Time for an update. I'm posting it here because I prefer to keep my own lathe alignment discussion and activity in one place. But ya, lots of discussion about alignment tools and methods elsewhere.
My test bar arrived - yes, very very late. Almost so late as to be irrelevant.
I was going to start by qualifying the bar but I was simply too impatient. In hindsight, that was a mistake.
Instead, I began by checking the level of my lathe bed. It is still perfect in the horizontal plane - no detectable twisting. My level is an adjustable unit with a claimed rating of 0.0001 over 10". Reversing the level across the length of the bed resulted in the same reading at all points - slightly tipped toward the operator approximately 4 tenths.
I don't yet know if my bed is bowed at all. Wear is unlikely as the ways are induction hardened, have been kept well oiled, and show no visible sign of wear or even slight polishing anywhere. In my opinion, it is still in virtually new condition. But regardless, I did not check it for bowing (arching) because my first interest was in a rough evaluation of the test bar.
Next, I chucked a regular steel 1.25" bar and made a 6 inch dumbbell. I would have preferred 10 inches, but 6 is what I had and I just wanted a quick and dirty test to see where the lathe is. I used a shear tool to do the final cuts on the collar ends to minimize cutting loads. (yes, I know there is still a load that must be accounted for, but again - I just, wanted a quick check.
This crude two collar test showed less than a tenth variance on both the vertical and horizontal axis as well as on diameter. I conclude that my spindle is reasonably aligned with my bed for at least the first 6 inches of the bed - maybe a bit more because the saddle sits to the right of the indicator mounting.
Next, I removed the bar and my chuck. Then I checked the runout on the inside of the MT5 taper on my spindle. There was zero movement of the tenth indicator anyplace on the taper from one end to the other. When I say zero, I mean no movement of the needle at all - it was rock solid and smooth. I've never checked it before because I've never used it. So I confess that I was very pleasantly surprised at that result. I do love my lathe.
Next, I cleaned both mating tapers, installed the test bar, and gave it a few taps with a deadblow to seat it. I did not check the fit. In hind sight, that was another bad decision.
Next, I mounted the indicator to measure bar runout. You can only begin to imagine my shock at seeing about 3 thou runout at the end of the bar.
So I removed the bar and reseated it. Same result. Next, I marked the spindle and the bar, removed it, and rotated it 90 degrees relative to the spindle. The runout moved with the bar. I repeated this several times with the same result.
I stopped all further testing at that point for the day and went to bed with a distinct bad taste in my mouth.
In the morning I decided to qualify the bar on my surface plate. (I should have done this right up front - dummy me. ) I indexed the horizontal position with a center in both ends of the bar to stop any lateral movement, and put the bar into V-blocks. To my surprise the bar was pretty concentric (plus minus a half tenth) across the length of the test section and also pretty good across the majority of the taper. I say pretty good because there is a distinct high point (just shy of a half thou) on the taper about 3/4 of the way to the small end of the taper. Using Dykem, that same high point seems to dominate the contact with the spindle taper. (yes, I know Dykem isn't the best thing to use. But I was out of Prussian Blue.
FWIW, that high point shows no sign of damage or cause. It's just a random high point on the taper. I find this very odd. The bar is clearly ground and it's hard to see how that could have happened. Perhaps there is an inclusion under the surface that didn't cope well with the hardening process or that decided to swell after manufacturing. There is no sign of a ding or other impact related cause.
My preliminary conclusion is that the bar is useless for its intended purpose as is.
I have thought long and hard about trying to fix it. It's probably not that hard to do. Some stoning at the high spot might do the trick. The bar is not designed to take a load so there is no need for high contact coverage - just location. That's my opinion anyway.
Right now, I can only say that I am now doubly convinced that bars like this cannot be trusted for testing lathe concentricity without first testing the bar itself for concentricity of the test bar and Morse taper. I have yet to decide if it was ever even worthy of consideration for that purpose. The two collar bar is pretty darn simple and is not susceptible to errors of this nature.
In my mind, the final word is still a two collar test for a lot of reasons. But more to the point, given that it's the final word why not just go straight there and skip right over other methods that may or may not have misleading flaws.
After I stone the bar, I'll post an update here.
I'll also put it away for a year or so and see what happens to it over that time.