Jeans-JH: Mathematician review (Step 2)

Reviewer: PhD mathematician. Scope: derivations, notation, units, hidden assumptions. Physical verdicts belong to the physicist; where a mathematical result decides a verdict I say so.

Source note: no Source gap (all three books have text), but the equations in all three are OCR-garbled and were read from page images by the deep readers. I re-read these page images myself: Atomicity PDF pp.25, 28, 32 (the h-K formula, the half-retarded plus half-advanced potentials, the two-atom solution and the "source and sink" sentence); Gases PDF p.221 (eqs. 493-495 and the italic "small" sentence); Electricity PDF p.587 (sections 659-660). All three match the deep reads. [read: page images]

Tags: [read] source document; [calc] my own computation (script run in the project venv; numbers below); [web] with URL; [memory] with confidence.

Verdict summary

Preliminaries I verified once and use below

1. Gases, Ch. VII: gamma = 1 + 2/n (eq. 415) with n counting all momentoids including the three translational ones; Ch. IX: gamma = 1 + 2/(n+3) (eq. 522) with n the internal ones. Both are correct for their own n, and both are the same formula. The change of meaning of n between chapters is a notational hazard (claims.md and facts.json use both forms). [read: deep reads; eqs. 415, 522]
2. Jeans's R is the gas constant per molecule, that is Boltzmann's k. [read: Gases deep read, symbol list; and Lodge 1926 in the Investigator's files credits Jeans with "the 8 pi R factor"]. So (8 pi R T/c^3) nu^2 d nu is the Rayleigh-Jeans law with k, and exp(-h nu/RT) is exp(-h nu/kT); nothing is mixed (see Corrections).
3. Gaussian units throughout the 1911 and 1925 books: flux of a charge e is 4 pi e; alpha = e^2/(hbar c). [read; memory high]

---

Claim 1: Real gases do not follow equipartition because their molecules radiate (1904)

a. Mathematically consistent? Partly. - Consistent: the specific-heat bookkeeping. gamma = 1 + 2/n (415); for thousands of spectral lines n is thousands and gamma is indistinguishable from 1 (my worked number: 1000 lines as 2000 extra quadratic terms gives gamma = 1 + 2/2003 = 1.0010 against 1.667 observed for argon). [read: deep read sect. 1; calc] - Consistent: the replacement of Liouville's theorem, D rho/Dt = rho Delta F (433), with Delta F > 0 when E and F are positive-definite quadratics. That is exactly the statement that a dissipative flow contracts phase-space volume; the "cluster" is an attractor. Fine as mathematics. [read: p.196-197] - Weak: the step from "points concentrate on one normal state" to "the normal state has the internal modes nearly empty". Jeans admits in section 234 that the general theorem gives only that a normal state exists; emptiness of the internal modes is shown only for two toy models (two vessels; loaded spheres with a friction term, eq. 468: epsilon H = (1/3) beta nu K^(3/2)). The 1904 book says so itself: "we must be content to rest the remainder of our investigation upon hypothesis". [read: deep read sect. 7] - Weak: section 219, the cubic terms neglected in eq. 445 are of the same order as the retained term (Jeans's own admission). The "infinite n" limit rests on order-of-magnitude estimates (eq. 441). [read: p.202] - Hidden assumption that decides everything: the ether is a zero-temperature sink (the far face of the wall in the two-vessel model is "kept at a zero temperature"), so alpha (radiative loss) never has a matching gain. A gas in an enclosure at its own temperature absorbs what it emits, and the steady state is equipartition again. Jeans concedes this in print: his Nature reply (27 Apr 1905) says equipartition "may be the true final ratio" and that the transition "is very slow ... millions of years would hardly suffice". [read: Investigator file, files/montambaux/04; [web] as logged in research/2026-10-08-montambaux.md] So the 1905 claim is a time-scale claim, and the strong form in claims.md (resolution of gamma by non-conservation) is not what he finally defended. - A true special-case failure that is mathematically instructive: section 233, the epsilon -> 0 limit does not reproduce equipartition (the normal-state curve H^2 proportional to K^3 does not tend to 2K = 3H). This is a singular limit and is correct as a statement about those equations. [read: p.212]

b. Consistent with experiment? The gamma tables (Hg, Kr, He, Ar near 5/3; diatomics near 7/5) are what any theory must reproduce, and they are correct data. The tables do not discriminate between Jeans's mechanism and quantum freeze-out. His own caveats: Ar gives n+3 = 3.20, He 3.07 (rounded to 3 without comment); CO2 6.98, N2O 6.96, ethylene 7.95; Cl2 6.08, Br2/I2 about 6.8 break the diatomic rule of 5, and Jeans hands them to chemists. [read: deep read sect. 9.5, 10] Predicted fall of gamma with T (Leduc: CO2 1.3190 at 0 C, 1.2827 at 100 C) is borne out, but dissociation, aggregation and, in quantum theory, thermal excitation of vibrations all predict that sign too, so it does not select his mechanism. [read: p.244; memory high on quantum] Real measured vibrational relaxation times: CO2 (nu3 mode) about 7 microseconds at 1 atm and room temperature [web: https://pubs.rsc.org/en/content/articlelanding/1966/tf/tf9666202655 and https://pubs.rsc.org/en/Content/ArticleLanding/1967/TF/TF9676302093, found by search, abstracts only]. A transfer time of microseconds means a calorimeter or a sound wave of ordinary frequency would see the vibrational heat capacity of CO2 if the molecule were classical and in equilibrium; the low-temperature absence of that heat capacity therefore cannot be a slow-transfer effect. It is the quantum freeze-out (kT much smaller than h nu), not a rate effect. [calc/memory high on the quantum side]

