← Gallery

Measured, not computed

The binding energy is a measurement. RealNucleus computes one. That difference is the whole basis for judging the model's scientific merit.
What this is. A summary of a long discussion on the *epistemic status* of nuclear binding energy — and why keeping "measured" and "computed" strictly apart is fundamental to judging RealNucleus. Two-edged, as these notes always are: it credits the real distinction the model rests on, and marks the one place the "it's just a name" reading overreaches.

The binding energy is measured

The procedure is pure measurement plus arithmetic — no nuclear model of any kind:

  1. Measure the proton mass (Penning trap).
  2. Measure the neutron mass (via n + p → d + γ — a measurement chain).
  3. Measure the mass of the nucleus (Penning trap).
  4. Subtract: $B = (Z\,m_p + N\,m_n - m_{\text{nucleus}})\,c^2$.

Every input is a weighed mass. No strong force, no shell model, no liquid drop, no geometry, no fitted parameter enters. So the binding energy is an empirical fact about the nucleus — a measured mass deficit — not the output of any theory. There is no more reason to doubt it than to doubt a weight read off a scale.

One honesty note: it is energy, not merely a label

The identification "mass deficit = energy" is not an arbitrary naming. It rests on E = mc², verified directly to ~10−7 (Rainville 2005), and the binding energy is independently measurable as an energy: the deuteron's 2.2246 MeV "missing mass" is also the measured γ-ray energy; larger nuclei give it as photodisintegration thresholds and reaction Q-values (product kinetic energies). Mass-route and energy-route agree. So the deficit is a cross-checked, doubly-anchored energy — "binding energy" is the right name, not a rebranding.

RealNucleus computes one

RealNucleus does something categorically different: it sets up a model (unit charges in domains, Coulomb only) and computes a binding energy — measuring nothing about the nucleus itself. It is a genuine predictive, explanatory theory: it attempts the why (a mechanism), where the mass deficit gives only the what (the value). Standard nuclear models (liquid drop, shell, χEFT) also compute binding — so "computing it" is not unique to RealNucleus; the distinction there is parameter economy (RealNucleus ~1 vs 5–30). But against the measurement, the distinction is stark and clean: one is weighed, the other is derived from a model.

This is why the test is fair and unforgiving. RealNucleus's binding is a computed prediction; the benchmark it is judged against is a measured, model-free fact. The target is solid empirical ground, so the model cannot hide behind theory — it either lands on the measured number or it does not. That is the ideal scientific setup, and it is exactly what makes the verdict trustworthy.

The verdict this test delivers

nucleiRealNucleus (computed)measuredreading
alpha-conjugate (He-4, C-12, O-16)~107%, parameter-freeexactreal, impressive
He-3, tritium, Li-6 (odd)21–248%, geometry-dependentexactundetermined

Against the measured floor, RealNucleus wins cleanly on the alpha-conjugate line — one calibration, pure Coulomb, ~107% — and is structurally undetermined off it: for the odd nuclei the computed binding swings by a factor of ten with the assumed geometry, and the tritium/He-3 ratio comes out ~2.3 where the measured near-degeneracy is ~1.1. The sweet spot is the even alpha-conjugate closed shells — not the magic numbers (it gets non-magic C-12 right) and not mere N=Z balance (it misses balanced Li-6).

The honest bottom line: RealNucleus is a parameter-light theory that computes binding from Coulomb geometry, judged against measured mass deficits — the firmest benchmark physics offers. Its wins are unhidden because nothing was fitted; its misses are unhidden because the target was measured, not modelled. That asymmetry — computed model vs. weighed fact — is the correct and fundamental frame for judging its scientific merit: real and remarkable on the alphas, open and unfitted everywhere else.

← Back to the Gallery · companion: computability and the only arbiter