TO SHIP CONSTRUCTION. iit 



(d) Minor tests, such as (i) cold bending of welds, (2) impact tests of welded 

 specimens. 



(e) Chemical and miscroscopic analysis. 



Tests were carried out on specimens as large as possible, particularly in respect 

 to the static determinations of elasticity, ultimate strength and elongation, some of 

 the test specimens being designed for a fatal load of just under 300 tons. The ad- 

 vantage of these large specimens was that the effect of workmanship was better aver- 

 aged and the results were more comparable to the actual work likely to be met with 

 in ship construction. 



Figure i, Plate 148, illustrates the method adopted to measure the modulus of 

 elasticity by means of a strain meter designed by Dr. James Montgomerie, princi- 

 pal surveyor for Lloyd's Register in Scotland. The photograph shows the holders 

 which were specially designed with a view to securing, as far as practicable, an 

 even pull across the breadth of the plate. 



Readings were taken in way of the weld and at various points on the plate itself, 

 both adjacent to and well clear of the weld, the points at which observations were 

 taken being clearly shown in Fig. i, Plate 149. 



Typical results, illustrating the extensions as measured along the line C, are 

 shown in Fig. 2, Plate 149, from which it would appear that the extensions in way of 

 the weld do not show any marked difference from those at various points in the plain 

 plate, the lines showing extensions in way of the weld lying among the others with- 

 out disclosing any distinctive features. Measurements were also taken with the strain 

 meter set at right angles to the line of pull, the readings in this case, of course, rep- 

 resenting contractions (Fig. 3, Plate 149). 



With a view to confirming this result a set of specimens of smaller size was pre- 

 pared and tested. Automatic stress-strain diagrams showing the extensions up to the 

 point of fracture are shown in Fig. 4, Plate 149, while Fig. 5, Plate 149, represents 

 on an enlarged scale the extensions within the elastic limit, the distance between the 

 points of attachment of the instrument being 8 inches in each case. The curves ex- 

 hibit the same general characteristics as those obtained from the large specimen and 

 would appear to justify the inference that there is very little difference between the 

 modulus of elasticity of the welded samples and that of the plain plate. 



With alternating stresses the specimens were relatively of small size. For the 

 rotating test pieces, circular rods, machined from a welded plate, were used, the di- 

 ameters selected being i inch and ^ inch. These bars, about 3 feet in length, were 

 attached to a lathe headstock, and a pure bending moment in one plane was ap- 

 plied by means of two ball races to which known weights were attached. The 

 material of the bar was thus exposed alternately to maximum tension and to equal 

 maximum compression once in each revolution. The machine was run at about 1,060 

 revolutions per minute. 



Bars of identical material were tried in pairs, one specimen welded and the 

 other unwelded, and the number of revolutions before the specimens parted was 

 observed for various ranges of stresses varying from ±15 tons to ±6 tons per 

 square inch. 



