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

Reviewer: PhD physicist (Step 2). Project "Willams-Crookes". One source document: Crookes's Sheffield lecture to the British Association, 22 August 1879, reprinted in Popular Science Monthly vol. XVI (23 pages, text read in full by the deep reader; no source gap). I re-read the dark-space passage (text layer, lines 112-140) and the molecule-count passage (lines 850-910) of the extracted text myself, and I use the deep reader's page-image values for the three OCR errors. Tags: [read] = the lecture or the project's own files; [web] = URL given; [memory] = my recall with confidence; [calc] = my own calculation, shown.

How to read this for learning: the lecture is a set of live demonstrations. Nearly every observation Crookes reports holds up. What changed is the explanation of what the stream is. He thought it was a stream of negatively charged gas molecules (a "fourth state of matter"). It is a beam of electrons, which J. J. Thomson identified in 1897. Most verdicts below split into "the observation" and "Crookes's explanation".

Corrections to Step 1

I checked claims.md, facts.json and the deep read against the extracted text and against the Investigator's notes.

1. Step 1 is accurate on the three OCR corrections (6,895 not 6,595; 1/20,000,000 not 1/6,754,000; 10^24 and 10^18 molecules, not 10^15 and 10^12). The extracted text layer still shows "6,595" (line 271) and "2 millionth" (line 853, an OCR slip for "1"); I rely on the deep reader's page-image reading. [read]
2. Step 1 and the deep read say the lecture gives no pump or gauge. True [read]. The Investigator supplies the pump (Gimingham's multiple-tube Sprengel pump, 1876) from an 1890 obituary and secondary sources [web: https://ietarchivesblog.org/2017/05/10/charles-and-edward-gimingham-light-bulb-pioneers/]. That is a plausible attribution, not something in the lecture. Do not read it back into the lecture as fact. The gauge Crookes used (a McLeod gauge) is [memory, medium]; I did not verify it, and no verdict depends on it.
3. facts.json lists ten "key_claims" while claims.md lists six. This is fine (claims.md is deduplicated). I treat claims.md as the list of record. facts.json folds the straight-line/shadow claim and the phosphorescence claim into separate items; I keep them separate in claim 3 and claim 4 below.
4. facts.json "mainstream_conflict" says the beam is "now treated as electrons, discovered later". Correct, and I add the quantitative reason in claim 3 below. It also says "Crookes's observations agree with later accounts". Mostly true; the exceptions are the mechanism of the paddle wheel (claim 5) and the dark-space identification (claim 2).
5. The Investigator's lead that Busch (2023) attributes a photographic-plate fogging remark to the 1879 lecture is not in the project's copy [read: deep read]. I searched the extracted text for photograph/plate and found nothing relevant to fogging; this lead remains open. It is not needed for any verdict.
6. The Investigator's plasma lead needs one correction of emphasis, not of fact: modern sources do call Crookes's "fourth state" the first identification of plasma [web: https://www.plasmas.org/what-are-plasmas, as reported in sites.md]. But Crookes's defining property was a collisionless neutral gas, and a plasma is defined by free charges and collective behaviour. The phrase was borrowed for a different property (claim 1).
7. One arithmetic cross-check that supports Step 1: Crookes's "1/20,000,000 atm ~ a hundredth of an inch in a barometric column three miles high" is self-consistent. 0.254 mm / (3 miles = 4.83 x 10^6 mm) = 5.3 x 10^-8 = 1/19 million. [calc]
8. Step 1 did not flag that Crookes's molecule count rests on an old estimate that is about 40 times too high (claim 6). That is new.

Reference numbers used below [calc]

Air at 20 C (T = 293 K), molecular diameter d about 3.7 x 10^-10 m (a textbook value, [memory, high]), 1 atm = 101,325 Pa, k = 1.381 x 10^-23 J/K.

