Quantum Chromodynamics (QCD) is the modern theory of the strong interaction. QCD allows the coupling of gluons, the gauge particles of the interaction, among them and therefore the theory with only gluons is non trivial. Glueballs, the bound states of gluons, are a beautiful consequence of self coupling. I review the recent developments on glueball spectroscopy with a special emphasis on constituent models. The problematic on the mixing with nearby mesons leads to difficulties in the identification of physical states. I therefore restrict my models to the pure gauge theory . The pure gauge spectrum of quarkless QCD was investigated by Morningstar and Peardon on a lattice. They identified 15 glueballs below 4 GeV both in C = + and C = −. In a constituent picture, the low-lying states are bound states of two gluons with C = +. A negative charge conjugation requires at least three gluons. In the usual picture, the gluon is a heavy spin-1 particle since the gauge boson gains a dynamical mass induced by non-perturbative effects. This consideration leads to gluonium pictures for two- and three-gluon glueballs which are extrapolation of meson and baryon systems. However I show that the usual LS-coupling scheme cannot reproduce the lattice spectrum. This formalism is indeed irrelevant to handle transverse particles such as gluons. The implementation of the helicity formalism for two-gluons glueballs leads to a spectrum in perfect agreement with the lattice results even without the inclusion of short-range spin-splitting potentials.