We have calculated the energy band structure of a linear model chain of selenium atoms on the basis of the ab initio four-component Hartree-Fock-Dirac equations to verify that the MOLFDIR program, extended for periodic polymers, is working well. The comparative calculation using the same relativistic basis set at the non-relativistic Hartree-Fock crystal orbital level shows that both results are similar, because the relativistic effects are expected to be small. Contrary to the finding of Schmidt and Springborg, we are convinced that poly(Se) is not a metal but an insulator. Additional evidence has been obtained from non-relativistic calculations with different basis sets. Their results may be an artefact of the density functional approach. We agree, that our relativistic basis set is not large enough to give the correct size of the band gap.