Fighter, Magic-User, Cleric ,Thief , Elf (as Fighter and Magicuser) and Dwarves and Halflings which were also Fighters. Were the initial Mix of character classes for me and my fellow players and a little while later I snagged a copies of Greyhawk and Blackmoor (before we got the PHB) and were separating race and class and had paladins, assassins and monks also. With the PHB we had more races and more classes.
I've been playing a little S&W:white-box with just Fighter, M-U and Cleric as classes and it does make the game feel different. Which got me thinking about the role classes have in defining the game and the campaign setting.
Take the Basic D&D model that was clearly expressed in the second basic set where they clearly separated the main classes and the demi-human classes (it had been done in the first basic set buty it was a little fuzzy). This makes it pretty clear the 4 human classes classes are different and Dwarves, Halfling and Elves just aren't the same thing as humans. The range of classes defines the roles the players will have and defines the setting.
AD&D actually screwed stuff up by making the Fighter the weakest of the Fighting Classes. Paladins and Rangers are clearly better Fighters then the Fighter, sure they have stricter requirments to qualify but with that many different character methods of AD&D it wasnt' unlikely to qualify and it always made the Fighter seem inadequate as it was the class to settle for when you weren't buff enough to qualify for Ranger or Paladin.
Sure there are lots of role-play reasons not to play Rangers or Paladins but it takes a while for that to really matter to players and DMs.
Unearthed Arcana added some special abilities for Fighters in the form of weapon specialization but didn't make it clear they were the reserve of Fighters alone and added Cavaliers and Barbarians to the Fighter class mix. There lot's to like and not like about the Cavalier and Barbarian but my main gripe is that once again they are better fighters then the Fighter. The campaign was now further defined by the presence of snotty knights and uncooperative fighters with a lot of new special abilities.
Now "what's wrong with options?" you might ask and "nothing at all" would be my reply. But the options available should define and add but they shouldn't' undermine and weaken other options.
The dreaded 3.x opened all the classes to all the races (and all the monsters too) which didn't bother me much at all mechanically but it did rip down definitions. What does it mean to be a Ranger when there can be Rangers among any group of people or a Paladin or a Monk? Not much really just a different ability package. I should note Fighters were the best Fighting class again at least until splat-books and 3rd party stuff flooded the game with so many options there were no clear choices and no definitions in any of the choices. If you have a campaign with 20, 30 or 40 classes which often overlap each other in abilities the definition of the character class becomes a virtually pointless one (not to mention the impact of willy-nilly multi-classing). Definitions are smeared and names become meaningless.
So where am I getting at here? From observing the cases above (and more i am not covering here) it's the range of classes a DM has available in his campaign that defines how the players are going to experience the campaign. It's going to trump all sorts of lovingly detailed campaign setting design because it will be experienced by the PCs through the character classes.
So my fellow DMs and would be DMs I recommend a careful selection of available classes and a tweaking and invention of them as you wish to define your campaign. A campaign with the classes of Charlatan, Seer, Alchemist, Scribe, Aristocrat and Tough is going to play differently from one that contains the classes of Archer, Knight, Knave and Sorcerer. Of the two ranges just given which one would makes more sense in an Arthurian inspired campaign and which would work better with foppish dandies and miscreants in a decadent metropolitan setting? Doesn't require much thought to figure it out does it? That's the power class availability and definition has on defining a campaign.
Stop trying to hammer Druids, Rangers, Paladins and such into every campaign and expect to have your campaign feel different. Drop some, create some, define them for they will be the main tool in how your campaign is realized by the players.
Level limits, racial selections, skills or proficiencies I'll leave those for another time.