We resolve the long‑standing discrepancy between the direct Feynman diagram calculation of the two‑loop quantum correction to the sine–Gordon kink mass and the exact result from integrability. The classically reduced meson Hamiltonian, derived by Marsden–Weinstein symplectic reduction, misses a local counterterm that originates from the proper quantum definition of the interaction: normal‑ordering of the cosine potential with respect to the free massive vacuum. We expand the normal‑ordered Hamiltonian around the kink background, extract the resulting c‑number and mass‑renormalisation counterterms, and show that they provide the logarithmic structure required to cancel the divergence of the bare sausage diagram. The finite part of the counterterm, however, is not fixed by the internal consistency of the reduced theory alone; it must be determined by an external condition. Following the classic work of Dashen, Hasslacher, and Neveu, we impose that the two‑loop kink mass equals the exact integrability result +^2/ (16). This calibration, far from being an ad‑hoc fudge, is the modern analogue of the finite renormalisation performed by DHN, now explicitly traced back to the operator normal‑ordering of the Hamiltonian. Complete Mathematica code is provided, and all numerical checks confirm cutoff independence and the exact mass.
Kalmykov et al. (Tue,) studied this question.