• 1 millionth of an atmosphere = 0.101 Pa (Crookes: 0.00076 mm Hg [read: page image p.7]; 0.00076 mm Hg = 0.101 Pa, agrees).
• Mean free path (average distance between molecular collisions) lambda = kT / (sqrt(2) pi d^2 p). At 0.101 Pa: lambda = 6.6 cm. At 5 x 10^-8 atm (0.0051 Pa): lambda = 1.3 m.
• Number density n = p / kT. At 1 atm: 2.5 x 10^19 per cm^3. At 10^-6 atm: 2.5 x 10^13 per cm^3.
• Bulb of 13.5 cm diameter: volume (pi/6)(13.5)^3 = 1288 cm^3 (Crookes: 13.5^3 x 0.5236, same).

Claim 1. At about one millionth of an atmosphere the residual gas is a "fourth state of matter" (radiant matter), as far from a gas as a gas is from a liquid.

What Crookes says [read: p.2-4]: below a certain pressure the molecules' mean free path becomes comparable to the vessel, "we have no longer to deal with a continuous portion of matter ... but we must here contemplate the molecules individually". He borrows the name and the four-state scheme (solid, liquid, gaseous, radiant) from Faraday's 1816 and 1819 lectures [read: p.1-2].

Mainstream view, stated fairly. The physical regime Crookes describes is real and important. When lambda is larger than the vessel, gas molecules fly wall to wall and almost never meet each other. Modern physics calls this the free-molecular (Knudsen) regime, and it governs vacuum technology, spacecraft and the upper atmosphere. Crookes was right that continuum gas dynamics fails there and that molecules must be treated one at a time. He was also right that new effects appear (the light radiometer works only in a pressure window where lambda is comparable to the vane spacing; [memory, high]). What mainstream physics does not accept is the label "fourth state". A state (phase) of matter is separated from its neighbours by a phase transition or a qualitative change of constituents. A gas pumped to lower pressure changes continuously: lambda just grows as 1/p. There is no threshold at "one millionth". Crookes says himself that the best effects occur at "roughly" one millionth and then fade "until the spark refuses to pass" [read: p.7], a gradual crossover, not a transition. Also, the stream he studied was not the residual gas at all. It was electrons coming off the cathode (claim 4). The modern "fourth state" is plasma, defined by free charges. Langmuir named it in 1928 [web: https://en.wikipedia.org/wiki/Plasma_(physics), per sites.md]. Crookes's tube does contain a plasma (the glow and positive column at higher pressure), but the radiant matter he demonstrated is a beam, not the plasma.

a. Mathematics: there is no mathematics, only a kinetic-theory statement lambda is inversely proportional to the number of molecules [read: p.2], which is correct. Internally consistent. His analogy "as far from gas as gas is from liquid" is rhetoric and cannot be checked. b. Experimental consistency: the observations (phenomena change character as pressure falls) are consistent with all vacuum-tube experience. The state label is not an experimental result. c. Tested? The observations were reproduced by every cathode-ray worker after him. Nobody tested "fourth state" as a thesis, because the cathode-ray work reinterpreted the phenomena. Contemporary dissent: Goldstein and the German ether-wave school; Puluj (1883) rejected the fourth state, calling it, in Busch's report of his view, "transcendental" with an inclination towards spiritualism, and proposed charged particles torn from the electrodes [web: https://pmc.ncbi.nlm.nih.gov/articles/PMC10311274/, per sites.md]. That is a rival's characterisation (Puluj had a priority and theory stake). Crookes's own later statement, quoted in obituaries, is "What I then called 'Radiant Matter' now passes as 'Electrons'" [web: Dyer 1920, J. Chem. Soc. obituary, as reported in sites.md; original place and date not found]. This is the inventor himself revising the claim. d. If true: very little would have to change for the weak form (it is already mainstream). The strong form, a distinct phase, would need a new phase boundary in the pressure-temperature diagram of gases; none exists.

Verdict (split): - Weak form (a collisionless molecular regime exists below a pressure where lambda exceeds the vessel size, with new phenomena): Established. - Strong form (a fourth state of matter distinct from gas, made of the residual gas molecules): Historically superseded but instructive. It was a reasonable organising idea in 1879 with Faraday behind it; the beam turned out to be electrons and the regime turned out to be a gas limit.

