I would like to better understand how FDISK in DOS 95 and WINDOWS 95,
version 4.00.1111, handles the partition tables, boot records, FATS,
and data area during partitioning.

My computer came with 2 hard disks both configured as "master" the
1-st with an active primary partition (C) and 5 logical drives
(D,E,F,G,H). The 2-nd disk had no primary partition but an extended
partition with 4 logical drives (I,J,K and L)

Things got messed up and I had to "FDISK" and reinstall windows 95.
Before doing so I saved the boot records of all the logical disks
since FDISK would erase them. After FDISK-ing, keeping all sizes of
the partition (or logical drives) the same, I used interrupt 13 of DOS
and "DISKEDIT" of Norton Utilities to reinstall the boot records
(overwriting them over the F6-filled sectors FDISK had put in their
place.) Most of the "FAT"s were not erased since FDISK did not erase
them and I did not have to FORMAT, but some sectors in the "FAT"s were
overwritten by hexadecimal "F6"s.

Almost everything seemed ok except that partition D was now the first
partition of disk 2 and the letters of all other partition (or logical
drives) had been changed. Since this messed up my batch files,
therefore when I later found out that if disk 2 has no primary
partition, only an extended partition - then disk D would again be the
2-nd partition of physical disk 1 - therefore  I repeated the
"FDISK"ing, saving and reinstalling the boot records several times
trying different arrangements.

I do not recall exactly all I did or when I discovered more errors as
follows:

1. One or more of the above procedures had caused the 1-st partition
of disk 2 to start at cylinder 1 instead of cylinder 0, but cylinder 0
also had copies of the partition table and boot records. All other
logical drives were shifted forward by one cylinder.

2. CHKDSK reported lost clusters. This was traced to F6-filled sectors
in the "FAT"s. By substituting the unaffected portions of FAT-2 and
portions of Norton's IMAGE file into Fat-1 it was possible to
reconstruct FAT-1 to the point where no obvious errors were apparent
but some odd behavior remained.

3. The 1st sectors of many sides of cylinder 1 of disk-2 were
overwritten with "F6"s. Many other 512-byte-portions were also
overwritten with "F6"s

My questions are:

1. If FDISK moves the partition table, boot records and "FAT"s from
cylinder 0 to cylinder 1 how can the "FAT" find the files which are
now one cylinder closer? Or does FDISK shift the whole disk in a
manner similar to the software program Partition Magic?

2. If certain portions of the data area are overwritten by "F6"s how
come I have not yet found a program that is corrupt? How come windows
does not crash?

3. Which portions of the disks does FDISK overwrite with "F6"s?

4. Do FDISK versions earlier than 95 also overwrite disk portions in
the data area with "F6"s? Or do they limit their overwriting to the
locations where the boot records were?