What is not known. It is here, not hidden away in the other pages.
The cheat table at 0x5B02 has a routine for both, so the Game Master knows which variables to poke. But we do not have their ROMs, and the msxblue manual does not list them either.
Something can be said about RC-737 from what its routine does (0x5DF1): it divides by eight, caps the lap at two and writes the lives straight into 0xE053 without going through the universal cheat. About RC-736 (0x5DCB) only that it assembles a byte with the lap in the high nibble and the stage in the low one, and that 0x5E47 gives it special treatment — putting a 4 into 0xD312 — that no other game gets.
With the ROMs to hand this would close in an afternoon.
0x628C0E 00 18 06, glued behind the eight VDP registers. The loop at 0x6279 stops at eight and never reads them, and searching for those four bytes across all 16,384 of the cartridge turns up not one reference.
They look like two more registers from another version: the 0x0E and the 0x18 would fit as R2 and R6 of a different VRAM layout. That is a guess, and with no other versions to look at it goes no further.
We know nobody calls them — checked over all 16,384 bytes, looking for 16-bit references and relative jumps — and what they used to do. What we do not know is which build they are left over from. There are three of this cartridge, and the other two are the 1985 Japanese ones, relocated with respect to each other. Comparing them instruction by instruction would say whether those fragments were alive in either.
0x54D1 looks at three bits of 0xD317: bit 0 is set by MODIFY PLAYER NUMBER, bit 1 by MODIFY STAGE NUMBER, and bit 2 is set by neither menu. Its routine, 0x5561, exists and does something.
Either some path we have not walked turns it on, or it belongs to an earlier version.
They are identified — two Japanese ones from 1985 and the 1986 European one, which is the one disassembled here — and they are known to be the same program recompiled and relocated, not different versions. What has not been done is comparing them instruction by instruction to find out what changed between 1985 and 1986. The French translation and the amateur MSX2 memory-mapper patch both derive from the European one.
RC-755, 128 KB, and outside this disassembly. It knows more games and brings its own editor. If the first one identifies by a 256-byte sum, the obvious question is what the second does when a game it does not know is plugged in beside it.
0x4124 tries four things in a chain and only one has to pass: a cartridge header, the ld (0xFD9B),hl in the first 256 bytes, the second header at 0x4010, or the sum being in the table. A third-party cartridge that happened to install its interrupt with that same instruction would pass the second test, and the Game Master would try to cheat it using the default header at 0x6034.
What happens then has not been tried, and it is worth trying.