A Systematic Trend in the Observed vs. Predicted Radii of Transiting Brown Dwarfs
Transiting brown dwarfs provide a direct empirical test of substellar evolutionary models, and with more than 50 now known we can carry out this test at the population level. We compared the observed masses and radii of 31 field-age, weakly irradiated transiting brown dwarfs to the Sonora Diamondback, Sonora Bobcat, and ATMO 2020 grids, and a subset of 23 of them to the irradiated models of Mukherjee et al. (2026). In every comparison we find a radius-dependent trend, with fits of $R_{\rm pred}$ versus $R_{\rm obs}$ yielding slopes of $0.3{-}0.4$, shallower than the one-to-one relation at $3.7{-}8σ$. The mean inflation is $5{-}14\%$ depending on the grid, reaching $\sim20\%$ among the largest objects. We also measure an intrinsic dispersion of $8{-}12\%$, two to three times the measurement uncertainty. While the mean inflation offset varies by a factor of three among grids, this dispersion does not, indicating that it is a property of the population and not the models. We find no correlation with mass, equilibrium temperature, orbital separation, tidal heating rate, or age, and only a marginal ($1.7σ$) trend with host-star metallicity. Most notable is the absence of any trend with irradiation, which argues against hot Jupiter-like stellar heating as the primary cause. We instead suggest that brown dwarfs span a wider range of atmospheric boundary conditions than current grids represent. The brown dwarf radius inflation is comparable to that in fully convective M dwarfs, and it is likely the transiting counterpart to the luminosity discrepancies seen in dynamical-mass benchmarks. If the trend extends below the deuterium-burning limit, it would bias directly imaged planet mass estimates, overestimating luminosity-age masses by several to ten percent and requiring spectroscopic-gravity masses to be revised upward by $\sim20{-}45\%$.