Calibration Question

I read the posts on calibration (here and elsewhere) and have a question that I didn't see an answer for...

I ran the calibration routine in Utilities just to get the numbers.  They were off a good bit so, I did the calibration method in the sticky (thanks, Carl).

I printed a test part on the four corners and in center.  The parts are fairly equal in dimensions but, the boundary lines were hit-and-miss (thin to non-existant in places). Regardless, the parts were ok.  I checked the gap again with sheet of paper and they were basically even and barely dragging the paper.

I ran the calibration utility again to see what the number were: 1 & 3 errored.  So, I re-calibrated per the sticky.  It was still okay but, I tweaked them just a tad then ran the utility again to see the results.  They are 35, 205, 165  that's better than error.

My questions are:
1.  Should I disregard the numbers and not bother with checking them after calibrating?
2.  How can they be so different and produce a fairly good part while being off so much that the indicate err?
3.  To get a baseline distance of 250mm, where/what is the reference point (I'm assuming were talking 0.250mm and referenced from the Nozzle or the Thermistor?  [it's the thermistor that touches the test points]