Comparison between a rigorous molecular crystal treatment of nonlinear opti
cal response and a cavity model approach to third harmonic generation shows
that a cavity field should be included at the output frequency. The argume
nt for omitting this cavity field, that there is zero macroscopic field at
the output frequency, fails to take proper account of polarization at the o
utput frequency outside the cavity. (C) 2001 American Institute of Physics.