Recurrent novae (RNe) are cataclysmic variables with two or more nova eruptions within a century. Classical novae (CNe) are similar systems with only one such eruption. Many of the so-called CNe are actually RNe for which only one eruption has been discovered. Since RNe are candidate Type Ia supernova progenitors, it is important to know whether there are enough in our galaxy to provide the supernova rate, and therefore to know how many RNe are masquerading as CNe. To quantify this, we collected all available information on the light curves and spectra of a Galactic, time-limited sample of 237 CNe and the 10 known RNe, as well as exhaustive discovery efficiency records. We recognize RNe as having (a) outburst amplitude smaller than 14.5 - 4.5 * log(t_3), (b) orbital period >0.6 days, (c) infrared colors of J-H > 0.7 mag and H-K > 0.1 mag, (d) FWHM of H-alpha > 2000 km/s, (e) high excitation lines, such as Fe X or He II near peak, (f) eruption light curves with a plateau, and (g) white dwarf mass greater than 1.2 M_solar. Using these criteria, we identify V1721 Aql, DE Cir, CP Cru, KT Eri, V838 Her, V2672 Oph, V4160 Sgr, V4643 Sgr, V4739 Sgr, and V477 Sct as strong RN candidates. We evaluate the RN fraction amongst the known CNe using three methods to get 24% +/- 4%, 12% +/- 3%, and 35% +/- 3%. With roughly a quarter of the 394 known Galactic novae actually being RNe, there should be approximately a hundred such systems masquerading as CNe.
Download