Mathematician review: Crookes, "On Radiant Matter" (1879)

Role: PhD mathematician (Step 2). I check arithmetic, units, scaling and hidden assumptions. Tags: [read] = the project's document or deep read; [web] = URL given; [memory] = recalled, with confidence; [calc] = my own computation (python, shown). The project has one document, a 23-page lecture of live demonstrations with no tabulated data and no formulas except arithmetic on molecule counts [read: deep read]. There is therefore little mathematics to check in the strict sense. I test the claims for consistency of their quantitative statements, and I add order-of-magnitude calculations from accepted laws so the reader can see what the demonstrations imply. No source gap: the text was read in full. Where the deep read and the OCR differ, I use the page-image values (highest vacuum 1/20,000,000 atm; red line 6,895; 10^24 molecules in the bulb) [read: deep read, OCR table].

Corrections to Step 1

1. claims.md claim 6 says "about 10^18 molecules" remain at one millionth of an atmosphere. That matches the printed 1,288252,350000,000000 (1.29 x 10^18) [read: page image p.22 via deep read] and my recomputation (see Q6). Correct as stated. But claims.md does not say that Crookes's underlying density (10^21 molecules per cm^3 at one atmosphere, from Johnstone Stoney) is about 37 times too high; the modern Loschmidt value is 2.65 x 10^19 per cm^3 at standard temperature and pressure [web: https://www.amusingplanet.com/2021/02/johann-josef-loschmidt-and-avogadros.html, search summary]. The lecture's figures are therefore internally consistent but numerically off by that factor (see Q6). The Stoney 1868 value itself I could not check on the web (the search returned nothing on it); I take 10^21 from the lecture footnote [read].
2. facts.json and claims.md describe the "dark space" claim as Crookes's own and untested. The Investigator's note that modern sources call the "fourth state" plasma is correct as a lead, but Crookes's definition is a collisionless neutral-gas regime. I confirm the Investigator's caution (sites.md, priority 3). The mismatch matters for claim 1 verdict (see below).
3. Step 1 and the Investigator's lead on the Harvard page: Harvard's text [web: research note] says Thomson 1903 used Starke's experiments. Wikipedia's "one revolution per minute" is unsourced by page; I reproduce the order of magnitude below (momentum force of order 10^-7 N per mA) but have not seen Thomson's own calculation. [memory, low] on his exact figures.
4. Deep read states the hundredth-of-an-inch comparison without checking it. I checked: 0.01 inch / 3 miles = 5.26 x 10^-8 = 1/19.0 million, which matches 1/20,000,000 [calc]. This supports the page-image reading of the denominator (and refutes the brief's 1/6,754,000, which would give 1.5 x 10^-7). The "barometric column" must then be a column of air of uniform density three miles high, not mercury. That is my reading, not Crookes's stated meaning.
5. Not a Step 1 error but worth fixing in later steps: the deep read says "Mr. Gimingham" built the hinged cross; Wikipedia credits Plücker 1869. The Investigator flags this. It is outside my remit; I note only that Crookes's text [read: p.12] credits Gimingham for that particular hinged apparatus.

Preliminary: the pressure scale

Crookes's footnote: 1 millionth atm = 0.00076 mm Hg; 1315.789 millionths = 1 mm [read: page image p.7]. Check: 760 mm / 10^6 = 0.00076 mm; 1 / 0.00076 = 1315.789 [calc]. Consistent. In SI: 10^-6 atm = 0.1013 Pa. 1/20,000,000 atm = 5.07 x 10^-3 Pa [calc]. Crookes gives pressures as fractions of an atmosphere, so a reader must convert; no unit errors found.

