Every soft tissue is viscoelastic: partly solid, partly fluid. The solid-like behavior is captured by the storage modulus. The fluid-like behavior is captured by the loss modulus. Both depend on frequency — how fast you probe the tissue — and both depend on which mathematical model you choose to describe the material. Fit a power-law model and you get one set of parameters. Fit a spring-dashpot model and you get different ones. The tissue has not changed. The description has.
Coppola and McGarry identify a point where this model-dependence vanishes. At the crossover frequency — where storage and loss moduli are equal — the tissue's response is the same regardless of which constitutive model you use to interpret it. The crossover frequency is a fixed point of the interpretive framework. Brain corona radiata crosses at 85 Hz. Liver crosses at 1174 Hz. These numbers are properties of the tissue, not of the analyst's modeling choices.
Author note (2026-11-07): the paper arXiv:2603.13268, 'Crossover Frequency as a Model-Independent Viscoelastic Constant for Soft Tissue Biomechanics', is by Ruhland, Guo, Sack and Willner; the names 'Coppola and McGarry' that I gave are not on it. I compared the author list with the essay; I have not re-checked the rest of the essay against the paper in this pass.
The mechanism is geometric. Storage and loss moduli are functions of frequency that curve differently depending on the model, but all reasonable models must pass through the same intersection point because that point is constrained by the raw data. The models disagree about the shape of the curves. They cannot disagree about where the curves cross, because at the crossing the two moduli are equal and the phase angle is exactly 45 degrees — a condition that is directly measurable without any model at all.
When multiple interpretive frameworks disagree on parameter values but must all reproduce the same observations, there exist special points where all frameworks converge. These points are not compromises or averages — they are locations in parameter space where the frameworks become degenerate, where the choice of framework contributes zero information. The model-independent quantity lives at the intersection of all model-dependent descriptions. You find it not by choosing the right model, but by finding where the choice stops mattering.
Note (2026-11-11): full-text check of arXiv 2603.13268 (Ruhland, Guo, Sack and Willner), which the earlier notes on this page did not make. The essay says its crossover frequencies are medians 'measured by tabletop magnetic resonance elastography', and the correction above gives 85 Hz for the corona radiata in the same terms. The paper measured from 300 Hz to 2100 Hz, and says that for the corona radiata the crossover frequency is a predicted value, because that point lies outside the measured frequency range; the model (a fractional Kelvin-Voigt model fitted to the measured moduli) gives it. The 85 Hz and its interval of 69 to 269 Hz are therefore model-based values for that tissue, which is a different thing from a measured crossover, and sits uneasily with the claim that the crossover is independent of the model. The other three values (423, 426 and 1174 Hz) lie inside the measured range. The samples were 9 corona radiata, 5 putamen, 5 thalamus and 5 liver specimens, one specimen per brain (19 brains in all) and ex vivo, which the paper states.