Thanks Daryl and Bill, I have ..
More than enough to be certain, thanks. That takes a significant worry off the table.
It looks like there is plenty of material to cut 1/2 a thousandth off the taper and 10 thousandths or so or the flat. Obviously I need to get my bearings fixed or replaced before attempting to modify the spindle. I am going to assume the inside taper is not modified, of course I will confirm that, so I can use that to rate the bearings.
It will be damaged. it has "features". Even when de-burred, the pits will prevent easy use of an indicator. Not much help directly read, sorry. More follows as to what worked for me.
I need to read up on spindle removal. I have already had the long nose setscrew out of the headstock, so I know I can get it out.
This may be an open ended question, but after I pull the spindle, if the bearings are not obviously bad, had does one determine if they are good? Since I have the 0.0004 runout we can probably conclude that something is not proper as is.
Helps to reed Moore on foundations of accuracy, Connelly on rebuilding, all you can find on PM.
My add is that conventional mag-base holders for indicators are not good enough once looking for millionths. A stiffened rig is helpful. Bipod, if not tripod.
D1-3 having a few dings that bumped my better indicators about - just as the spindle bore taper had done? Once the spindle taper was cleaned up, my first go at better measurements used the protruding ground surface of a virgin-new Stark #12 Jarno dead-centre.
I'm an Accountant and Statistician, too, so I built a list of
many readings.
I had what I called a "D" shape to the array.
A deviation that did not clock with the spindle, but moved, and in a regular precession.
Family circumstances intervened. Time passed. When next I went back to it, I had one of Brian Miller's nice test bars. Cylinder, now, not the flank of a cone.
Same effect, better stability, and I was convinced I had detected a failing rolling element in the bearing. Total deviation is 130 Millionths, full rotation. And it moves.
Now - BECAUSE the fault DOES precess? It would not show anywhere near as badly on a turned workpiece had the indicator tip been a cutting tool.
Think that through. Effect would be similar to a "moving" spring pass, rotating around the circumference, smoothing out the OD.
And .. it means the lathe is still more usable than not.
Given how rolling-element bearings fail - one or more elements before the rest, not with perfectly distributed wear?
I suspect many among us have been "living with" this sort of thing to one level or another, and for long years.
Bearings costing as they do? The "future" type of work and amount of it in store for our 10EE? Face it.. going mano-a-mano with a modern CNC spindle is probably a right silly thought.
For most of us, it might not be a
bad idea to keep right on so doing. "JF living with that."
The house needs a new roof more than I need a 50-millionths spindle, anyway.
YMMV