Thompson-JJ (J. J. Thomson): Primer

Primer author (STEP 4), 2026-10-09. Built from arbiter.md (the final verdicts), the three reviews (physicist.md, mathematician.md, contrarian.md), the deep/ reads, sites.md and research/. No new verdicts are given here. The folder name spells the name "Thompson"; the physicist is Sir Joseph John Thomson (1856-1940), without a "p".

QE revision (2026-10-09). Updated after qe.md (verdict MINOR) and the arbiter's "QE revision" section. Changes: Thomson's own measured e/m and e history (§3.1, corrects an earlier sentence that said his 1897 value "agrees"); the Kaufmann table framing (§4, §7, question and answer 2, example.py Part D: two of the three printings label the columns the other way from the 1906 book, so the labelling is not settled); Barkla's datum printed as both 0.0002 and 0.00025 (§5, example.py Part E); verdicts for Part IV of the vortex essay, the two "universal" positive carriers and the metal specific-heat paradox (§6); Thomson's pre-1913 null searches, Nelson's "tritium" reading, the Le Sage-by-X-rays proposal and Wall 1924 (§7, §9). Second update (2026-10-09, after the QE recheck): deep reads 7 (Electricity and Matter ch. VI, Le Sage and the 50,000 years) and 8 (Conduction pp.657-662, Haga-Wind and Marx) are done. §4, §6 row 3c and §7 now give the arbiter's refined grounds for the claim 3 strong-form verdict, which is unchanged (arbiter QE revision 10).

Source tags: [read] the project's own documents (page numbers are PDF pages unless "printed p." is said), [web] with URL, [memory] recalled with a confidence level, [calc] arithmetic done in this review (reproduced by example.py), [Arbiter] / [Physicist] / [Mathematician] / [Contrarian] for a reviewer's finding.

Source note (carried from claims.md and the arbiter). There is no source gap for the central claims: all eleven documents are Thomson's own books and papers (1883-1921 editions) with readable text, mostly OCR of scans; equations were read from page images. Three limits travel with every conclusion below [Arbiter]: 1. Read by digest only, with no mathematics checked: most of Conduction of Electricity through Gases chapters I-XX (about 660 pages; after QE, chapters V-VI on e/m and e, chapter XI on Barkla and Bumstead, chapter XX pp.657-662 on Haga-Wind and Marx, and the Kaufmann table pages were deep read), Electricity and Matter chapters I and III-VI (chapter II, the Kaufmann table, was deep read after QE; chapter VI, with the Le Sage passage in §7, has since been deep read [corrected 2026-10-09: this note earlier said chapter VI was read only by digest and from the text layer]), the early chapters of Rays of Positive Electricity, Part I of the vortex-ring essay, and the Textbook of Physics. Corpuscular Theory chapters I and IV-V and vortex Part IV were deep read in the QE revision [Arbiter, QE revision]. 2. Not obtained: Thomson's 1897 papers, his 1913 Nature letter (preview only), his 1922 Proc. Roy. Soc. A paper (title and date confirmed, result unread), Kaufmann's original papers (they decide the Kaufmann table labels), Barkla's 1904 paper (it decides the 0.0002 against 0.00025 datum), Reynolds 1879 and Knudsen 1909-10 as primary papers, and the 1914-1924 papers by Strutt, Merton, Egerton and Allison-Harkins. What is said about these rests on bibliographies and secondary summaries. 3. Several verdicts lean on textbook history that no reviewer web-checked: Laue-Friedrich-Knipping 1912, Duane-Hunt 1915, Moseley 1913, Geiger-Marsden 1909, Millikan 1909, Aston's neon-22 (1919) [memory, high]. They are flagged where used.

Framing. Thomson is not an outsider. He won the 1906 Nobel Prize in Physics, ran the Cavendish Laboratory, and is the person most associated with the discovery of the electron [memory, high]. This project is useful to a learner for a different reason: in one man's books you can watch a first-rate experimenter be decisively right (the electron, ionisation, mass spectrometry, neon-22, the H3+ ion), reasonably wrong (X-rays as pulses, the "plum pudding" atom), and occasionally wrong in a way his own numbers could have caught (the count of electrons per atom, one table of mass against speed, one vortex stability verdict). It also holds one episode that the record never closed: helium and neon appearing after he bombarded salts with cathode rays.

---

1. The idea in one paragraph

Before 1897 most physicists took the atom to be the smallest piece of matter. Thomson showed that the "cathode rays" in an evacuated discharge tube are streams of tiny negatively charged particles, which he called corpuscles and we call electrons, that they are the same whatever metal or gas they come from, and that each is about two thousand times lighter than a hydrogen atom. He concluded that electrons are a part of every atom. From there he built a programme: the electron's mass might be nothing but the energy of its own electric field; X-rays might be sharp pulses of field thrown off when an electron is suddenly stopped; an atom might be a ball of spread-out positive electricity with electrons arranged inside it in rings, and the way those rings fill up might explain the periodic table. Later he sorted charged atoms by mass with electric and magnetic fields, which became the mass spectrometer, found that neon has a second, heavier form (neon-22), and found a particle of mass 3 made of three hydrogens. He also reported helium and neon coming out of salts after he bombarded them, and said plainly that he could not tell whether this was trapped gas or atoms breaking up. Some of these ideas are now textbook physics, some were replaced, and one question is still open on the record.

---

2. What you need to know first

Units: emu and Gaussian. Thomson writes in centimetre-gram-second electromagnetic units (emu). In emu one unit of charge is 10 coulombs, so the electron's charge is 1.602 x 10^-20 emu (1.602 x 10^-19 C). The Gaussian electrostatic unit (esu) of charge is related by e_esu = c e_emu, with c = 3.00 x 10^10 cm/s, so e = 4.80 x 10^-10 esu. A practical rule: any Gaussian formula containing e^2/(m c^2) becomes e^2/m in emu [Physicist; Mathematician]. The electron's charge-to-mass ratio e/m is 1.759 x 10^7 emu/g (1.759 x 10^11 C/kg). Read Griffiths, Introduction to Electrodynamics, Appendix C on units, for the conversions [memory, high].

