Version 2.0 (erratum / self-correction). This release retracts and corrects the central claim of v1. Version 1 reported that single ore-forming systems show GOE/GUE-class level repulsion in pulse timing, interpreted as an intrinsic charge-and-release memory and unified with other deep-Earth systems. That interpretation was wrong, and we withdraw it here. What was wrong. A high nearest-neighbour spacing ratio (r-bar around 0.5 to 0.7) is two-ended: it is also produced by a narrow-marginal scalar renewal clock (a relaxation oscillator) that carries no sequential level repulsion whatsoever. A shuffle test (observed r-bar versus the same intervals randomly reordered) shows the ore-system signal is a property of the interval marginal distribution, not of their sequence: shuffling barely changes r-bar. A zero-repulsion i.i.d. gamma clock with shape set by each province's coefficient of variation, matched to finite N, brackets every observed r-bar. By the Bohigas-Giannoni-Schmit conjecture, a scalar charge-release reservoir is an integrable one-degree-of-freedom system and is therefore structurally not GOE. For the non-magmatic orogenic-gold "control", the within-deposit intervals are differences between independent geochronometers on the same deposit, so the apparent rigidity is a closed-data (compositional) artifact of the v1 normalisation, reproducible by pushing a zero-repulsion null through the same pipeline. What is retained. The superposition result survives: pooling independent ore sources drives the spacing ratio toward the Poisson value (r-bar to 0.386). This is real and is kept. Version 1 is preserved as a prior version of this record and archived inside the repository. Transparent self-correction, not silent revision.
Ruqing Chen (Sun,) studied this question.