c. Tested? Jeans's mechanism for equilibrium specific heats was not tested in his lifetime; it was overtaken. By 1916 he writes in the 2nd edition that the quantum theory has dissipated the specific-heat "cloud", "at best only partial, but ... probably ... satisfactory" and leaves the classical difficulties "unsolved". [web: Investigator, archive.org dynamicaltheoryo00jeanuoft, research/2026-10-08-archive-org.md] Later mathematicians proved a rigorous classical result of the weak form: exponentially slow energy exchange for fast modes in model systems, the "Boltzmann-Jeans conjecture" (Benettin, Galgani, Giorgilli 1989; Benettin-Carati-Gallavotti 1995). [web: Investigator, files/benettin-carati-gallavotti-1995-jeans-landau-teller.pdf, research/2026-10-08-unimi-boltzmann-jeans.md] They prove time scales, not a different equilibrium.

d. If true? Equilibrium statistical mechanics would have to fail for systems coupled to radiation (classical canonical equilibrium would not give equipartition). Mainstream instead says: equilibrium heat capacities are quantum; classical equipartition is an approximation valid for kT much greater than level spacing. Jeans's 1904 view survives, mathematically, only as the adiabatic-invariant freezing of fast modes over finite times.

Verdict: split, as in the summary table. Strong form Historically superseded but instructive. Weak form Established for model systems (mathematically proved by others; Jeans's contribution is the conjecture and the exponent). As an account of real gases Tested, not supported (a 10^-6 s relaxation time against a claimed 10^330 years).

---

Claim 2: Slow feeding of internal modes, exp(-2 p beta) (1904)

a. Mathematics. - The forced-oscillator result is correct. For a unit-mass-like oscillator a phi'' + b phi = U(t), the mean energy gain per encounter is (1/2) b (X^2 + Y^2) = |F(p)|^2 / (2a), with F the Fourier transform of the force at the oscillator's own angular frequency p (eqs. 479-484). I checked the algebra with b = a p^2. [read: pp.214-216; calc] - The pole argument is correct. For the Lorentzian pulse U = I c / (pi (c^2 + t^2)) the transform is exactly I e^(-p c) (numerically verified: p = 1, 3, 6 give 0.36788, 0.049787, 0.0024788 = e^(-p) etc.), so the energy is (I^2/2a) e^(-2 p c) (eq. 495). For a sech^2 pulse (the force shape of an exponentially repulsive potential) the first pole is at beta = pi tau/2 and the transform is pi p tau^2 / sinh(pi p tau/2), so the same law holds with beta = pi tau/2. [calc] This is the Landau-Teller adiabatic factor of 1936. [memory high; web: Benettin et al. file] - Hidden assumptions, all stated or recoverable: U analytic with isolated poles (Jeans notes a force that is exactly zero outside 0 < t < tau cannot be analytic); no resonance (q = p terms excluded, "obviously not true" in that case); the pole nearest the real axis dominates. Jeans says "that the result is in general true there can be little doubt" and does not prove it for general U. [read: pp.217-219] - Printed slip: the italic conclusion on p.204 (PDF p.221) says dissipation is slow if the product of collision time and frequency is "small". The mathematics on the same page (e^(-2 p c)) requires "large". Page image confirmed. [read: page image p.221] Corrigenda do not list it. - Numbers, all re-checked: p = 2e15 and a collision time of 8e-15 s give p t = 16, e^-16 = 1.1e-7; with penetration one-half of the radius p t = 800, e^-800 = 10^-347.4; 10^337 s = 10^329.5 years; 1201 e^-1200 = 10^-518.07; 13 e^-12 = 8.0e-5. [calc] All match the printed values. - The decisive point: the claim of "slow transfer" rests on the choice of penetration depth. With 1/100 of the radius Jeans himself finds e^-16 "much too great to agree with observation"; only at one-half radius does e^-800 appear, and he says there is "not the slightest evidence" for the molecular-action spheres either way. [read: deep read sect. 5] The exponent is p times a collision time; it is not a prediction. For a realistic soft repulsion of range L = 0.2 Angstrom at 500 m/s-class thermal speeds the collision time is about L/v = 4e-14 s; with p = 2e15 this gives p t = 80 and a factor of roughly e^-80 to e^-250 depending on the factor pi, still absurdly small for optical modes (so optical modes are indeed isolated from collisions; this is consistent with visible light not being thermally excited) but 10^330 years is not a measurement. [calc, my own estimate; L chosen by me]

b. Consistent with experiment? The exponential suppression is real and measured in the infrared range: vibrational relaxation times grow roughly exponentially with frequency and are microseconds for CO2 at room temperature [web: Faraday Trans. abstracts above]. The "temperature of incandescence" prediction (between 0 C and 28,000 C, within a factor below 10 in speed) is a bracket spanning the whole range of interest; Jeans says himself it "may seem ... almost worthless as a confirmation". [read: p.224] Jeans's check of "28,000 C": C = 5e5 cm/s for air corresponds to 273 x 100 = 27,300 K. [calc]

c. Tested? Not as a prediction. The later rigorous theory (Benettin et al.) tests the mathematics in models, not the 1904 numbers. The Schuster-Hemsalech observation, the vacuum-tube estimate (2.2e7 cm/s, ratio 2.2e15) are qualitative agreements with the frequency dependence. [read: sect. 7] None of this was a controlled test.

d. If true? Nothing in mainstream mathematics changes: the formula is standard. What would have to change is the claim that it explains equilibrium heat capacities, which is not the case.

