tivity of the liquid without appreciably affecting the quartz.
R. W. Wood has studied the iridescent colours seen when a precipitate
of potassium silicofluoride is produced by adding silicofluoric acid
to a solution of potassium chloride, and found that they are due to
the same cause, the refractive index of the minute crystals
precipitated being about the same as that of the solution, which
latter can be varied by dilution.

_Anomalous Dispersion._--In some media the usual order of the colours
is changed. This curious phenomenon was noticed by W. H. Fox Talbot
about 1840, but does not seem to have become generally known. In 1860
F. P. Leroux discovered that iodine vapour refracted the red rays more
than the violet, the intermediate colours not being transmitted; and
in 1870 Christiansen found that an alcoholic solution of fuchsine
refracted the violet less than the red, the order of the successive
colours being violet, red, orange, yellow; the green being absorbed
and a dark interval occurring between the violet and red. A. Kundt
found that similar effects occur with a large number of substances, in
particular with all those which possess the property of "surface
colour," i.e., which strongly reflect light of a definite colour, as
do many of the aniline dyes. Such bodies show strong absorption bands
in those colours which they reflect, while of the transmitted light
that which is of a slightly greater wave-length than the absorbed
light has an abnormally great refrangibility, and that of a slightly
shorter wave-length an abnormally small refrangibility. The name given
to this phenomenon,--"anomalous dispersion"--is an unfortunate one, as
it has been found to obey a regular law.

In studying the dispersion of the aniline dyes, a prism with a very
small refracting angle is made of two glass plates slightly inclined
to each other and enclosing a very thin wedge of the dye, which is
either melted between the plates, or is in the form of a solution
retained in position by surface-tension. Only very thin layers are
sufficiently transparent to show the dispersion near or within an
absorption band, and a large refracting angle is not required, the
dispersion usually being very considerable. Another method, which has
been used by R. W. Wood and C. E. Magnusson, is to introduce a thin
film of the dye into one of the optical paths of a Michelson
interferomet