Nonlinear Decline-Rate Dependence and Intrinsic Variation of Type Ia Supernova Luminosities

Published B and V fluxes from nearby Type Ia supernovae are fitted to light-curve templates with 4-6 adjustable parameters. Separately, B magnitudes from the same sample are fitted to a linear dependence on B-V color within a post-maximum time window prescribed by the CMAGIC method. These fits yield two independent SN magnitude estimates B_max and B_BV. Their difference varies systematically with decline rate Delta m_15 in a form that is compatible with a bilinear but not a linear dependence; a nonlinear form likely describes the decline-rate dependence of B_max itself. A Hubble fit to the average of B_max and B_BV requires a systematic correction for observed B-V color that can be described by a linear coefficient R = 2.59 +- 0.24, well below the coefficient R_B ~ 4.1 commonly used to characterize the effects of Milky Way dust. At 99.9% confidence the data reject a simple model in which no color correction is required for SNe that are clustered at the blue end of their observed color distribution. After systematic corrections are performed, B_max and B_BV exhibit mutual rms intrinsic variation equal to 0.074 +- 0.019 mag, of which at least an equal share likely belongs to B_BV. SN magnitudes measured using maximum-luminosity or CMAGIC methods show comparable rms deviations of order ~ 0.14 mag from the Hubble line. The same fit also establishes a 95% confidence upper limit of 486 km/s on the rms peculiar velocity of nearby SNe relative to the Hubble flow.