Verdict: exp(-2 p beta) Established. Application to freeze optical or infrared modes in real gases over geological times: Tested, not supported. Temperature of incandescence: Not testable as stated.

---

Claim 3: Kinetic theory fixes planetary atmospheres and molecular sizes (1904)

a. Mathematics, all re-derived. - Escape: x = 3 g a / C^2 with C the rms speed equals m g a/(kT) = G M m/(k T a) (since C^2 = 3kT/m). At the exobase radius 2a the usual Jeans parameter is lambda = x/2, and the isothermal density ratio e^(-x/2) times the Maxwell tail (1 + lambda) e^(-lambda) gives n_surface (1 + x/2) e^(-x): the e^(-x) of Jeans's eq. 767 and its (1 + x) are the same expression up to the bookkeeping of the factor (1 + x/2) versus (1 + x). [calc] - Orders: loss fraction 1e-2 (1+x) e^(-x) per second gives 4.9e-8 at x = 15 (half-life 0.45 yr) and 2.9e-14 at x = 30 (half-life 0.76 million years). Jeans prints "about two million years" for x = 30; the printed fraction gives 0.76 million years (half-life) or 1.1 million years (mean life). A small slip, same order, as the deep reader flagged. [calc] - Planet table g a ratios (M/a): all agree with the printed values (Sun 3059, Moon 0.048, Mercury 0.174, Venus 0.825, Mars 0.208, Jupiter 27.3). [read; calc, deep reader] - Retention depends on a much sharper requirement than "millions of years": for the age of the solar system (4.5e9 years) the loss fraction must be below about 1e-17 per second, needing x near 40 on Jeans's own formula. At x = 30 his criterion is a million-year criterion, not a retention-over-the-age-of-the-planet one. [calc: x = 40 gives half-life 1.3e10 years] - My x values at stated temperatures [calc; surface radius as Jeans uses]: Earth at 500 K: H2 30.3, He 60.2; Earth at 1000 K: H2 15.2, He 30.1, N2 211; Moon at 273 K: H2 2.5, He 5.0, H2O 22.4, N2 34.8, CO2 54.7; Moon at 390 K: N2 24.4; Mars 210 K: H2 14.6, He 29.0; Mercury 440 K: H2 5.0, He 9.9, N2 69. Jeans's statement for the Moon is conditional ("at a temperature much above zero") and is right as written: at 273 K N2 and CO2 are retained by his own rule, water vapour is lost; at a lunar daytime 390 K N2 falls in the gray band. Mars retains N2, H2O, CO2 and loses H and (marginally) He: as Jeans says. - Molecular size: the viscosity, conduction, diffusion and Boyle's-law radii (1.0-2.3e-8 cm) are standard kinetic-theory order-of-magnitude numbers; the solid/liquid-density upper limits I could reproduce to within 0.06 except ethyl chloride (the deep reader's computation). The Mossotti equation as printed, mu - 1 = (1/2) pi N sigma^3, lacks a factor 2 relative to the number he uses: with N = 4e19 and the printed relation the radius is 0.836e-8 cm; with K - 1 = 2(mu - 1) it is 1.054e-8 cm as printed. A misprint in the equation, not the number. With the modern N = 2.69e19 the radius would be 1.20e-8 cm [calc; memory high on N]. The hydrogen molecule mass 2.25e-24 g (true 3.35e-24 g) propagates into the density-limit radius.

b. Experiment. Observations of the Moon and Mercury (essentially airless) agree with the loss rule; Jeans escape is accepted today as one mechanism. Present-day Earth loses hydrogen at about 3 kg/s with Jeans escape contributing about 10-40 percent (the rest charge exchange and polar wind) [web: https://en.wikipedia.org/wiki/Atmospheric_escape]. Jeans's premise that Earth's outer atmosphere is "certainly less than 230 C" is wrong: exospheric temperatures range from about 740 to 1350 K [web: https://en.wikipedia.org/wiki/Thermosphere]. At 1000 K his own rule gives x(H2) = 15.2, at his "appreciable loss" edge, which is the modern conclusion that Earth does lose hydrogen. The mathematical machinery was right; an input temperature was not.

c. Tested? Observationally yes by later planetary science and by direct kinetic-theory measurement of transport coefficients; replications are numerous. Stoney's dissent (Phil. Mag., June 1904) is reported by Jeans and left unanswered. [read: p.342] Single-source on this page; the letter itself is not in the folder.

d. If true? Nothing changes; it is mainstream. Corrections are refinements (Bryan's, Chamberlain-type exospheric work).

Verdict: Established as a method and order of magnitude. Premise on Earth's outer atmosphere temperature and the value of N are Historically superseded but instructive. The molecular radii are Established only as order of magnitude (hard-sphere model; Jeans says "almost worthless except as regards order of magnitude" for some vapours).

