We've talked about hit points before, many times. Everyone has. I'm not about to go over that ground again, or start (another) debate about which is better, hit points as luck or hit points as damage resistance. Let's just assume that some people treated it one way and some treated it the other way. And Gygax seemed to be more on the luck side from the start, but that means nothing: we all disagree with Gygax on at least one point.
But here's a thought that occurred to me in the middle of a debate about whether a one-minute combat turn should be subdivided into individual rounds or phases. As you may know, the Holmes Basic book uses six-second combat rounds. This seem to be derived from combining the Chainmail one-minute combat turn (which contains a number of mêlée rounds) with the comment in U&WA about "ten rounds of combat per turn", which many other people interpret as ten one-minute rounds per ten-minute exploration turn.
But think about this: we know the one-minute D&D round interpretation is supposed to include multiple attacks, feints, parries, blocks, and maneuvers, resolved with a single die roll (by those of us who are more abstract) or ten die rolls (by those who prefer one roll per action.) We know also that, in Chainmail, one hit equals one kill, in contrast to D&D with its 1d6 damage rolls (later changed to different damage ranges based on weapon.)
Maybe the damage roll is not the strength of the blow, but how many six-second mêlée phases involved actual attacks, as opposed to feints or defensive maneuvers? Assume that in one minute, whoever wins initiative gets one six-second action, then the other side gets a six-second action, and so on, alternating, Thus, the person who goes first gets attacks or feints on odd-numbered phases (1, 3, 5, 7, 9) and the person who goes second gets attacks/feints on even numbered phases. A roll of 1 means only one of your attacks was "real", the rest were feints or were meant to wear down your opponent as you looked for an opening; a 5 means you were ferocious, battering your opponent repeatedly; a 6 means you were so aggressive, you snuck in an extra attack, perhaps as part of a riposte. Hit points would thus be how many last-minute dodges and parries the combatant has in reserve; combatants are assumed to be dodging, blocking, and parrying all the time, but sometimes a defender knows "I almost died just then, but I just barely escaped with a scratch."
The benefit of this interpretation is that it adds more detail to one-roll-per-minute combat without adding extra rolls. The fighter engages in combat with a goblin. The fighter has 4 hp and the goblin has 3 hp. The fighter goes first, succeeds on the attack roll, rolls a 2 on 1d6; this translates into something like "feint, thrust, feint, thrust, feint". The goblin's attack roll also succeeds, but the damage roll is a mere 1, which means mostly feinting or recovering and only one major blow. Neither combatant is dead, so the GM could describe the round this way: "Bob and the goblin exchange a couple swipes, then Bob makes a stab at the goblin's heart, which just misses, grazing its ribs instead. The goblin throws a half-hearted slash at Bob's head to buy some time, but Bob keeps pressing him. Then, the goblin makes an unexpected chop at Bob's head, which Bob just barely blocks with his shield; Bob tries to hack off the goblin's arm for that affront and almost succeeds; they circle each other warily for the rest of the round."
You don't have to literally assign each blow to a specific numbered combat phase, or interleave attacks in such a rigid manner; I just did that for the example, to show how you could describe two good blows by the fighter versus one good blow by the goblin, instead of just describing one blow each.