28 Frederick Guthrie on the Fracture of Colloids. 



should give the same crack-figure as the cooling of the cir- 

 cumference : and, as a matter of experiment, it is found that 

 the figures are very similar. In order to cool the circum- 

 ference of a heated circular plate with some approach to 

 uniformity, an annular trough was constructed by cementing 

 concentrically two glass cylindrical vessels of the same depth 

 but different diameters, one inside the other, and over-filling 

 with mercury, so that the convex surface of the metal pro- 

 jected. The glass was held in a wooden clip widely stretched, 

 so that the axis of the clip being vertical the plate was hori- 

 zontal. Held, when uniformly hot, immediately above the 

 mercury, it was let drop and pressed down in the middle by a 

 piece of wood. The fracture is in this case instructive ; for 

 while in fig. 8 (a) the old type got by heating the centre is 

 resumed in b and c, the fracture is either influenced or even 

 accompanied by the circular fracture along or near the line of 

 greatest temperature-difference. 



§ 10. It is clear that heating in the central regions should, 

 produce a similar fracture to that brought about by cooling 

 around the circumference, and cooling at the centre a similar 

 fracture to that caused by heating the circumference. 



On heating a circular plate at the circumference by means 

 of a " rose " air-gas burner, it breaks with far greater violence 

 than when fracture is produced by central heating. The 

 parts are scattered at least three times as far in the former as 

 in the latter case. The form of the fracture is essentially 

 radial ; but the fragments, even when the primitive type is 

 widely departed from, present wonderful symmetry. A 

 noticeable point in this fashion of fracture is the invariable 

 appearance of two pieces on opposite sides of the centre whose 

 form is approximately rectangular ; that is, their sides are half- 

 cords instead of radii. This form suggests that there are two 

 chief centres of maximum fracture, and that the bounding 

 radii of the two s}'stems are parallel. In fig. 9 the pieces 

 marked a represent these singular pieces. Out of seven plates 

 which have been broken in this way, there is not one in which 

 this feature is absent. 



§ 11. As to cooling a hot plate in the centre, I find such 

 extreme difficulty in reproducing the inverse conditions of 

 heating a hot plate at the circumference that I have rarely 

 succeeded in reproducing the same type of fracture. 



Also it is seldom the case that a sheet of glass cracks during 

 heating at its edge. More frequently a sheet of glass which 

 has been heated at its edge cracks when cooling. The crack 

 then appears to follow that isothermal line along which there 



