> Aha, that's what I thought because 63 sounded extremely familiar.

Indeed. Windows does the same too. It avoids the first 63 tracks,
and the last one. The first sector of the first track receives the
boot block and the first track is reserved for the second stage
boot (the first sector loads from first track a second-stage boot
block if required).

The modern hard-disks do something else too : the number of cylinders,
heads.. is virtual. They have so much sectors that they increase
the number of heads (and other fields) to non-real values because
of BIOS legacy support.

So either the first track is really 63 sectors and you have to skip
them to avoid it, or it could now be seen as some kind of "magic" value
by hard-disk integrated electronics and designating the first track
of the disk (the first track could have much more than 63 sectors of
512 bytes, but if you specify your partition start at sector "63" the
drive might have it really start on track 2).

So perhaps you are skipping more than just 63 sectors, because of this
virtualzation. Any modern hard-disk expert around ? :o)

We should be able to check that by doing a raw access to the drive, but I
don't feel like sending such commands to any modern drive with my data on
it.

That could be interesting. FFS tries to keep some structures in the same
cylinder area to get efficiency (well, it did historically) but since
virtualization has become the norm, I don't know if anyone did study the
impact on that optimization attempt at cylinder grouping by FFS.

--
unzip ; strip ; touch ; grep ; finger ; mount ; fsck ; more ; yes ;
fsck ; umount ; sleep