REFRANGIBILITY of LIGHT. 47 
green rays are the moft refrangible. But as the angle formed 
by thefe moft and leaft refrangible rays, would be much 
greater in the concave if the whole refractions were equal, the 
whole refraction is here reprefented to be precifely that which 
is requifite for giving the fame inclination of the green rays to 
- the united red and violet rays, which takes place after refrac- 
tion through the convex. Hence, as thefe refractions are equal 
_and oppofite, they deftroy each others effect. The rays proceed 
after refraction without any divergency from unequal refrangi- 
bility ; and the aberration from the fpherical figure being alfo 
corrected by means of the concave glafs lens, which is more 
denfe than either of the fluids, they are converged to the fame 
point. 
Tue conftruction reprefented in thefe figures, is not, how- 
ever, the moft perfect and convenient for the purpofe. The 
beft method is to divide the concave glafs neceflary for re- 
moving the fecondary colour, by making two of the lenfes, or 
all three of them, concave menifcufes. But throwing the 
whole concave glafs into one lens, and exhibiting the compound 
convex and compound concave lenfes feparately, anfwers beft 
the prefent purpofe of explaining the principle on which the 
aberration from unequal refrangibility may be totally removed. 
On the fame account, the difference of the difperfive power of 
the two fluids, is reprefented greater than it is in reality. 
‘Havine completed an object-glafs of this kind, I carefully 
examined whether any colour was yet difcernible. For though 
the red and violet and green rays were now united, it was a 
thing poffible, that rays of other colours might {till have a {mall 
inclination to thefe. But I could difcover no colour by the 
moft rigid teft; and therefore conclude the refraction ‘of all 
the rays of the fpectrum to be now equal. If there be any de- 
viation from this equality of refraction, it is infenfible; and 
infenfible- errors, in thofe cafes where fenfe is the only judge,. 
may be accounted no errors at all. 
I 