---

Claim 4: The electron is purely electromagnetic; bodies physically contract (1911)

a. Mathematics. Checked item by item (the deep read's SymPy checks, plus mine). - Electromagnetic mass: magnetic energy outside radius a, (e^2 u^2/8 pi C^2) integral of sin^2 theta r^-2 dr ... = e^2 u^2/(3 a C^2), so T = (1/2) m u^2 with m = (2/3) e^2/(a C^2). [read; calc series: T = e^2 beta^2/(3a) + 2 e^2 beta^4/(15a) + 3 e^2 beta^6/(35a), matching the printed coefficients (1/3, 2/(3.5), 3/(5.7))]. Electric energy W of eq. 678 tends to e^2/(2a). a = 2e^2/(3 m C^2) = 2.00e-13 cm with e = 4.65e-10 esu and m = 8e-28 g. [calc] - Hidden inconsistency not noted by Jeans: the electrostatic self-energy is U = e^2/(2a), and Einstein's mass-energy relation would give mass U/C^2 = e^2/(2 a C^2); Jeans's mass from the kinetic energy (and from the field momentum) is (2/3) e^2/(a C^2). The ratio is exactly 4/3 [calc]. Mass from the energy and mass from the momentum of the field disagree. This "4/3 problem" is the reason Poincare (1905-06) introduced non-electromagnetic stresses to hold a Lorentz electron together; once those stresses carry energy, "m0 = 0, all mass electromagnetic" no longer follows. Jeans's own text asserts both m = m0 + (2/3) e^2/(a C^2) as the low-speed mass and E = m C^2 as a universal exchange rate in 1925 without noticing that they conflict for a charged shell. [read: deep reads; memory high on Poincare] - Larmor radiation formula: emission rate (2/3) e^2 (acceleration)^2 / C^3, with the standard angular integral (sin^2 theta over the sphere gives 8 pi/3). Correct. Jeans's printed field X = [e u'']/(C^2 r) drops the angular and transverse factors (and the potential gradient), but the Poynting integral gives the standard total. [read: deep read 3d; calc] - Biot-Savart element law (section 624): the field of a moving point charge at low speed, e u sin theta/(C r^2), summed over a current element, is exactly the Biot-Savart element formula. Jeans's claim that this proves the element law "for every element" rather than only round a closed circuit is right for the field of an element (the superposition over moving charges is exact at low speed). It is not a proof that the Ampere force between two elements is correct; Ampere's, Grassmann's and Biot-Savart-Lorentz element forces differ by terms that vanish on closed circuits, and Lorentz force between charges selects one. [memory high] This distinction matters for any project that uses Ampere-type force laws. - Section 659-660 derivation is correct as far as it goes: if a purely electrostatic system is in equilibrium at rest then its contracted counterpart (x' = x (1 - beta^2)^(-1/2)) is in equilibrium when moving at u; E-field components are X proportional to the rest-frame value and Y larger by gamma, the standard transformation. [calc: consistent with the printed forms up to the sign convention as printed]. What does not follow: that this equilibrium is "inevitable". The proof shows that S' is an equilibrium configuration, not that a system set in motion reaches it (it needs stability and the dynamics of acceleration), and it needs the premise "all forces in the ether are electrical" (all molecular forces transform like electrostatic forces). Jeans states that premise explicitly, so the claim is conditional. [read: page image p.587] - Michelson-Morley (section 661): t1 = (2l/C)(1 - u^2/C^2)^(-1) with longitudinal contraction becomes t1 = (2 l0/C)(1 - u^2/C^2)^(-1/2) = t2 for equal rest lengths. Correct algebra. "Almost positive proof": it is a consistency, not a proof. Contraction of the longitudinal arm alone is not the only way to get a null for equal arms, and with unequal arms the contracted-arm hypothesis alone does not give a null at all velocities; the Kennedy-Thorndike experiment (1932) is the standard test, showing that contraction alone is not enough and a matching time effect is needed. [memory high; not web-checked this run] So "almost positive proof" overstates what the algebra shows; I flag this as a hidden-assumption point (equal arms). - Lorentz's deformable electron masses as printed, longitudinal m0 + (2e^2/3aC^2)(1 - beta^2)^(-3/2), transverse with exponent -1/2, are exactly what follows from field momentum p = (2e^2/3aC^2) gamma u (so transverse = p/u, longitudinal = dp/du). [calc] They are the same factor-gamma form as special relativity, which is why Bucherer's agreement with them cannot by itself show that the mass is electromagnetic: the data fit total mass times (1 - beta^2)^(-1/2) whatever its origin. [memory high] - Misprints and slips in the 1911 text (all confirmed): .33 for .033 esu in section 615; "even" for "odd" for the undisplaced Zeeman line (section 635); "contracted" and "uncontracted" swapped (section 657); "In 1908, Kauffmann" against a footnote to Annalen vol. 19 (1906). The last is confirmed: Kaufmann's Ann. Phys. 19 paper is 1906 and Bucherer's 1908 [web: https://en.wikipedia.org/wiki/Kaufmann%E2%80%93Bucherer%E2%80%93Neumann_experiments]. The strictly printed Zeeman restoring force "-p^2 x" lacks the factor m (read as m p^2). Also: Jeans uses "local time" for the retarded time t - r/a, not Lorentz's t - v x/c^2. [read; deep read]

b. Consistent with experiment? Zeeman delta p = eH/(2mC) (normal triplet) is verified (anomalous Zeeman needs spin, not in 1911). Hall effect: confirmed. Radiation of accelerated charge: confirmed. Radiation pressure 4.3e-5 dyne/cm^2 for sunlight: Lebedew and Nichols-Hull 1901, correct. [read; calc: pressure = energy density]. Bucherer/Kaufmann: the 1908 Bucherer result was disputed by Bestelmeyer for years; decisive precision came from later experiments (Neumann 1914, Glitscher 1917 on hydrogen fine structure, Rogers 1940). [web: Wikipedia KBN page above] Jeans writes "found to agree exactly" and "Abraham must be discarded": stronger than the data bore at the time.

