I did a little ‼science‼ on silk vs. leather, it appears to show that silk is slightly better than leather.
In 15 unarmed ten-vs-ten human battles, there were 6 wins for those wearing leather and 9 for silk.
In 16 ten-vs-tens with silver short swords, it came out a 7-9 victory for silk.
The victory margins seemed a bit weird, though...when leather won, it tended to be by a larger margin (i.e. with more surviving humans) than when silk won. The total survivor numbers were 39-53 in the unarmed battles, which still looks like a win for silk, but less decisively, and 49-53 in the sword battles, which is still technically a win for silk, but is likely a statistical dead heat, within the margin of error.
To declare a clear victor, you'd like to be able to say "if you wear X, you're more likely to win, and you're less likely to lose dwarves", but I don't feel like I can say that.
In contrast, with a test of silk vs. plant cloth using dwarves wielding silver short swords, silk was the clear victor in number of successes and an even clearer victor in survival margin, 11-4 and 64-20.
In all of these tests the humans or dwarves are wearing (in the chosen material):
1 shirt
6 cloaks
6 hoods
1 gloves
1 mittens
2 chausses
2 trousers
I went with multiples to increase the effect of the cloth armor, so the difference would (hopefully) show up better.