An improved theoretical long-range dispersion plus induction dipole moment surface for N-2-Ar is presented, and detailed numerical simulations are used to test its predictions against experimental far-IR and mid-IR spectra using two previously-reported potential energy surfaces. As was found in earlier work on the mid-IR spectrum using a pure induction dipole surface, the MMSVmod potential of Jager [J. Chem. Soc. Faraday Discuss. 97, 105 (1994)] yields distinctly better agreement with both experiments than does the XC-3 potential of Dham [J. Chem. Phys. 103, 8477 (1995)]. However, the new dipole surface yields slightly poorer agreement with certain features of the experimental mid-IR and far-IR spectra, which suggests that the existing theoretical values and derivatives with respect to the bond length of some of the permanent moment and (hyper)polarizability properties of N-2 need improvement.