132 SECTIONAL ADDRESSES. 



indicate the nature of the load— thus I cannot say if it broks under 

 a torsional load (possibly torsional vibration) or under a bending load 

 (possibly due to some periodic variation of thrust on one of the pro- 

 peller blades as it passed an obstruction). Until the nature of the load 

 which caused the failure is known, it is very difficult to take steps to 

 ^ard against similar accidents. For the most urgent reasons, there- 

 lore, we require to be able to understand the fracture, as in nearly all 

 aircraft problems men's lives hang on the answer. 



Turning now to the question of the most suitable material, I will 

 take as an example the material for the crankshaft of an aeroplane 

 engine. A few months before the Armistice there were difficulties in 

 getting sufficient supplies of the high-grade nickel-chrome steel forgings 

 then in general use for shafts, and proposals were made to use a plain 

 carbon steel. Such a steel would be about 30 per cent, weaker, accord- 

 ing to the ordinary tests. A conference of leading metallurgists and 

 engineers was held to discuss the suggestion. No one present ventured 

 to predict whether the weaker steel would answer or not, or whether 

 the dimensions would have to be increased or not. It was pointed out 

 that a French engine was now using 50-ton steel with better results 

 than when using the 100-ton steel for which it was designed, no 

 changes in dimensions having been made. Such a reduction of strength 

 might be understood in ordinary engineering where there are large 

 margins of safety, but in an aeroplane engine, in which every ounce 

 of metal is cut off which can be spared, they show how completely 

 ignorant engineers are of what the suitability of material depends on. 



As another example, AVhy are oxygen cylinders annealed — repeat- 

 edly ? Annealing reduces the steel to its weakest condition. I believe 

 the fondness for annealing is due to our ignorance of the properties 

 we i-equire. Perhaps the quality of steel which an engineer fears most 

 is brittleness. He believes that annealing will soften it and reduce the 

 brittkness ; so he anneals, blindly. The fact is that we do not know 

 what brittleness is — we cannot define it — we cannot measure it — 

 though there are endless empirical tests to detect it. Till we know 

 what it means and can measure it we are in a miserable position. 

 During the war I was consulted on what could be done to reduce the 

 enormous weight of oxygen cylinders, and I advised that experiments 

 should be made on the high-quality alloyed steel tubes we were using 

 in aircraft construction. The department dealing with these tubes 

 took the matter up, and alloyed steel cylinders, properly heat-treated, 

 were made. These were, I believe, a success, and only weighed a 

 small fraction of the old-fashioned cylinders. But my suggestion was 

 little more than a guess, and no means was known of accurately testing 

 the suitability of the material, so they were only accepted after passing 

 any number of empirical tests, consisting of various kinds of rough 

 usage, to see if they would crack or burst. Surely an engineer should 

 ■be able to sav whether a cylinder is safe without dropping it from 

 tTie roof or rolling it down the front-dcor steps to see if it breaks. 



These examples refer only to different grades of the same material — 

 steel — ^bnt how far worse off we are when the problem is whether some 

 other alloy would be suitable to replace steel. Proposals have been 



