7fi Mr. J. J. Guest on the Strength of 



given of the elastic-limit effect in simple tension. As any 

 defects in the material will contribute their effect, the result 

 is that it is practically impossible to select the yield-point, or 

 to determine with any degree of precision the point where 

 the curve first deviates from the straight line. 



10. First reason for use of Thin Tubes as specimens. — To 

 avoid this masking of the phenomena, I conducted my ex- 

 periments upon thin tubes, the walls of which were from 

 2V ^° 23 °f ^he ra dius in thickness. I also tested two solid 

 bars, more to ascertain, from the agreement of the elastic 

 constants with their known values, that no large mistake had 

 been made in the measurement of the parts of the apparatus. 

 An inspection of the torsion stress-strain, or rather torque- 

 twist curves (fig. 15), for these bars will show the practical 

 impossibility of locating the yield-point (or the elastic-limit, 

 if it be a true phenomenon), particularly if we bear in mind 

 Mr. Love's theorem that a small round flaw will double the 

 shearing-stress in its neighbourhood, which if the flaw were 

 near the surface would produce the elastic-limit effect at about 

 half the true yield-point torque. 



11. Results of previous Torsion Experiments. — Published 

 results of torsion experiments (see Unwin, ' Machine Design/ 

 vol. i.) make the elastic strength of iron and bronze in torsion 

 to be from 0'625 to 0-735 of the strength in tension: for 

 copper the ratio is about 0*545. The ratio of the ultimate 

 strengths is a little higher. The ultimate strength in torsion 

 should be determined by the use of the formula T = |7r^ >3 , 

 which considers the whole of the material across a normal 

 section to be in the same state of stress, but as the employ- 

 ment of this formula is not universal, the ratio of the ultimate 

 stresses is open to suspicion. 



12. Previous experiments upon the Yield-point under Com- 

 Lined Stress. — Beyond the numerous experiments upon torsion, 

 the only experiments upon the yield -point which I have 

 been able to find are those made upon pianoforte-wire by 

 Mr. McFarlane for Lord Kelvin's article upon Elasticity in 

 the Encyclopaedia Britannica. These showed that a simul- 

 taneous tension lowered the yield -point in torsion, but 

 numerical data are not given. Lord Kelvin concludes that 

 as a tension lowers the torsional yield-point, a compression 

 would raise it; but his argument does not appear to be rigorous, 

 and I venture to think that either tension or compression 

 would lower it. 



13. Eaperiments upon Ultimate Strength under Combined 

 Stress. — Of experiments upon the ultimate strength under 

 combined stress, there are the well-known experiments of 



