Back when I first heard about sealing the front of the chambers to prevent a chainfire, I wondered how this could happen. I had a 1851 Navy .36 and when I rammed the balls into the chambers a ring of lead would be shaved off each one so it seemed unlikely that flame could get around what amounted to a solid lead plug. At one point I also had one of the CVA cannons and therefore some fuse. The fuse gave me an idea. I loaded all six chambers of the Navy and then removed the cylinder from the gun. The percussion caps I used at the time required at firm push to seat them on the nipples so I used them as I normally would. I intended to pour powder into the mouth of each chamber, onto the bare seated balls, and then ignite it from a safe distance by using a piece of cannon fuse. Along about then it occurred to me that if I was wrong and did manage to create a chainfire that the cylinder wasn't likely to sit there pointing up but might flip over and spew balls in random directions. I used a hole cutter to make a slightly over-cylinder-sized hole in a piece of scrap 2X8. Now I was ready. I filled each loaded chamber mouth with powder, set the cylinder into the hole in the board, stuck a foot-long piece of cannon fuse into one chamber mouth, lit it, and ran for cover. There was a "poof" and a mushroom cloud of smoke but no main charges fired. After waiting a couple of minutes as insurance, I went back and repeated the test. More smoke but still no discharges. I concluded that when a ball of large enough size to cause a ring of lead to be shaved off on loading was used, there was no need to further seal the chamber mouth. I did continue to apply a homemade mixture of Crisco and beeswax over the balls to act as a lube but never bothered with cereal fillers or over-powder wads. Loading a cap n' ball is slow enough without adding extra unnecessary steps.
It would be interesting to see if a chainfire could be created by using caps that require pinching to remain in place. It would seem to me that in order to be valid the testing would have to be done using the fully assembled revolver locked into a machine rest and triggered from a safe distance. Other variables that could be introduced include nipples with oversize holes due to erosion and a weak hammer spring that could allow the hammer to be blown back slightly on firing. It seems logical that either of those could increase the likelihood of a chainfire. Of course, this test would also mean having a revolver that the owner doesn't mind risking damage to! Maybe I can come across one that has been severely abused by a previous owner....
John