\ • . 
104 Dr. Brewster on new properties of heat , 
crystallization is shown in Fig. 50, (PL V.) where ABCD is a 
plate of glass cooled upon a cylinder of iron at its centre. See 
Fig. 19, (PL III.) and 29, (PL IV.) 
When the section of the glass is a polygon of any number 
of sides, the form of its fringes may be easily deduced from 
the principles which have already been established. When 
the section is a triangle, no regular figure is seen. If the 
triangle is equilateral, the lines which bisect the angle, and 
those which are perpendicular to the sides, are inclined to 
each other 120°. So that the axes of the crystals are not 
symmetrically related to the rectangular axes of the particles 
of light. When the glass is a sphere, the axes are all directed 
to its centre. 
Proposition XXXIX. 
To ascertain the probable mechanical condition of the parts of the 
glass that produce the different sets of fringes. 
1 <v ‘ ; 1 ; * 
I have not felt myself authorised to deduce, from any of 
the preceding results, the mechanical condition of the parts 
of the glass which produce the different sets of colours. It is 
obvious that in the case of a red hot plate of glass, cooled in 
the open air, there is a variable density diminishing from all 
the edges inwards, but in the propagation of heat along a cold 
plate, there is no direct argument to prove, that such an in- 
creasing density exists at any of the edges, excepting the one 
adjacent to the source of heat. A similarity, however, in the 
mechanical conditions of the two plates, may be safely inferred 
from the perfect similarity of their optical properties. The 
central part of the crystallized plates, which produce fringes 
of an opposite character, are in a state of dilatation decreasing 
l 
