Abstract
<jats:p>The Bretonnet–Silbert (BS) pseudopotential, with linearised Weeks–Chandler–Andersen (LWCA) perturbation theory and Rice–Allnatt (RA) transport theory, is a static, pairwise route to the magnitude of the shear viscosity of liquid sp metals. We use the miscible Pb–Bi alloy as a composition-axis test of this static description, over 21 compositions (xBi = 0–1) within 500–1200K, benchmarked against the Assael and Guzachev measurements. Because Pb and Bi have nearly equal hard-sphere diameters (σBi/σPb = 0.983), the alloy is near-degenerate: mixing adds cross correlations and a composition degree of freedom. The temperature-gradient deficit found for the pure metals, a calculated |dη/dT | too small by about a factor of two over the physical single-phase window (900–1200K) and up to a factor of three once the sub-liquidus formal extrapolations are included, is not relaxed along this composition axis, so within this framework the single-component approximation does not by itself supply the missing slope. The binary three-pair RA sum collapses onto the effective single-component route to within 0.27%, and a eutectic Ornstein–Zernike calculation with a Malijevský–Lab́ık bridge narrows the deficit by only a few percent. The soft-tail fraction ηs/η = 0.53–0.58, a common fixed-cutoff diagnostic, is nearly composition-independent. The pipeline captures the order of magnitude and the monotonic composition background of η(xBi) but does not reproduce the experimental non-monotonicity, which the source attributes to a metastable melt microheterogeneity outside the equilibrium closures tested here. Because the near-degeneracy makes the binary structural corrections intrinsically small, the test bounds the single-component approximation for this chemically similar pair rather than for alloys in general. It adds composition-axis evidence to the prior interpretation that the missing temperature dependence is consistent with a dynamic contribution beyond the static description.</jats:p>