Abstract
The standard approach for including laser line-shape effects in calculated coherent anti-Stokes Raman spectra is shown to be incorrect. The correct approach involves extra terms that become important if the laser linewidth is larger than the Raman linewidths. The standard approach can lead to errors in measurements of concentration or nonresonant susceptibility. A closed-form solution for the special case of Lorentzian laser line shapes is presented. Measurements using broadband and narrow-band lasers support the modified theory but are inconsistent with the standard theory.