174 
MR. A. A. GRIFFITH ON 
per sq. inch at 540° C. At lower temperatures the rates of increase, both of this “ solid 
stress ” and the viscosity, were enormously greater. At 540° C. a fibre took about 
70 hours to reach the steady state. 
Table I. below gives the values of the surface tension obtained from these experiments. 
That for the temperature 1110° C. is the mean of five determinations by the drop 
method. The remaining figures were obtained from the sag of fibres. 
Tabl# I.—Surface Tension of Glass. 
Temperature. 
Surface Tension. 
0 C. 
lb. per inch. 
1110 
0-00230 
905 
0-00239 
896 
0-00250 
852 
0-00249 
833 
0-00254 
820 
0-00249 
801 
0-00257 
760 
0-00255 
745 
0-00251 
15 
0-0031* 
So far as they go, these figures confirm the deduction that the surface tension of glass 
is approximately a linear function of temperature. Moreover, as the actual variation 
is not great, the error involved in assuming such a law and extrapolating to 15° C. is 
doubtless fairly small. The value so obtained, 0-0031 lb. per inch, will be used in 
the present application. 
Rigorously, expressions (13) and (21) above are true only for small cracks in large 
flat plates. In view, however, of the difficulties attendant on annealing and loading 
large flat glass plates, it was decided to perform the breaking tests on thin round tubes 
and spherical bulbs. These were cracked and then annealed and broken by internal 
pressure. The calculation cannot be exact for such bodies, but the error may obviously 
be reduced by increasing the ratio of the diameter of the bulb or tube to the length of 
the crack. It will be seen from the results of the tests that the variation of this ratio 
from two to ten caused little, if any, change in the bursting strength, and hence it may 
be inferred that the error in question is negligible for the present purpose. 
The cracks were formed either with a glass-cutter’s diamond, or by scratching with 
a hard steel edge and tapping gently. The subsequent annealing was performed by 
heating to 450° C. in a resistance furnace, maintaining that temperature for about 
one hour, and then allowing the whole to cool slowly. The question of the best annealing 
temperature required careful consideration, as it was evidently necessary to relieve the 
* By extrapolation. 
