as exhibited in its propagation along plates of glass. 103 
A, B. C, D, Fig. 49, ( PL V. ) is also peculiar. The glass cools 
more rapidly there than at any other part, and therefore a 
higher tint is developed at the angles, than towards the 
middle of the plate. 
2. On the fringes produced by square pieces of glass. 
If the breadth of the glass plate is equal to its length, as in 
Fig. 32, ( PL IV.) all the four diagonal portions nearly meet, and 
therefore, when the lateral and terminal fringes are developed, 
the central part is altogether black, as the central fringes 
have entirely disappeared. When the line AB is parallel or 
perpendicular to the plane of primitive polarisation, the diagonal 
fringes appear as in Fig. 33, (PL IV.) being always separated 
from each other by a black space in the form of a cross. This 
black cross is a necessary accompaniment of the diagonal 
fringes, for it follows, from the reasoning in Sect. I. of this pro- 
position, that all the crystals situated in the central lines, AB, 
CD, have their neutral axes in the directions AB, CD, and 
therefore cease to depolarise the incident light when the 
diagonal fringes are in full perfection. 
3. On the fringes produced by cylindrical pieces of glass. 
As the heat radiates most copiously from the heated cy- 
linder, in lines perpendicular to its surface, that is, in lines 
directed to its axis, it follows that the axis of all the elementary 
crystals will be directed to the axis of the cylinder. The uni- 
formity of the radiation in every part of the cylinder, will 
produce an uniformity of structure, which will develope si- 
milar tints at similar distances from the axis, and thus produce 
fringes concentric with the cylindrical circumference. The 
effect of a radial crystallization combined with an angular 