Claim 1: A "fourth state of matter" at about a millionth of an atmosphere

a. Internal consistency. The argument is a scaling argument: the mean free path (average distance between collisions) is inversely proportional to the number density of molecules [read: p.2]. This is correct kinetic theory: lambda = kT / (sqrt(2) pi d^2 p), with d the molecular diameter and p the pressure. Worked numbers [calc, d = 3.7 x 10^-10 m, 293 K]: at 1 atm lambda = 6.6 x 10^-8 m; at 10^-6 atm lambda = 6.6 cm; at 1/20,000,000 atm lambda = 1.3 m. The 13.5 cm bulb therefore has lambda comparable to its size at 10^-6 atm, which is exactly Crookes's stated regime ("comparable to the dimensions of the vessel", [read: p.4]). So the quantitative core of claim 1 is consistent with accepted kinetic theory. The label "fourth state" is a definitional move, not a derivation: "as far removed from a gas as a gas is from a liquid" has no metric. A state of matter is normally defined by a phase boundary or distinct equations of state. Rarefied gas at low pressure is described by the same ideal-gas law with a different transport regime (free-molecular flow, now called the Knudsen regime, Knudsen number Kn = lambda / L > 1). There is no discontinuity; the passage from continuum to free-molecular flow is gradual. So "state" is a loose usage, not an inconsistency. b. Experiments. Free-molecular behaviour at these pressures is established (Knudsen flow, high-vacuum technology). What Crookes saw (glass fluorescence, rays) is established as electron-beam physics (see claim 3). Crookes's later identification of "radiant matter" with electrons [web, Dyer 1920 obituary quote in the Investigator's research note: https://pubs.rsc.org] is the inventor's own reinterpretation; I did not see the original. c. Tests. Not applicable as a definition. The measurable parts (pressures, mean free path) rest on the pump and gauge, which this text does not describe [read: deep read]. The Investigator reports the multiple-tube Sprengel pump (Gimingham) and that the pressure range is stated in a secondary source [web: Wikipedia, via sites.md]. No pressure gauge calibration is available to me. d. If true. Nothing in mainstream physics changes. Mainstream already has the corresponding regime (free molecular flow) and, for the ionised case, plasma. The modern plasma use of Crookes's phrase [web: plasmas.org, plasma-universe.com, per Investigator] is a different defining property (ionisation, collective behaviour), as the Investigator notes. Verdict: Historically superseded but instructive (the term "fourth state" was later reassigned to plasma and the low-pressure regime is ordinary rarefied gas dynamics). The scaling statement itself is Established.

Claim 2: The dark space around the negative pole is the mean free path of the residual gas

a. Consistency. Crookes's reasoning is: the dark space grows and shrinks as the vacuum is varied "in the same way that the mean free path" does, so it "is" the mean free path [read: p.3]. Same scaling (inverse in pressure) does not prove identity: it proves only proportionality. The statement "thickness ... is the measure of the mean free path ... of the molecules of the residual gas" [read: p.3-4] is a hidden-assumption problem on three counts: (i) the carriers that matter are charged particles (electrons) with a different collision cross-section from the neutral molecules; (ii) the dark space is a region of strong electric field (the cathode fall), whose length is set by the condition that an electron gains enough energy to ionise, so it depends on voltage as well as pressure; (iii) a single mean free path is the average distance to the next collision, not the length over which collisions are absent. Modern value: the product of pressure and cathode-fall length for normal glow in air with aluminium is of the order 0.2 to 0.5 Torr cm [memory, medium]; my web check for this number failed (the search returned only qualitative statements, e.g. https://en.wikipedia.org/wiki/Glow_discharge, the dark space is "sometimes referred to as the cathode fall"). With that value, at 1 Torr the dark space is a few millimetres; the neutral mean free path at 1 Torr is about 0.05 mm [calc: 6.6e-8 m x 760 = 5.0e-5 m]. So the dark space would be roughly 50 to 100 mean free paths, not one. This is [memory, medium] for the dark-space number. It means claim 2's identity is quantitatively wrong by about two orders of magnitude, while the 1/p scaling is right. Crookes gives no number to test: only "about an inch on each side" at an unquantified "good exhaustion" [read: p.3]. I cannot refute or confirm the inch figure. b. Experiments. The cathode-dark-space dependence on pressure and voltage was measured by later workers; the scaling with 1/p is correct (Paschen-type similarity: dark space length times pressure roughly constant) [memory, medium]. The "mean free path" identification was not borne out. c. Tests. Crookes's own check "confirmed by experiment" is the qualitative change of dark-space size with exhaustion; no measured pairs are given [read]. Others later measured it (see b) but I have no named source in the project's folder. d. If true. The theory of glow discharge would have to equate cathode-fall thickness with a neutral-gas mean free path. Mainstream has instead the electron-multiplication (Townsend) picture. Verdict: Split. The scaling (dark space grows as pressure falls) is Established. The identity "dark space = mean free path" is Refuted by calculation in its literal form (order of magnitude off, using the memory-medium cathode-fall product; the number needs a web check by the Arbiter) and is Historically superseded but instructive as a qualitative picture.

