Gary,
Since you have been so patient and informative, I will expose more of my ignorance on this matter.
First, it is fascinating.
Second, with tolerances such as this, are the metals responsive enough such that one could hold his/her hand on the object for a period (transferring body heat) and skew the reading/s?
Wally
your a lot smarter than you think you are! That is a well known problem. We use metals that are well known to be stable (grow or shrink very evenly). Now we are using carbide gauge blocks a lot simply because it dosn't change size much. Plus we often will setup in a temperature controlled area that has no drafts (air). The floors are usually 12" or more thick. You bring a part into the room, and it will set there at least 24 hours before being checked. Everything in there acclimates. I was in a shop once that had a slab of granit setup that was three feet thick. Above it was a gantry type piece of equipment that measured parts. They were mostly checking spindles for large boring complexes like LAMB. It was fifty millionth capable, and big enough to put two half ton trucks ontop of! I ask the guys how the lapped in that plate after it was set, and they said they really didn't know as the folks that built the measuring machine did it over a three week period. Told my boss that if he ever bought a slab of granit that big I'd kill him! The deal is that if the spindle rotated it automaticly doubles the error built into it. So if it has .0002" TIR; the best it can do is .0004". But when you add in the error in the housing and bearing packs your probably looking at .00045" before you ever rotate anything. Now the housing dosn't move much (it will flex a little), so your probably looking at .00065" TIR. Not good enough for many super precision ball bearing sets. That's why we often split one ten thousandth of an inch. But if you ever hear the term arc second, you know the folks are serious about their measurments. Surface plates are not really measured in inch or millimeters as I spoke of. I used those numbers because most everybody knows what they are. Surface plates are measured in tenths of an arc second. (a tenth of an arc second is 4.8 millonths of an inch). Sounds crazy, but it really fairly easy to measure. You do it with an autocollumator. Lasers are not an accepted method even though some shops use them. The laser beam will bend! 99.99% of the time a laser is good enough for any measurment. An autocollumator works exactly the sameway, but uses a regular light beam (much like a flashlight) shinning into a calibrated mirror. The source measures the deflection of light from the mirror as it returns. This is extremely accurate, and readilly reads down to a little less than a tenth of an arc second (the highth of a line drawn ontop a surface with a lead pencil is almost always right at .000020" by the way). Most folks think that granit plates are more stable than cast iron plates, but that's just not true. They are similar. Both will bend and flex all over the place if not used correctly. But the granit plate is far easier to maintain, and in the long run comes out much cheaper. But the problem with granit is plate quality. Most folks you see are buying the solid black ones. And most of these are known as composit plates. This simply means that fine pieces of granit are epoxied together to form a large slab. The plates are very stable, but also softer. As a rule they are not for use in a production area, but for a gauge room. You will also see grey ones and pink ones. These are much harder than the black ones, and can sometimes be a bear to lap back in. If you see white lines running thru a granit plate, you know these are fault lines. They are much harder than the rest of the plate, and with a large amount of weight place between the line and the outside of the plate it will flex and often take a set (30lb. will do this in the right place). When the plate is setup in the area it's being used you have to let it acclimate for a few days before checking it; if it's much bigger than 6' x 9'. Then you level it within .001" per foot. The actual plate sets on three points, so it's already on a perfect plain.
Cast iron plates are usually made from menonite. This is a form of cast iron that is extremely stable, and dosn't change size of shape much. The older the plate is the better it is (cast iron continouisly ages). I think the biggest cast plate I've scraped is a 4" x 6', and most are 18" x 24" ones. I used to keep old cast iron plates in racks stored outside to let them age. When I had a call for a plate I'd just go get one and have it plained flat; then of course scrape it in. It sounds bad, but it's a lot better than not having a job!
I once came into an inherited project of rebuilding a specially built grinder that plunge ground an irregular curve on a piston cup used in a tank transmission. The shape of the curve was ridicullas at best, and the accracey was flatly stupid! Nothing was square (everything came in at an angle). You were in fear of moving the machine to a better place to work on it, and always worried that somebody was going to bump into it. I ended up sending the slided out to be ground by a specialized outfit that had a grinder capable of grinding within .0003" in 23 feet!!! Then I mogliced everything else to the slides. Then I had them grind me two masters that were about six inches longer than the base of the machine, and Mogliced the base to the two masters. The wheel head wasn't all that bad, but was an oddball. I sent it back to the factory, and let them rebuild it while I did the rest. The grinder cut better parts than when it was new. There are lots of ways to cheat, but you just gotta know a few of them
gary