Charged particles in electric and magnetic fields. A charge q moving with velocity v through fields E and B feels the Lorentz force F = q(E + v x B) (SI). In a magnetic field alone it moves on a circle of radius r = m v/(q B), so the bending measures momentum per charge; an electric field measures energy per charge. Crossing the two gives v and e/m separately. This is how Thomson measured e/m in 1897 and how his positive-ray parabolas work. Griffiths, chapter 5 (magnetostatics), covers it.

Field energy and field momentum. An electric field stores energy density (epsilon_0/2)E^2 (SI); a moving charge also has a magnetic field, and the combined fields carry momentum density epsilon_0 E x B. A charged sphere moving slowly therefore carries field momentum proportional to its speed, which acts like extra mass. Griffiths, chapter 8 (conservation laws), derives field momentum.

Radiation from an accelerated charge (Larmor). A charge e with acceleration f radiates power P = (2/3) e^2 f^2 / c^3 (Gaussian) or q^2 a^2/(6 pi epsilon_0 c^3) (SI). A charge stopped suddenly therefore sends out a thin shell of radiation. Griffiths, chapter 11 (radiation), derives it.

Thomson scattering. A free electron in an incoming electromagnetic wave is shaken by the electric field and re-radiates. The fraction of the beam's power it removes is described by an area, the Thomson cross-section sigma_T = (8 pi/3) r_e^2, where r_e = e^2/(4 pi epsilon_0 m c^2) = 2.818 x 10^-15 m is the "classical electron radius". sigma_T = 6.652 x 10^-25 cm^2 [memory, high; computed in example.py]. This is the formula behind the worked example. Griffiths, chapter 11 (problems on scattering), or any modern-physics text.

Gas counting: Avogadro, Loschmidt, Faraday. One mole of an ideal gas at 0 C and 1 atmosphere fills 22,414 cm^3, and one cubic centimetre holds the Loschmidt number, L = 2.687 x 10^19 molecules. Faraday's constant F = 96,485 C per mole is the charge carried by one mole of singly charged ions; it equals N_A e, Avogadro's number times the electron charge. A modern-physics text (Krane, Modern Physics; or Serway, Moses and Moyer) covers these.

Kinetic theory: speed, effusion and heat capacity. In an ideal gas at absolute temperature T, the molecules' mean kinetic energy is (3/2) k_B T, where k_B = 1.38 x 10^-16 erg/K is Boltzmann's constant, so the mean speed rises as the square root of T. Two consequences appear in §6 and §7. (1) Thermal transpiration: two vessels of the same gas joined by a tube much narrower than the molecules' free path settle not at equal pressure but at p1/p2 = sqrt(T1/T2), because the number of molecules crossing per second goes as p/sqrt(T) on each side. (2) Equipartition: every free particle in thermal equilibrium carries (3/2) k_B T, so a gas of free particles adds (3/2) k_B per particle to the heat capacity. The Dulong-Petit rule says a solid's atoms carry 3 k_B each (about 6 cal per mole per degree) [memory, high]. Any intro thermodynamics or kinetic-theory text (e.g. Schroeder, An Introduction to Thermal Physics) covers these.

The modern atom in one paragraph. An atom has a tiny positive nucleus (radius about 10^-13 cm) holding Z protons and about A - Z neutrons, where A is the mass number (close to the atomic weight). Z electrons surround it in quantum shells, and Z, the atomic number, sets the chemistry. For light elements Z is close to A/2 (carbon 6/12, nitrogen 7/14, oxygen 8/16); hydrogen (1/1) is the exception. Isotopes are atoms with the same Z and different A. Krane, Modern Physics, covers the nuclear atom and Rutherford scattering.

Linear stability. To test whether an equilibrium is stable, displace it slightly, linearise the equations of motion and look at the exponents of the small motions, x proportional to exp(s t). If every s^2 is negative the motion oscillates (stable at linear order); if any s^2 is positive it grows (unstable); if the largest s^2 is exactly zero the linear test cannot decide and higher-order terms must. Any ODE text (e.g. Strogatz, Nonlinear Dynamics and Chaos) covers it.

---

3. The textbook view

3.1 The electron (claim 1)

Electrons are universal, negatively charged, with e/m = 1.759 x 10^7 emu/g and e = 1.602 x 10^-19 C (4.80 x 10^-10 esu); the electron's mass is 1/1836 of a hydrogen atom's [memory, high].

How the numbers were reached is itself a lesson, and Thomson's own 1906 book tabulates it [read: Conduction PDF p.135 and p.145, page images; Arbiter, QE revision 3]:

Thomson's first value was about 2.3 times too low. It gave the right order of magnitude and showed the same e/m for every gas and electrode metal, which is what established the electron as universal, but not the right value. The method he criticised (using the voltage across the tube) gave the better numbers, and he concedes this through Seitz's 1902 work. The same book carries both a mass of "1/770" of hydrogen (Art. 59, from 7.7 x 10^6) and "1/1700" (Art. 78, from Kaufmann's and later values), without reconciling them. His charge followed a similar path: 6.5 x 10^-10 esu in 1898 (cloud method, about 35% high), revised to 3.4 x 10^-10 in 1901-02; H. A. Wilson got 3.1 x 10^-10 in 1903 (about 35% low); Millikan's oil drops (1909-13) gave close to the modern 4.80 x 10^-10 [read: Conduction; memory, high for Millikan and the modern value]. The Zeeman effect (1896-97) gave e/m of the right size independently [memory, high]. The normal Zeeman effect: a magnetic field H splits a spectral line into three, with angular-frequency splitting omega_+ - omega_- = H e/m (emu), exactly as Lorentz's classical model of an orbiting electron gives [read: deep/Corpuscular s.8].

