. 
¥ 
8 W. Gibbs on a normal Map of the Solar Spectrum. 
chart, has been determined, the wave length may be found from 
Table III. — , 
length of a given line may be found with very nearly as much 
precision as by direct measurement with a ruled glass. For in 
a large spectroscope with five or six prisms of high dispersive 
ower, the angular separation of two lines differing in wave 
ength by only a single millionth of a millimeter, is so great that 
the interval may be readily divided into hundredths. Of course 
the magnitude of this interval for the same system of prisms 
will differ in different parts of the spectrum, being greatest for 
the most refrangible rays. But the interval will always be suffi- 
cient to give us the advantage of determining a small difference 
in magnitude by the measure of a considerable distance in space. 
To the best of my knowledge this method of determining wave 
lengths is new. 
The spectroscope enables us however to determine wave 
lengths by another method which, though not new in principle, 
has not.received the attention which it deserves, and has been 
applied in practice only in certain cases. I refer to the employ- 
ment of the interference bands of Talbot, an application first 
made by Esselbach* in the measurement of the wave lengths o 
the ultra-violet rays. The method in question exhibits an ex- 
traordinary degree of precision when a spectroscope with several 
prisms of high dispersive power is employed, since in this case 
interference plates of considerable thickness may be used so as 
to produce very numerous dark bands. By measuring the num- 
ber of bands between the given spectral line and two other spec- 
tral lines whose wave lengths are known, the wave length of 
the line in question may at once be found.t This method pos- 
sesses the farther advantage, that by employing several inter- 
ference plates of different thicknesses or different kinds of glass, 
a number of independent measurements may be made, the mean 
of which may be taken. A number of observations may also be 
made with a single plate and different pairs of comparison lines. 
_ The chart which is herewith presented to the Academy must 
be regarded simply as a first approximation to a normal chart of 
the spectrum. It contains 187 lines, the wave lengths of which 
have been accurately determined. The chart is drawn to ascale 
of millimeters, and in entering the lines of the spectrum upon 
* Pogg. Ann., xcviii, 513. 
¢ See Miiller’s Lehrbuch der Physik und Meteorologie, Bd. i, p. 850. 
