Backing up a step, I'm not really sure how the ROMs should work in my machine. I can't fault Ansgar's disk image if I can't access my disk due to improper ROM function.
I have two machines to work with - a 9825A and a 9825T (by the board set, but it seems to act only as a B at this point, ignoring the top board). Both machines power up to the 'lazy T' prompt and seem to operate normally within the 'regular' realm. If I type 'list' and press EXECUTE, the A shows 23228 bytes (correct for Option 002 as listed inside the printer cover) and the T shows 31294 (as opposed to the label which claims 61670 bytes; the label lists the chassis number which matches that on the bottom of the machine).
Should the 'list' result change when ROMs are inserted? It doesn't on either of my machines. The manuals for the String/AdvProgramming ROM (98210A) claim that the two functions combined take 56 bytes of RWM (read-write memory). Entering dimA$ on either machine gives error 22 (bad dimension), with or without the ROM inserted. I've tried several different ROMs (Matrix, 98211A, and Disk, 98217A) and all the slots on both machines without any difference. On the A, I can pull the Operating System ROM carrier from the base of the machine - this prevents the machine from booting but at least verifies that the ROM connector board is electrically attached to the rest of the system. On the T, I've eyeballed the cable but haven't checked in any other way.