Any theory has its limits of applicability including the phase-matching condition. Then we may ask whether a criterion exists which tells us when the effective frequency doubling may be discussed "phase-matchingly" and when it must be discussed " non-phase-matchingly". In this manuscript based on the actual discrete atomic structure of crystal,the limits of applicability of phase-matching condition described by Franken and Ward on the effective frequency doubling behavior are discussed. Roughly speaking a phase-matching treatment is adequate,i.e.,sufficiently accurate,whenever all relevant optical wavelengths are large compared to the lattice constant of the crystal.