I hope you are accepting my comments in the spirit they are offered. Having worked as a toolmaker for years, I'm well aware of the tiny things that control the outcome of tool modifications. Once you move to dimensions measured in tenths, it becomes rather difficult to keep control, in particular when you're trying to generate surfaces that really beg to be ground. Your results are impressive, and I concur----well done.John Hill wrote:I will test the perpendicularity when I find, or make, a suitable bar but until then I will continue to fret and chew my finger nails!
If you make the decision to make a test bar. make it between centers. Because you're looking for miniscule differences, it must be quite precise. Turning by chucking, with the use of a center, isn't in keeping with good practice and can yield a less than straight and round bar. A ground bar would be preferred, but don't trust any *ground* stock, which often has been ground in a centerless sander and may or may not be round and straight (drill rod comes to mind).
The bar you make can have slight taper ( a few tenths, max), for what you're looking for is just differences in readings (TIR). If you can duplicate the .0003" reading a distance away from the collet, perpendicularity was not lost, and you have a simple concentricity issue. It is then a judgment call, if you feel it is close enough for your use, or not. Less would be desirable, but it is certainly functional as is.
If the reading is different, before concluding that perpendicularity was lost, remove and replace the bar to see if the readings are duplicated. If they repeat, you can trust the results. If not, explore to see why they do not. Remember, what you're looking for is quite small---so don't discount the slightest thing.