2(3 Frederick Guthrie on the Fracture of Colloids. 



glass, about 3J inches in diameter, be loosely balanced hori- 

 zontally between the lips of a wooden clip and brought with 

 its centre over an air-gas burner so that the top of the flame 

 is about an inch below the glass, the latter almost invariably 

 cracks at least into three pieces ; and when the pieces are three 

 in number they as invariably have the form shown in fig. 2. 



The remarkable symmetry of each of these, and their 

 similarity to one another, show that the shape is not an 

 accident of the glass. The constant features are (1) that the 

 two main cracks join before reaching the circumference, 

 (2) that there are in each crack, reckoned from this con- 

 fluence, three concavities towards the centre of the circle, the 

 first being nearly straight, (3) that there is a little kick 

 given by the crack as it leaves. 



Out of sixty-four specimens of fracture produced under 

 these conditions ten showed this species of two-crack fracture. 

 The shapes of the cracks are perfectly similar to those given ; 

 and the symmetry is sometimes such that the side pieces may 

 replace one another after inversion so perfectly that it is 

 scarcely possible to tell that they are misplaced. 



§ 4. The same method of heating may result in the pro- 

 duction of a great variety of forms ; but they are all derived 

 from the above type. In fig. 3 are shown a few of the more 

 simple. 



The three-crack figures (a, a) are about as frequently 

 formed as the two-crack figures. I find eight of the three- 

 crack out of sixty-four similarly treated specimens. Per- 

 fectly similar forms were got when the plate was laid on a 

 retort-ring or supported on three corks and heated in the 

 same manner. A special series of experiments was moreover 

 made to see if the position of the clip had any influence upon 

 the attitude of the crack. In figures 2 and 3 the mark f 

 shows where the plate was held. As to the effect of the posi- 

 tion in the original sheet of glass of the pieces experimented on, 

 as determining the attitude of the axis of cracking, the follow- 

 ing examination was made. Six pieces having been marked 

 as they lay in the sheet, forming a radial band, were cut out 

 and heated as above. The apex of crackage always appeared 

 somewhere on the semicircle which w r as towards the centre 

 of the sheet, but varied in this semicircle so considerably that 

 it is at present doubtful whether the position in the sheet in- 

 fluences the crack-axis. Fig. 4 shows the amount of varia- 

 tion ; the point of the arrow represents the apex of the crack- 

 curve. The lower figure represents the original sheet and 

 the positions of the several pieces in it. 



