1194 
in the direction of the central maximum of G, the corresponding 
angles of diffraction are (see table on page 1182) 
BAA C40 DEE DT 
respectively and the widths of the slit 
B 53.7 u,- 0-56 4u, sE 2m SO. 
The mean width 65.4 u may be looked upon as giving a close 
approximation to the correct value. This gives 075 mm for the 
width of the side-face of the prism and .15 mm for the diagonal 
of the base. 
Very small crystals will produce very broad maxima, in com- 
\ 
parison to which 44’ — the difference between the maxima of red 
and violet — may be looked upon as small, in consequence of 
which the various colours will cover each other and nothing will 
be seen but white with a red inner edge. 
2. It may be useful to point out the analogy with the rain-bow. 
In that case large drops give narrow diffraction-maxima and distinet 
colours, small drops broad maxima, diluted colours and the rare white 
rainbow. Similarly with the halo: the larger the ice-crystals, the 
more distinct the spectral colours will be. The ‘‘white halos” are 
by far the most common. 
Still there are some very fundamental differences between rain- 
bow and halo. Whereas in the former case the wave-front becomes 
curved, it remains flat in the latter case. Whereas in the rain-bow 
the maxima are strongly developed, though only on one side by 
which the extremely common secondary bows on the side of the 
violet are formed, these maxima are comparatively weak in the 
halo and possible on both sides. They will have the best chance of 
being seen in the dark region inside the red, but in the white on 
the outside they will but seldom sueceed in making themselves 
visible. 
3. In connection with the colours of halos the shape of the 
crystal is of some importance. Let us consider a crystal plate with 
a broad side-face but of small height. The width of the side-face 
determines the width of the slit which plays a part in the formation 
of the ordinary ring, the height determines the width of the slit for 
the circle of 46°, as this halo is formed by a refracting angle of 
90°. A plate of the above shape is specially suited to the production 
of colours in the ordinary circle, but unsuitable as regards the large 
circle. With an elongated prism the colour-production in the circle 
of 22° is again dependent on the width of the side face, but for 
the circle of 46° the determining dimension is now the short diagonal 
