
—_—_S SC OOO 
a a i a el ee a | 

Bright Spectrum Lines. 495 
shown in fig. 3, and due to multiple reflexion between the 
silvered plates we have transmitted a number of rays whose 
path differences are in arithmetical progression, and the 
theory of the method is exactly as that sketched above. By 

this method the bright interference-bands are much brighter 
than those obtained with the Fabry and Perot method, and 
hence the components more readily seen. On account of the 
larger incident angle @ of the light upon the plates, the fringes 
observed are quite a distance from the centre of the system, 
and are therefore close together, causing the necessary 
adjustments to make the plates parallel more difficult than 
with the method above where the centre of the system is 
used. Observations were made with these two methods and 
the results agreed extremely well. 
As mentioned in the introduction a method proposed by 
Lummer appeared during these experiments. He employs 
only a long glass plate with parallel faces and passes light 
into it by means of a prism at such an angle that it emerges 
at almost the critical angle. The method is very similar to 
the method above. In Lummer’s methods, however, since 
the thickness of a given glass plate is fixed, the positions of 
the components relative to one another cannot be determined, 
and if the faces of the plates are not perfectly parallel many 
anomalous results, as those indicated above, may be obtained. 
Results. 
On the basis of what has preceded the following results 
have been obtained. A number of sources of light were 
employed—metallic vapours in vacuum-tubes rendered Jumi- 
nous by the discharge from a large induction-coil, metallic 
vapours in a Bunsen flame and in an electric arc, and, lastly, 
the electric spark between electrodes of the metals. This 
latter source was found to be very unsatisfactory. Of the 