c. Tested? Yes: Kaufmann (1901-06), Bucherer (1908), Bestelmeyer (1907), Neumann (1914), Rogers (1940). Replication outcome by modern standards: relativistic mass increase confirmed; purely electromagnetic origin not supported. Michelson-Morley null: confirmed repeatedly.

d. If true? A real ether with real contraction and a purely electromagnetic electron would require that special relativity's kinematic reading be wrong; but it is mathematically equivalent in all observable predictions (Lorentz ether theory), differing only in interpretation. The EM-mass part would require that no non-electromagnetic stresses exist, which fails the 4/3 problem.

Verdict: sub-claims as in the summary. Overall: Historically superseded but instructive; the derivations in Chapters XIX-XX are sound mathematics, the conclusions are those of 1911 Lorentz ether theory.

---

Claim 5: Energy atomic, quantisation in matter-radiation exchange, free radiation wave-like (1925)

a. Mathematics, piece by piece. - Sun: L/c^2 = 4.26e9 kg/s = 2.56e8 tonnes per minute; Jeans's "250 million tons a minute" is correct. [calc] But the data are a rate of continuous loss. "Atoms of energy mC^2" is an inference from "matter is atomic": a loss rate does not demonstrate discreteness. Non sequitur as presented. - Printed "rate of exchange taken to be C" where C^2 is meant: a one-power slip (OCR or print), noted by the deep reader. [read] - Hydrogen collapse: with Larmor's formula, dr/dt = -(4/3) r_e^2 c / r^2, so the initial shrink rate at the Bohr radius is 1.13 m/s (Jeans: "over a metre a second", correct) and the lifetime is a0^3/(4 r_e^2 c) = 1.56e-11 s. [calc] Jeans's "about 10^-10 s" is high by a factor of 6; fine as an order of magnitude statement, and he does not derive it. Established by calculation (accepted laws applied correctly). - Quanta in sunlight and starlight: sunlight energy density 4.3e-5 erg/cm^3 divided by 3.7e-12 erg per visible quantum gives 1.2e7 per cm^3 ("ten millions"); 6th magnitude star, about 1e-10 W/m^2, gives about 1 quantum per m^3. [calc] Wien: 1/(5.4e7) s gives c tau = 5.55 m ("several metres"); a million wavelengths of H-alpha is 0.66 m ("two feet"). [calc] All arithmetic correct. - Free electron cannot absorb or emit a quantum (sect. 8): the conclusion is correct and standard. The cleanest proof is four-vector: (P_e + P_gamma)^2 = P_e'^2 gives P_e . P_gamma = 0, impossible for non-zero frequency. Jeans's energy-only argument is valid in this sense: if the energy balance (1/2) m v^2 + h nu = (1/2) m v'^2 holds in every inertial frame, one may choose the frame in which v' = 0. A boost by w changes the balance by (h nu_w - h nu) + m w (v' - v); this vanishes only if m (v' - v) = (h nu/c) n (photon momentum) and the Doppler shift is first order. [calc: symbolic expansion] So "the energy rule holds in every frame" is not an independent premise: it is equivalent to momentum conservation with photon momentum h nu/c. The theorem is true because radiation carries momentum h nu/c; the argument therefore imports a quantum (particle-like) property that Jeans elsewhere denies to free radiation. The step is correct; the claim that only energy is needed is slightly misleading. The deep reader's remark that the frame choice does the work of momentum conservation is refined by this. - "A radiation field produces no acceleration in a free electron" (section 8c): does not follow. Absorbing a whole quantum is forbidden; but the classical equation m dv/dt = e E0 cos(omega t) gives a quiver v = (e E0/m omega) sin(omega t) with zero net work per cycle, so classical theory also has zero net absorption by a free electron, and the electron still accelerates (and scatters: Thomson scattering, the low-frequency limit of Compton). Jeans conflates "no net absorption" with "no acceleration". [calc/memory high] He printed the same statement a year earlier (1924 Report section 87: the electron "cannot feel the force Xe"), so it is a settled position of his, not a slip. [web: Investigator, research/2026-10-08-archive-org.md] - Wave-or-bullet dilemma: "If quanta were feet or metres in length, they would not be indivisible. Every click of a camera shutter would cut several in halves." The hidden assumption is that "indivisible" means "spatially small" and that a quantum must be a localised parcel. A single-quantum state can be an extended normalised wave packet whose detection (not whose energy) is all-or-nothing; the dilemma has a third alternative which was not available in 1925 mathematics. [memory high; not checked on the web]. The coherence-length numbers themselves are fine. The Compton-Simon and Bothe-Geiger experiments are dated April 1925 (before the 11 May lecture) and 1 September 1925 (after). [web: Investigator, Wikipedia as logged in research/2026-10-08-google-and-other-sites.md] - Relation of "reversible implies equal weight to convergent waves": the wave equation is time-reversal invariant, the selection of divergent solutions is a boundary condition, not a law. Jeans's logic here is sound. [read: p.26-28]

b. Experiment. Photoelectric threshold and one-electron-per-quantum, Compton shift (h/mC)(1 - cos theta) = 2.43e-12 m (1 - cos theta), and the energy-momentum correlation (Compton-Simon, Bothe-Geiger) are all explained by quantised radiation in modern theory. Jeans's own text concedes that "experiments are [not] as yet precise enough to decide" on Compton. [read: p.31]

c. Tested? The astronomical part was superseded by hydrogen fusion (Eddington, 1920 suggestion; [memory high]); the cavity-radiation, Einstein-law and Bohr parts are the mainstream quantum theory of 1925.

d. If true? The strong form (no photons in free space, quantisation only in exchange) would require QED-type fluctuation phenomena (antibunching, single-photon interference with all-or-nothing detection) to be explained without field quanta.

