Re: Blade grind - flat vs cambered
ThomD
>"I believe it is the opposite--that a few thou is crucial. My test is what I said above--the join line should literally disappear when the boards are brought together with little to no pressure applied."
I would agree with that, as I sorta said in the part that followed the quote. I probably shouldn't have said a couple of thou, that is a pretty thick increment. I hope I didn't already tell this story, because telling twice kinda undermines my point... But I recently had to turn a < 3/8" stub on the metal lathe. I just eyeballed it and it came out .371" I was chuffed, but not totally stunned. I think we can feel and see a lot more than we let on. So yeah, 2 thou is a lot, five times the thickness of my thinnest shavings. (Or maybe not, see below)
My point was more theoretical, in the good sense. If you are talking about the subtle correction you are referring to. Something you don't appear to be overtly controlling, just suggesting it may be there, then it sounds like a difference that is in the rub. Very small. At some point the difference one might identify will fall within whatever the invisibility increment is. You are talking about something that you think is happening, but don't know is happening affecting stuff you can easily see, like ugly glue lines.
I do see what you are getting at, but it sounds like this random correction could just as easily work against as for. So we have something that is maybe in the tens (you reject 2 thou, probably with reason), essentially unmeasurable without a very nice toolroom level of measurement, and it is supposed to be compensating for errors that are easily in the thous or above - angular error. The reason this sounds attractive is that you believe there is an otherwise unexplainable result - really nice gluelines. So let's look at that.
I can think of two examples where results are way out of line with inputs:
1) Dovetails. I do ok when I am in practice. I can wow an audience. A Cosman wow for people who have never seen Cosman. :) The reality is that how I finish the parts can make a big difference to the presence of gaps. Usually the baseline of my cuts is good, but I can make it bad simply by how I plane of file the ends. I can also make it good. I am talking about post glue-up finishing, not filling in or patching holes. I don't think this is what happens when joining an edge because I do that totally differently, but I do think it shows that accuracy in parts is not the whole story. How we use glue and clamps can make a difference. That may tell us that more is happening than we at first see. But it is happening after all the joints are cut.
2) Soundboards. This is demanding edge joint work, and it requires accurate surfaces, but that is not all. The joint has to totally disappear, but it also has to be as strong as possible. One technique that works is to place the two pieces side by side, and just rub them till they take. At first they don't seem that tight, and aren't, but the glue pulls them together as they dry and the joint is ultimately invisible. So invisible that one better be careful to mark the joint because even the ends aren't always enough to show the position of the joint.
So who knows. There may be a variety of things that knit the edges together.
Note: I drew it out and a 2 thou eror on the edge of a pair of one inch boards 12 inches wide would cause a "cup" of .02" in the glued up pieces, which actually isn't all that bad. Another factor to consider.