CHAP. XII.] 



LATITUDE BY STARS 



287 



Let S = sum of the differences between each observation and 

 the mean of all the observations, irrespective of 

 sign. 

 N = number of observations. 



Then- 

 Probable error of one observation 



S 



Probable error of arithmetical mean of all] ^ 



observations j "^2 



Value for meaning with others, or " Weight of) ^^ 

 the Set " / " ^ 



This question of giving values is a very difficult one, and Difficulty 

 many experienced observers prefer to take a direct mean, both ya^i^jng 

 for the result by each individual and when finally meaning 

 the different observers' results, omitting those altogether which 

 seem least worthy of trust, and giving an equal value to 



everything else. It is very much a case of judgment under 

 the circumstances ; on some occasions one system will be best, 

 and at others the other, and we must so leave it, simply 

 remarking that whenever observations vary enough to make 

 the consideration of this question necessary, the final result 

 must always be regarded with suspicion. 



An example is given on p. 288 of latitude arrived at by 

 giving values. 