Verdict: split as in the summary. The recognised-correct core of this lecture is that a free electron cannot absorb or emit one quantum.

---

Claim 6: Time-symmetric field, non-radiating atom, whole quantum from emitter to absorber (1925)

a. Mathematics. - The wave equation solution phi = (1/r){F(t - r/a) + Phi(t + r/a)} satisfies it for any F, Phi (SymPy by the deep reader). With Phi = F, near the origin phi tends to 2F(t)/r, and matching to e/r gives F = e/2, which is where "([rho] + {rho})/2r" comes from. Page image confirmed. [read: page image p.28] - Zero mean radiation: for the half-sum field the far-field Poynting flux is (A^2/r^2) [cos^2(omega t - k r) - cos^2(omega t + k r)] = -(A^2/r^2) sin(2 omega t) sin(2 k r), whose time average is exactly zero, with an instantaneous ebb and flow, matching Jeans's words. [calc: symbolic mean = 0; advanced wave's B sign reversed]. This agrees with Dirac's 1938 result that the radiation-damping force on a charge is half the (retarded minus advanced) field, which vanishes for the half-sum field [memory high]. Established. Limits: strictly periodic (or recurrent) motion; infinite field energy and no selection of Bohr orbits, which Jeans lists himself. [read: p.29] - Two-atom solution Psi = integral [rho]/r + integral {rho'}/r': it is a legitimate solution of the inhomogeneous wave equation for the two prescribed charge systems (a sum of a retarded and an advanced solution). Mathematical limits Jeans does not state: 1. A solution for prescribed sources is not a solution of the coupled field-and-charge dynamics. With a purely retarded field the emitter feels radiation damping and loses energy, and the absorber, with a purely advanced field, gains it. For bound atoms in steady "similar motions" that means the energy loss must be replenished by something outside the equations; Jeans assumes similar motions. 2. The sphere-averaged flux vanishes only for equal strengths. The hydrodynamic "source and sink" analogy is not exact for waves. In the scalar analogue, the retarded wave from atom 1 alone gives outward flux 1/r^2 in all directions at large r, and the advanced wave from atom 2 alone gives equal inward flux; the cross terms oscillate with angle. So at large distance there is outgoing and incoming energy everywhere, cancelling only in the integral, not locally. A true source-sink flow field is confined and falls off as a dipole. Jeans's sentence "gradually drawn in ... finally concentrates wholly upon it" is a narrative that the stationary formulas do not display; no calculation of the energy flow is given. [read: page image p.32; calc/analysis, my reasoning on the scalar analogue, not a numerical field plot] 3. The model needs the two atoms' periods to coincide (resonance). Jeans concedes that in nature only for very low frequencies does any period of an atom coincide with the radiation emitted. [read: p.33-34] - Compton alternative: no equations are given; the "small change in the coefficient h/mC" is not specified; the check cannot be done from the lecture. [read: p.30-31]

b. Experiment. The scheme keeps event-by-event energy conservation (whole quanta), unlike Bohr-Kramers-Slater. Whether a spreading emitted quantum fits Compton-Simon's directional correlation is unanswered in the record. Tolman's 1926 review found the experiments "greatly increased the difficulties of the wave theory", states the scheme as requiring "convergence of spherical waves into an absorbing atom", and does not say Jeans erred. [web: Investigator, Tolman, Physical Review 1926 p.1048, via Google Books, research/2026-10-08-reviews-1926.md] Loring (Chemical News, 1927) took the convergent-wave photoelectric idea as "promising" while noting Jeans did not offer it as a solution. [web: Investigator]

c. Tested? No test of the scheme as such is on record; neither dismissed nor tested. It is part of a lineage (Tetrode 1922, Lewis 1926, Fokker 1929, Wheeler-Feynman 1945) whose standard histories do not mention Jeans 1925 (Wikipedia; Kennefick 1997 has zero mentions). [web: Investigator]. That silence is not a contradiction. Related modern work using a retarded wave from a source plus an advanced wave from a sink (Zhao, "mutual energy flow", arXiv:1606.08710) is outside the mainstream journals; I list it as related work and have not evaluated its mathematics. [web: Investigator, not checked by me]

d. If true? Electrodynamics would need an absorber-conditioned boundary rule (future boundary condition) in place of purely retarded potentials, and the quantisation of the atom would still have to come from somewhere (infinite field energy, no selection of orbits).

Verdict: Zero mean radiation of the half-sum field: Established. Two-atom source-sink as the mechanism of quantum absorption: Never properly tested (no neutral test exists, not dismissed without a test; Jeans offers it as a conjecture that "has nothing to fear"). Compton alternative: Not testable as stated.