Keep in the record: modern plasma sources credit Crookes with the "fourth state". Theosophy took the lecture as support for "more refined states of matter" (Blavatsky; Theosophy Wiki snippet [web: https://theosophy.wiki/en/Radiant_Matter, page blocked, snippet only]). That reception is outside science. The lecture's closing line about "the border-land where matter and force seem to merge" [read: p.23] is a metaphysical flourish with no measurable content.

Claim 2. The dark space round the negative pole is the mean free path of the residual gas molecules, so at high vacuum the molecules travel from the pole without collisions.

What Crookes says [read: p.3-4]: the dark space grows and shrinks with the vacuum "in the same way that the mean free path of the molecules lengthens and contracts ... We naturally infer that the dark space is the mean free path ... an inference confirmed by experiment." In Fig. 1 it extends "about an inch on each side of the pole" at "good exhaustion" (pressure not stated) [read: p.3].

Mainstream view. What Crookes saw is now called the cathode dark space (the Crookes dark space), the region next to the cathode where the electric field is strong and electrons are being accelerated but have not yet gained enough energy to excite the gas, so there is little light. Its thickness d scales roughly as 1/p, which is the same scaling as lambda, so his observation (grows as the vacuum improves) is correct and his scaling instinct is good. The identification is not: the dark space is set by how far an electron must travel to reach the energy needed to ionise or excite, in a strong field near the cathode, not by the free path of neutral molecules. In modern terms the two lengths are of the same order but not equal. [memory, medium-high: standard glow-discharge physics, e.g. Raizer, Gas Discharge Physics.] At the pressure of Crookes's best phosphorescence, lambda for air is 6.6 cm [calc], comparable to a 13.5 cm bulb, so for neutral molecules his claim that they travel the vessel without collisions is right to the right order of magnitude. At 1/20,000,000 atm lambda is 1.3 m, longer than any of his tubes [calc]. So the weak claim in the claim statement ("at high vacuum the molecules travel without collisions") is correct, and in fact more strongly so than he could show. What is wrong is saying that the visible dark space measures that distance.

Note on the number: "an inch" (2.5 cm) of dark space against lambda = 6.6 cm at 10^-6 atm. These are not equal and Crookes gives no pressure for the tube, so no quantitative check of his equality is possible from the lecture. [calc, read]

a. Mathematics: the only formula is lambda inversely proportional to molecule number [read: p.2], correct. The step "dark space ~ 1/p, so dark space = lambda" is a plausible but unproved identification. Not a calculation error. b. Experimental consistency: consistent with the 1/p scaling of the cathode fall region in glow discharges ([memory, medium-high]). Inconsistent with the later finding that the beam is electrons, which have a different (much longer) free path than molecules. c. Tested? Crookes says "confirmed by experiment" but shows only the qualitative growth of the dark space with exhaustion. No measurement of dark-space width against pressure is in the lecture [read]. Later discharge physics gave the correct model. Stokes refereed the two 1879 technical papers and recommended publication, and on the second suggested "a further clarifying experiment" (Royal Society RR/8/151 and RR/8/152 [web: https://makingscience.royalsociety.org/items/rr_8_151]; the remarks themselves were not read by the Investigator and my attempt to fetch RR/8/152 failed with a certificate error). This shows a leading physicist took the work seriously in 1879; it is a record of reception, not a test. d. If true as stated: the glow-discharge sheath would have to be set by neutral-gas kinetics rather than by electron acceleration and ionisation; this would contradict discharge theory built from Townsend onwards.

Verdict (split): the scaling observation (dark space widens as pressure falls) is Established; the identification of the dark space with the neutral-molecule mean free path is Historically superseded but instructive.

Claim 3. The stream leaves the negative pole in straight lines normal to its surface, independent of the positive pole, casts sharp shadows, and is deflected by a magnet like projectiles.

