Really, the letters s, p, d and f don’t stand for anything. They may be thought of as representing sharp, principal, diffuse and fundamental (or faint), but there’s nothing sharp about an s-orbital or diffuse about a d-orbital.
The azimuthal quantum number l (properly a script l) represents the shape of an orbital (an area of electron density probability). The value may range from 0 to (n-1), where n is the principal quantum number or “energy level”. If l = 0, the orbital is called s, traditionally; if l = 1, it is called p, and so on.
The names “sharp”, “principal”, “diffuse” and “fundamental/faint” come from early work on spectral analysis of hydrogen and other elements. Lines in the spectrum were described qualitatively as “sharp”, “diffuse”, and so on.
We now understand that the lines arise from a transition between two orbitals – orbitals with different azimuthal quantum numbers. The ‘sharp’ lines are from p --> s transitions, the ‘principal’ are from s --> p transitions, the ‘diffuse’ are from d --> p transitions and the ‘faint’ are from f --> d. See here.
Since there’s no consistent correlation between the orbitals involved in the transition and the type of line observed, it’s not strictly accurate to name the orbitals s, p, d and f, but the names have stuck.
Eurograff: I don’t like the idea of “shells” and the labels K, L, M, N… It’s a rather Bohr-Rutherfordian concept (though I think it refers to de Broglie’s model, actually); outdated since the early 20’s. Numbered energy levels (the principal quantum number n seem more appropriate for dealing with quantum-mechanical electrons, which are most certainly not confined to anything resembling a shell. =)