1195 
of the base which is }“3 times or 1.7 times longer than the width 
of the side-face. A crystal of that kind is therefore more suited to 
the production of colour in the circle of 46° than in that of 22°. 
And as a matter of fact a number of halos enumerated on pages 
1176 to 1178 show striking differences in the degree of colouring in the 
two circles: 95 belongs to the former kind, 7, 12, 27 and 31 to the 
latter. 
4. A further important conclusion seems to me justified, although 
I have not tested it in detail. As we have found (p. 1185), in the 
neighbourhood of the minimum deviation we can turn the incident 
beam or, what comes to the-same, the crystal over a comparatively 
large angle before its having any influence on the diffraction-fringes. 
But if that is true, the difficulty disappears which lies in the necessity 
of having to assume a constant, vertical axis in the usual explanation 
say of the circum-zenithic are. *) 
The “strikingly pure colours”, the “pure violet” of which Brsson 
speaks, are a consequence of diffraction, but not of a constant 
direction of the refracting edge. 
5. We also found (page 1185) that in the external minimum a 
much smaller variation was admissible. The same will hold with 
regard to the next maximum: another ground, therefore, to expect, 
that diffraction-rings outside the main circle will be very great 
exceptions. 4 
6. In the large circle of 46° the difference in the minimum for 
red and violet is 2°6’*): the spectrum is thus spread out over an 
angle three times as wide as in the circle of 22°. But the slit 
COS 2 
: cos 67°51! 
becomes smaller in the ratio ——~— —= SE 
COS 2 cos 40°55 
is thus enhanced in the ratio ij. With a favourable shape of the 
crystal, the effect may be increased another 1.7 times and the con- 
ditions so become 2'/, times more favourable as regards production 
of colour in the circle of 46°. This agrees with the fact, that in 
this circle striking colours have been seen comparatively frequently. 
7. In the formation of halos where the light no longer passes the 
erystal at right angles to the refracting edge, which corresponds to 
a broadening of the beam, tbe diffraction pattern agrees with that 
of a larger crystal with the light moving in a plane at right angles 
to the refracting edge. The chance of colour is increased. In agree- 
ment with this the tangential curves to the circles of 22° and of 46° 
(cireum-zenithic curve) are pretty frequently distinctly coloured. 
460 
= =. The colour-effect 
1) See a.o. L. Besson, lc. 
*) PernTER, |. c. p. 354. ‘ 
