Sword 12/13 century or not?

[werecow]

Originally Posted by Will M

I think it is unscientific to assume such swords found are automatically labelled fakes, it's like being guilty before proven innocent.

Except law is not science. We work from the presumption of innocence in law but that is not the case in science. In science, ideas are very often immediately suspect and will even be discarded without consideration because they lack evidence and plausibility and are simply not worth our time and money. They are "guilty" of being inconsistent with science and we don't bother trying to prove their innocence.

It's simply about which option is more likely a priori. Failing to take that prior plausibility into account is a problem you see a lot of in pseudoscience and fringe science. It is what distinguishes "science based" approaches from "evidence based" approaches (see here for a discussion; in this case pertaining to medicine, but the same principle applies to all fields of scientific knowledge). Basically, science has the additional requirement that the hypothesis has to be consistent with a hard won and strongly supported theoretical framework that is based on years or even centuries of rigorous work and evidence gathering. If it doesn't, then it must have exceptionally strong evidence in its favor, or else the more likely explanation is that the hypothesis is in error since it goes against an extremely robust body of established knowledge.

To come back to swords: If we know that there are many more artificially aged reproductions out there than real examples and that it is exceedingly rare for unseen specimens to pop up out of nowhere, then the default assumption should be that what we have is a reproduction, because that is statistically the most likely. That's a major reason why we want provenance dating back to before the 2000s.

To give a reductio ad absurdum example, suppose we knew of a single real example of a 12th century sword and a 999,999 fakes, then the a priori chance of a never before seen example being real is 1 in a million whereas the a priori chance of it being a fake is 999999 in a million. We could probably all agree what the default assumption should be in that case. It doesn't mean it never happens (it happened at least once, after all), but we better cross our t's and dot our i's if we want to make our case.

In real life we can of course only guess at the real numbers, but I don't think it's controversial to say that the forgeries significantly outnumber the real examples that have never been described before, and in the absence of strong evidence and well established provenance, the same principle applies.

Anyway, sorry to drone on but the science demarcation problem is my other hobby.