My own belief is that chain fires happen almost exclusively from the front from either sloppy bullet/ball seating, spilled powder, or most likely bullets or balls backing out of charge holes under recoil. That said, a conical bullet has more ways to be improperly seated than a round ball including tapered heels, badly shaped rammer cavity and bullet grooves that can scrape burned and unburned powder into the charge hole, especially if the grooves are greased. An oversized ball readily cuts a nice lead ring. An elongated oversized conical cannot cut a neat ring, but rather partially cuts a ring and swages itself into the charge hole. It's all probability and statistics as to how chain fires happen, but the mechanics are pretty simple. Rear chain fires could happen if an incandescent ember bounced around several 90 deg turns into a nipple recess and got under the cap skirt. Not likely but possible. Incandescent gas expands toward low pressure. Both embers and gas molecules obey Newton's laws. Upsets at the chamber mouth are another matter. A ball or conical bullet backing out exposes powder, possibly several charge holes at once. A loose projectile not aligned with the barrel also cannot be accelerated like a normal on, explaining why revolvers and their careless loaders can survive a chain fire as the bullets plop on the ground short of target. Any cause of chain fires is theoretically possible, but physics and probability favor front of cylinder causes, and conicals slightly increase those odds.