Data from four independent superconducting gravimeters of the IGETS network on four continents are analysed: Wuhan (wu004, China, 1999–2012), Pecny (pe050, Czech Republic, 2007–2014), Apache Point (ap046, USA, 2010–2018), and Sutherland (su037, South Africa, 2000–2022) — 492 station-months in total with over 31. 7 million data points. The method measures the tidal phase difference Δκ = κ (S2) − κ (M2) between the solar (S2) and lunar (M2) tidal waves. The Moon (τ ≈ 1. 28 s ≈ 0) serves as an internal reference. Two mutually exclusive predictions are compared: the geometric model (τ=0, predicts Δκ=0) and the retarded model (τ≈499 s, predicts Δκ≈+498 s). Result: no expected 498-second phase shift was detected at any station. The retarded model is rejected from 20. 4σ (Sutherland) to 137σ (Apache Point). After correction of residual systematics, Δκₜrue ≈ 0 for Pecny and Apache Point. Analysis of the total solar eclipse of 22 July 2009 (Wuhan) confirms the main result: in the ±30-minute window the geometric model (τ=0) is better on 84% of the time. The result establishes an experimental fact: the tidal signal shows no observable retardation of the order of 498 s in any of the four independent observation series. All source data are publicly available (IGETS open archive). The reproducibility package (Python pipeline, bootstrap with fixed seed=42) is included.
Yurii Trokhimchuk (Mon,) studied this question.