226 SCIENCE PROGRESS. 



change comparing them with material which has never been 

 stressed at all. Until this obscure point is cleared up we 

 shall never be able to understand thoroughly this question. 



Commercial Testing. In concluding this brief notice of 

 the present condition of testing iron and steel, it may be 

 worth while to state what it is desirable that the commercial 

 testing, carried out only to determine whether the quality or 

 a given parcel of material is up to the standard, should aim 

 at. There is no doubt that simultaneously with the tensile 

 tests what may for convenience be termed the workshop 

 tests should always be rigorously insisted on, namely, for 

 tires, axles, rails, etc., the drop tests ; and for plates, bars, 

 etc., the bending tests. These tests do at times detect 

 faults which would not so readily be brought to light 

 by the simple tensile tests of samples of the material only, 

 but they should always be accompanied by the latter test. 

 If they are properly carried out and an accurate record of 

 the results obtained is kept they must throw much light on 

 many important questions, but if, as is so often the case, 

 they are done in a careless fashion with no precautions to 

 eliminate possible errors, they will probably simply produce 

 results so discordant as to be useless, and puzzling and un- 

 satisfactory to both producer and user. 



I have already dealt to some extent with the question 

 of the form of test bars. Personally I would advocate for 

 all flat specimens a uniform length of 8 inches in the 

 parallel part, and a width of i*6 inches for all thicknesses 

 up to i inch ; beyond this it is better to have turned 

 specimens. The ends should if possible be at least 2 

 inches wide and long enough to secure a good grip of the 

 wedges, though for moderate thicknesses perfectly plane 

 parallel strips will give quite trustworthy results. For all 

 ordinary turned specimens a ratio of length to diameter of 

 10, but in the case of tires, axles, gun steel, etc., a ratio of 

 2\ to 1 will probably be found best, the ends being formed 

 either with a collar to pull from or screwed. Though for 

 scientific work it is advisable to pull through a spherical 

 bearing, it is not of much consequence for ordinary work ; 

 the ends must be exceptionally badly formed and held to 