Claim 3: Molecules leave the negative pole in straight lines normal to its surface, cast shadows, and are deflected by a magnet like projectiles

a. Consistency. Straight-line travel, sharp shadows, and normal emission from a cup or hemi-cylinder pole (focus at the centre of curvature) are geometrically consistent with a collisionless stream when lambda exceeds the vessel size (Q1 numbers). The magnetic part has a flawed analogy. Crookes compares the magnet to gravity and the molecules to "iron bullets from a mitrailleuse" [read: p.16]. Gravity gives a parabola whose curvature is independent of the projectile's mass and, to first order, depends on speed only through the flight time. A magnetic force F = qv x B is perpendicular to the velocity, does no work and bends the path into a circular arc of radius r = m v / (q B). Slower particles bend more tightly, which fits his observation that adding gas, slowing the stream, increases the curvature [read: p.17]. His own stated reason, "the magnetism has longer time on which to act" [read: p.17], is only part of it: r itself shrinks with v. So the qualitative observations are consistent with the physics, but the projectile analogy is not the right mathematics. Worked example [calc]: an electron accelerated through 10 kV has v = 5.9 x 10^7 m/s (v/c = 0.20, so non-relativistic to a few per cent). In a field B = 0.01 T (100 gauss, plausible near a horseshoe magnet) r = m v / (e B) = 3.4 cm. A 30 cm tube would show clear bending, as in his Fig. 14. A neutral air molecule (29 atomic mass units) with the same kinetic energy but carrying the electron's charge would have r about 230 times larger, 7.8 m; a neutral molecule has no magnetic force at all. So the observed deflection quantitatively demands a charge-to-mass ratio of the electron's order, not that of a gas molecule. This is a point the lecture could not make (no mass or velocity given) and it settles which "molecules" Crookes was seeing. Crookes says it would be "of great interest" to find the law of deflection [read: p.17]; the law is r = m v / (q B), which was derived and tested later (J. J. Thomson 1897 [web, Wikipedia via sites.md]). b. Experiments. Hittorf 1869 and Goldstein 1876 report straight lines, shadows and normal emission [web: PMC10311274, Busch 2023, per the Investigator]; Crookes was not first on these observations (the lecture itself credits no one on them). Magnetic deflection: established. c. Tests. Hertz 1883 and others showed that electrostatic deflection was hard to see because of gas conduction; Schuster later did see it [web: Wikipedia via sites.md; [memory] medium]. Thomson 1897 gave e/m. All consistent with a stream of charged particles, not neutral molecules. d. If true. As Crookes stated it (the carriers are "molecules" of the residual gas), the molecular picture would require that gas molecules are charged and have e/m near the electron's. That fails, see (a); the mainstream picture of electrons is what the effects require. Verdict: Split. The phenomena (straight lines, shadows, normal emission, magnetic deflection) are Established. The interpretation that the carriers are gas molecules is Historically superseded but instructive (Crookes himself later accepted electrons, per his obituary quote [web, not seen in original]).

