OF ARTS AND SCIEXCES : APRIL 14, 1863. 121 



apparatus march within xtj^iTtt nearly of accordance during whole 

 minutes. 



" Notwithstandinsf the assurance I had gained in the measurement 

 of time, I was surprised to prove in my results discordances, which 

 were out of proportion to the precision of my means of measuring. 

 After long research, I discovered the source of error in the micrometer, 

 which did not allow of the degree of accuracy willingly attributed to it. 

 To meet this difficulty, I have introduced into the system of observa- 

 tion a modification which amounts simply to a change of the variable. 

 Instead of measuring micrometrically the deviation, I adopt for it a 

 definite value in advance ; suppose ■^^J of a millimetre, or 7 entire parts 

 of the image ; and I seek by experiment to find the distance between 

 the mark and the turning mirror necessary to produce this deviation ; 

 the measures extending over a length of about a metre, the last frac- 

 tions have a magnitude directly visible, and leave no room for error. 



" By this means the apparatus has been purged of the principal 

 cause of uncertainty ; henceforth the results accorded within the limits 

 of errors of observation, and the means are settled in such a way, that 

 I am able to assign confidently the new number, which appears to 

 me to express nearly the velocity of light in space, namely, 298000 

 kilometres in a second of mean time." 



This value reduced to statute miles shows that the velocity of light 

 is 185177 miles in a second ; which is less by 6336 miles than the 

 velocity for light usually admitted into science, namely, the velocity 

 obtained from the aberration of light. This discrepancy between the 

 result of experiment, and that astronomical determination which comes 

 nearest to it, is three times greater than the variation between the 

 velocity deduced from aberration and that derived from eclipses. 



Foucault states that the extreme difference of the results of various 

 trials amounted to only y^^ of the whole quantity, and that the mean 

 result can be trusted to the fraction of -g-^-^. Moreover, the aberration 

 of 20".45, adopted by astronomers, cannot be supposed at fault by more 

 than iw^u *^f the whole. Neither the velocity by Foucault's experi- 

 ment, nor the value of aberration, can be charged with a possible error 

 of 3 per cent, or of any error approaching to this large discrepancy. 

 How is the new velocity of light to be reconciled with the old value of 

 aberration ? I have said that aberration establishes only the ratio 

 between the velocity of hght and the velocity of the earth. If this 

 ratio cannot be tampered with, and if one term of it (the velocity 



VOL. YI. 13 