3.2 Electromagnetic mass (claim 2)

For a sphere of radius a carrying charge e on its surface and moving slowly at speed V, the field momentum is p = (2/3) (e^2/(a c^2)) V (Gaussian), so the field adds an inertia m_em = 2 e^2/(3 a c^2), which in emu is (2/3) e^2/a [read: deep/Corpuscular s.1; Physicist: integral checked]. The same sphere's electrostatic energy is U = e^2/(2a), so U = (3/4) m_em c^2 rather than m_em c^2. This mismatch is the "4/3 problem"; it is removed only by adding non-electrical stresses that hold the charge together (Poincare, 1905-06) [memory, high]. In special relativity every mass, electrical or not, gives momentum p = gamma m v with gamma = 1/sqrt(1 - v^2/c^2), so a mass that rises with speed says nothing about whether it is electrical [Physicist; Arbiter]. Experiments since show the electron behaving as a point to below about 10^-16 cm [memory, medium; not web-checked; the verdict does not depend on it, Arbiter].

3.3 X-rays (claim 3)

X-rays make gases conduct by knocking electrons out of molecules (ionisation). The continuous X-ray spectrum is bremsstrahlung, "braking radiation" from decelerated electrons, whose classical power is Larmor's formula. X-rays are also electromagnetic waves of wavelength about 0.01-1 nm: crystals diffract them (Laue, Friedrich and Knipping 1912), elements emit sharp characteristic lines (Barkla; Moseley 1913), and the spectrum has a sharp short-wavelength cutoff (Duane-Hunt 1915) that only the photon picture explains [memory, high; not web-checked]. The energy arrives in localised packets (photons), which is why a weak X-ray beam ionises only a few molecules among very many.

3.4 The atom (claim 4)

Alpha particles fired at thin gold or platinum foil are occasionally thrown back (Geiger and Marsden 1909); a spread-out positive sphere cannot do this, so Rutherford (1911) concentrated the positive charge in a nucleus [memory, high; not web-checked]. The number of electrons equals the atomic number Z, about A/2 for light elements, fixed by Barkla's X-ray scattering (around 1911) and by Moseley's X-ray lines (1913) [memory, high]. The periodic table's periods 2, 8, 8, 18, 18, 32 come from quantum numbers and the exclusion principle [memory, high].

3.5 Mass spectrometry and isotopes (claim 5)

Ions of charge e and mass m pass through parallel electric and magnetic fields and strike a plate. The electric deflection x goes as e/(m v^2), the magnetic deflection y as e/(m v), so y^2/x = (e/m)(B-constant^2/A-constant): all ions of one e/m lie on one parabola whatever their speed [read: deep/Positive Rays s.1]. At a fixed x, y is proportional to 1/sqrt(m). Neon has stable isotopes of mass 20, 21 and 22; neon-22 is about 9% of natural neon [memory, high; not web-checked]. H3+ (three protons sharing two electrons) forms in hydrogen discharges by H2+ + H2 -> H3+ + H, is stable as an ion and is important in interstellar chemistry [web: Kragh 2011, https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1600-0498.2011.00237.x (abstract)]. A neutral H3 molecule has a repulsive ground state: it exists only in short-lived excited (Rydberg) states, which fall apart into H2 + H in about a picosecond for the lowest ones [web, Arbiter QE revision 6, search summaries: https://en.wikipedia.org/wiki/Triatomic_hydrogen ; https://thesis.library.caltech.edu/6664]. Tritium, the hydrogen isotope of mass 3, was not discovered until 1934 [memory, high].

Nuclear energy scales: to pull an alpha particle (a helium-4 nucleus) out of sodium-23 takes 10.47 MeV and out of potassium-39 takes 7.22 MeV [web, Arbiter: Audi-Wapstra via https://www.nndc.bnl.gov/ensnds/23/Na/23na_a_aP.pdf and mass excesses at https://atom.kaeri.re.kr/cgi-bin/nuclide?nuc=K-39]. Both nuclei are stable against alpha emission.

3.6 Vortices (claim 6)

In an ideal (frictionless) fluid, vorticity is carried with the fluid and vortex rings are permanent; real smoke rings follow the same mathematics for a while [memory, high]. For n equal point vortices placed evenly on a circle, the linear stability exponents satisfy s^2 proportional to -k(n-k)[2(n-1) - k(n-k)] for k = 1 ... n-1 [Arbiter; Mathematician]. The largest values for n = 3 to 9 are -4, -8, -12, -9, 0, +32, +80 [calc, Arbiter]: stable up to six, exactly neutral at seven, unstable from eight. Nonlinear analysis (Cabral and Schmidt 1999; Montaldi and Tokieda) shows the seven-vortex ring is in fact stable [web, Arbiter: https://rcd.ics.org.ru/RD2004v009n04ABEH000294/], and a 2008 water-vortex experiment matches the stable and unstable ranges [web: https://link.aps.org/doi/10.1103/PhysRevLett.100.174503].

---

4. Where Thomson departs

First, the real physics his starting points rest on. Every one of Thomson's bold steps starts from something true. A moving charge really does carry field momentum and so has inertia from its field (section 3.2). A suddenly stopped charge really does emit a thin shell of radiation (Larmor). Free electrons really do scatter X-rays in proportion to their number (Thomson scattering, his own formula). A positive-ray parabola really does weigh atoms. Ideal-fluid vortex rings really do obey the equations he solved. Where he departs, he usually takes a correct calculation one step further than the data could carry, and in 1897-1907 that was often a reasonable bet.

The single most instructive departure, because it can be checked with a pocket calculator and because Thomson's whole atom model was built on it, is the count of electrons per atom. That is the worked example.

---

5. Worked example: how many electrons are in an air molecule?

What Thomson did. In chapter VII of The Corpuscular Theory of Matter (1907) Thomson counts the electrons in an atom by three methods. The first uses C. G. Barkla's X-ray scattering measurements. Thomson's argument [read: page image PDF p.159, printed p.144]:

the energy of the secondary radiation ... is equal to (8 pi/3) N e^4/m^2 times the energy in the primary radiation passing through unit volume ... N is the number of corpuscles per unit volume, e the charge on a corpuscle, and m its mass.

