The original Kodak research indicated an average reflectance of 12-14%, and an average subject brightness range (IIRC) of 128:1 -- for outdoor subjects near Rochester, N.Y. Needless to say, there are wide variations even in
averages: go to Greece, for example. In general, brightness ranges decline as you go north and increase as you go south.
The 18% grey card is a Munsell mid-tone, i.e. if you show someone a series of grey cards of different densities, they will choose 18% as a mid-tone. Originally the 18% had nothing whatsoever to do with exposure determination; Kodak recommended taking a reading off a Kodak-yellow paper packet (honestly!).
Even a poor exposure meter often represented an enormous improvement over many people's ability to estimate exposure. As Bill says, they all work pretty well, not least because there is quite a lot of latitude in exposure, especially with neg/pos, and there is no such thing as an objectively correct exposure. There is however such a thing as a perfect exposure, which is the one you want.
Spot meters -- used properly, to read shadows, not mid-tones -- were another big jump again. The first successful spot meter, the SEI, didn't even have a mid-tone index, because no film speed system is keyed to a mid-tone: negatives are keyed to shadows, and slides (and digital) to highlights.
Another name for the incident light system is the 'artificial highlight' system, which gives the game away. An incident reading has nothing to do with an 18% grey,
EDIT: except insofar as an 18% grey is related to a bright diffuse highlight (typically 90% reflectance, i.e. 5x that of an 18% grey card): it will merely ensure that the highlights aren't blown.
Broad-area reflected light meters are based on the 12-14% assumption, but multi-sector meters may weight the exposure in one direction or another according to the different readings of the different sectors.
There's more about this on the site. Start with
http://www.rogerandfrances.com/subscription/ps expo neg.html and go on from there.
Cheers,
R.