FAT16 question

[Bas]

I'm still slowly plodding away at a disk image generator and now at the point where it spits out FAT16 formatted partitions that I could start writing files to.

My question, as I use MS-DOS 6.22 for reference. I generate a _50MB disk and partition that. I format it from DOS using FORMAT /S to get a minimal bootable reference disk.

The FAT is weird though. Clusters 0 and 1 are reserved as expected but I expect cluster 2 to be the start of IO.SYS. it's not, though. Instead IO.SYS starts at cluster 0x44.

What causes this gap? And how is it calculated? I can't find it in any docs.

[Bas]

A theory: clusters are counted from the start of a volume. A FAT volume contains a BPB/VBR, usually 2 copies of the allocation table and the root directory. These structures are large-ish. After all these structures, the first available data cluster is quite a way into the volume, or the metadata structures would get nuked, so 0x44 doesn't sound all that odd anymore.

I have yet to check this.. but hoping it rings a bell for anyone. Are clusters actually counted from the start of the volume, or do they start at the first sector of the data area?? Docs are still inconclusive.