So after a second look it's apparent that the EPROM pinout for the '16 and '32 are roughly compatible, the active low select signal from pin 5 of Z22 joins pins 20 & 21, resulting in A11 being driven low along with the !CE pin. After fitting the missing pull-up R15 the machine boots with the EPROM in place woohoo!
The EPROM contains a tiny routine which preserves registers then fills the screen with the letter A. Registers are restored and a return executed.
I tried executing the code:
but the machine hung. PEEKing address 12288 gives 255. As does peeking the remaining 2048 addresses
I removed the chip and once again verified it in a programmer. I also replaced the chip with a different brand - a Fujitsu this time - with the same symptoms.
I've looked at the chip select for the EPROM and it pulses as expected if I execute a BASIC loop to PEEK from it, so I think that part of the equation is sound. I'll stick the tickler on there this afternoon and see what it reveals.