Canonical structure of Yang-Mills theory

I consider the problem of defining canonical coordinates and momenta in pure Yang-Mills theory, under the condition that Gauss' law is identically satisifed. This involves among other things particular boundary conditions for certain dependent variables. These boundary conditions are not postulated a priori, but arise as consistency conditions related to the equations of motion. It is shown that the theory indeed has a canonical structure, provided one uses a special gauge condition, which is a natural generalisation to Yang-Mills theory of the Coulomb gauge condition in electrodynamics. The canonical variables and Hamiltonian are explicitly constructed. Quantisation of the theory is briefly discussed.