Besides Murray's History of Chess (already referenced), you may also try:
The Oxford Encyclopedia of Chess Games (by Levy and O'Connell) tried to reproduce all known chess games before ... was it 1800? As it was intended for modern readers, it uses modern notation, but most scores have a source reference, which you might be able to trace -- though I suspect many go to Murray's manuscript notes. (Don't rely on its index, though.)
It is also a question of how you define 'chess' and 'game'. You might be defining it in a way that limits your inquiry already from the start. Several early manuscripts on chess problems have 'notation' but not for complete games. You can find the Bonus Socius collection in Murray, for example, (also in van der Linde) and there's at least one modern reproduction of the original manuscript, in case you want to go at that level. But you should have at least some facility for reading medieval latin or french for that.
And I'm sure there are Arabian manuscripts as well, though those are outside my sphere of interest.