rs On Obsey^vations made with a Rigid Spectroscope. [June 20^ 



made. Acknowledging with him that the variation due to any other cause 

 than temperature must be sought for in the residue left on eliminating 

 the temperature-correction, I yet venture to differ with him regarding the 

 degree of accuracy to which we can trust the temperature-correction. 



"Eeferring to Prof. Stokes's numerical results, I should ohject to 

 found any theory upon the first-noted reading •71. 



" The rigid spectroscope was unfortunately brought to K-ew only one 

 day before the temperature experiments (instituted more particularly 

 for the magnet ographs) commenced, and only two readings were taken 

 before the temperature was raised. I believed it only right to include 

 these two in the account given of temperature observations, but I take 

 this opportunity of saying that I do not attach much value to the 

 mean of these two. 



" If these two be excluded, Prof. Stokes's Table will be modified in 

 the following manner 



Mean 

 temperature 

 of set. 



Mean reading 

 observed. 



Deviation from 



mean temp, 

 of the four sets. 



Calculated 

 reading. 



Observed 

 — calculated. 



75-5 



2 01 



+ 11-2 



2-26 



-•25 



51-5 



1-13 



-12-8 



1-13 



•00 



76-7 



2-48 



+ 12-4 



2-31 



+ •17 



53-6 



1-31 



-10-7 



1-23 



+ •08 



giving an error much less than that shown by Prof. Stokes. 



" If now we compare together the temperature-corrections for 30° of 

 range, as determined at the Minories and at Kew, we find — 



Difference 

 from mean. 



(A) Temperature-correction for 30° at the Minories =1-32 ... —-06 



(B) . „ „ Kew 1-44 ... +^06 



Mean 1-38 



" I should imagine the set of observations taken on board to be com- 

 parable in number and accuracy with either (A) or (B), and I should 

 expect, on the supposition that the temperature-correction may he deter- 

 mined equally well hy land and hy sea observation, to obtain a result 

 differing from the mean of (A) and (B) by a quantity something like "06. 

 If temperature-correction be the same at land and at sea, I think 

 therefore that the residual difference observed at sea cannot be attri- 

 butable to an imperfect estimation of temperature-correction. 



" Prof Stokes next suggests that the progressive change of the instru- 

 ment may have been greatest at the first part of the voyage. Allowing 

 as the greatest possible extreme though very improbable case, that it 

 all took place before the equator, that would still leave a difference of 

 •09 revolution to be accounted for. 



" But this supposition cannot evidently be entertained ; indeed, as the 



