[JohnBHanson]

I would check my conclusion first and then it is up to you. I have previously found the Gemini naming convention a little confusing in its correspondence to the physical format. Also cylinder mode should be slightly faster as there is less head movement. In addition the 765 controller can read both sides of the same cylinder in one command which is probably why cylinder mode is the surviving format on PCs etc.

When I did my matmos/alphatronic format I decided to allow it to boot from both 512 byte sectors and 256 byte sectors on DD/HD format. The original format was 256 byte sectors. This gives me the maximum options from the boot rom. For the HD system I built it with 512 byte sectors - with 18 sectors/track for 3.5 inch this easily mapped into its 1k sector buffer. In addition it meant the low level format was IBM PC and hence making it possible for me to create HD disks using a PC with a USB floppy drive.

If you are wanting to keep your boot roms the same you are restricted to having sector 0. This removes any possibility of being able to create disks on HD 3.5 inch drives.

My choice would be to confirm my conclusion and if correct re-publish rather than inventing a new Gemini format it will be less confusing in the future.

However looking at multi.mac just after label readrw double is in cylinder mode while
single density is in side mode. This corresponds to what you have on your rack system.

So your rack system is the correct format at the moment. Its just that this does not seem to correspond to the definition of GEMQDDS in the multiformat system - and it appears the multiformat system has it wrong!