What Crookes shows [read: p.5-18]: shadows from a hinged aluminium cross; a rod-like phosphorescent line along a long tube, curved "like a flexible wand" by a horseshoe magnet; a cup-shaped pole that focuses the rays; a heavier, slower deflection when a trace of gas is admitted; and a contrast with the low-vacuum discharge, which snaps back to the positive pole while the high-vacuum stream "continued its path in the altered direction". The projectile analogy is the mitrailleuse (an early machine gun) with the magnet playing the part of gravity [read: p.16].

Mainstream view. Every observation is correct and is textbook cathode-ray physics: straight-line propagation in a field-free region, emission normal to the cathode surface, shadow casting, focusing by a concave cathode, deflection by magnetic fields (the Lorentz force F = qv x B). Hittorf (1869) and Goldstein (1876) had reported straight lines, shadows and normal emission earlier [web: https://pmc.ncbi.nlm.nih.gov/articles/PMC10311274/, per sites.md], so priority for this item is shared and Crookes's contribution is a more systematic and public demonstration. Plücker and Hittorf had earlier reported magnetic deflection of the discharge (same Busch source). Wikipedia credits the hinged-cross tube to Plücker 1869; the lecture credits Gimingham's workmanship [read: p.12], and Plücker died in 1868 [memory, high]. The attribution is unresolved; it needs a primary source.

Where "molecules" fails, by calculation. A particle of charge q and momentum p = sqrt(2mE) moves on a circle of radius r = p/(qB) in field B. For a 10 kV beam and a field of 0.01 T (100 gauss, a plausible horseshoe-magnet field a few cm from the poles; the field is not stated in the lecture): - electron: p = 5.4 x 10^-23 kg m/s, speed 5.9 x 10^7 m/s (0.2 c), r = 3.4 cm [calc]; - singly charged nitrogen molecule (mass 4.65 x 10^-26 kg): r = 7.6 m [calc]. A tube is about a metre long, so only the electron gives the strong "wand-waving" bending Crookes shows; a gas-molecule beam would bend by a few millimetres at most. The ratio of radii is sqrt(m_N2 / m_e) = 227. This is consistent with Thomson's 1897 measurement that the carriers have a mass about 1/1800 to 1/2000 of hydrogen [web: https://en.wikipedia.org/wiki/Cathode_ray]. So the claim "these molecules are deflected by a magnet like projectiles" is right as a description and wrong as to the carrier. The projectile picture itself (velocity, slowing by residual gas causing more curvature) is qualitatively correct for electrons: collisions with the residual gas reduce the electrons' energy, which reduces r = p/(eB), and the curve tightens. Crookes's reading of that effect was right.

a. Mathematics: no equations in the lecture. The projectile analogy is physically sound for any charged particle in B. b. Experimental consistency: fully consistent with later cathode-ray work (Hittorf, Goldstein, Hertz, Perrin, Thomson, Lenard). Hertz's 1883 failure to see electrostatic deflection was due to poor vacuum, as he later accepted (surface charges masked the field) [web: https://en.wikipedia.org/wiki/Cathode_ray]. This is a case where a negative experiment was an artefact; it is relevant to how dissent against Crookes arose (the ether-wave school argued from Hertz's null result). c. Tested? Yes, by Crookes and by many later workers, with replication (the Harvard demonstration tube reproduces the shadow/deflection family of effects [web: https://sciencedemonstrations.fas.harvard.edu/presentations/crt-paddle-wheel]). The lecture itself gives no numbers (no field, voltage, deflection angle), and Crookes says it would be "of great interest" to find the law of the deflection [read: p.17]; he did not give it in this lecture. d. If true: nothing would have to change for the observations. For the claim that the carriers are gas molecules, the Lorentz-force numbers above would have to be wrong by a factor of 227 in radius.

Verdict (split): the observed behaviour (straight lines, normal emission, shadows, magnetic deflection) is Established. The inference that the moving objects are the gas molecules is Refuted by calculation (the deflection radius for molecules is about 200 times larger than observed) and was later superseded by the electron identification. Priority is shared with Hittorf and Goldstein.

