Ifith .. 28-04 +0-08 -7-70 0-23 



M") 



Therefore, if an error of 7-05 sec. in the azimuth constant has occurred, the azimuth 

 will now be zero, and the corresponding time calculation 



dt 



July 12th .. .. 19 -09 sec. . rates 



> . . . . 0-21 sec. losing. 



14th .. .. 19-50 



L. .. 0-49 



15th .. .. 19-90 ,. -4 



L. .. 0-62 



16th .. 20-34 J 



In order to keep a better check on the possible irregularities in the action of " S.C.," 

 during the two series C and D, the pendulum clock was compared five times daily with 

 each of the three mean time box chronometers " E," " F " and " H." By this menus 

 it was hoped, if the rate of " S.C." did not remain constant, to pick out the times at 

 which the rate changed, and thus be able to apply a rate appropriate to the period 

 during which the pendulums were swinging. As an illustration of the working of this 

 method, we find by comparison of " S.C." and " E." at the time of sights that : 



(1) For the first pair of sights, July 12th to 14th, " S.C." was gaining on | 



" E" 230 -32 sec. daily. 



(2) For the second, July 14th to 15th, " S.C." was gaining on " E " 



235-81 sec. daily. 



(3) For the third, July 15th to 16th, " S.C." was gaining on ' E " 



236-15 sec. daily. 



From (a) we see that 



(1) " S.C." was losing 0-14 sec. daily 



(2) 0-49 



(3) ,, ,, 0-23 ,, in the same intervals. 



49 



