Although the 3 bullets were similar to within .001" in the dimensions that you measured, one possibility is that they may differ in nose diameter at a given point on the ogive, which means the caliper insert won't seat at the same spot. I'm not claiming this is what's actually happening, just pointing out that there are possible reasons for what you observed/measured. I would suggest taking repeated measurements with the bullet in question and at least one of the other three bullets. That way, you can develop a feel for the precision you are able to get when taking the measurements several times with the same bullet. It's good to have some feel for this so that when you get one measurement that differs significantly from the others (i.e. .007" difference), you have a better idea of whether it might be a measuring technique issue, or some dimensional difference between the bullets that isn't a bullet dimension we typically measure, but affects the output nonetheless.
I am also in agreement with LVLAaron. The "distance to touching" measurement is a reference point, nothing more. With the tools we typically use to take such measurements, the accuracy of such measurements may sometimes be questionable. Nonetheless, as a reference point it will work just fine if the accuracy isn't way off (i.e. within a couple or three thousandths or so). A single bullet measurement will work just fine for that purpose. I am stupidly OCD, so I make a randomly selected set of 10 bullets from every individual Lot# as my "Measurement Set", and take/record measurements with all 10 of them. I then use the average value as my reference point. The main reason I do this (at least in my mind) is to minimize the effect of bullet dimensional variance and user error. But you can do the same thing with a single bullet, especially if that one bullet gives a measurement that is reproducible over several repetitions.
The average measurement I use for "touching" might actually be .003" into the lands, or it could be .003" off. But when used as a reference point only to set up a seating depth test, it won't matter. The target will ultimately reveal which seating depth provided the best grouping, and the CBTO measurements of loaded rounds can be reproduced with very good precision. If the bullet is really .009" off the lands instead of .012" off because it was actually .003" into the lands for what I measured as "touching" won't matter. I can still reproduce the correct CBTO measurement for loaded rounds that also produced the best grouping.
www.brownells.com