Barkla finds that the energy of the secondary radiation from one cubic centimetre of air at atmospheric pressure is about .00025 that of the primary radiation passing through it. Hence we have (8 pi/3) N e^4/m^2 = .00025. Now e = 1.2 x 10^-20 and e/m = 1.7 x 10^7. Substituting these values, we find Ne = 10. If n is the number of molecules in a cubic centimetre of air at atmospheric pressure and 0 C we know ne = .4.

and on the next page [read: PDF p.160, printed p.145]: "Thus N = 25n, so that the molecules of air contain on the average about twenty-five corpuscles. The molecular weight of nitrogen is 28; this result suggests that the number of corpuscles in the atom of nitrogen is equal to the atomic weight of nitrogen."

The same argument in the 1906 book, with a different datum. Conduction of Electricity through Gases (1906) gives the same count one year earlier, but its sentence quotes Barkla's datum as "about ·0002 that of the primary" and the equation on the next line uses "= ·00025"; the two are never reconciled [read: Conduction PDF p.339, page image; QE S1; Arbiter, QE revision 2]. Barkla's own 1904 paper (Phil. Mag. (6) 7, 543), which would say which figure he reported, was not obtained. Step 8 below redoes the count with 0.0002.

Symbols. N = number of electrons per cm^3 of air; n = number of molecules per cm^3; e = electron charge (emu); m = electron mass (g); e/m = charge-to-mass ratio (emu/g); R = 0.00025 = Barkla's scattered fraction for 1 cm^3 (equivalently, per cm of beam path through a 1 cm^2 cross-section).

The physics is right. In emu, e^2/m is the classical electron radius r_e (because e_emu^2/m = e_esu^2/(m c^2)). So (8 pi/3) e^4/m^2 = (8 pi/3) r_e^2 = sigma_T, the Thomson cross-section, and Thomson's equation says simply: scattered fraction per cm = (number of electrons per cm^3) x (cross-section per electron). That is correct physics, still used today [Physicist; memory, high].

Step 1: solve for N e. Write e^4/m^2 = e^2 (e/m)^2. Then (8 pi/3) (N e) e (e/m)^2 = R, so

N e = 3R / (8 pi e (e/m)^2).

Step 2: Thomson's numbers. e (e/m)^2 = 1.2 x 10^-20 x (1.7 x 10^7)^2 = 1.2 x 10^-20 x 2.89 x 10^14 = 3.47 x 10^-6. And 3R/(8 pi) = 0.00075/25.13 = 2.98 x 10^-5. So N e = 2.98 x 10^-5 / 3.47 x 10^-6 = 8.6, not 10 [calc; Mathematician; deep/Corpuscular s.1.2].

Step 3: divide by n e. The quantity n e is the charge carried by the molecules in 1 cm^3 if each carried one electron's charge. It does not depend on e: n e = F / V_m = (96,485 C/mol / 10 C per emu) / 22,414 cm^3/mol = 0.430 emu/cm^3 (Thomson used 0.4). With Thomson's own inputs N/n = 8.6/0.4 = 21.5 electrons per air molecule. Rounding N e up to 10 gives 25, which is closer to nitrogen's 28. (A possible reason for the 10: his chapter I value of the charge, 1.03 x 10^-20 emu, does give N e = 10.0 [calc]. The book does not say so [deep/Corpuscular s.1.2].)

Step 4: see how the answer depends on e. e/m and n e were both measured directly; e was the poorly known number. From Steps 1-3,

N/n = 3R / (8 pi e (e/m)^2 (n e)),

so with the measured quantities fixed, N/n is proportional to 1/e (example.py Part C2 checks this with SymPy). Thomson's e was 1.2/1.602 = 0.749 of the true value, so his count was too large by the factor 1/0.749 = 1.34.

Step 5: redo it with modern constants. e = 1.602 x 10^-20 emu, e/m = 1.759 x 10^7 emu/g, n e = 0.430: - e (e/m)^2 = 1.602 x 10^-20 x 3.094 x 10^14 = 4.96 x 10^-6 - N e = 2.98 x 10^-5 / 4.96 x 10^-6 = 6.02 emu/cm^3 - N/n = 6.02/0.430 = 14.0 electrons per air molecule [calc].

The 21.5 -> 14.0 change splits into three factors: the charge (x 0.749), the slightly larger modern e/m (x 0.934) and the modern n e (x 0.929): 21.5 x 0.749 x 0.934 x 0.929 = 14.0 (example.py Part C2).

Step 6: compare with the two hypotheses. Dry air is 78.1% N2, 20.9% O2, 0.93% Ar by molecule. Electrons per molecule if each atom has Z electrons: 0.781 x 14 + 0.209 x 16 + 0.0093 x 18 = 14.5. If each atom had A electrons, as Thomson proposed: the mean molecular weight, 29.0. The datum gives 14.0: atomic number, not atomic weight.

Step 7: run it forward as a check. Scattered fraction per cm = sigma_T x L x (electrons per molecule): - with 14.5 electrons: 6.652 x 10^-25 x 2.687 x 10^19 x 14.46 = 2.58 x 10^-4, Barkla's 0.00025; - with 29.0 electrons: 5.18 x 10^-4, twice what Barkla measured [calc; Arbiter].

Step 8: the 1906 datum, 0.0002. Because N/n is proportional to R, every count scales by 0.0002/0.00025 = 0.8: - Thomson's constants: N e = 6.88, N/n = 6.88/0.4 = 17.2 (against 21.5 with 0.00025); - modern constants: N/n = 14.0 x 0.8 = 11.2; - forward check: 14.5 electrons per molecule predict 2.58 x 10^-4, about 30% above a quoted 2.0 x 10^-4, while 29 electrons predict 2.6 times it [calc; QE; example.py Part E]. Either printed datum gives 11-14 electrons per air molecule, near the 14.5 of the atomic numbers and far from the 29 of the atomic weight. (The arbiter's QE revision also quotes 20 and 16 for Thomson's constants; those divide by the modern n e = 0.43 instead of his 0.4.)

