16 Transactions of the Society. 



^entration of stress due to too rapid change of section, and the 

 possibility of a tool mark upon the specimen acting as a starting 

 point for a crack, were entirely avoided. The strain effects were 

 observed at intervals under the Microscope upon a longitudinal 

 face of the piece which had previously been polished, and, if 

 necessary, lightly etched, a faint outline only of the structure being 

 developed for the purpose of locating, whilst not obscuring, the 

 strain effects. The speed of rotation was in all cases about 400 

 revolutions per minute. The mechanical results have been else- 

 where discussed. 



Probably tlie most interesting series, which at the same time 

 comprises a number of typical features, w^s one in which the steels 

 were heated for various lengths of time at chosen temperatures, 

 ranging from 600° C. to 1200° C, and allowed to cool slowly with 

 the furnaces. The microscopic strain effects in these specimens 

 can be divided into two types, which, however, gradually merge 

 into one another. In the case of steels tested in the rolled con- 

 dition, or annealed at temperatures not exceeding about 750° C, 

 the surface becomes greatly rufHed \\\)on. fatiguing, and there 

 appear very numerous short crooked outcrops of surfaces, upm 

 which slip has repeatedly occurred. An example is shown in 

 plate IV. iig. 1 : it represents a piece of steel A which had been 

 annealed at 620° C. The stress in the field of the photograph was 

 19 tons per square inch, alternately in tension and compression, 

 and the number of alternations, previous to photographing, 100,000. 

 Plate Y. fig. 2 shows another field of the same specimen at a 

 higher magnification, after rupture, which occurred with 252,300 

 alternations. Specimens which have been heated to, and slowly 

 ■cooled from higher temperatures, however — and which therefore 

 have become more or less overheated — show a much less ruffled, or 

 practically quite unruffled surface, with relatively few outcrops of 

 surfaces of repeated slip ; these outcrops are less crooked and 

 longer than those in the previous class. l*late IV. fig. 3 gives a 

 general view of an incipient crack in a specimen of steel A, which 

 had been heated to, and slowly cooled from 900° C. The stress 

 was 19 tons per square inch, and the specimen had borne 75,000 

 alternations. Plate V. fig. 4 shows another field upon the same 

 specimen near to the fracture, which formed after 114,300 cycles 

 of stress. 



Several years ago Professor Ewing and the author were engaged 

 in studying the nature of Liiders' lines, which, as is well known, 

 appear obliquely to the direction of stress upon a suitably pre- 

 pared surface of certain metals, when stressed beyond the elastic 

 limit. With further loading a line becomes a band of increasing 

 width, within whose limits the surface of the specimen is visibly 

 ruffled. With moderate magnification the wave-length of these 



