494 Mr. J. Barnes on the Analysis of 
6 and 7 show the effect upon the bands when the interference- 
plates are only a very small degree from being parallel, they 
being displaced from parallelism by merely raising one of 
the mercury adjusting tubes less than a centimetre. 
In figs. 8 and 9 the plates are separated 0°5 mm., in 
neither case are the plates parallel, in 8 they have an angular 
separation of over 1”. These photographs also show the 
interference-bands produced im the plates themselves super- 
imposed upon the other. 
It is to advantage in these observations to obtain all the 
light possible, thus a broad source is always employed. 
The interfering rays from the different points of the source 
can only produce a clear interference pattern in the focal 
plane of the objective ; in any other plane the interference- 
bands will be wide and hazy. Figs. 10, 11, and 12 illus- 
trate this point. The whole slit is covered with the exception 
of two points separated 4 mms. from one another in a 
horizontal direction. Fig. 10 shows the effect when the 
photographic plate is placed about 1 cm. inside the focus of 
the objective. Fig. 11 when the plate is placed 2 ems. 
beyond the focus. Fig. 12 when the plate is exactly in the 
focal plane ;\ then the fringes produced by all points of the 
source are coincident and give clear and sharp interference- 
fringes. One can easily see that if the whole source were 
used instead of two points, the bands in 10 and 11 would be 
wide and hazy, so that if any of the bands due to the com- 
ponents of the radiation were present they would probably 
be entirely obliterated. 
To set the telescope at infinity is easy, but the adjustments 
necessary to obtain the silvered surfaces parallel are more 
or less difficult and can only be obtained with practice. The 
plates are parallel when the fringes are sharp and the 
illumination equally distributed over the series of rings due 
to the components. 
A Reflecting Interferometer. 
In the Fabry and Perot interferometer a large amount of 
light is lost due to reflexion from the surface of the silvering 
in contact with plate A (fig. 1), so that only a small per- 
centage is transmitted. To eliminate this defect the plates 
were mounted according to the following fig. 3. Plate A 
was heavily silvered and polished on its inside surface and 
mounted on the carriage of the interferometer as described 
above. Plate B has its inside surface almost completely 
silvered, its reflecting power being about ‘9, and is mounted 
in the other frame. Light is incident upon the plate A, as 

