PHYSICS: C. BARUS 55 
To secure the best conditions for sharp strong fringes the two slit images 
seen in the telescope must be of equal intensity, and this depends on the half- 
silver N. On testing a number of plates it is usually easy to find one which 
fulfills this condition nearly enough. The fringes are still good, even when 
the intensity of images is noticeably different. The secondary images due to 
the reflection from glass faces are either weak or (if thick plates are used) 
these reflections may be blotted out by small opaque screens suitably ^fixed to 
the mirrors m and N. It is interesting to observe that with proper adjust- 
ment these secondary reflections carry their own fringes, some of which are 
modified in like conditions more rapidly than the main set. Intersecting 
fringes producing a beaded structure and fringes moving in opposite direc- 
tions are also observed. 
2. Character of the achromatic fringes. — Since the achromatic fringes are 
quite symmetric, consisting of a central white or fringe, flanked on either side 
by 3 or 4 colored fringes rapidly decreasing in intensity, it is obvious that 
(practically) they must consist of superposed monochromatic confocal hyper- 
bolae. This may be well shown in the present apparatus, where the fringes 
are stationary and are displayed relative to horizontal and vertical lines of 
symmetry. To carry out the experiment, it is best to insert a single plate 
compensator (say C, fig. 6, 5 mm. thick) normally into the rays 85 and 21, 
preferably in the same vertical plane. When the plate is not perfect, it may be 
necessary to adjust for coincidence of slit images. If now this plate is rotated 
about a horizontal axis (normal to the lines 85, 21) the fringes walk laterally 
through the broad coincident slit images, in such a way as to clearly outline 
a moving design of the form given in figure 2. In other words, as a first ap- 
proximation (for the case is, of course, essentially more complicated) the 
achromatic fringes may be assumed to be a family of confocal equilateral 
hyperbolas, referred to given horizontal and vertical axes. When the rays 
85, 12 are at the same level, the broad slit image is in a position of symmetry 
relating to the hyperbolae, figure 2. When this is not the case, the image is at 
55 or s's\ with the fringes very rapidly becoming horizontal. Since this design 
is similarly carried out with decreasing coarseness from red to violet, it is 
clear that a single characteristic central achromatic fringe results, invaluable 
for purposes of displacement interferometry, from its smallness and since from 
the breadth of slit it can be made so intensely luminous. When the path 
difference of the rays 85, 12, in figure 1 is changed by the micrometer or by 
independent compensators, the figure 2 shifts bodily up or down the slit 
image. It is also obvious that when the fringes with white light are hori- 
zontal, they must appear as horizontal black bands in the spectro-telescope, 
regardless of the width of slit used ; and hence these fringes also are excessively 
luminous, while their displacement may be referred to a definite wave length. 
If the interferometer is not selfadjusting, the axes of figure 2 are as a rule 
inclined, and fringes are obtained in all angles of altitude needing special 
adjustment. The spectrum fringes then demand a fine slit but are also hori- 
