Following the suggestion by Walsh, Carswell, and Weymann that the ratio of the observed wavelengths of the C IV and C III) (lambdalambda 1549, 1909) emission lines in the double QSO 0957+561 differs significantly from the expected value, the distribution of this ratio in a sample of 189 other QSOs is examined. The mean value of the ratio is consistent with the expected one, and the dispersion is consistent with typical measuring uncertainties. The result indirectly strengthens Wills and Netzer's conclusion that O IVk) lambda1402 is the principal or sole contributor to the 1400 A emission feature in QSOs. An unexpected finding, with no ready explanation, is that the ratio of the C IV and C III) wavelengths is correlated with emission-line redshift, at the 4 sigma level.