THE PHENOMENA OF RUPTURE AND FLOW IN SOLIDS. 
173 
The specific heat of glass is greater at high than at low temperatures, but the 
temperature coefficient is not large. Hence its surface tension may be expected to be 
nearly a linear function of the temperature, and extrapolation should be fairly reliable. 
This was found to be the case with the glass selected for the present experiments. 
In the neighbourhood of 1100° C. the surface tension was found by Quincke’s drop 
method. At lower temperatures this method was not satisfactory, on account of the 
large viscosity of the liquid glass; but between 730° C. and 900° C. the method described 
below was found to be practicable. Fibres of glass,‘about 2 inches long and from 
0-002-inch to 0-01-inch diameter, with enlarged spherical ends, were prepared. These 
were supported horizontally in stout wire hooks and suitable weights were lnmg on 
their mid-points. The enlarged ends prevented any sagging except that due to 
extension of the fibres. The whole was placed in an electric resistance furnace main¬ 
tained at the desired temperature. Under these conditions viscous stretching of the 
fibre occurred until the suspended weight was just balanced by the vertical components 
of the tension in the fibre. The latter was entirely due, in the steady state, to the surface 
tension of the glass, whose value could therefore be calculated from the observed sag 
of the fibre. In the experiments the angle of sag was observed through a window in 
the furnace by means of a telescope with a rotating cross wire. If w is the suspended 
weight, d the diameter of the fibre, T the surface tension, and 6 the angle at the point 
of suspension between the two halves of the fibre, then, evidently, 
x . d . T . sin = w. 
For this method of determining the surface tension to be valid, it is evidently necessary 
that the angle of sag shall reach a steady value before the development of local 
contractions, arising from the instability of liquid cylinders, becomes appreciable. 
That this requirement is satisfied is shown by the following experimental results. After 
heating for two hours at about 750° C. the angle of sag of a particular fibre was 18°-25. 
Two hours later it had increased by less than 0°-l. The temperature was then raised 
momentarily to 940° C., and quickly reduced again to 750° C. The angle was then 
found to be 20°-2. After two hours further heating at 750° C. the angle had decreased 
to 18°-4, agreeing within permissible limits of error with the former value. That 
is to say, substantially the same limiting angle of sag was reached whether the initial 
angle was above or below that limit. 
Above 900° C. it was found that the viscosity was insufficient to enable an observation 
to be made before the fibre commenced to break up into globules. Below 730° C., on 
the other hand, observations made on fibres of different diameters were inconsistent, 
the apparent surface tension being higher for the larger fibres. The obvious meaning 
of this result is that below 730° C. the glass used was not a perfect viscous liquid and 
hence the method was inapplicable. The transition from the viscous liquid state was 
quite gradual. The maximum tension (apart from surface tension) which could be 
permanently sustained, was zero at 730° C., 1-3 lbs. per sq. inch at 657° C., and 24 lbs. 
2 b 2 
