In message <199603011602.LAA26027@Collatz.McRCIM.McGill.EDU> you write: 
> Indeed, this raises the unpleasant possibility that some disk maker may
> have decided to _not_ make cylinder number the most-significant part of
> the absolute sector number, in which case _all_ the assumptions about
> disk layout go right out the window.

MS-DOS's Norton Utilities uses this ordering (head:cylinder:sector) when
discussing disk geometry.

If anyone used this kind of ordering for sector values, adjacent cylinders
would not be adjacent linear sector numbers.  This would make it difficult
to share the disks sanely with other OSs that are cylinder-centric. Mfr's
which do this should be shot in the firmware :)

Since many IDE drive/controller/bioses now use LBA for addressing, and
reported geometries on (e.g. IDE drives) are often translated, we probably
can take this opportunity to cut a lot of cruft, speed some things up,
and minimize the area which cares about such details.

This is not to mention controllers which already do reordering, drives
which do bad sector forwarding, etc.