Claim 4. The carriers are negatively electrified (two parallel streams repel, so they are not current carriers), and their impact produces phosphorescence in glass, diamond, ruby and other minerals.

What Crookes shows [read: p.5-7, p.18-19]: glass glows apple-green (soft German), blue (English) or dark green (uranium); diamond "the most sensitive substance I have yet met"; ruby glows a rich red with one intense line "a little below the fixed line B ... a wave-length of about 6,895" (page image p.7; the units are not stated); alumina does the same, citing Becquerel (1859). The two-negative-pole tube (Fig. 18) shows the second stream push the first one away, and he concludes the streams act "not like current carriers, but merely as similarly electrified bodies".

Mainstream view, charge. Correct: the beam is negatively charged. Two parallel beams of the same charge feel electric repulsion and magnetic attraction; the ratio of the magnetic to the electric force is (v/c)^2, so for v = 0.2 c it is 4 percent and the net force is repulsive [calc; standard Griffiths E&M result]. Crookes's logic (a pair of parallel current-carrying wires would attract, a pair of like-charged beams repel) is therefore sound: the repulsion shows that what is seen is a charge distribution moving fast, not a neutral-wire current. Practical caveat: in a gas the beam creates positive ions that partly neutralise its space charge, and I would expect that to reduce the repulsion; the lecture gives no angle, voltage or pressure for Fig. 18 [read], so the effect cannot be quantified. Later independent tests of the sign: Schuster (electrostatic deflection, 1890), Perrin (charge collection in an electroscope, 1895), Thomson (e/m, 1897) [web: https://en.wikipedia.org/wiki/Cathode_ray; primary papers not obtained by the Investigator]. Hertz's null result in 1883 was a poor-vacuum artefact.

Mainstream view, phosphorescence. Light emitted by materials under electron bombardment is cathodoluminescence (sometimes called "phosphorescence" in the 19th-century sense; strictly fluorescence if the light stops when the beam stops). Soft German glass, diamond, ruby (Cr3+ in alumina), phenakite and emerald glowing under a beam is routine. The best intensity near 1 x 10^-6 atm (0.1 Pa) is reasonable: at higher pressure the beam is stopped in the gas, at lower pressure the discharge cannot be sustained, as Crookes states [read: p.7].

Check of the ruby line. The ruby R lines (Cr3+) lie at 694.3 nm and 692.9 nm (6943 and 6929 Angstrom) [memory, high]. Crookes's "about 6,895" is therefore 0.7 percent below the true value if the unit is Angstroms. This is within what one would expect from a pocket spectroscope read against the fixed line B (B at about 6870 Angstrom, [memory, medium]; the line is "a little below B" in his words, meaning toward the red side in wavelength order—i.e. close to B). The figure is a rough visual estimate and the unit is [memory, medium] only; I would not read precision into it. This is a good example for a learner: a number in a lecture, read from a pocket spectroscope, is right to about 1 percent. The later "radiant matter spectroscopy" papers (1883-1887) used the characteristic emission as an analytical tool for rare earths [web: archive.org identifiers listed in sites.md]; the technique is a precursor of cathodoluminescence spectroscopy.

a. Mathematics: no formulas; the repulsion argument is qualitatively correct (v/c squared). b. Experimental consistency: fully consistent with Schuster, Perrin, Thomson; with modern cathodoluminescence of ruby and diamond [memory, high]. c. Tested? Charge: yes, independently and quantitatively by 1897. Phosphorescence: yes, by Crookes in many materials and later widely. Crookes's own later statement that radiant matter "now passes as electrons" is the inventor's own revision of this claim [web, see claim 1]. d. If true: nothing. The correction is that the carriers are electrons rather than molecules (Thomson 1897). "Negatively electrified molecules" is superseded.

Verdict (split): negative charge and the cathodoluminescence are Established. "Negatively electrified molecules of the residual gas" as the identity of the carriers is Historically superseded but instructive.

Claim 5. The stream carries strong mechanical force and heat: it turns a paddle wheel and radiometer vanes (also when driven by a red-hot platinum ring), and at its focus melts wax, glass and iridio-platinum.

