i6o EASY ANB ACCURATE METHOD of 



look direftly at the fun, the adjuftment may be made, 

 or the index-error found, with great eafe and exa£tnefs.. 

 If the images of the fun fhould be too bright for the 

 eye, one of the coloured glaffes belonging to the inftru- 

 ment may be held before the eye. The two images of 

 the fun, thus viewed, will appear nearly of the fame 

 Ihade, fince the one from the index-fpeculum, which 

 would otherwife appear the brighteft, will lofe part of 

 its light by palling through the horizon-glafs. 



After all, I am of opinion that this adjuftment may 

 be made with equal accuracy, and much more eafe, by 

 anv well defined objedl on land, as the edge of a chim- 

 ney, theroof of a houfe, or the like, at a fufficient dif- 

 tance. 



Any one may fatisfy himfelf that this is the cafe, by 

 repeatedly meafui-ing the error of adjuftment in the com- 

 mon way, viz. by moving the index till he produces an ap- 

 parent coincidence between the objeil feen direftly and 

 by refiedtion. If a chimney, or tlie like, be the object 

 viewed, he ^^ill fcarce ever find any of thefe errors to 

 differ from the mean error more than otic minute ; and 

 the difference of fuch errors, when the fun is ufed, will 

 not be lefs, but generally greater. 



If the diftance between the index-fpeculum and the 

 line of direct vifion (viz. a line joining the eye-hole and 

 horizon-glafs) fhould not exceed three inches, which it 

 feldom does, then the parallax of the inftrument will not 

 amount to a quarter of a minute, and may therefore be 

 fafely neglected; provided the obje£l viewed be at the 

 diftance of two thirds of a mile. If a fuitable objedl at 

 fuch a diftance cannot be readily found, then you may 

 take one at any given diftance, and compute the paral- 

 lax to be allowed for that diftance, thus — Multiply the 

 conftant number 95 (the nat. tang, of i' to rad. tV) by 

 the diftance, in inches, of the centre of the index-fpe- 

 ?iulura above the line of direct vifion, and dividing the 



produ(5l 



