My two (additional) cents. You start with Bloch's theorem (say, 1D lattice $L=Na$). You see that you can write the eigenfunctions of the (lattice) periodic Hamiltonian as Bloch functions, which are indexed by $k$ and $n$, where the former obeys the BvK boundary conditions. You further see that you only need $k$ from the 1BZ. Let's do so; you end up with $E_n(k)$ and $\psi_{nk}$ and have the "usual" band structure. Fine so far?
Now sometimes it seems convenient to extend the *definition* or *construction* of these entities to $k$ not only from the 1BZ. One obvious thing to do is to say that…