cr7 -- thanks; on my screen I didn't scroll down to that. So here's what my copy (from BS-EN 1993 version) of the 2768-2 standard says:
5.2.5 Coaxiality
"General tolerances on coaxiality are not specified.
NOTE: The deviation in coaxiality may, in an extreme case, be as great as the tolerance value for circular radial run-out given in Table 4, since the deviation in radial run-out comprises the deviation in coaxiality and the deviation in circularity."
and Table 4 tells us that, for your tolerance class K, that the radial runout can be up to 0.2.
But then there's another hiccup to this. Paragraph 5.2.6 says that, for this type of unspecified runout, "if the features are of equal nominal length, either may be taken as the datum."
Both of these cylinders have a length (depth) of 5 mm, so there could be two readings.
I agree that it's too shaky of a question. But let's see what other folks have to say.
John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems