520 ACCOUNT OF ASTRONOMICAL OBSE^lVATrONS 



it is likewise evident, from what appears in Table IV. 

 that were the mean of any number of northern and south- 

 ern stars, to be taken separately at any two places of 

 different latitudes, and in the order here described, the 

 deviation would tend the same way as in the above 

 exposition ; and this, it may be supposed, by a certain 

 quantity, thrown in from a distant zenith^ on account of 

 refraction, which cannot be corrected here for the quan- 

 tity P .r, nor done away by that due to the small arcs 

 D y near the zeniths. Also that the latitude of ^ will 

 fall too low by a certain quantity (P D — P x) D x -^ 



J__t_^,where D .vis the w^hole error affecting the de- 

 clination of D\ and the divisor «, will be in some pro- 

 portion of the error affecting the whole arc D ^ (always 

 in eixcess) from the extremities of which the mean la- 

 titudes of Z and ^ were deduced. 



IS. With the mean latitude given in Table II. we 

 can therefore be no otherwise satisfied than from the fol- 

 Jowino; consideration, which as far as it affects our lati- 



tude seems to reduce the error Z> ^-f ^ to a mere 



n 



nothing. 



J 9. Regidus, a star of the first magnitude, no doubt 

 attentively observed from every part of Europe, owing 

 to its being very near the ecliptic, and situated only 9 

 minutes and 14 seconds from the zenith of the Madras 

 Observatory, gives a latitude differing only from the 

 mean results in Table II. by 'I0'\l2. This, no doubt, 

 will be admitted to be a strong indication, that its decli- 

 nation, such as laid down in the tables, is very accu- 

 rate. Regulus may therefore, without inconveniency, 

 be taken as a visible point in the heavens, from which 

 ru lay down the position of the other stars ; by this 



