Joerg Knappen:

> You may call it general principle. The following conventions are often
> employed by physicists:

> Variables:         math italics
> Vectors:           bold math italics
> Operators:         upright
> Vector Operators:  bold upright
> Tensors:           sans serif

Hmm, I was not aware of the distinction between vectors and vector
operators using bold italics or bold upright.  I can't remember any
such a specification in the IUPAP document [1], although I would
agree that it might serve a useful purpose if employed consistently.

[1] International Union of Pure and Applied Physics, S.U.N. Commission:
    Symbols, Units and Nomenclature in Physics, Document U.I.P.~20 (1978),
    Physica~A, Vol. 93 (1978), S. 1--60.

As for the actual specifications, I'd like to point out that tensors
should acctually appear in bold sans oblique, not just sans serif.
In practice, however, the conventions used by several publishers may
differ form the IUPAP recommendations to greater of lesser extent.
Setting vectors in bold upright (instead of bold math italics) and
tensors in bold sans upright (instead of bold sans oblique) is seen
quite often, and therefore allowing for some choice between different
conventions while using the same functional markup in all cases would
seem the most practial solution for document portability.

> Constants are usually also in math italics (generally for physical
> constants like speed of light $c$), only \emph{numbers} like e, i, and pi
> occur upright depending on the publishers style. Differentials are also
> often upright.

I'd rather consider this a distinction between physical constants
(which aren't treated any different from physical variables) and
mathematical constants (which indeed should be set in upright).

> If you want to see a journal which follows all those conventions, look at
> Il nouvo cimento.

There are probably a lot more physics journals that try to follow
the IUPAP conventions as far as possible.  At least, I've seen a
couple of style guides (including e.g. Elsevier and IOP) that do
suggest the use of upright d, e, i, etc.

Cheers, Ulrik.