1867.] 



made with a Rigid Spectroscope. 



17 



of tlie latitude, or of the temperature which marches with it, and may be 

 removed by supposing the temperature-correction applied by Mr. Stewart 

 to have been too great by about 



0-10 for 20^, 

 a quantity well within the limits of uncertainty. 



Pig. 3. 



" I think therefore that the statement made iu the paper — that a 

 residue so and so exists after all known causes have had their effects 

 eliminated — is too boldly advanced. 



" The inference I should be disposed to draw from the observations 

 is : — 



" That the influence of the variation of gravity (or rather, I believe 

 it is supposed to be, of the potential of the Earth's attraction) does 

 not exceed, in passing from lat. 45° to the equator, a change of 

 refraction for the yellow of the spectrum equal to about three-fourths 

 of the interval of the D-lines ; and that even this small apparent 

 change may be referred with great probability to known causes. 



" As to future observations I should say— 



"I. Let the observations be repeated on the homeward voyage (I 

 presume the survey is not yet finished, and the ' Nassau ' is still in 

 the Straits of Magellan). If the instrument has shaken down into 

 permanence the result will be different (I mean the result ^^/^corrected 

 for change of zero), and one source of uncertainty will be removed. 



"II. Should the observations taken iu the homeward voyage lead 

 to the same result, repeat the observations for temperature-correction 

 by means of the change naturally occurring with change of season, 

 relying chiefly on observations taken during pretty uniform weather of 

 whatever kind. " Tours very truly, 



" J, P. Gassiot, Esq:' " a. a. Stokes." 



"Kew Observatory, Slsfc May, 1867. 

 "Mt dear Sir, — I have read Prof Stokes's remarks on your pro- 

 posed communication regarding the rigid spectroscope, and before 

 adverting to any point on which we may have a difference of opinion, 

 it may be well to remark that we both agree that the experiment in its 

 present state is not decisive — in any case more observations must be 



VOL. XVI. C 



