With the air coming form beneath the coal bed, its oxygen content is mostly consumed. Like the combustion mixture in a gasoline engine, the mix of gas:air has to be in the sweat zone to ignite. Too rich (GAS:air) no ignition and conversely, too lean (gas:AIR) and no ignition. To get the mixtures on either side of the sweet spot to burn, you have to change something else. Like engaging the choke on a carbureted engine or changing the timing curve. I might be restating what Uglysquirrel said using an example I can relate to - maybe others can as well (?). The wood gives off gasses/smoke that don't ignite due to too rich of a mixture in the zone above the coal that is richer in CO & CO2 than O2.
This also might help explain what happens when a new load of coal is placed on top of the fire. It's not just exposure to the flame that sets off the gas above a new load but includes the ratio of gas:air. The side of the fire we keep exposed allows more air up thru and a zone of gas above the fire exists right there that allows the fire to spread smoothly over the new coal that's off-gassing at a higher rate (richer) than the old part of the fire. My stove is a top loader and I don't have secondary air inlets above the fire to use like Devile does. If I'm rushed I'll dump an entire hod and cover the entire old fire. I'll crack the ash door for a better part of a minute to allow more air in then close it. At this point, I'll lift the top lid a few inches and the gas might gently light off with the fresh air drawn into the fire box. If it doesn't, I'll blow into the gap (from a safe distance
) and it usually lights off, maybe with a gentle POOF! Nothing nearing an explosion like in the discussion started a while ago by Wood'nCoal --->http://nepacrossroads.com/about2130.html