If this A.D. has not been easily understood or reproducible, there are a few areas of concern that could alter normal function of the entire trigger mechanism. I guess anything affecting normal flow of motion from one part to another could cause improper action such as an accidental discharge.

After cocking, the trigger is pressed to the rear. The slack is taken up in the first press of the trigger rearward until the second stage is reached where the sear is engaged to be released from the ledge of the firing pin rod. There is pivot at the "hinge" of the trigger to cause this sear release with response from the trigger spring pushing rearward at the front of where the sear attaches to the receiver. The rear most part of the trigger must be clear to allow for it to move upward to allow for the hinge to pivot and release the sear at the point of contact with the firing pin rod to allow the spring release of the striker in the bolt to hit the bullet primer.

So, if anything affects any part of these actions, something abnormal and potentially unexpected can happen. That could include any debris temporarily in the way of the sear spring, pressure laterally on any part of the sear or trigger to prevent smooth movement, debris in the way of the rear part of the trigger that would prevent pivot of the sear and trigger at the hinge, metal burrs or rust preventing smooth and predictable motion of any of these parts or mechanism.

I also could see where there could be "premature release" of the sear from the firing pin rod ledge if either of these is not squared off and improperly ground or worn down and not permitting solid engagement.

It would be nice to see pictures, Fred, of the trigger and sear to see if anything is worn abnormally. The trigger spring could easily be replaced, as Butlers recommended, and make sure the trigger seat is clear front and back.

This solves nothing but just serves as an additional attempt to evaluate the situation.

Thanks again Fred.

I tried it this morning and no failures again. Right now I am thinking that I have a perfect storm of everything mentioned so far. I did stick a knife blade into the sear slot to see if the sear was centered. I could scootch to the right and left so if it is binding it is not extreme. It does have a little bias wanting to be to the left, so BRs first inclination of something touching the sear could be correct. I probably have a weak spring and can check it against the one just purchased from Whig. I had this thing apart and together so many times I probably forgot to lube the trigger when installing it the last time. Also I did not alter the sear in anyway, I can see some micro dings that should have been removed from the sear with a fine Arkansas hone. Also as discussed in the trigger modification thread I did not hone off any dings on the edges of the sear knuckle/hinge.
So right now the plan is to disassemble and look for all these obvious things, and reassemble correctly and hope there is not one more unknown we forgot. I looked for my small tube of gun grease and could not find it, that makes it kind of obvious I forgot to grease things doesn't it.
So it will take a couple of days to get some more grease, so that is more time to think more on this.

Got a question since the trigger was not catching the cocking piece that meant the opening cam for pulling it back was releasing it relatively slowly. How slowly does the firing pin have to be released so there is no hang fire or unintended ignition. Can you pull the trigger and let the pin down slowly enough that the pin will not dent the primer? Easy experiments to try with a primed case with no powder.

I did the check on the safety pulling the cocking piece off the sear. That was an early concern since this Krag has the bolt with the extreme wear on the lug. Everything was OK there. OOH, that is one more thing to check, make sure that in cycling the bolt it always comes back far enough to catch the trigger sear

" OOH, that is one more thing to check, make sure that in cycling the bolt it always comes back far enough to catch the trigger sear."

That one was a red herring, pulling the cocking piece far enough back to catch the sear is automatic during the lifting the bolt handle. Short cycling the bolt will cause no harm unless you do not pull it back far enough to catch the next shell.
I tried some other things like rotating the bolt sleeve while the bolt is pulled back, impossible the Krag inventors and armory really nailed this design.

Whig,
Sorry I did not see your reply, my 2 were just a few minutes apart. Since the sear failed to catch the cocking piece when closing the bolt the speed that the firing pin can fall is controlled by the closing cam. On this AD I closed the bolt very slowly as we watched the bolt catch the cartridge and start pushing the round into the chamber, I did speed up towards the finish. The firing pin speed is still controlled by the opening cam. so it was still very slow compared to the normal free fall. Imagine holding the trigger as you close the bolt, the pin will move forward as controlled by the opening cam. Same speed as when opening except in the opposite direction. Surprised when I think about it that it would even ignite the primer. If you close the bolt slow enough there has to be a threshold were it is too slow to ignite.

BR,
I had already done the test to see if the safety did pull the cocking piece off the sear. It did pass. My other tests were to make sure the bolt wear/setback wear not causing any issues. Design is too robust so no issues can be created by twisting on the cocking piece, bolt or sleeve when opening the bolt even tried the same test when the bolt was fully back.
I do have a purchased firing pin with some wear on the cocking piece and can see where this could cause some issues. Will open this post up in a few minutes to attach a photo.

I am putting the 2 photos in a separate reply to make it easier to find for other searching for AD. The photos show wear on the point of the cam following surface, more extreme wear could cause the cocking  piece not to pull back far enough to catch the sear. Not my issue but could be helpful to others. Photos are now upgraded.

These photos are not from the problem sporter, but are from a junk firing pin used for experiments. The worn areas have a line drawn around the wear. The roughness in this area would be bad for a good bolt as well a possibly getting worse. For others diagnosing an AD event this will be something to check. Worse wear in this are could cause the sear surface on the cocking piece to not engage the sear.

Were you able to compare the A.D. fired empty case with another fired case from the same rifle? Maybe a high primer or different indentation in the primer. Just a thought.

OK, primary cause spring was not standard, weaker because of smaller wire diameter, also longer and larger diameter than original. One end is slightly larger diameter than the other. Could be the reason it worked for years then did not. The end that is larger diameter can bind in the hole in the sear. Not sure which side was in first as I did not pay attention when I pulled it out.

In one item I have a probable cause plus a contributor. Will still address the other potential problems and investigate a couple that you all suggested. Brought over my working Mag Light to see if parts can be used to repair my HawkEye bore scope. I will continue to look for other contributing factors till I can not find anymore.

The spring that was pulled is on the left, close to 1/4 inch diameter. The spring from the newly purchased Krag on the right .227 diameter and .570 long. Close enough or should I wait for a new one?

The spring is noticeably stronger and the trigger will pull back to the full retract when released. For anyone else if your trigger does not retract when you pull it to the end of the first stage then release, you may have a modified/wrong spring or a serious lack of lube with maybe gunk holding things up.
Did clean up the sides of the trigger channel per BR suggestions. Looking with the repaired bore scope from the sides there was no evidence of rubbing but the sides are now smoother and overflow accuglass has been removed. At least I will never have to think about that one again.
Still planning to take the front step off the trigger to lengthen first stage travel. No down side to that plus a little extra sear engaged initially is a good thing. Several more small things to address.