Claim 4: The molecules are negatively electrified (parallel streams repel) and their impact makes glass, diamond, ruby and other minerals phosphoresce

a. Consistency. The parallel-streams argument [read: p.18-19]: current carriers act like parallel wires and attract; charged bodies repel. This is valid for his comparison class: two neutral wires carrying parallel currents attract. But a free beam of charges moving in parallel at speed v feels electrostatic repulsion F_e and magnetic attraction F_m with F_m / F_e = v^2 / c^2 (Gaussian-style derivation: current element field B = v E / c^2). Worked [calc]: for 10 kV electrons v^2/c^2 = 0.039, so the net force is repulsive, 96% of the electrostatic value. So the test distinguishes "current in a wire" from "free charged beam", and the beam result (repulsion) is what the physics predicts. Hidden assumptions: Crookes assumed the rays sit in neutral space; in reality positive ions from ionised residual gas partly neutralise the beam (space-charge neutralisation), and the net force depends on pressure. The text gives no angle, voltage or pressure for Fig. 18 [read: p.19]. So his repulsion is qualitative; I cannot compute the expected deflection. The claim of negative electrification is consistent with the sign of deflection in a magnet (the direction would be given by the sign of q) but he gives no direction convention [read]. Phosphorescence: the lines of ruby are given as "about 6,895" with units not stated [read: page image p.7; deep read]. If in Angstrom units this is 689.5 nm. Modern Cr^3+ ruby R lines are at 692.9 and 694.3 nm [memory, medium; not web-checked]; so the lecture's value is 0.7% low, within pocket spectroscope accuracy. An exact unit check cannot be done as units are not stated. The brief's 6,595 is an OCR error [read: deep read]; I confirm 6,895 is the more plausible value. b. Experiments. Negative charge of cathode rays: Perrin 1895 collected charge; Thomson 1897 measured e/m [web: via sites.md Wikipedia and Busch 2023]. Cathodoluminescence of ruby, diamond, and glass is established. c. Tests. Perrin, Thomson, Schuster are named in secondary sources; the primary papers were not obtained by the Investigator. Replicated widely. d. If true as stated (negatively charged "molecules" of gas). Nothing in mainstream changes for charge sign; but the "molecule" identification does not stand (see claim 3). Puluj 1880-83 proposed particles torn from the electrodes, rejecting the fourth state, [web: Busch 2023]; this is contemporary dissent that was correct about the particles not being gas. Verdict: Split. Negative electrification and impact luminescence: Established. "Negatively electrified molecules of the residual gas": Historically superseded but instructive.

Claim 5: The stream carries strong mechanical force and heat (paddle wheel, radiometer vanes, red-hot platinum ring, melting at the focus)

