well, some swords actually soften over time.
http://pueblo-ops.cz/wp-content/uploads/2013/01/10.11648.j.am_.20130204.12.pdf
metal aging leads to recrystallization, grain growth, and softer metal in bronze and iron and steel. maybe a different reaction for hardened steel. But, age hardening is typically used in the literature as a synonym for precipitation hardening. This occurs over 10 or so hours at high temps. Maybe there is low temp precipitation hardening, but then it would be in competition with the aging process and relaxing of steel.
That's a good article, but unfortunately the researchers base much of it on the huge flaw of assuming that the modern reproductions are directly comparable to the ancient pieces, when to begin with, the modern alloys are going to be more pure, inclusion-free by ancient standards and formulated under much more strict controls (unless the reproduction smith smelted his own ore the old fashioned way, but I didn't see it mentioned). As we all know, tiny amount of elements or contamination can have a big affect on an alloy's properties.
Another thing is that they keep saying that the spontaneous recrystallization is "softening" the metal over time, but what they describe is more of a drop in toughness - almost like a mis-translation. I would think that grain growth is likely to make the blade more
brittle, as Kohai999 said about the 500 year-old Japanese swords.
The effects of aging/precipitation-hardening in beta titanium alloys is
much different in a billet that has been solution-treated, than in an annealed or plain forged bar, or a non-beta-metastable ti alloy. I would expect this to be true in martensitic steels as well, as you mused above. The alloys in question in the article - bronze, or iron, must be hardened mainly through work and not heat-treatment, just like alpha or alpha-beta ti. I wish the article went into detail about the differences in mobility between substitutional and interstitial alloying elements during spontaneous recrystallization. I'm sure there are different effects depending on which elements are in question, and the level of saturation, as well.
According to the article, the aging processes that happen over time at ambient temperatures are different from the forces at play during heated/accelerated aging. Is that what you mean by low-temp precipitation hardening being in competition with aging and relaxing of steel? One thing I completely agree with in the article, is that the recrystallization that happens over time, happens a lot faster than is usually assumed.
Unfortunately, I only have an anecdotal explanation for the phenomenon of recrystalization over time, from my welding instructor who was very savvy when it came to metallurgy and heat-treatment. He was teaching me and my blowhard group of scallywag students to weld cast iron, which isn't the easiest thing to do. Somewhere in the blur of lectures between shop hours, he said that old iron engine blocks that had been sitting undisturbed for decades were desirable to hot-rodders. He said that over time the cast iron undergoes a slow recrystallization that makes the blocks more
rigid and flex less, which is good for forced-induction or high-compression engines. Is it true? I don't know I like 80's cars with pop-up headlights! :]
There wasn't much information to draw upon over the course of figuring out the nuances of forging beta ti into swords and heat-treating them. What came about was a process of carefully discerning what's happening to the blade as it's processed, trusting the observations as "real" even when they can be as subtle as instinct, and then later on researching to figure out why the observed results were seen. It's been a useful way of developing a reliable, explainable process when there was none.