Constraining broad-line regions from time lags of broad emission lines relative to radio emission

In this paper, a new method is proposed to estimate the broad-line region sizes of ultraviolet (UV) lines RuvBLR. It is applied to 3C 273. First, we derive the time lags of radio emission relative to broad emission lines Lya and C iv by the z-transformed discrete correlation function (ZDCF) method. The broad lines lag the 5-, 8-, 15-, 22- and 37-GHz emission. The measured lags tuvob are of the order of years. For a given line, tuvob decreases as the radio frequency increases. This trend results from the radiative cooling of relativistic electrons. Both UV lines have a lag of tuvob=-2.74+0.06- 0.25 yr relative to the 37-GHz emission. These results are consistent with those derived from the Balmer lines in Paper I. Secondly, we derive the time lags of the lines Lya, C iv, H, H beta and Ha relative to the 37-GHz emission by the flux randomization/random subset selection (FR/RSS) Monte Carlo method. The measured lags are tob=-3.40+0.31- 0.05, -3.40+0.41- 0.14, -2.06+0.36- 0.92, -3.40+1.15- 0.20 and -3.56+0.35- 0.18 yr for the lines Lya, C iv, H, H beta and Ha, respectively. These estimated lags are consistent with those derived by the ZDCF method within the uncertainties. Based on the new method, we derive RuvBLR= 2.54+0.71- 0.35 to 4.01+0.90- 1.16 and 2.54+0.80- 0.43 to 4.01+0.98- 1.24 light-years for the Lya and C iv lines, respectively. Considering the uncertainties, these estimated sizes are consistent with those obtained in the classical reverberation mapping for the UV and the Balmer lines. This indicates that their emitting regions are not separated so large as in the classical mapping of the UV and optical lines. These results seem to depart from the stratified ionization structures obtained in the classical mapping.