376 Mr Pocklington, On a method of increasing the 
a plane polarised beam, the plane of polarisation depending on 
the amount of retardation that one of the interfering beams may 
have undergone owing to a displacement of one of the mirrors or 
to the interposition of some body in the path of one of the 
beams. As the intensity of the resultant beam is independent of 
the retardation, nothing will be observed when the emergent 
light is examined with the naked eye ; but if a nicol is used, all 
light polarised in a certain plane will be quenched, and the inter- 
ference phenomena will be seen as usual. For example, if one of 
the mirrors is slightly inclined to the image of the other in the 
inclined plate, a series of parallel bands is seen. If the analyser 
be turned through any angle, the effect produced on the 
phenomena is the same as if one of the two interfering beams had 
been retarded by a proportional amount (a retardation of a wave- 
length corresponding to a rotation of 180°), and the bands will 
move uniformly across the field. I have verified this experiment- 
ally with an interferometer having an unsilvered inclined plate, 
and find that when white light is used, the bands as they move 
change their colour in accordance with the position they occupy, thus 
e.g. each band as it comes to the place of the central band becomes 
black, and after passing it reverses the order of its colouring. 
In order to secure the maximum sensitiveness it would theoreti- 
cally seem best to work with each mirror coincident with the 
image of the other in the inclined plate. The state of polarisa- 
tion of the emergent beam is then uniform, and the plane of 
polarisation can be determined as in any form of polaristrobo- 
meter. The error, under favourable circumstances, should 1 not 
exceed 3', which corresponds to an error of 1/3600 of the breadth 
of a fringe when the observations are made in the usual way. 
Even if we assume that the twentieth part of a fringe can be 
measured in the latter case, the accuracy can be increased nearly 
two hundred-fold by the use of circularly polarised light. For 
many of its uses the interferometer is already sufficiently sensi- 
tive, but for experiments on the drift of the ether, and perhaps 
also on the influence on transmitted light of the layer of air that a 
surface of glass condenses on itself, the increased sensitiveness 
would be useful. 
3. In arranging the details of the apparatus we see that the 
inclined plate cannot be silvered, on account of the difference in 
the retardation it produces in the two perpendicularly polarised 
parts of either beam. Further, one of the interfering beams has 
undergone four refractions, and the other two if no compensating 
plate is used, or six if one is used. The beams are hence of 
different intensities. Since the two parts of a beam are 
1 H. Landolt, Handbook of the Polariscope. 
