223 Observations on the D'rectio-i and Intcusit'j of tho [April 



been obtained: we perceive, iliat, em/loyin tlie need'es 1 and 3 

 T. G. T., an ei TiT from ihe mean of a Tmv so ends oii' v would be commit- 

 ted: anti that eniplovin^^ I. 2. 3, an I \, V. G. T. or I, ;iiid 2 .1. C, in either 

 case our re>u!is vvoul I, as far as coiistan. error is concerned — possess 

 nearly the same de ree of a "ci ra y as if made from ihe whole 8 ne«-tllea. 



In tbo ll-'port of the B iiish Asso. ia ion relaiive lo ihe m iLHieric ob- 

 servations uiide in IrelautI, it is nientionevl, " of the eight r-sulls ob- 

 " taint d wiib neeillt- (P), ihere is on'y one wld h diilers from tbe mean 

 *' of the e '.•lit 1). a single m nule." Such acciira y, however, cannot be 

 the boast i f the Madras observations ; for, as it wili b.; seen presenii v, 

 a.s///7/e r/eu,ee. instead of a single minute , will occasionally be thrust- 

 ing ll 111. \n"w en observaiiou and ihe trmh. lint, not to be too has- 

 ty in siateiiit^i'is. ii may I)e as well lo m n'i-ri, that, by reason of the fee- 

 bleness of mi4netic force in India, arising' from proximity to the niag- 

 netic eq laior on the one h Old, and high temperature on the other, one 

 carefui y- Ilia le s«M of dip obstMval h»ns under evcrv advantage, cannot be 

 expected to arrive neaier than within •> or IJ minaies of ihe truth. On 

 looking over the whole of the resuNs. however, discordances are found 

 to an amount f.ir exceeditig this, - on- indcid which would render the ob- 

 servations alto^eihcr lise'ess, were there nu m ^ans left of ai'counting 

 and makiui^ a.lowani-e lor the same. I became aware of the di^conlances 

 which affect our observaduiis, when at work on their redac ion at I'ri- 

 vandium: at first I Iclt ini lin( d to attrilmie the dis'-rep-in . ies to local 

 cau-.es and error, but a little retleciion convinced me that I inust give up 

 such a supposition, since, from th'^ carr we had taken in every particular, 

 the one was highly imj)robahle and the oilier impossible; for we had 

 been c.irt^ful to observe in sheds or under trees, far removed from local 

 di-turbin-!; causes, and had not omitted to ay asid^- our keys and the buc- 

 kles from our stocks and braces, & •. ; so that we must look to some other 

 cause for an explanation of the discrepancies. 



In the COUPS'- of our travels, I had noticed at one place ( Carryshandy 

 I think), that, on applying llie mH^^nel to one end of the needle to ascer- 

 tain the po-ition of the poles, the edge o (fig. 1) attra-led the needle and 

 a' repelled it -, but, ou a;;'ain apph'ing the magnet tt) a, its pole had become 

 changed to the s line deuoaiinaiion as*?'; and on immediately vibrating 

 the needle, iis in ii ation h id altered about 50 minutes. Having at Tri- 

 vandrum met with a similar anomaly, 1 w.is led to enquire, whether tbe 

 discrepancies whi -h appt ared in some of the observations could not be 

 accounted for in tliis way: might not the magnet, in the act of reversing 

 the poles, have pressed U[)on one edge, and have established a pole at 

 a or a' instead of the end e ? To resolve this difficulty, I availed myself 



