Self-bending of light: comment.

This Comment addresses an observed effect that we believe has been inaccurately identified by Xing et al.1 The reported effect is not self-bending, as they suggested, but rather a cross-phase modulation effect. The self-bending effect2 has, for 24 years, been defined as a degenerate process whereby a beam of wavelength A induces an intensity-dependent index change that deflects the trajectory of light (at A) when the transverse intensity profile is triangular or asymmetric. In the first observation of self-bending3 a wedge was used to achieve large beam deflections, reminiscent of the configuration of Xing et al. ,' although in their case nanosecond pulses were used and anti-Stokes generation was not necessary. In fact, short pulses are not needed for self-bending to be demonstrated, as evidenced by the first cw measurements made in sodium vapor.4 The effect observed by Xing et al. is attributed to cross-phase modulation of the anti-Stokes beam by the pump beam. This mechanism permits a strong pump beam to affect the propagation of light at another frequency (or polarization), as first reported in Ref. 5 and recently investigated in Ref. 6. The self-bending mechanism would require a high-intensity antiStokes beam (which was not reported in Ref. 1), whereas the cross-phase modulation would not.