198 MR. A. A. GRIFFITH ON THE PHENOMENA OF RUPTURE AND FLOW IN SOLIDS. 
strength is, in fact, no greater than it would be, according to the theory, if the test- 
pieces contained cracks several thousand molecules long. 
(5) It has been found possible to prepare rods and fibres of glass and fused quartz 
which have a tenacity of about one million pounds per square inch (approximately 
the theoretical strength) when tested in the ordinary way. The strength so observed 
diminishes spontaneously, however, to a lower steady value, which it reaches a few 
hours after the fibre has been prepared. This steady value depends on the diameter 
of the fibre. In the case of large rods it is the same as the ordinary tenacity, whereas 
in the finest fibres the strength diminishes but little from its initial high value. The 
relation between diameter and strength is of practically the same form for glass fibres 
as for metal wires. 
(6) If it is assumed that intermolecular attraction is a function of the relative 
orientation of the attracting molecules, it is possible to construct a theory of all the 
phenomena mentioned in (3), (4) and (5) above. In the case of crystalline substances 
the theory also appears to explain yield and shearing fracture ; elastic hysteresis ; 
elastic afterworking ; the fracture in tension of ductile materials and the flow of brittle 
materials under combined shearing stress and hydrostatic pressure ; the drop in stress 
which occurs on the initiation of yield in ductile substances ; fatigue failure under 
alternating stress ; and the relatively slight effect of surface scratches on fatigue 
strength. In the case of non-crystalline materials the theory explains elastic after¬ 
working and the great strength of short columns in compression. 
(7) The theory shows that the application of the mathematical theory of homegeneous 
elastic solids to real substances may lead to error, unless the smallest material dimension 
involved, e.g., the radius of curvature at the corner in the case of a scratch, is not many 
times the length of a crystal. 
(8) It should be possible to raise the yield point of a crystalline substance by 
“ refining” it, until at the ultimate li m it of refinement the yield stress should be of 
the same order as the theoretical strength. It should also be possible similarly to 
increase the tenacity. Up to a certain stage the fatigue range should be unaffected by 
refining, but thereafter it should increase in the same degree as resistance to static stress. 
(9) The theory requires that a thin film of liquid enclosed between solid boundaries 
which it wets should act as a solid. Experimental confirmation of this has been obtained. 
In conclusion, the author desires to place on record his indebtedness to many past 
and present members of the staff of the Royal Aircraft Establishment for their valuable 
criticism and assistance, and also to Prof. C. F. Jenkin, at whose request the work 
on scratches was commenced. 
[Note .—It has been found that the method of calculating the strain energy of a cracked plate, 
which is used in Section 3 of this paper, requires correction. The correction affects the numerical values 
of all quantities calculated from equations (6), (7), (8), (10), (11), (12) and (13), but not their order of 
magnitude. The main argument of the paper is therefore not impaired, since it deals only with the order 
of magnitude of the results involved, but some reconsideration of the experimental verification of the 
theory is necessary.] 
PRESENTED 
2 B 0C T !920 