What the example shows. 1. Thomson's physics (Thomson scattering) and his data source (Barkla) were sound. His conclusion "number of corpuscles = atomic weight" came from a charge 25% too low plus a rounding of 8.6 to 10, both pushing toward the answer he expected [Contrarian: "the count was steered by the prior", "this is not fraud"]. If the 1906 book's "about ·0002" is Barkla's real figure, a third push is added: the equation used a datum 25% larger than the one quoted, and with 0.0002 his own constants give 17, well short of nitrogen's 28 [QE; Arbiter, QE revision 2]. 2. The same datum worked with a better charge gives about A/2, the atomic number. Barkla himself reached about A/2 around 1911, and Moseley's X-ray lines fixed Z in 1913 [memory, high; not web-checked]. This is why the arbiter's verdict on "N = atomic weight" is Refuted by calculation: accepted laws applied to his own datum rule it out. 3. Barkla's other finding, scattering proportional to density, is real and Thomson read it correctly as "number of electrons proportional to atomic weight". It holds because Z/A is close to 1/2 for all light elements; it fixes the proportion, not the number. Hydrogen breaks it (Z/A = 1), so his example "oxygen has sixteen times the corpuscles of hydrogen" is really eight times [memory, high]. 4. How firm is the 14? Barkla's datum is quoted as "about", one significant figure, and Thomson prints it as both 0.00025 (1907, and in the 1906 equation) and 0.0002 (1906 sentence), so the honest answer is 11-14, and 14.0 against 14.5 is better agreement than the datum deserves. But the test is between 11-14 and 29, a factor of two or more, which a one-figure datum does decide. (At soft X-ray energies, coherent scattering from the electrons of one atom can add to the total for heavier atoms [memory, medium]; for air, the light-element case Barkla and Thomson chose, the simple sum works, as Step 7 shows.) 5. Why it mattered: an atom with A electrons, each carrying 1/1836 of a hydrogen mass, still holds only about 1/1800 of the atom's mass in electrons. Thomson saw this himself (Conduction, Art. 355; Corpuscular Theory preface: the corpuscles cannot carry all of an atom's mass) [read; Physicist], which is why the arbiter rates his extension of "all mass is electrical" to whole atoms as Refuted by calculation by his own counting.

Run it yourself. .venv/bin/python review/Thompson-JJ/example.py prints every number above (Parts A-C2, and Part E for Step 8) and the Kaufmann column check used in question 2 of "Check yourself" (Part D, which also prints how each of the three printings labels the columns and the slow-speed e/m the test assumes).

---

6. Evidence table

Verdicts are the arbiter's, using the scale in prompt.txt. "Tested?" and "Replicated?" summarise the reviews; the sources are in arbiter.md.

---

7. What a learner should take away

What is solid. The electron as a universal constituent of matter; X-ray ionisation; the classical radiation of a decelerated charge; Thomson scattering (his formula, still the textbook result); the parabola method and with it mass spectrometry; neon-22; the H3+ ion; the field-momentum calculation of electromagnetic mass; the core vortex-ring mathematics. These are not small things: the mass spectrometer and the electron are among the most productive results of twentieth-century physics.

What is wrong, and why each mistake teaches something. - Electrons per atom = atomic weight. The physics was right; a constant was wrong and a rounding leaned toward the expected answer. Lesson: when a conclusion depends on a poorly known constant, write out how the answer scales with it (N/n proportional to 1/e here) before trusting it. - All mass is electrical. A correct calculation (field momentum) was promoted to a complete theory by choosing the radius to make it fit. A parameter fixed by the hypothesis cannot then confirm the hypothesis. The same calculation contains the 4/3 problem, which Thomson did not notice and which pointed toward relativity. - The Kaufmann table. Three printings carry the same two columns: Electricity and Matter (1904-09) and Corpuscular Theory (1907) call one column Kaufmann's, Conduction (1906) calls the other one Kaufmann's, and neither column is reproduced by the formula he prints. An internal-consistency test (e/m times the true mass ratio must be constant) favours the 1906 labels, but only as a fit and only if Kaufmann used one slow-speed e/m of about 1.95 x 10^7 (example.py Part D). (An earlier version of this primer, following the arbiter's first pass, said the test settled the labelling and that the 1907 book had swapped the headings; the QE found the 1904 printing agrees with 1907, and the arbiter withdrew "settled".) Lesson: a consistency test can narrow a question the text cannot decide, but state its assumptions; here only Kaufmann's own 1901-03 table settles it. - Thomson's own first e/m. His 1897 deflection value, 7.7 x 10^6, was 2.3 times too low, and his first charge 35% too high; the right values came from Kaufmann, Simon and later Millikan (§3.1). What made the 1897 result decisive was not its value but its constancy across gases and metals. Lesson: a discovery can rest on a robust pattern while its first number is wrong. - X-rays as pulses. Reasonable when only the absence of refraction and polarisation was known. Thomson even absorbed the first diffraction evidence: in 1906 he read Haga and Wind's slit broadening, equivalent to a wavelength of about 2 x 10^-8 cm, as the thickness of the pulse [read: Conduction p.662]. By Fourier analysis (writing any signal as a sum of sine waves), a train of irregular thin pulses is the same thing as a continuous spread of wavelengths, so diffraction of that spread alone cannot tell the two pictures apart. What overturned the "not waves" form was what crystal diffraction (1912) made measurable: sharp, periodic characteristic lines and a sharp short-wavelength cutoff (Duane-Hunt 1915) [memory, high; Arbiter QE revision 10]. Lesson: before calling an experiment decisive, ask whether the rival picture can absorb it. Yet his "specks" argument about localised energy was pointing at something real. - Seven vortices. His exact equations gave a zero; rounding made it look negative. When a linear test lands exactly on zero, it has not decided. - The plum-pudding atom. A model consistent with everything known in 1904 that a single new kind of experiment (large-angle alpha scattering) overturned. Thomson never used the name "plum pudding" [web via Investigator, Hon and Goldstein 2013 per Wikipedia]. - Two universal positive carriers. In 1907 he read two positive spots, e/m about 10^4 and 5 x 10^3, seen "whether the tube is filled originally with air, hydrogen, or helium", as two units of positive electricity common to all gases, and he argued against hydrogen contamination [read: Corpuscular pp.35-38]. His own later parabola method showed each gas gives its own ions. Lesson: an "it appears in every gas" result is exactly what a common contaminant also produces. - The vortex gas (1883, Part IV). Thomson's vortex atom made one sharp, falsifiable laboratory prediction: thermal effusion with an exponent m > 1 instead of kinetic theory's 1/2, which he called a "crucial experiment". The measurement (Reynolds 1879, Knudsen 1909-10) gave 1/2 [web, Arbiter]. He also predicted that the mean speed of a monatomic vortex gas falls as it gets hotter, which molecular-beam measurements contradict. His valency rules ("at most six" links) are applied inconsistently and rest on the seven-vortex rounding error. Lesson: this is how a theory should fail, by naming the experiment that could kill it.

