816 
Proceedings of the j Royal Society 
their passage through quartz is a function of the wave-length, and 
is roughly represented by the formula 
A 
p ~\ 2 ’ 
where A is a constant depending on the quartz. This formula is 
only approximate, however, and one object of the experiments 
described below was to ascertain how closely the rotation might 
be represented by three terms of the equation 
ABC 
p= 'x? + w + T« + • • • • 
Another object was to test the amount of accuracy attainable by 
the exceedingly simple and direct method employed ; and a third 
was to find how nearly the constants in the assumed formula agree 
with one another in different specimens of quartz. 
A quartz plate 8 ’9 mm. in thickness was first used with ordinary 
Nicol’s prisms, as polariser and analyser. The quartz plate was 
fixed immediately in front of the analysing prism. A beam of sun- 
light, having passed through the polariser and the quartz, proceeded 
through the upper half of the analyser. Another beam was re- 
flected through the under half by a mirror placed between the 
polariser and the quartz. Both beams being then passed through a 
spectroscope, two spectra were obtained in juxtaposition, the Fraun- 
hofer lines in the one exactly corresponding to those in the other. 
The upper one exhibited in addition the usual dark band caused 
by the rotatory action of the quartz upon the plane polarised 
ray. 
The analyser being now moved round until the centre of the dark 
band coincided with a Fraunhofer line in the under spectrum, the 
angle recorded gave the total rotation for that ray, minus some 
multiple of 7 r. That position of the analyser, for which all rays 
were extinguished when the quartz was withdrawn, was of course 
taken as the standard position from which all angular deviations 
were reckoned. The following results were obtained from observa- 
tions with the rays D, E, b , F : : — 
