This sounds like something I've seen before but never fully investigated.
A 16-64K board behaves as expected when it is fitted with the second revision BIOS (10/19/81 U33=5700671).
You can remove banks of RAM, setting switches per documentation, and things work as expected.
If a chip fails, you get the expected error code.
But when the third revision BIOS (10/27/82 U33=1501476) is fitted to a 16-64K board, things change.
First, I have never been able to get the motherboard memory working unless the following is true:
1. All four banks populated, and
2. Switches 3 and 4 of SW1 set to OFF.
3. And with no RAM on an expansion card, switches 1 to 5 of SW2 are set to ON.
If you deviate from points 1 and 2 above, depending on the deviation, you then start to see "10xx 201" errors (usually followed by "PARITY CHECK 1").
Don't ask me why this is. It is certainly not in the documentation that was supplied in the BIOS upgrade kit that IBM supplied (I have a kit).
So fully populate the motherboard banks and set switches 3 and 4 of SW1 to OFF.
You should then see an accurate 201 type error per the following:
16/64K board
------------
00xx to 03xx = Bank 0 (the XX indicates which chip [or chips] have failed)
04xx to 07xx = Bank 1 (the XX indicates which chip [or chips] have failed)
08xx to 0Bxx = Bank 2 (the XX indicates which chip [or chips] have failed)
0Cxx to 0Fxx = Bank 3 (the XX indicates which chip [or chips] have failed)
XX is 00 indicates the parity chip
XX is 01 indicates the bit 0 chip
XX is 02 indicates the bit 1 chip
XX is 04 indicates the bit 2 chip
XX is 08 indicates the bit 3 chip
XX is 10 indicates the bit 4 chip
XX is 20 indicates the bit 5 chip
XX is 40 indicates the bit 6 chip
XX is 80 indicates the bit 7 chip
XX is something else indicates multiple chip failure, eg. 24 would indicate bit 5 and bit 2 failure.