Gear Inspection Chart Evaluation; Specifying Unusual Worm Gear Sets

Question: When evaluating charts from a gear inspection machine, it is sometimes found that the full length of the profile traces vary, and that sometimes they are less than the length of active profile (above start of active profile-SAP) by up to 20%. This condition could be caused by a concentricity error between tooth grinding and shaping, or by unequal stock removal when grinding. (See Fig. 1.) Is it possible that some of the variation is coming from the inspection machine? How can variation from the inspection machine be reduced?