SuttaCentral
Hi,
As I understand it, we have several very old fragments of the Pali Canon, and the oldest complete Pali Canon we have is from the modern period.
I am writing this topic to find out what are the oldest complete versions of the Pali Canon that we have preserved, as well as the oldest fragments that we have preserved, and I would like to know how scientists have been able to cross-reference these documents to find out what the original Pali Canon was about.
Thank you in advance.
It would be interesting to see what passages have textual variants.
Going back to the 14th or 15th century for certain books, and a few isolated fragments older than that.
The oldest complete book is the Colombo Cullavagga (14th century).
Yes, the oldest identified (mostly) complete set of manuscripts in Sri Lanka dates from the 18th century. But this has not yet been studied.
The versions of the canon that we use today were created in the 19th and 20th century on the basis of manuscripts that are from the 18th and 19th centuries.
I don’t think we’d call them scientists but editors. Typically modern editors will try to get a representative cross-section of manuscripts from different countries. The idea is that since the transmission in these countries is to some degree separated, comparing them will show how much textual drift has occurred. Editors will then reconstitute what seems to them to be the best text while noting variants in footnotes.
The text is further checked against the traditional commentaries, which were compiled in the 5th century and which provide an external reference that can establish many passages.
Again, checks against non-Pali sources can be made, althoughh of course the differences are much greater there. See for example my recent post for an example of this. At nearly 2000 years old, the oldest physical non-Pali manuscripts are in Gandhari, some of which are close parallels with the Pali.
All the evidence points to a transmission with a high degree of fidelity to the text. We can, however, open up further ways of investigating such things, for example by digitizing exact replicas of the oldest manuscripts and comparing them to our modern editions. Closer study, perhaps aided by machine learning techniques, can also help to clarify various issues. It is also possible that further acheological finds will expand our understanding.
Thank you very much Bhante for your clarity.
I realize that it would probably be a ridiculous amount of work, but a future enhancement to the site that I think would be worthwhile would be to indicate what passages have textual variants, what the alternatives are, and the rationale for the preferred choice. That will keep you busy for many lifetimes 
Already have:
These are a partial set of variants as pulled from the Mahasangiti edition.
The only rationale that matters is “because the editor thought so”.
Ask and ye shall receive. Fantastic!