The Hidden Second Law of Thermodynamics behind the Boltzmann-Grad Limit
We show that the Boltzmann-Grad (BG) limit, $N\varepsilon^{d-1}=α=\text{const}$, is not a neutral scaling condition but already encodes the Second Law of Thermodynamics, resolving Loschmidt's paradox. Newton's First Law is the bridge: applying the BG condition locally to two adjacent cells at different temperatures, we decompose their evolution into a free-streaming period $δt_f$, governed by inertia alone, during which molecules cross the interface ballistically and still carry their cell of origin's velocity distribution; and a collisional adjustment period $δt_a$, governed by Newton's Second Law, during which the locally mixed, non-Gaussian population relaxes toward a new Gaussian equilibrium. Using the one-dimensional Gaussian velocity distribution, we show that isotropy alone produces non-zero net molecular and momentum flux across the interface, directed from the warmer to the cooler cell -- the microscopic origin of the pressure gradient driving the system toward uniformity. Treating relaxation as a sum of random collisional increments, the Central Limit Theorem fixes the collision count $n^*$ required for convergence, giving $δt_a \sim n^*\,\ell/\bar v$. This decomposition resolves Loschmidt's paradox: his reversal acts on a fixed, identifiable set of molecules, while the temperature field Boltzmann's equation tracks is carried by an ever-changing local population; the step relating the two -- convergence under the Central Limit Theorem -- is itself a coarse-graining that discards microscopic information by construction, recoverable only at an unavoidable thermodynamic cost, by Landauer's principle. Temporal irreversibility is present from the moment the BG framework, and the coarse-graining it requires, are adopted -- not generated during derivation.