CHAP. I.] CENTRING ERROR 9 



Centring error may also be obtained by careful measurement 

 of the angles between stars. The correct apparent distances 

 must be found in the same manner as in clearing a lunar dis- 

 tance ; the true distance being first calculated from their de- 

 clinations and right ascensions, but if stars in the same vertical 

 plane can be chosen, the apparent distance can be arrived at 

 by simple application of the refractions. 



There are other methods, involving more calculation, which 

 need not be described. 



The centring error is determined at Kew Observatory for 

 certain angles by fixed collimators, and is given on every Kew 

 certificate, but it must be remembered that in any case it can 

 never be considered as determined for good. Including, as it 

 does, errors from so many causes, it does not remain perfectly 

 steady, but its amount should be ascertained from time to time 

 for any sextant which is to be employed for accurate determina- 

 tion of positions, for circumstances often prevent the use of 

 methods whereby it as well as other errors are ehminated. For 

 instance, a latitude may have to be obtained by altitude of the 

 sun only, when, without knowledge of the centring error, it may 

 easily be incorrect to as much as a minute, or even more. 



As an example, the author's Troughton sextant had at 120° 

 a centring error of — 20". After a fall and repair by the maker, 

 it was + 50". 



To find the error caused by the refraction, through non- Errors of 

 parallelism of the sides, of the coloured shades. Shades. 



Measure the diameter of the sun, with different combinations 

 of the shades. Take out the pin which supports one set of the 

 coloured shades, and replace the shades reversed, so that the face 

 before next the index glass is now away from it. Remeasure 

 the diameter of the sun with same combinations as before, and 

 half the difference of the measurements of each set will be the 

 error due to the shade reversed. 



These errors can be neglected in sea observations, and if 

 coloured eye-pieces are fitted as recommended above, the 

 shades are not required when the artificial horizon is made 

 use of. 



A level fitted to revolve on the pillar on the Index bar on Spirit 

 which the magnifying glass pivots, and working in the plane j^^g^^ Ba,r. 

 of the sextant, is a useful addition. It is set by actual experi- 