What Crookes shows [read: p.13-21]: a mica-paddle wheel runs up a glass railway when the upper vanes are struck (reversing the poles reverses the motion, and it can run "up hill"); an "electrical radiometer" turns when its fly is the negative pole, which he explains as recoil of the pole from the molecules driven off it; vanes at 45 degrees spin at "extreme velocity" under the induction coil, and "as fast" when the coil is disconnected and the platinum ring below is heated red-hot by a battery; a focus melts a patch of wax, perforates glass, and melts iridio-platinum.

Mainstream view, heating. Fully correct, and a calculation shows the energy budget is ample. A 10 kV beam at 1 mA carries 10 W [calc: V x I]; Crookes's coil currents are not given [read], so this is an assumed, typical order of magnitude. Focused on a patch of 1 mm^2 the beam could heat a metal until it radiates sigma T^4 per unit area: at 2300 K a black body radiates about 1.6 W per mm^2 [calc], so 10 W on a few mm^2 can reach the melting point of an iridio-platinum alloy (about 1800 C; [memory, medium] for the alloy fraction, not stated in the lecture). Electron-beam melting is a modern industrial process; so heating of the focus is Established. Glass is perforated by atmospheric pressure once spot-softened, as Crookes describes.

Mainstream view, mechanical force. The wheel does turn; a present-day tube does it at 40 kV [web: https://sciencedemonstrations.fas.harvard.edu/presentations/crt-paddle-wheel]. The effect is real; the mechanism Crookes proposes (momentum of the bombarding stream) is not the main cause. Momentum rate of the electron beam: force F = (I/e) p = (6.2 x 10^15 electrons/s)(5.4 x 10^-23 kg m/s) = 3.4 x 10^-7 N at 10 kV and 1 mA [calc]. A mica-and-aluminium wheel of about a gram weighs about 10^-2 N, and rolling it uphill needs a force a large fraction of that. So momentum from the beam is about four to five orders of magnitude too small unless the current is far larger than a coil would give (assumed, not in the lecture). On the other hand a radiometric force (molecules leaving a hot surface faster than they arrive, so pressure on the hot face exceeds that on the cool face) in free-molecular conditions is about p x (delta T / 2T) per unit area: at 0.1 Pa and a 50 K difference, 8 x 10^-7 N per cm^2 [calc; standard radiometer estimate], which is comparable to or larger than the momentum force, and a vane face struck by a beam is heated. That is exactly the conclusion in the record: Starke (1900) measured the effect while varying current, pressure, surface and angle and "concluded that thermal effects were greater than the mechanical impulse"; Thomson (1903) calculated that electron momentum would turn the wheel at about one revolution per minute, and attributed the motion to the radiometric effect [web: https://sciencedemonstrations.fas.harvard.edu/presentations/crt-paddle-wheel; Starke, Ann. Phys. 3 (1900) 101; Thomson, Conduction of Electricity Through Gases, 1903, p.501; primary texts not obtained]. The platinum ring demonstration is the key piece: heat alone, with no cathode stream, spins the vanes. In the lecture Crookes takes it as showing that a hot wire sets "radiant matter" in motion [read: p.15]. In mainstream terms it shows that thermal molecular effects (radiometer force, plus thermionic emission from a hot wire, which is small at red heat, [memory, medium]) are enough, which supports the thermal explanation of the other vane demonstrations. Crookes's own explanation for the original light radiometer is not given in the lecture [read: deep read]; the thermal mechanism of Maxwell and Reynolds is the accepted explanation [memory, high].

a. Mathematics: no equations in the lecture. Energy and momentum numbers above are mine and rest on assumed currents and voltages. b. Experimental consistency: heating, yes. Force: the observation is consistent; the explanation by bombardment momentum is not consistent with measured forces (Starke) and the momentum estimate (Thomson). c. Tested? Yes: Crookes (demonstration), Starke 1900 (measurement), Thomson 1903 (calculation), Harvard (modern replication of the motion). The mechanism dispute is a case of an actual experimental test, not dismissal. d. If true as stated (momentum of the beam drives the wheel): the electron momentum would have to be about 10^4 to 10^5 times larger than p = sqrt(2mE), or the beam current vastly larger; neither fits.

