There has been some talk from time to time about the generation of surfaces and angles by the three piece method(s). I wanted a right angle reference, so I decided to do it.
It was somewhat easier than I expected, but took a while. I did 'cheat" a bit, because while I used Connoley's "one over one" method, I interspersed spotting and scraping with a granite flat as a reference.
That kept things flat, and meant the angle was the target when comparing between the pieces, and flat was the target when spotting on the flat. it also avoids the strange cases that some folks, including Evan, brought up from time to time as a pitfall of the process.
I am by no means sure that the practical process of generation would even allow those oddities, but no matter, the granite flat avoids them.
I also cheated in that I wanted ONE right angle..... so I sliced an angle plate into three pieces, and used the unreinforced parts as "reference tools", being the "other two" of the three. I really don't care if they warp later, they can be used again if I want another angle piece.
I kept the piece with the reinforcement as the "real angle".
I'd show the final spottings, but I find that the marks don't show up well in pictures unless the "blue" is very heavily applied, which is just misleading.
Here they are:
The main angle at an earlier stage, with spottings against both another angle and the plate (for bottom) rather heavy blue.... at this stage I believe there were about 25 actual spots per square inch, but distribution was not great, I had probably gotten a bit too aggressive in one or two areas: