edit: Unrelated, but peculiar enough for me to comment on - that whole deal with the Thief in 1e
losing the ability to wear armour above leather? Bizarre!

Of course, that's only if normal men really can wear all types of armour, whether that's indeed as written, or as interpreted...
Yep. There's no armour "proficiency" until 3e AFAICT. Thieves can't use their special abilities in armour better than leather.....but this isn't spelled out until UA, where modifiers are given for these abilities when wearing other forms of armour. So, yes, by strictest reading of "core 3" 1e AD&D, one could potentially discover hidden thieves (and assassins, magic-users, and illusionists) by finding out who is unwilling to try on that chain mail shirt.......
Interestingly enough, there are at least two 3.x adventures I am aware of that begin with 0-level PCs, and 1e had a system for starting characters as 0-level folks, who gain their first character class level in-game. It was published in Greyhawk Adventures, if memory serves.
Finally, [MENTION=22779]Hussar[/MENTION]: The term "Normal Man" is a game construct, like "Fighter" is a game construct. No one in their right mind says, "Sorry, Joe, you can't run away. You're a fighter." Likewise, no one in their right mind mandates that a character steal from the party because his class is "Thief".
Remember all of those discussions about how 3e classes can represent various things? How about those discussions about how 4e classes can represent various things? How about how a "warforged ninja" doesn't have to literally be a warforged, or a ninja? Remember all of those things?
Well, "Normal Man" is the same. There are normal men who don't use the "Normal Man" nomenclature, and there are men who use the "Normal Man" nomenclature that are exceptional (i.e., Joe the Blacksmith, Normal Man, 8 hp, 18 Strength).
And, as already demonstrated, I can not only make a 1e AD&D F1 who is effectively statistically identical to a turnip farmer, but I can make one that is statistically inferior.
Are you still entrenched in your position that someone who is effectively statistically identical or statistically inferior to a turnip farmer cannot be a turnip farmer?
RC