An anomaly the classical model found itself: the specific heat of free electrons. In Corpuscular Theory chapter IV Thomson works out how many free corpuscles silver needs to reflect light as Rubens and Hagen measured: more than 1.8 x 10^24 per cc. If each carries the classical gas energy, (3/2) k_B T (§2), they alone would need about 5.7 cal per degree per cc, but the whole metal takes only 0.58 cal (the Dulong-Petit value). He states this as a contradiction and concludes that free-corpuscle conduction is "at most only a part" of metallic conduction (p.90; drawn first from thermoelectric and melting data) [read: Corpuscular pp.90-100, page images; Arbiter QE revision 5]. He was right that something was wrong: the answer, twenty years later, was that electrons obey Fermi-Dirac statistics, so only those near the top of the energy distribution can take up heat, and the electronic heat capacity of silver at room temperature is about 1% of the lattice's (Sommerfeld 1927) [memory, high/medium]. His Wiedemann-Franz ratio (thermal over electrical conductivity), which matched pure metals within about -5% to +10%, contains no mass or charge sign, so, as he himself wrote, it "would hold if the carriers were bodies having a much greater mass" [read; Arbiter]. Worked number: 5.7 / 0.58 = about 10, a factor no adjustment of a classical model could remove.

Thomson's own speculation: gravity from penetrating Röntgen rays (Le Sage). In Electricity and Matter chapter VI Thomson asks whether radium could draw its energy from "some form of radiation which is passing through all bodies on the surface of the earth". He recalls Le Sage's theory of gravity (a sea of very penetrating "ultra-mundane corpuscles", partly absorbed, so that two bodies shield each other and are pushed together) and Maxwell's objection (the absorbed energy would keep the bodies white hot). He then proposes that "we might therefore replace Le Sage's corpuscles by very penetrating Röntgen rays", since electromagnetic waves carry momentum: two bodies would then attract "inversely as the square of the distance", and in proportion to the product of their masses if absorption is proportional to density. He notes two difficulties himself: gravity would travel at the speed of light, "whereas, astronomers believe they have established that it travels with a very much greater velocity"; and the re-emitted rays would have to be more penetrating than the incoming ones, with an enormous energy turnover. He concludes that "the magnitude of the energy radiated from radium is not a valid argument against the energy being derived from radiation", and places the energy inside the atom for another reason: radioactivity is "a transient property" of each active product [read: Electricity and Matter 1909, PDF pp.174-179, printed pp.157-162, text layer; also digests of the 1904 and 1907 printings]. A reader's margin annotation in the project's copy of the 1904 printing, on the chapter III page where Thomson describes the stop-pulse as the Röntgen ray (PDF p.75, not in chapter VI), links that passage to an external cause of radioactivity; the QE connects the annotation to this chapter VI passage, which is the place Thomson discusses an external source [read: 1904 scan text layer p.75; QE]. No reviewer assessed this proposal and it carries no verdict. It has now been deep read [corrected 2026-10-09: this sentence earlier said the deep read was not yet done]: the reader finds the reasoning locally consistent, but the "simple calculation" of energy turnover per gram (p.177) is asserted and never shown, and the radiation postulated for radium (absorbed only by radioactive bodies) is not the same as the one postulated for gravity (absorbed by all matter in proportion to density) [read: deep/ELECTRICITY-AND-MATTER-1909 section F; Arbiter QE revision 10]. For the reader: push gravity of this kind is still discussed in alternative physics; the mainstream objections are the heating and drag problems Maxwell and others raised, and today the measured speed of gravity, which matches light speed (gravitational waves, 2017) [memory, high], removes Thomson's own first objection. In the same chapter he estimates radium's life at "about 50,000 years" from its heat output and Rutherford's alpha speed [read: printed p.154] (radium-226 has a modern half-life of about 1,600 years, a mean life of about 2,300 years [memory, high], so his estimate was about twenty times too long).

