Archival Hubble Space Telescope (HST) observations of the Seyfert 1 nucleus of NGC 3227 obtained with the Space Telescope Imaging Spectrograph (STIS) are re-examined in order to constrain a viable photoionization model for the broad-line region (BLR). The results imply that the BLR is a partially ionized, dust-free, spherical shell that is collapsing, supersonically, at the free-fall velocity due to its proximity to a supermassive black hole. The BLR is ionization bounded at the outer radius, coincident with the dust reverberation radius, and transforms into an X-ray emitting plasma inside the Balmer reverberation radius as the central UV-X-ray source is approached. Only 40 M⊙ of Hydrogen are required to explain the Balmer emission line luminosity, but it is compressed by gravity into a column measuring 5.5×1024 atoms cm-2. Assuming radiatively inefficient accretion, the X-ray luminosity requires ~10-2 Msun yr-1. However, the mass inflow rate required to explain the luminosity of the broad Hα emission line is ~1 Msun yr-1. The very large disparity between these two estimates indicates that 99% of the inflowing gas must be re-directed into an outflow, and on a very short timescale corresponding to ~40 years. Alternatively, the radiative efficiency of the inflow has been overestimated, or the X-ray luminosity has been underestimated; a distinct possibility if the BLR is indeed Compton thick.