We recover PH3 in the atmosphere of Venus in data taken with ALMA, using three different calibration methods. The whole-planet signal is recovered with 5.4{sigma} confidence using Venus bandpass self-calibration, and two simpler approaches are shown to yield example 4.5-4.8{sigma} detections of the equatorial belt. Non-recovery by Villanueva et al. is attributable to (a) including areas of the planet with high spectral-artefacts and (b) retaining all antenna baselines which raises the noise by a factor ~2.5. We release a data-processing script that enables our whole-planet result to be reproduced. The JCMT detection of PH3 remains robust, with the alternative SO2 attribution proposed by Villanueva et al. appearing inconsistent both in line-velocity and with millimetre-wavelength SO2 monitoring. SO2 contamination of the ALMA PH3-line is minimal. Net abundances for PH3, in the gas column above ~55 km, are up to ~20 ppb planet-wide with JCMT, and ~7 ppb with ALMA (but with signal-loss possible on scales approaching planetary size). Derived abundances will differ if PH3 occupies restricted altitudes - molecules in the clouds will contribute significantly less absorption at line-centre than equivalent numbers of mesospheric molecules - but in the latter zone, PH3 lifetime is expected to be short. Given we recover phosphine, we suggest possible solutions (requiring substantial further testing): a small collisional broadening coefficient could give narrow lines from lower altitude, or a high eddy diffusion coefficient could allow molecules to survive longer at higher altitudes. Alternatively, PH3 could be actively produced by an unknown mechanism in the mesosphere, but this would need to be in addition to cloud-level PH3 detected retrospectively by Pioneer-Venus.
Download