a. Consistency. Two separate quantitative problems. (i) Momentum. A beam of electrons carrying current I at speed v transfers momentum per second (force) F = (I / e) m v if all are absorbed. Worked [calc]: I = 1 mA at 10 kV gives F = 3.4 x 10^-7 N, i.e. about 0.34 micronewton, about 3.4 x 10^-8 kg force or the weight of 0.03 mg. A mica paddle wheel with milligram vanes on glass rails would hardly move on this force at friction scales of a rail. So the observed wheel motion is hard to reconcile with momentum from electrons at induction-coil currents. This is the Starke 1900 / Thomson 1903 conclusion (momentum insufficient, motion mostly radiometric heating) [web: https://sciencedemonstrations.fas.harvard.edu/presentations/crt-paddle-wheel, per the Investigator's note; I did not read Starke's paper]. If the "molecules" were really gas molecules of mass 29 u, the momentum per unit charge would be larger by sqrt(M / m) = 230 at the same energy, which is what Crookes needed to explain the force; but they are not (claim 3). So the mechanism he gave (impact of heavy molecules, and recoil from the pole [read: p.14]) is not supported quantitatively. (ii) The hot-ring demonstration cuts against his reading. With no induction coil, a red-hot platinum ring on a battery spins the vanes as fast as the coil does [read: p.15]. With no accelerating field the only energy source is heat. At 10^-6 atm, lambda = 6.6 cm [calc], far larger than the vane to ring gap, so the force comes from free-molecular thermal effects (gas molecules leaving a hot surface faster than a cold one, "thermal transpiration" and radiometric force) rather than collisional momentum from a "stream". Crookes himself saw the heated-wire result and read it as radiant matter excited by heat [read: p.15]. Both readings exist side by side; the lecture's own result supports the thermal one. Heating and melting: power budget [calc]. A black-body surface at 2300 K radiates 1.6 x 10^6 W/m^2 (sigma T^4, sigma = 5.67 x 10^-8). A target of 0.1 cm^2 per side (two sides) at emissivity about 0.3 (my assumption, [memory] low confidence) radiates about 10 W. A beam of 1 mA at 10 kV carries 10 W; so a focused beam of a few mA at 10 to 30 kV could in principle reach the temperature of an iridio-platinum alloy (melting near 1800 degrees C [memory, medium], not given in the lecture [read]). That makes the melting demonstration order-of-magnitude feasible, with no violation of energy conservation, but it requires a focused spot and currents the lecture does not give. No temperature, current or alloy percentage is stated, so no check beyond this estimate is possible. b. Experiments. A modern 40 kV paddle-wheel tube moves as described [web: Harvard link above]. Electron-beam heating and melting of metals is established technology (electron-beam welding and melting) [memory, high]. c. Tests. Starke 1900 (experiment) and Thomson 1903 (calculation) address mechanism [web: Harvard]. I have not read either primary paper. They are interested neither in confirming nor refuting Crookes; no stake is known. d. If true as stated (momentum from molecular bombardment). Mainstream would need larger momentum transfer than electron beams carry; it does not need changing, because the heating mechanism accounts for the effects. Verdict: Split. The effects (wheel turns; focus melts metal) are Established. The mechanism "molecular impact momentum" is Tested, not supported (Starke 1900 experiments, Thomson 1903 calculation, as reported by one web source that I could not trace to the originals), and Crookes's own hot-ring result indicates a thermal contribution. The bare momentum estimate above is a rough supporting calculation, not a refutation.

Claim 6: The effects are the same for hydrogen, carbonic acid and air (only onset pressure differs), and at a millionth of an atmosphere a 13.5 cm bulb still holds about 10^18 molecules

a. Consistency. Arithmetic check [calc]: sphere volume = (pi / 6) x 13.5^3 = 1288.25 cm^3, printed 13.5^3 x 0.5236 = 1288.25 [read: page image p.22, footnote]. With 10^21 molecules per cm^3 (Stoney) the bulb holds 1.288 x 10^24 (printed 1,288252,350000,000000,000000), and 1.288 x 10^18 at one millionth; printed 1,288252,350000,000000. Both correct arithmetic from his premise. Filling time at 10^8 molecules per second: 1.288 x 10^24 / 10^8 = 1.288 x 10^16 s = 4.08 x 10^8 years [calc: 408.14 million years]; printed 408,501731 years (grouping in sixes = 408,501,731) [read: page image p.22]. My value is 408.14 million years against his 408.50 million: a 0.09% difference. He used 365 days per year with rounded seconds; with 365.25 d I get 408.14, with 365 d: 1.288249e16 / (365 x 86400) = 408.5 million [calc, recomputing quickly from his seconds]. So his figure agrees with 365-day years. Internally consistent. The premise is the problem. The modern Loschmidt constant is 2.65 x 10^19 per cm^3 at standard temperature and pressure [web: amusingplanet link above; consistent with 2.687 x 10^19 at 0 C [memory, high]]. Using it [calc]: the bulb holds 3.46 x 10^22 molecules at one atmosphere, 3.5 x 10^16 at 10^-6 atm, and filling at 10^8 per second takes about 11 million years, not 408 million. So Crookes's headline figures (10^24, 10^18, 408 million years) are inflated by a factor 37 by an old estimate; the "same order of magnitude" lesson (that even a millionth of an atmosphere still contains billions of billions of molecules) survives: 3.5 x 10^16 is still huge. The comparison with the Sun's age is not mathematically meaningful as stated; Crookes himself says "figures ... cease to have any meaning" [read: p.23]. The "three hundred trillions a second" figure he quotes at p.22 to 23 is conditioned on an unexplained picture of smaller molecules and I cannot check it [read]; I do not rely on it. Same-gas statement: claim that "phenomena are identical, only they commence at different pressures" [read: p.21] is consistent with scaling by mean free path: lambda at a given pressure differs by gas (diameter d^2 in the denominator), so onset pressure shifts. Qualitatively right. No data. b. Experiments. Number density at low pressure follows the ideal gas law n = p / kT, so the physics is established. At 10^-6 atm and 293 K, n = 2.5 x 10^19 x 10^-6 = 2.5 x 10^13 per cm^3 [calc, ideal gas, Loschmidt 2.65 x 10^19 at 273 K scaled by 273/293 gives 2.5 x 10^13], consistent with 3.5 x 10^16 in 1288 cm^3 within the temperature factor (3.5 x 10^16 / 1288 = 2.7 x 10^13 at 273 K). c. Tests. Crookes presents no separate test. Gas independence of cathode rays was later confirmed by Thomson 1897 (same e/m for different gases) [memory, high]. d. If true. Nothing changes: this is standard kinetic theory with an outdated molecule count. Verdict: Split. Gas independence and the structure of the counting argument: Established. Specific counts and the 408-million-year figure: Refuted by calculation (accepted Loschmidt value gives 3.5 x 10^16 molecules and 11 million years, a factor 37 lower); the error is in the 1868-era input, not in Crookes's arithmetic.

