270 



Observ lit ions oft f/i>' Direction onJ fnt 'iis'itji of iJie 



[AvniL 



Place. 



Obsorvod v.iliir> 

 of A. 



No»japafain 



Manargoovly 



Piittoocottah . . . 

 MuuHineK'goody 

 Kalheunaiy, . . . . 



llamnad 



Pauinban 



Canyb.Iiamly . . . , 



Vavliiiatrain 



Pow ani 



Teutocorin 



Palamcottali . . . . 



Nagracoil 



Tdvaiulruin 



Qiiilon 



Allepoe 



Balghatly .., 



Clietwaye 



Penauey 



Calicut 



Telliclieriv , 



,9050 

 ,9:71 

 ,9717 

 ,982(5 

 ,982 1 

 ,9775 

 ,9953 

 ,9718 

 ,9730 

 ,97(53 

 ,99.)8 

 ,9872 

 ,9898 

 ,98()7 

 ,97()5 

 ,9771 

 ,9929 

 ,9882 

 ,9958 

 ,9946 

 .9901 



,09921 



l)iiroroncc. 



-t- ,0038 



+ ,0032 



H- ,0037 



— ,0019 



— ,0023 



— ,0054 



— ,0156 



— ,0150 

 + ,0008 



— ,0039 

 4- ,0037 



1- ,0025 



— ,0020 



If to each of the above values we were to give a weight proportionate 

 to the number of cylinders employed, we shoidd obtain a mean value 

 possessing the least probable error of observation; but, since in this 

 case the errors due to local causes would not be fairly dealt by, it be- 

 comes a Cjuestion — if a more correct mean value would not be attained, 

 by taking the mean without reference to the number of observations 

 employed ; to pursue a middle course, however, we may safely give 



to 1 observation a weight 1 



_ 3 or 4 2 



and — 5 or 6 • 3 



(when, assuming the horizontal intensity at Madras = 1.) 

 The total magnetic force at the magnetic equator ■= ,9906 

 or, rejecting the observations at eight stations made only with cy- 

 linder T. G. T. No. 3, we get the horizontal intensity of the mag- 

 netic equator — ,9921 : — On comparing this with the several consti- 

 tuent values, we obtain the column " difference exhibiting the amount 

 of eiTor of observation, mixed up with the effect of local disturbances: 

 the largest of these (that at Sadras), might possibly be accounted for by 



