188 The Rev, Dr. Robinson on the Constant of Refraction. 



To construct a table of refractions from this formula, we require the nume- 

 rical values of -, of fi at some given temperature and pressure, and of e the 



expansion of air for one degree of Fahrenheit. The last of these has almost 

 universally been taken from Gay Lussac, who found that a unit of any gas or 

 vapour at the freezing point of water, became 1.375 at the boiling point. But 

 the experiments of Rudberg have shown that this number is too great, and that 

 the true increase is 1.365. I have, therefore, used this coefficient, notwith- 

 standing the opinion of some whose authority is of much weight, that even Gay 

 Lussac's number should be increased on account of the moisture of the atmos- 

 phere. But the expansion of vapour is the same as of dry air : if water be 

 present, it does indeed seem greater, because heat increases the quantity as well 

 as the bulk of the vapour, and a correction to this effect is necessary to the 

 barometric measurement of heights. In respect of refraction the case is other- 

 wise ; aqueous vapour and dry air refract alike under equal pressure and tem- 

 perature ; when, therefore, more vapour is added to the atmosphere, the effect 

 is the same as if so much dry air were added as is equivalent to its tension. 

 Observation leads to the same conclusion ; for the illustrious astronomer of 

 Kbnigsberg found that the coefficient which satisfies the variations of refraction 

 is 1.00364. — Tab. Reg. p. Ix. The only way in which the hygrometric state 

 of the atmosphere can affect refraction is by changing the value of I, or by 

 varying the arrangement of the strata. The latter of these cannot be taken into 

 account, and the former is, in this climate, insensible within the limits of this 

 inquiry. 



The value of / used is that given by Arago and Blot in their experiments on 

 the refractive power of air. They give It for centesimal ; but as their experi- 

 ments were made at the mean temperature 10° cent, or 50° Fahrenheit, the 

 normal temperature of most refraction tables, their result is not affected by the 

 error of Gay Lussac's expansion. 



There remains only the refractive power of air, which may be investigated 



necessary to employ it) can be computed by the sliding rule. A table of ? for every minute of the 



first 6 hours is almost essential to the use of the equatorial, and if my first table and the second 



cos^IT 

 X . ., were added to it, the refraction can be as easily computed as on the meridian, 

 sm't •' 



