460 ON THE ABSORPTION OF LIGHT BT COLOUBED MEDIA, &e. 



artist, for another purpose, and were probably taken from the 

 same melting pot. Meanwhile, it ought not to excite sur- 

 prise, that the dispersions deduced by this method should con- 

 siderably exceed all former estimates. It will be recollected, 

 that they are founded on observations of rays situated rigor- 

 ously at the extremities of the spectrum. These rays elude 

 all ordinary observation in the solar spectrum, and are too 

 feeble to exert any sensible influence on the colours of the 

 edges of objects, in the usual mode of compensation. This 

 latter, indeed, being merely comparative, assuming as known 

 the dispersion of a standard prism, its results must be affected 

 by all the uncertainties attending the determination of this ele- 

 ment, which, if obtained by actual measurement of the solar 

 spectrum, must, as I have before observed, necessarily err con- 

 siderably in defect : add to which, the method of compensa- 

 tion, owing to the " irrationality of the coloured spaces," can 

 only give results corresponding to the union of the two bright- 

 est and most strongly contrasted colours, which may differ 

 considerably from those corresponding to extreme rays. The 

 values of p for the crown and plate glass, Nos. 7. and 8.,^may 

 possibly be somewhat incorrect, from the smallness of the re- 

 fracting angles of the prisms used. — I have the honour to re- 

 main, dear Sir, with sincere regard, yours, 



John F. W. Herschel. 

 Slough, July 24. 1822. 



XXXII. 



