We have determined accurate values of the product of the mass-loss rate and the ion fraction of P ^ { 4 + } , \dot { M } q ( P ^ { 4 + } ) , for a sample of 40 Galactic O-type stars by fitting stellar-wind profiles to observations of the P v resonance doublet obtained with FUSE , ORFEUS/BEFS , and Copernicus .
When P ^ { 4 + } is the dominant ion in the wind ( i.e. , 0.5 \lesssim q ( P ^ { 4 + } ) \leq 1 ) , \dot { M } q ( P ^ { 4 + } ) approximates the mass-loss rate to within a factor of \lesssim 2 .
Theory predicts that P ^ { 4 + } is the dominant ion in the winds of O7–O9.7 stars , though an empirical estimator suggests that the range from O4–O7 may be more appropriate .
However , we find that the mass-loss rates obtained from P v wind profiles are systematically smaller than those obtained from fits to H \alpha emission profiles or radio free-free emission by median factors of \sim 130 ( if P ^ { 4 + } is dominant between O7 and O9.7 ) or \sim 20 ( if P ^ { 4 + } is dominant between O4 and O7 ) .
These discordant measurements can be reconciled if the winds of O stars in the relevant temperature range are strongly clumped on small spatial scales .
We use a simplified two-component model to investigate the volume filling factors of the denser regions .
This clumping implies that mass-loss rates determined from “ \rho ^ { 2 } ” diagnostics have been systematically over-estimated by factors of 10 or more , at least for a subset of O stars .
Reductions in the mass-loss rates of this size have important implications for the evolution of massive stars and quantitative estimates of the feedback that hot-star winds provide to their interstellar environments .