196 
MR. A. A. GRIFFITH OH 
the thermal vibrations must overcome the minimum attractions of the molecules in each 
group. It is clear, therefore, that at this temperature the substance will be unable to 
withstand, shearing stresses. At the same time it cannot vaporise, as the molecules 
must still be held together in chains by their maximum attractions. In other words, 
the transformation which has been discussed is simply the liquefaction of the solid. 
This view of the phenomenon of melting indicates that the molecules of liquids are 
in general arranged in groups or chains, of a length comparable w r ith that of the 
structure ascribed to solids in the preceding work, or, say, 10 4 molecules. 
If, therefore, a liquid be contained in a solid boundary which it wets, the ends of 
these chains may be expected to attach themselves to the solid; and if at any point 
the distance between the bounding walls is less than the length of the chains, some of 
the latter will attach themselves to both walls and hinder the free flow of the liquid 
and the relative movement, if any, of the boundaries. At such a point the liquid will 
act as a solid under any stress which is insufficient to break the chains. 
This has been verified experimentally. The apparatus consisted of a polished steel ball 
1 inch in diameter, and a block of hard tool steel containing a circular hole about 4 inches 
long. The hole was carefully ground, after hardening, to a diameter about 0*0001 inch 
greater, at its smallest part, than the diameter of the ball. When both were dry the ball 
passed freely through the hole. If, however, they were w T etted with a liquid, consider¬ 
able pressure was necessary to force the ball through. This resistance possessed the 
characteristic “ stickiness ” of solid friction, and was exactly the kind of resistance 
which would have been expected in forcing the dry ball through a hole which was too 
small for it. 
To show that the resistance was a true “ solid stress ” and not due merely to viscosity, 
the apparatus was on one occasion left for a week, with the weight of the ball supported 
by the stress in the liquid (paraffin oil). The hole was vertical, so that there was no 
normal pressure between its surface and the surface of the ball. During this period 
no motion whatever could be detected. 
It is essential to the success of these experiments that the ball and hole should be 
thoroughly wetted by the liquid. For this reason the liquids used have been chiefly 
paraffin oil and lubricating oils, but on one occasion the effect was obtained with water. 
The present theory suggests a reason for the very low tensile strength of liquids. 
If a liquid is composed of a random aggregation of chains of molecules, it may reasonably 
be expected to contain regions of dimensions comparable with, but smaller than, the 
length of the chains, across which no chains run. Rupture of the liquid will evidently 
occur by the enlargement of these cavities. Now the tension, R, necessary to enlarge 
a spherical cavity of diameter, D, in a liquid of surface tension, T, is given by 
R = 4T/D. 
In the case of water, the tensile strength, R, is about 70 lbs. per sq. inch at ordinary 
temperatures, while T is about 0-00042 lbs. per inch. Hence the cavities, if spherical, 
must be at least 0-000024 inch in diameter. This is of the order indicated by the theory. 