---

Claim 7: h C / 2 pi = K (4 pi e)^2, quantum from charge atomicity (1925, conjecture)

a. Mathematics. hC/2 pi = hbar c, and in Gaussian units hbar c = e^2/alpha, so K = 1/(16 pi^2 alpha) = 0.8678 with alpha = 1/137.036. [calc; page image p.25 shows the equation exactly] The relation is the fine-structure constant rewritten; since hbar c and e^2 have the same dimensions, some dimensionless K must exist. The mathematical content of Jeans's remark is only that 16 pi^2 = 157.9 is within 15 percent of 137.04 (K is 13 percent below 1). A fair test of how surprising that is: of 88 products 2^a pi^b (a from -4 to 6, b from -3 to 4) only 1 lies within 20 percent of 137 and 2 within 20 percent. [calc; the family is my arbitrary choice]. So the near-coincidence is not trivially cheap in that family, but a 13 percent miss in a dimensionless number is not an identity, and no derivation supplies the 16 pi^2.

The same relation appears a year earlier in the 1924 Report with the added caution that the atomicity of e "will not lead to the quantum-theory, for if it did the quantum-theory would have been fully developed long ago" (OCR loses the prefactor). [web: Investigator, research/2026-10-08-archive-org.md] The 1925 text says "this suggests" and "it seems possible": a conjecture, labelled so by him. [read: page image p.25-26]

b. Experiment. alpha is measured to about 10 digits; the statement "alpha is about 1/(16 pi^2)" is false at 13 percent level, so K is not 1. Why alpha is near 1/137 is unexplained in mainstream physics. [memory high]

c. Tested? Not testable; no mechanism.

d. If true? Would require a derivation of alpha from charge quantisation (and the tubes-of-force picture, which Jeans himself rejects as physics on the evidence of beta-particle scattering). [read: page image p.25]

Verdict: Established arithmetic, Not testable as stated as physics.

---

Investigator findings that bear on the mathematics (kept whatever the verdict)

1. Jeans, Nature 71, 607 (27 Apr 1905): equipartition may be "the true final ratio"; the claim is time scale ("millions of years would hardly suffice"). Weakens the strong form of claim 1 in claims.md. [web, research/2026-10-08-montambaux.md]
2. Jeans, Dynamical Theory of Gases 2nd ed. 1916: hands the specific-heat problem to quantum theory. His own revision of claims 1-2. [web: archive.org dynamicaltheoryo00jeanuoft]
3. Jeans, Report on Radiation and the Quantum-Theory (1914 and 1924): the slow-transfer argument restated ("millions of years", citing Gases Ch. IX); the 1925 lecture's Lorentz 1906 and Taylor 1909 (2000 hours) arguments and the h-(4 pi e)^2 relation already in print with references; no advanced or convergent waves in 1924, so the time-symmetric step is new in 1925. [web, research/2026-10-08-archive-org.md]
4. Jeans, Mathematical Theory of Electricity and Magnetism, 4th ed. (1919) adds relativity, 5th (1927) keeps the Bucherer reading "wholly electromagnetic" (a = 1.874e-13 cm). [web]
5. Boltzmann-Jeans conjecture and the Jeans-Landau-Teller approximation: rigorous results (Benettin-Galgani-Giorgilli 1989; Benettin-Carati-Gallavotti 1995 in files/) proving exponentially slow energy exchange in model systems. This is independent mathematical support for claim 2's mechanism; it says nothing about equilibrium heat capacities. [web]
6. E. P. Adams's 1901 paper (Phil. Mag. [6] 2, 285): Harvard, two sets of sixteen copper spheres, 75-85 rev/s, measured magnetic effect of convection, not radiation; Jeans's radiating-spheres argument is his own calculation, and the lecture's "brass spheres" is slightly inaccurate. [web] Mathematically this means the premise "the spheres radiate" in claim 6 is a calculation by Jeans using the Poynting flux, not an Adams measurement.
7. Reviews by Adams (Bull. AMS 1926; favourable, brief, by the experimenter and a colleague), Tolman (Phys. Rev. 1926), Loring (1927), Lodge (Nature 118, 81, 1926: credits Jeans with the 8 pi R factor). [web] Hudson 1989 dates Jeans's move to the quantum side to 1910. [web]
8. Leads not obtained: Milne 1947 obituary, Nature 117, 46 review (paywall), G. N. Lewis 1926 (scan), Ritz-Einstein 1909 note, Royal Society JHJ/2 letters (1901-1907, written while he finished the gas theory), HRC Richardson letters. These would bear on how Jeans himself weighed the dissipative theory and the 1925 scheme. Absence of a second source for the 1925 absorber idea in later histories is a silence to analyse (not a contradiction); the Investigator's records on Tetrode and Ritz could decide whether Jeans knew of the lineage.

