Sorry I have been trying to get my lathe level all morning, what?
I'm not sure exactly where you are, so let me summarize and I apologize if I assume something incorrectly:
Getting the lathe to be level is a means to an end, and that end is that the tailstock and the spindle share a centerline and both are parallel to the carriage motion along the bed. The result is that turning a cylinder will create a cylinder with the same diameter on both ends--that is--without taper.
Twisting the bed along its length will rotate the tailstock center laterally with respect to the spindle center, which is the same as offsetting the tailstock. This is a bigger effect than the twist has directly on the carriage. We know its happening when we perform a two-collar test and get different diameters at both ends of a cylinder.
We can achieve a flat bed using a level, but the objective isn't necessarily a level bed, but rather a straight bed. If the lathe refuses to be level because of a crown on the floor, trying to make it level might be harder than just making it straight--so that the level reads the same on both ends even if not level.
My statement was really aimed at going beyond the leveling capability of the 98, for which each mark is 0.005"/foot. We can probably estimate fifths between each mark, or maybe quarters, so when comparing both ends of the bed, we can depend on an accuracy probably close to a thou per foot or maybe a smidge worse. If we are turning a 3-foot cylinder--the longest my lathe can turn between centers--that could mean that the bed is twisted by maybe 0.003 over that length (if the read error is 0.0015 wrong in one direction at the headstock and 0.0015 wrong in the opposite direction at the tailstock). That's rather precise!
At that point, the next step, it seems to me, is to perform a two-collar test and do the final adjusting to eliminate taper directly rather than trying to do better with levels, because now you are directly measuring the error you are trying to eliminate by leveling. If the tail end of the two-collar workpiece is thinner than the head end, the tailstock is pushed back--drop the front tail leg a tiny bit. If it's fat, go the other way. Of course, before doing that, the tailstock has to be aligned laterally to the headstock when it's up at the headstock (versus at the tail of the bed), and you can do that by trapping a razor blade between centers--when the are aligned, the razor blade will be held between them at right angles to the centerline.
The 199 is extremely finicky and requires a lot of settle time. Until the bed is pretty close to level with a 98, the 199 will move the bubble so far off-center that you won't be able to see it. Breathing on it will make the bubble move. That's what I meant by driving a person crazy.
Rick "this is a lot harder when there is bed wear, like I have" Denney
