Thanks for the picture, Mitchejc - that's exactly what I was meaning. X1 and X2 may be "level" in the sense of horizontal to the limits you can achieve with self-levelling epoxy, but they do not have to be in the same plane, just parallel. It's the same error and corrected the same way as not having the feet of the gantry exactly co-planar. It's not ideal, but once set up, your friends will never know the difference...

I have to agree that epoxy is a second-best. If we could use self-hardening water, I might believe that the surface would be ideal, but even low-viscosity epoxy doesn't run that well, and I suspect that you also get surface imperfections because of internal stresses caused by shrinkage. If I had a surface plate 1.8x1m, I would consider scraping (proper engineers' scrapers and everything, I've used them before) the top of the rails to it. In my case, I was looking to take out a dip in the 100x50 box sections of something around 1.5mm. I don't know why it was that bad, whether the box section was warped when I bought it or if it is welding distortion or what, but it would need quite a lot of filler and/or filing to get that flat to acceptable limits, and it just ain't that easy to do. As pointed out, even if you could use a straight edge to get the top level along its length, measuring twist is really difficult even though the profile rail manufacturer's specs are pretty tight and you need to get it right (or at least consistent along the length of rail).