A .005 variance using the Hornady Comparator & Lock/Load Gauge isn't uncommon.
There are so many moving parts and surface contact pressures to master in using it to its fullest potential that, in your place, I'd simply average the results.
First you have the primary rod with shell casing attached that has to be pressed against the shoulder area of the chamber and held precisely in place while you push a rather sloppy fitting bullet forward and "feel" its contact with the lands. That contact has to be precisely the same every time.
Then you have to tighten the set screw to hold the center rod in place - hopefully without changing the amount of pressure you exerted on the bullet - and hope that tightening the screw doesn't pull or push the center rod in either direction or that you didn't inadvertently add or reduce pressure in the process.
Then you have to withdraw the pieces (and sometimes reassemble them because the bullet pulled free of the case neck) and take a measurement using calipers that may not be laboratory grade.
The Hornady Comparator & Lock/Load Gauge works OK for setting up OAL where variables of .001 - .002 aren't usually a big deal but they're not intended for measuring throat erosion.
Try making a casting of the chamber.