What is genuinely open. 1. Helium and neon from bombarded salts. Thomson's controls were real: helium yield grew with bombardment time; salts dissolved and re-evaporated ten or twelve times gave "apparently undiminished supplies of helium"; a split-aluminium blank exposed to helium for two days differed by about the scatter; potassium salts, especially KI, gave an "exceptionally large" supply [read: Positive Rays pp.149-150]. The controls rule out surface gas better than gas held inside the crystal or the glass [Physicist; Contrarian]. Transmutation by his electrons is ruled out by energy (section 6, row 5f), but the observation itself has never had an independent test of his protocol on record. The deciding records: his 1922 Proc. Roy. Soc. A 101:290 paper on "deflagrated metals" (result unread), the Strutt, Merton and Egerton papers, and his 1910-22 notebooks at Cambridge University Library (MS Add. 7654) [web via sites.md]. The deciding experiment: bombard the same salts in helium-free ultra-high vacuum, with blank runs and helium-3/helium-4 isotope analysis, and look for the residue atoms Thomson himself named (beryllium minus helium = 5, carbon minus helium = 8) [Arbiter]. 2. Absence of evidence after 1922. Thomson stopped publishing on bombardment helium. The competing explanations, as the arbiter sets them out: he concluded it was contamination and dropped it (no statement found); Aston's mass spectrograph and nuclear physics displaced the work (explains the displacement, not why his named test was never reported); he ran the residue test and it was negative or inconclusive and unpublished (notebooks exist; no source says so); suppression (no evidence: both sides published in Nature and Proc. Roy. Soc., and Thomson himself was the senior critic of transmutation). Interested parties on both sides had reputations at stake: Thomson and Aston (the method), Ramsay, Collie, Patterson, Masson and Baly (priority for transmutation) [web via Investigator: Morrisson, Modern Alchemy]. Secondary verdicts differ: Redgrove (1922) says the discharge-tube claims "failed to be satisfactorily confirmed"; a Royal Society referee called Collie-Patterson-Masson's results "currently unexplained" [web: https://makingscience.royalsociety.org/items/rr_21_53]; Nelson, an alternative-science author, writes "The matter never has been resolved" [web: https://www.alchemywebsite.com/nelson2_6.html]. Related later claims in the same tradition: Wendt and Irion's exploding tungsten wires (1922), now cited in low-energy nuclear reaction (LENR) lists, and Paneth and Peters (1926), who retracted [web via Contrarian]. 3. What was the stored mass-3 gas? Thomson kept it for weeks, destroyed it with light plus oxygen and with hot copper oxide, enriched it through palladium, and recorded his own negative: "No lines which could be ascribed to X3 were detected" [read: Positive Rays pp.141-147]. He kept the name X3 until 1937 [web via Investigator, thesis snippet]. No independent test of the stored gas is on record. Nelson, an alternative-science author, identifies X3 as "tritium" [web via sites.md: https://www.alchemywebsite.com/nelson2_6.html]; that is his characterization, pointing to a nuclear origin for the mass-3 line. Against it: tritium was not known until 1934, Thomson himself never called it that, and the mainstream assignments for a mass-3 line in hydrogen are H3+ and HD+ [memory, high; Arbiter QE revision 6]. 4. Which Kaufmann column is observed, and how did Thomson get the other? Two of three printings call 3.09, 2.43, 2.04, 1.83, 1.65 Kaufmann's; the 1906 book calls 3.1, 2.42, 2.0, 1.66, 1.5 Kaufmann's. Neither is his stated model. If the 3.09 column is Kaufmann's, it is already gamma-like, which is a hypothesis for Kaufmann's 1901-03 papers to test, not a finding [Arbiter, QE revision 1]. 5. His late models (1924-28): light quanta as "anchor rings" of closed electric lines of force moving at c, and the electron with a "dual structure" of tubes plus a guiding wave train [web via Investigator: The Structure of Light, 1925; Beyond the Electron, 1928]. No reviewer read them; whether they made any quantitative prediction is unknown. The mainstream view is that free fields do not hold closed "anchor ring" shapes [Physicist]. 6. Part I of the vortex essay (a circular ring stable for all small displacements; an elliptic vibration taking 0.289 of the time to travel one circumference) has not been checked by anyone. Part IV now has verdicts (§6, rows 6d-6h). 7. Barkla's real figure (0.0002 or 0.00025) is in his 1904 paper, not obtained (§5, Step 8). It moves the count between 11 and 14, not toward 29. 8. Thomson's 1906 account of X-ray diffraction (Haga and Wind's slit broadening, which he reread as the pulse thickness) and of Marx's finding that X-rays travel at the speed of light is in the folder (Conduction pp.657-662) and has now been deep read [corrected 2026-10-09: this item earlier said "read only by digest"]. Thomson reports Marx's result with no number and does not use it for or against pulses. The arbiter weighed the read and kept the claim 3 verdict, with refined grounds (§6 row 3c; Arbiter QE revision 10). Still open: the cited Haga-Wind, Wind and Marx papers themselves are unread, and how later work interpreted Wind's slit effect is unchecked. C. N. Wall's 1924 extension of the Stokes-Thomson pulse theory (Phil. Mag. 48:378) is another unread lead on claim 3 [web via sites.md]. 9. Before 1913 Thomson already suspected X-rays could release atomic energy. From Bumstead's 1906 finding that lead heats nearly twice as much as zinc for equal X-ray absorption, he wrote that X-rays "may furnish a means of tapping the stores of inter-atomic energy", and suggested keeping radium in contact with lead to look for new elements. He also reports "a large number of experiments" that found no radioactivity in metals after hours of strong X-rays, and similar negative radium tests with Bumstead, with no detector sensitivities given [read: Conduction pp.333-344, page images; Arbiter QE revision 6]. So his 1913 caution about helium from salts followed his own earlier null searches. Whether Bumstead's result held is not in the folder.

---

8. Check yourself

1. In the worked example, show that N/n is proportional to 1/e when e/m and n e are held at their measured values. If Thomson had used Millikan's 1913 charge (about 4.77 x 10^-10 esu [memory, medium]) with his other inputs, roughly what count would he have obtained?
2. Thomson's 1906 table gives Kaufmann's measured e/m as 0.63, 0.77, 0.975, 1.17, 1.31 (x 10^7 emu/g) at five speeds, and two columns of mass ratios: A = 3.1, 2.42, 2.0, 1.66, 1.5 and B = 3.09, 2.43, 2.04, 1.83, 1.65. One column is "observed by Kaufmann"; the 1906 book says A, the 1904-09 Electricity and Matter and the 1907 book say B. Which column does a consistency test favour, what does the test assume, and why does it not settle the question?
3. Using U = e^2/(2a) for the electrostatic energy of a charged sphere and m_em = 2 e^2/(3 a c^2) for its field mass (Gaussian), find U/(m_em c^2). What does the answer say about "all mass is electrical"?
4. A 30 keV electron hits a potassium-39 nucleus. What fraction of the 7.22 MeV needed to remove an alpha particle does it carry? What does this decide, and what does it not decide, about Thomson's helium observations?
5. For n equal vortices on a circle, the stability exponents satisfy s^2 proportional to -k(n-k)[2(n-1) - k(n-k)]. Find the largest value for n = 6, 7 and 8, and say what each means.

