286 



National Resources Planning Board 



phase of aircraft design in so far as they were available. 

 The standard technique is first to design a part on the 

 basis of calculated strength, then build and test it, and 

 if the tests do not agree with predictions, revise the 

 design and build and test the modified part. This 

 process is continued as many times as necessary to 

 attain a satisfactory result. It is slow and expensive. 

 Theoretical methods are now reliable enough that the 

 majority of structural tests confirm predictions with 

 sufficient accuracy to require no revision. However, 

 new problems constantly present themselves — the 

 introduction of pressurized cabins recently gave rise to 

 several — and hence continual mathematical study is 

 required. A beginning has also been made in the use of 

 the principles of probability in setting up structural 

 loading factors. 



(5) Flutter 



We have already commented upon the impractica- 

 bility of studying this phenomenon by any means other 

 than the mathematical. The general equations are 

 complicated and have only been solved by making 

 important simplifying assumptions. The results are 

 serviceable for check purposes, but need further elabora- 

 tion. The importance of the problem increases pro- 

 gressively as more efficient planes are designed, and the 

 necessity for an adequate mathematical theory is 

 becoming critical. 



Future prospects: It appears inevitable that from 

 motives of economy the industry will rely increasingly 

 upon theoretical methods of design and that mathe- 

 matics will play a larger part in the future than at 

 present. It is also probable that for competitive reasons 

 the various companies will supplement government 

 research by fundamental studies of their own. Further- 

 more, in view of the present fragmentary state of aero- 

 dynamic theory, it would not be surprising if part of 

 the research effort was devoted to the improvement of 

 the basic theory itself. 



The reliability of these predictions is, of course, 

 conditioned by the financial prospects of the industry. 

 Just now, war orders are causing abnormal inflation of 

 earnings ; when these cease, retrenchment will be inevi- 

 table. The industry is not highly mechanized, however, 

 and hence its present cycle of inflation does not imply 

 so large an expenditure for plant as would be true in 

 most manufacturing fields. For this reason, the period 

 of deflation may prove to be one of large war profits in 

 the bank but insufficient orders to occupy the time of 

 many competent technical men whom the management 

 would be reluctant to let go. If this should occur, an 

 almost explosive development of research may take 

 place. 



Wliether the development is explosive or not, how- 

 ever, it is probable that the industry will soon become 



one of the largest employers of industrial mathemati- 

 cians. 



Industrial Statistics and Statisticians 



The subject of statistics enters the business world at 

 points quite distinct from those touched by the rest of 

 mathematics. Moreover, the types of business activity 

 to which it most frequently applies — insurance and 

 finance, economic forecasting, market surveys, elas- 

 ticity of demand against price, benefit and pension 

 plans, etc. — belong to the field of economics which is 

 the subject of a separate report, and need not be 

 touched on here. 



There are certain other respects in which statistical 

 theory could be of great service in industry, but they 

 have been exploited to only a limited extent. This 

 report must therefore point out these hopeful fields 

 rather than record achievements in them. 



Statisticians in Industry 



By "statistician" we mean a person versed in and 

 using the mathematical theory of statistics, not one 

 who collects, charts, and scrutinizes factual data. In 

 the business world the word is more often used in the 

 latter sense. 



There is a very great difference between the number 

 of statisticians in industry, and the number of men 

 interested in some form of statistics. How great the 

 discrepancy is will be clear from a comparison of the 

 membership of the American Statistical Association, 

 which devotes itself to the application of statistics in its 

 broadest sense, and of the American Institute of Mathe- 

 matical Statistics, which confines itself narrowly to the 

 development of statistical technique. The former lists 

 277 names with industrial addresses; the latter only 10. 



Statistics in Industry 



Dr. W. A. Shewhart, research statistician of the Bell 

 Telephone Laboratories, has delineated broadly and 

 succinctly the field in which statistics may be expected 

 to find application as foUows: 



Since inductive inferences are only probable, or, in other 

 words, since repetitions of any operation under the same essential 

 conditions cannot be expected to give identical results, we need 

 a scientific method that will indicate the degree of observed 

 variability that should not be left to chance. Hence it appears 

 that the use of mathematical statistics is essential to the develop- 

 ment of an adequate scientific method, and that mathematical 

 statistics may be expected to be of potential use wherever scien- 

 tific method can be used to advantage. 



More specificaUy, there are five recognizable types of 

 industrial engineering activity in wliich statistical 

 theory either is, or should be used. 



(a) In studying experimental data to determine 

 whether the observed variations should be regarded as 



