I've only ever seen pictures of two 1392464 keyboards. They did have different F keys. I just don't know if a sample of two statistically means anything. Could there have been other variations? I don't know. Changing the caps would also explain it. Might do it just to make it look like a normal keyboard.
Look at the two real 1392464's. For keycaps which have shifted and non-shifted legends on the same key, they all have the non-shifted legend printed high up on the keycap compared to normal M keycaps. It's all very consistent. Now look at the board on eBay. It has the same high non-shifted legend on the following keys:
`
1
2
3
4
5
0
[
]
\
;
'
,
.
It however has standard M keycaps (with the lower non-shifted legends) on the following keys:
6
7
8
9
-
=
/
I simply cannot believe that IBM would go to the trouble of changing the legends for some of the keys, but not for others.
Look at the legends on the keycaps that weren't swapped. None of them are DisplayWrite specific legends. The person swapping the keycaps likely failed to realize the difference in how non-shifted legends were positioned differently and so just swapped some of the keycaps and not all of them. Either that or they were lazy and just didn't bother.
Statistically speaking, two samples isn't much, but in this case I think it's sufficient to make it very obvious that the all the keycaps on this board are not original. The ones I mentioned with the high non-shifted legend obviously are original, and in addition it looks like the backspace, and the stepped keys are. Beyond that it looks like all other keycaps were swapped, with the possible exception of keys which are indistinguishable between a 1392464 and a standard M, such as alphabetic characters without any secondary legends, the space bar, and a few other keys.
In my opinion, it's safe to say with greater than 99% accuracy that this is most definitely a hodge-podge board.
