2021MNRAS.507..927M

The Milky Way hosts on average a few supernova explosions per century, yet in the past millennium only five supernovae have been identified confidently in the historical record. This deficit of naked-eye supernovae is at least partly due to dust extinction in the Galactic plane. We explore this effect quantitatively, developing a formalism for the supernova probability distribution in space and on the sky, accounting for dust and for the observer's flux limit. We then construct a fiducial axisymmetric model for the spatial supernova and dust densities, featuring an exponential dependence on galactocentric radius and height, with core-collapse events in a thin disc and Type Ia events including a thick disc component. When no flux limit is applied, our model predicts that on the sky, supernovae are intrinsically concentrated in the Galactic plane, with Type Ia events extending to higher latitudes. We then apply a flux limit and include dust effects, to predict the sky distribution of historical supernovae. We use well-observed supernovae as light-curve templates, and introduce naked-eye discovery criteria. The resulting sky distributions are strikingly inconsistent with the locations of confident historical supernovae, none of which lie near our model's central peaks. Indeed, SN 1054 lies off the plane almost exactly in the anticentre, and SN 1181 is in the second Galactic quadrant. We discuss possible explanations for these discrepancies. We calculate the percentage of all supernovae bright enough for historical discovery: ≃13 per cent of core-collapse and ≃33 per cent of Type Ia events. Using these and the confident historical supernovae, we estimate the intrinsic Galactic supernova rates, finding general agreement with other methods. Finally, we urge searches for supernovae in historical records from civilizations in the Southern hemisphere.