CAKI) TK'WSLATOi; i;X I'l'.KI M KXTS 



391 



Table XII — Analysis of Vaiuance — Summary — 

 With Card Support Bars 



* See Appendix and Reference. 



t Significant at 5 per cent level. 



$ Significant at 1 per cent level. 



(+) X.S. = Not significant at 5 per cent level. 



Cleail}', Card Position A gives rise to an undesirable assignable cause 

 which must be dealt with, and Card Position B while showing tilt as 

 an assignable cause has a dropping time at the extreme tilt well within 

 the allowable tolerances (see Conclusions). 



Balanced Loading 



The two main varial)les were total machine load in numbers of cards 

 (Xumber) and distribution of cards over bins (Loading). The results 

 of this experiment were so clear that little analysis was necessary. Al- 

 though the mean dropping time of cards distrilnited uniformly over the 

 twelve bins was statistically significantly different from the means of 

 cards from the three nonuniform distributions the magnitude of the dif- 

 ference 4.8 ms was not sufficiently large to cause concern. Furthermore 

 a decrease in mean dropping time was obsei'ved as the total load in- 



