Measuring runout on a lathe is a royal pain to do correctly. Simply measuring the spindle with an indicator is not the best way to do it.
Test bars and plates are expensive but they are what should be used to measure the runout on a machine among other things. An old toolmake I know has a couple of test plates that ride in the spindle of different lathes. Precision ground to near zero run out. He can place this plate in your spindle and measure the runout at a 10" diameter magnifying the error.
The other thing is to put a load on the spindle and see if the bearing are moving.
Now here's the deal. Most engine lathes are rated at .0007 TIR spindle run out or less. most tool room lathes are rated at .0001 TIR. At the spindle nose. Then there is the difference between id an od and at 12" out with a test bar. Parallelism with the carriage movement both top and side, same for the compound movement. Cam action of the spindle. where the spindle axis changes due to the forces applied to it by the running gears. Which is why belt drive machine are more accurate in spindle run out then geared head machines. You have removed one influencing force from the spindle.
That said, that is if your indicator can read .0001" TIR as most won't. Pull out the cert on accuracy and repeatability. it will shock you.
Thing is with a four jaw chuck your spindle could be out .005" and you could still dial it in to near perfect concentricity.
Thanks.
WAY beyond my ken to measure all that stuff. When I mention measuring runout in the future I'll be more specific that I'm just talking a no-load measurement right at the spindle with a .0001" indicator. I am pretty careful about it; I'll measure different parts of the nose taper, plunge or retract the indicator more to make sure I'm not in a funky spot in it's travel, and so on. But your points are very good ones.
Metrology is fascinating stuff. It'd be fun to take some classes. I saw there's a scraping/metrology class happening in Seattle soon.
Practically speaking, I get myself tied in knots trying to think it through, but, with a 4-jaw dialing out spindle runout it seems that you'd only be able to get it true to one spot in Z. For any machining operation requiring a cut parallel to the spindle (like chambering) as you moved away from the spot you'd dialed in, your error increases. Which would be less and less true, the less runout there was to begin with, so, less runout is ALWAYS better. No?
