SHEET NON-FERROUS METALS 



753 



are significant. By significant variations are meant variations 

 that can be assigned to definite sources, such as measuring errors, 

 defects in the metal tested, testing errors, etc. For the case under 

 consideration, a method of analysis was employed which allows for 

 small sample numbers. The method used is described elsewhere.^ 

 Figure 1 shows an engineering analysis of the tensile strength 



z 116,000 



LlJ 



5^ 114,000 



UJ 



^II2,000|^ 



z 



^ 110,000 



<I08,000 



= 1 



A B C D 

 MILLING LABORATORIES 



A B C D 

 MILLING LABORATORIES 



S 116.000 

 cc 



^ 114,000 



UJ 



^ 112,000 



UJ 



^ 110,000 



UJ 



< 108.000 



A B C D 

 TESTING LABORATORIES 



A B C D 

 TESTING LABORATORIES 



Fig. 1 — -Statistical Analysis of Round-Robin Tension Tests on Alloy B Nickel Silver. 



(a) Points are average of all specimens milled at laboratories indicated, and 

 tested by the four laboratories. 



(6) Points are average of all specimens tested at laboratories indicated, and 

 milled at the several laboratories. 



results for alloy B nickel silver plotted in two ways.^ The data in 

 Fig. 1 (a) were obtained by computing, for each of the laboratories, 

 the average tensile strength and the standard deviation ^ of all the 

 specimens milled by that laboratory and tested by each of the four 

 laboratories A, B, C and D. The data in Fig. 1 (6) were obtained 



^ See Modified Criterion No. 1 in Appendix I to the Report of Sub-Committee 

 XV on Die-Cast Metals and Alloys appended to the report of Committee B-2 on 

 Non-Ferrous Metals and Alloys, presented before the American Society for Testing 

 Materials, June 1929. 



^ The methods used are those described by W. A. Shewhart in a paper on "Quality 

 Control," The Bell System Technical Journal, Vol. VI, October 1927. 



