204 SCIENCE PROGRESS 



others being arrested by meeting the crystal boundaries. 

 When it happens that the directions of the slip-bands in two 

 adjacent grains nearly coincide, it is possible for a crack to be 

 propagated and as every increase in its length produces a 

 further concentration of stress, a crack once extended over 

 several grains tends to spread to the exclusion of neighbouring 

 smaller cracks. Ultimately the crack spreads through* the 

 whole rod with increasing velocity, owing to the increasing 

 intensification of local stress. 



The "crystalline" fracture of metals broken by fatigue is 

 thus accounted for. The glistening facets which are usually 

 regarded as crystal faces are in reality cleavage planes exposed 

 by the process just described. When produced by simple 

 alternations of stress, such a fracture is not accompanied by 

 any marked deformation of crystal grains whilst a fracture 

 produced by slowly applied tensile or bending stress preceded 

 by great deformation has an entirely different character, the 

 crystals being drawn out and torn rather than snapped asunder. 

 In the well-known instance of wrought iron, the presence of 

 brittle slag bands causes fissility in one direction, so producing 

 the characteristic " fibrous " fracture. 



The brittleness occasionally exhibited by masses of mild 

 steel, such as boiler plates, is not revealed by the usual tests 

 involving the slow and continued application of stress. It is 

 possible for a metal to show the required strength and ductility 

 in a tensile test and yet to be so brittle that a sudden blow will 

 break it without previous yielding. In order to guard against 

 such accidents, a special form of test is required in which the 

 application of the stress is such as to cause fracture in the 

 manner just described, that is, by rupture of single crystals 

 along their cleavage planes. Such tests are of two kinds, the 

 one involving repeated alternations of stress and the other a 

 suddenly applied shock — both kinds are susceptible of many 

 different modifications. An alternating stress test may consist 

 in bending the test-piece to and fro or in alternately stretching 

 and compressing it, whilst a shock test may be made in a variety 

 of ways, by means of a falling weight, a swinging pendulum or 

 a revolving arm. The test-piece intended to be broken by shock 

 is generally notched to localise the stress. Although consider- 

 able differences of opinion exist as to the most suitable form of 

 test, it is certain that either of those mentioned gives a more 