---

Corrections to Step 1 (and to earlier documents), for the orchestrator

1. deep/AtomicityQuanta-df752a.md, section 2c (the Boltzmann factor), says Jeans's RT is "per mole units used loosely" and mixes R and k. Wrong. In the Gases deep read symbol list R is the gas constant per molecule (modern Boltzmann's k), and in Atomicity the formulas (8 pi R T/c^3) nu^2 d nu and exp(-h nu/RT) are then exactly the Rayleigh-Jeans law and the Boltzmann factor. Lodge (Nature 118, 81) writes "8 pi R" in the same notation. Evidence: [read: deep/The-dynamical-theory-of-gases-b7b468.md, "R = the gas constant per molecule"; Investigator note]. Fix the 2c remark and the facts.json statement that repeats it if present.
2. deep/AtomicityQuanta-df752a.md, section 8b says the frame choice "does the work that momentum conservation does". More exactly: demanding the energy equation in every inertial frame is equivalent to momentum conservation with photon momentum h nu/c plus the Doppler shift (my boost algebra above). The free-electron theorem therefore rests on the quantum momentum of radiation. Not an error in the conclusion, a refinement.
3. facts.json (Gases entry, key claim on e^-800): "for optical frequencies and collisions of order 10^-14 s the transfer can be extremely small (e^-800)". The book gives p t = 16 (e^-16, "much too great") for 8e-15 s collisions; e^-800 arises only with a penetration of one-half of the molecular radius, a collision time 50 times longer, which Jeans says has "not the slightest evidence". Correct to state both numbers and the assumption.
4. claims.md claim 1 and claim 2: claim 1 gives "gamma = 1 + 2/(n+3)" (eq. 522, n = internal momentoids) while the same chapter's first formula is gamma = 1 + 2/n (eq. 415, n = all momentoids). Add the note that n changes meaning between Ch. VII and Ch. IX. Also add that claim 1 in its strong form was narrowed by Jeans himself in 1905 to a time-scale claim, and that observed n+3 values are not integers (He 3.07, Ar 3.20, CO2 6.98, ethylene 7.95, Br2/I2 about 6.8) and the polyatomic and halogen rows disagree with the integer rules, as Jeans admits (sections 271-273). Claim 2: note the misprint "small" for "large" in the italic conclusion of section 241 (page image confirmed).
5. deep/The-dynamical-theory-of-gases-b7b468.md (planetary atmospheres part), section 1.5: the Earth's outer-atmosphere premise ("certainly less than 230 C") is stated as Jeans's reading and is superseded: exospheric temperatures are about 740-1350 K [web: Wikipedia Thermosphere]. At 1000 K Jeans's own criterion gives x(H2) = 15.2, so hydrogen loss is predicted; Earth now loses about 3 kg/s of hydrogen [web: Wikipedia Atmospheric escape]. Add so the primer does not present the "no perceptible escape" statement as still valid. Also the half-life for x = 30: the printed fraction gives 0.76 million years (confirmed by my computation).
6. deep/AtomicityQuanta-df752a.md, section 2e: add that the classical Larmor collapse time is 1.56e-11 s and the initial shrink rate 1.13 m/s (so "over a metre a second" is exactly right and "10^-10 s" is high by a factor of about 6). [calc]
7. claims.md claim 4 states "Bucherer's experiments as confirming Lorentz's deformable electron over Abraham's rigid one". Add: Bucherer's 1908 result was contested by Bestelmeyer, and the decisive confirmations came later (Neumann 1914, Glitscher 1917, Rogers 1940); Jeans's date "1908" for Kaufmann is 1906 by his own footnote. [web: Wikipedia KBN page]
8. Investigator corrections 1-3 in sites.md (Adams at Harvard and copper spheres, Hudson's 1910, the Bothe-Geiger and Compton-Simon dates) are consistent with what I read and I endorse them: Adams's paper itself, and Hudson's abstract, both support the changes proposed there.
9. claims.md claim 6 states a two-atom solution "can ... carry a whole quantum from emitter to absorber". Qualify: Jeans gives the formulas of the field for prescribed sources, not the energy flow, and the source-sink analogy is not exact for waves (see claim 6 above).

Web checks done (budget about 6)

1. Exospheric temperature 740-1350 K: https://en.wikipedia.org/wiki/Thermosphere (read via action=raw).
2. Earth's present hydrogen loss, Jeans escape share: https://en.wikipedia.org/wiki/Atmospheric_escape.
3. Kaufmann 1906, Bucherer 1908, Bestelmeyer dispute and later experiments: https://en.wikipedia.org/wiki/Kaufmann%E2%80%93Bucherer%E2%80%93Neumann_experiments.
4. CO2 relaxation about 7 microseconds at 1 atm: https://pubs.rsc.org/en/content/articlelanding/1966/tf/tf9666202655 and https://pubs.rsc.org/en/Content/ArticleLanding/1967/TF/TF9676302093 (abstracts via search; I did not open the papers). The N2 vibrational relaxation time was not found; I do not give a figure for it.
5. Failed or not done: Kennedy-Thorndike statement tagged [memory high], not web-checked this run; the quantum-optics statements (single-photon wave packets, antibunching) tagged [memory high], not checked; Dirac 1938 radiation reaction [memory high].

Worked numbers (reproducible; script used jeans_checks.py, scratch)

• K = 1/(16 pi^2 alpha) = 0.8678.
• Hydrogen classical collapse: 1.56e-11 s; initial |dr/dt| = 1.13 m/s.
• Sun: 2.56e8 tonnes/min.
• Fourier of the Lorentzian pulse: exp(-p c) exactly; sech^2 pulse: pole at pi tau/2.
• e^-16 = 1.1e-7; e^-800 = 10^-347.4; 13 e^-12 = 8.0e-5.
• Energy ratio 4/3 between (2/3) e^2/(a C^2) and e^2/(2 a C^2).
• Escape x values above; half-life at x = 15, 30, 40: 0.45 yr, 0.76 My, 1.3e10 yr.
• Boost of the absorption energy balance: difference = (h nu_w - h nu) + m w (v' - v).