Lens AF calibration

This is really about lenses, but since most of the autofocus threads are here, I post here anyway.

I learned something new today, something I've never heard about before.
If a Canon lens is calibrated together with an EOS 1Ds Mark III, for example, and then work perfectly together with that camera (focus is as good as it can be), that's not at all necessarily any guarantee that the same lens will perform just as well on a 5D Mark II, to give another example!
Lenses are calibrated to work with a certain group of camera bodies, where the 1D-series is one group, but cameras like the 5D Mark II or 7D belong to another. Most Canon lenses can store information related to two different camera body groups. The newest lenses can handle three groups.
Since this information comes from a service technician, who does such calibrations, at an authorized Canon service center, I do believe it's correct.
As far as I understood this, you have to calibrate again, if you get another camera, belonging to a different group. I don't have any exhaustive list of the groups, unfortunately.

I also believe that had this been common knowledge (I've never heard about it before), 25% of all posts in all photo related forums could have been avoided.

But any test camera, regardless of model, should be calibrated to Zero - meaning perfect. Which should result in all calibrated lenses working within spec on any other camera that is properly calibrated. I am not sure the specific model has anything to do with it.

As far as bodies out in the real world, they may be slightly off zero but still within spec. The same goes for lenses. If you put a lens and a body together that are both a little off in the wrong direction, then you get noticeable problems.

For example, Say the tolerance for AF is +/- 5. You have a lens that is a off at -4 - so not great but still within tolerance. You then put it on a body that is slightly off the other direction at +4. Well add the two together and you have a smokin' setup because you have perfect calibration due to both offsetting each other. Now you sell your -4 lens to someone with a body that is off in the opposite direction as yours. Theirs is off by -3. They put the -4 lens on the -3 body, and end up being -7 off of perfect. That puts it out of the tolerance window and they would think your "sharp copy" of the lens was really soft.

Thanks Anders for putting this info here on POTN (I was completely occupied in the Swedish forum discussion)

Actually there are two separate parts. First the lens.
The thing is that, according to the Canon tech guy, most Canon lenses has the possibility to store AF adjustment in two separate registers (the newest lenses even have a 3 register capability).
The 1D series uses one of the registers and the other models uses the other register.
For the AF to work properly we have to assume that the body itself has been adjusted to +/- 0
We are going to dig further into this info since I think it is very vital to understand some of the AF problems that are known today.

Then the camera body.
I also heard about a new technique (equipment?) being used by Canon tech centers that will make it possible to not only adjust the AF focal plane but also groups of AF points vertical and horizontal.
This will of course give much better possibility to achieve an exact AF. This information was regarding the 1DMkIV. I have no idea, yet, if this also can/will be used on other models.