Bent/sprung bayonet blades

[mhb]

In examining a good many U.S. bayonets for the Krag and 1903 rifles, I've found a number of them have blades which are very noticeably not straight. It was my understanding that proof testing of new bayonets included a bending test, which required that the blade be sprung a certain number of degrees, and released, but that they were required to return to the original straight configuration without taking a permanent set.
I've been unable to find a good reference for the proof methodology, though, and a friend says that his study of the problem left him with the impression that the blade COULD take a set, up to a certain limit, and still pass proof. Thus, his belief is that some original bayonets could have been issued with a permanent set to the blade - my question in this case is: why not spring them back straight?
Can anyone furnish a good reference which addresses this issue?

Thanks;
mhb - Mike

[Embalmer]

Can't find the pic I took at springfield armory, but showed and explained of testing bayonets during the civil war fixing bayonets to rifle, pointing to ground and attempting to get it to bend. If did was no good ect. With a sketch/engraving of this. I'm sure they were still doing same test (though prob mechanical later on)

[Dan Shapiro]

Given that our military men can break an anvil with a rubber mallet, I'm wondering if those blades might have seen prior use as pry bars before they left military service.

[mhb]

I had seen the depiction of CW bayonet testing, and the description left me with the impression that they were NOT supposed to take a permanent set.
And Dan (a fellow refugee from the old ASA, apparently) is almost certainly right about how those bayonets came to be bent in service.
I still hope to find a good reference for the original testing spec. for post-CW bayonets, though.

mhb - Mike