732 



BELL SYSTEM TECHNICAL JOURNAL 



from a control viewpoint of Type V specification over the other 

 suggested specifications. We have seen in Fig. 4 of the previous 

 article on the control chart, footnote 1, that evidence of lack of control 

 may be obtained through deviations in one parameter and not in 



13 15 17 19 2\ 23 

 = NUMBER OF CELLS 

 Fig. 7 



25 27 29 



another. For example, in this figure the per cent'defective part of the 

 chart corresponded to Specification Type I. Only 4 of the 12 points 

 on this chart were outside the control limits whereas more than 4 points 

 were outside the control limits for every other parameter and for the 

 X^ part of the chart every one of the points was outside the control 

 limits. Of course it is to be expected that the x^ test would be much 

 more stringent than the test applied under Type I specification because 

 the control limits established under the Type I specification are merely 