---

9. Further reading

Mainstream textbooks and history - D. J. Griffiths, Introduction to Electrodynamics (4th ed.): chapter 8 (field momentum), chapter 11 (radiation, Larmor formula), chapter 12 (relativity), Appendix C (units). - K. Krane, Modern Physics: the electron, Rutherford scattering, X-rays, the nuclear atom and isotopes. - J. D. Jackson, Classical Electrodynamics: the self-energy and electromagnetic mass of the electron, and Thomson scattering (graduate level; read after Griffiths). - A. Pais, Inward Bound (1986): the history of the electron, X-rays and the nucleus [memory, high]. - J. Navarro, A History of the Electron: J. J. and G. P. Thomson (2012) [web via sites.md]. - H. Kragh, papers on the history of H3 and H3+: Centaurus 53:257 (2011); Phil. Trans. R. Soc. A 370:5225 (2012) [web via sites.md; abstracts only were read]. - P. G. Saffman, Vortex Dynamics (1992), for vortex rings and point-vortex polygons [memory, high]. - M. Morrisson, Modern Alchemy (2007), for the 1913 helium-neon controversy [web via sites.md]. - C. N. Wall, "The selection principle: a development based upon the Stokes-Thomson pulse theory", Phil. Mag. 48:378 (1924): a later extension of Thomson's X-ray pulse theory by another author; not read by any reviewer [web via sites.md, Crossref]. - R. C. Tolman (1922), on the thermodynamics of the discharge-tube helium claims, and the Trinity College Cambridge Thomson papers (fonds THMJ/THMF): leads on claim 5, not read [web via sites.md; Arbiter QE revision 6]. - For the vortex-gas test: O. Reynolds (1879) on thermal transpiration through porous plates, and M. Knudsen (1909-10) on narrow tubes; known here only through later papers [web, Arbiter]. - C. Kittel, Introduction to Solid State Physics, the chapter on the free electron Fermi gas, for how Sommerfeld resolved Thomson's specific-heat contradiction [memory, high].

Thomson's own key works (in the folder unless marked) - A Treatise on the Motion of Vortex Rings (1883, Adams Prize essay). - "On the Discharge of Electricity produced by the Roentgen Rays" (1896 note). - "A Theory of the Connection between Cathode and Roentgen Rays" (1898), with Stokes's Wilde Lecture, in Roentgen Rays (Harper's Scientific Memoirs, 1899). - Electricity and Matter (1904; 1907 and 1909 printings). - Conduction of Electricity through Gases (2nd ed., 1906), chapter XXI on electromagnetic mass. - The Corpuscular Theory of Matter (1907), chapter VII for the electron counts in the worked example. - Rays of Positive Electricity and Their Application to Chemical Analyses (1913); 2nd ed. 1921 online [web: https://archive.org/details/in.ernet.dli.2015.160862]. - A Textbook of Physics (with J. H. Poynting). - Not in the folder: "Cathode Rays", Phil. Mag. 44:293 (1897) [memory, high]; Nature 90:645 (1913) on helium and neon in vacuum tubes [web: https://www.nature.com/articles/090645d0]; Proc. Roy. Soc. A 101:290 (1922) [web: https://royalsocietypublishing.org/doi/10.1098/rspa.1922.0044]; The Structure of Light (1925); Beyond the Electron (1928) [web: https://archive.org/details/beyondelectron0000unse]; Lord Rayleigh, The Life of Sir J. J. Thomson (1942).

---

Answers

1. From (8 pi/3) N e^2 (e/m)^2 = R, N = 3R/(8 pi e^2 (e/m)^2); n = (n e)/e. So N/n = 3R/(8 pi e (e/m)^2 (n e)): one power of e is left in the denominator. Millikan's 4.77 x 10^-10 esu is 1.59 x 10^-20 emu; 21.5 x (1.2/1.59) = about 16 with Thomson's other inputs (and 14 once e/m and n e are also updated). Either way, about half the molecular weight.

2. The test favours column A. A slow electron has one fixed e/m; at speed the measured e/m is that value divided by the true mass ratio. So (measured e/m) x (true ratio) should be the same at all five speeds. With A: 1.95, 1.86, 1.95, 1.94, 1.96 (spread about 5%). With B: 1.95, 1.87, 1.99, 2.14, 2.16 (spread about 14%, drifting upward). The assumption: Kaufmann reduced all five speeds with one common slow-speed e/m, here about 1.95 x 10^7. That is higher than the slow-speed values of 1.77-1.86 x 10^7 in Thomson's own tables; with 1.86 the ratios would be 2.95, 2.42, 1.91, 1.59, 1.42, which match A only to within 3-5%. So the test favours the 1906 labels as a fit, not exactly, while two of the three printings label the columns the other way. Only Kaufmann's own 1901-03 table can settle it (example.py Part D; Arbiter, QE revision 1). Under either labelling, neither column is what Thomson's own formula gives.

3. U/(m_em c^2) = (e^2/2a) / (2e^2/3a) = 3/4. If all the mass were field energy, m c^2 should equal U; field momentum gives 4/3 as much. A purely electromagnetic electron is therefore not self-consistent; non-electrical stresses (Poincare) are needed, and relativity removes the need for the electromagnetic account altogether.

4. 0.030 MeV / 7.22 MeV = 0.0042, or 0.42%. Energy conservation forbids an electron of that energy from removing an alpha particle from potassium-39, whatever the mechanism, so transmutation by his cathode rays is ruled out. It does not tell you where the helium he measured came from: that is an experimental question (occluded or atmospheric helium is the standard explanation, not demonstrated for his experiments in the record), and Thomson himself did not claim transmutation.

5. n = 6: k = 3 gives -9 x (10 - 9) = -9 (the largest; all negative, stable). n = 7: k = 3 or 4 gives -12 x (12 - 12) = 0: neutral at linear order, so the linear test cannot decide (nonlinear analysis shows the heptagon is stable). n = 8: k = 4 gives -16 x (14 - 16) = +32: a growing mode, unstable. Thomson's "seven unstable" contradicts his own equations.