The Investigator's findings: what bears on the mathematics

Kept in the record whatever the verdict: - Bakerian Lecture 1879 and "Contributions to molecular physics in high vacua" (Phil. Trans. 1879): would contain the pressures, apparatus and "laws of magnetic rotation" the lecture lacks [web: archive.org, GMU PDF; not downloaded]. These could settle the quantitative gaps in claims 2 and 3. I have not read them, so my claims 2 and 3 verdicts rest on the lecture and my calculations. - Crookes's later statement "What I then called 'Radiant Matter' now passes as 'Electrons'" [web: Dyer 1920 obituary, per research note; original place and date not found]: the inventor's own reinterpretation; I treat it as consistent with my e/m calculation in claim 3. - Stokes's referee remark suggesting "a further clarifying experiment" (RR/8/152) [web, per research note; not read]: the contemporary expert saw a gap in the evidence; what the gap was is unknown to me. - Starke 1900 / Thomson 1903 on the paddle wheel: reported by Harvard's page; source of my claim 5 mechanism verdict. Primary papers not read. - Puluj 1883 (particles torn from electrodes, dismisses "fourth state"); Goldstein 1876 (ether waves); Hittorf 1869: rival and prior interpretations [web: Busch 2023]. Goldstein's ether-wave view was wrong about waves but right that the effects emanate from the cathode; Puluj's carriers (charged particles, not gas) are closest to the modern account. Puluj's characterisation of the fourth state as "transcendental" with spiritualist leanings is Puluj's, as Busch reports. - Busch 2023 quotes a photographic-fogging remark attributed to the 1879 lecture that is not in the project's copy [read: deep read; web: PMC]. I cannot resolve this from the folder. It may belong to a different printing or paper. - Plasma terminology: Crookes's phrase was later attached to ionised gas; mathematically the defining parameters differ (Knudsen number versus Debye length and ionisation fraction). Crookes's text describes the first, not the second. - Absence of evidence: the Investigator found no gap in the record; the notebooks (Science Museum MS/0410) are unavailable because they "register slight radiation" [web]. Those notebooks are the only source for numbers (pumps, gauge readings, currents) that would let a mathematician check claims 2 and 5 quantitatively.

Web checks done (budget about 6; 2 used)

• Loschmidt constant: confirmed at 2.65 x 10^19 per cm^3 (https://www.amusingplanet.com/2021/02/johann-josef-loschmidt-and-avogadros.html, search summary only). Stoney's 10^21 not verified on the web.
• Cathode-fall pressure-length product: check failed to return a number; the 0.2 to 0.5 Torr cm figure and the ruby R-line wavelengths remain [memory, medium]. The Arbiter should check them; claim 2's "off by about 100" and claim 4's "0.7%" depend on them.

Verdict summary