“…For R ρ < 3, <F H,mol,DNS > strongly depends on the rejection criterion for disturbed interfaces, which can lead to overestimates (as here, by 29%) as well as underestimates (without rejecting, 20%) of <F H,tot,DNS >. We note that Flanagan et al [2013] showed that at R ρ = 2, 2-D simulations result in a factor of 1.5 smaller heat fluxes than the corresponding 3-D simulation. This questions, in general, the validity of the 2-D DNS in accurately capturing fluxes at small R ρ .…”