" On the Inflexion oj Light" 429 



ui'{d) or (c). At least it would be satisfactory to see what de- 

 gree of inaccuracy in these measures might suffice to bring 

 the result within the limit before indicated. We may easily 

 make a sufficiently near estimate of this. 



Taking the millimetre = -03937 inch, we shall have the 

 value before assumed for A =• 00063 millimetre nearly. We 

 shall thus find for the limit, which makes the denominator 

 vanish, c 2 = *0044 b. 



If we assume Biot's values of (b), and calculate those of 

 (c) which are the least compatible with them, as just explained, 

 we shall find, nearly, 



Biot's Value of c. Difference. 



•25 -02 m so -0007 inch. 



•50 -05 -0019 



•75 -02 -0007 



These values of (c) were probably obtained by estimation ; 

 and errors even to the insensibly minute amounts here stated 

 would bring them to the limit, even supposing the values of 

 (b) unaltered. These however are of a nature open to consi- 

 derable uncertainty; (as indeed Biot confesses in the higher 

 numbers, Traite de Phys. iv. 764.) In experiments of this 

 kind it is almost impossible to determine, precisely, at what di- 

 stance the central point of the screen is at its maximum of 

 darkness. It is not necessary here to enter into further formal 

 computations of the amount of error necessary to reconcile the 

 alleged discrepancies ; since it is obvious that results of this 

 kind are not of a nature to be depended on for such minute 

 calculations as those of the values of the lengths of undulations. 

 Those of Biot were tried with a view to a different compu- 

 tation. In my own trials of such experiments, I have seldom 

 been able to feel certain of the distance at which the dark band 

 first appeared, or of that at which it attained its maximum, and 

 have often shifted the eye-glass through a considerable space 

 before any marked difference appeared. 



(3.) Mr. Barton's third objection is certainly more for- 

 midable in appearance than the preceding ; but some of the 

 same remarks will apply here as in the former cases : we shall 

 also find one or two other considerations, which will furnish a 

 complete explanation. 



With respect to the experiment of Newton, to which our 

 author here refers as at variance with Fresners theory, we 

 might recur to the general and obvious unreasonableness of 

 citing experiments made when the art of observing was in 

 its infancy, as tests of the refined theories of modern research ; 

 but the present is a peculiarly strong case : this particular ex- 



