"Julian Scarfe" wrote:

How can you possibly suggest that it would be more "practical" to
use kPa?

Because one digit less is needed, and conversions are easier when
powers of 1,000 are used as normally.

There is an installed base of tens of thousands of
altimeters in aircraft out there that are calibrated in mbar.

This is not about calibration, this is about expressing physical
quantities. Besides, if the installed base of equipment were decisive,
each of us would still use one's country's inch, pound, or whatever
local measures were in use long ago.

Describing them as hPa makes it clear what the unit is for someone
familiar with the SI, without risking accidents through unit
confusion.

Would it be clearer to use a non-recommended prefix than a recommended
prefix? Besides, your argument indicates a fundamental confusion. There
is only one SI unit of pressure, the pascal (Pa). That's part of the
beauty and practicality of the system. All the rest that is used to
express pressures relates just the way of expressing the numerical
value. For convenience, we can use multiplier prefixes of _the_ unit if
we like, or a multiplier of the number, consisting of a power of ten.

The preference to use powers of 1000 is just a preference because
practicality and pragmatism is sometimes more important than an
arbitrary recommendation. This is a perfect example of where
pragmatism should (and does) win.

The reason for preferring powers of 1,000, explicitly expressed in
several recommendations and standards, is its practicality, based on
the use of the system as a whole. If you take arbitrary special
aspects, you can always find arguments in favor of using non-SI units
or non-recommended SI expressions - but then you lose all the benefits
of a unified system. Using hPA is a half-hearted "solution" that
combines the trouble of transition (after all, it needs to be
introduced to people who didn't know it, and they need to be reminded,
and some people will inevitably misunderstand or forget) with the
effect of gaining almost nothing. (We _can_ convert millibars to
pascals too.)

--
Yucca, http://www.cs.tut.fi/~jkorpela/