Verdict (split): the focus heating and the fact that the wheel and vanes turn are Established. The strong mechanical claim (the force is the impact momentum of the stream) is Tested, not supported (Starke 1900; Thomson 1903; both reported in one secondary source, Harvard, and the calculation above agrees with them).

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

What Crookes says [read: p.21-23]: "Whether the gas originally under experiment be hydrogen, carbonic acid, or atmospheric air, the phenomena of phosphorescence, shadows, magnetic deflection, etc., are identical, only they commence at different pressures." For the count he uses a footnote citing Johnstone Stoney: 1 cm^3 of air holds about 10^21 molecules; 13.5^3 x 0.5236 x 10^21 = 1.29 x 10^24 in the bulb; at one millionth of an atmosphere, 1.29 x 10^18 remain; filling at 10^8 molecules per second would take 408 million years (the sun's life estimate) [read: page image p.22].

Mainstream view, gas independence. Correct, and this is a clue that the beam is not made of the gas. The electrons are produced at the cathode and travel in the near-vacuum; the gas only sets the pressure at which a discharge can run and the energy losses along the way, which matches "only the onset pressure differs". Later work (Hertz, Lenard, Thomson) found the same, which is what led Thomson to a universal particle. The word "chemical" in "At this low density the molecules retain their chemical characteristics" [read: p.21] is a statement about the residual gas, not about the beam, and is correct for the gas.

Mainstream view, the count. The method is sound and the arithmetic is internally correct (13.5^3 x 0.5236 = 1288 cm^3 [calc]; 1.288 x 10^24 / 10^8 per second = 1.288 x 10^16 s = 4.08 x 10^8 years [calc]). The input is wrong. The modern number density of air at 1 atm and 20 C is 2.5 x 10^19 per cm^3 (the Loschmidt number at 0 C is 2.69 x 10^19 [memory, high]), not 10^21; Stoney's figure is about 40 times too high [calc]. Corrected: the bulb holds 3.2 x 10^22 molecules at 1 atm and 3.2 x 10^16 at one millionth of an atmosphere [calc], not 1.3 x 10^18; the filling time becomes 10 million years rather than 408 million [calc]. For 5 x 10^-8 atm the bulb holds 1.6 x 10^15 molecules [calc]. The rhetorical point (an enormous number remain, and an enormous number still collide with the walls) survives a factor of 40. Crookes's honest hedge ("I have done the sum, but figures when they mount so high cease to have any meaning" [read: p.23]) stands.

a. Mathematics: internally consistent given its input; the input is out of date by a factor of about 40. The digit-grouping in sixes and the British long-scale "quadrillion" (10^24) and "trillion" (10^18) are a notation point; the figures are right in the lecture's own convention. b. Experimental consistency: gas independence is consistent with cathode-ray physics; the count is not consistent with the modern Avogadro/Loschmidt value. c. Tested? Gas independence was observed by Crookes and by later workers. The molecule number was not tested by Crookes; it was a borrowed estimate. d. If true: nothing for gas independence. For the count, nothing: it is a calculation with a wrong constant.

Verdict (split): gas independence of the beam phenomena is Established. The figure of 10^18 molecules is Refuted by calculation (it should be about 3 x 10^16 using the modern number density); the qualitative point it served survives.

Weighing the Investigator's findings (all kept)

• Crookes's own technical papers: the Bakerian Lecture (Phil. Trans. 170 (1879), DOI 10.1098/rstl.1879.0065), "Contributions to molecular physics in high vacua" (1879), "On Electrical Insulation in High Vacua" (1878), and the 1881 and 1883-1887 spectroscopy papers [web: archive.org identifiers in sites.md]. These were not read by the Investigator or by me, so this review is of the lecture only. They should hold pressures, apparatus and the "laws of magnetic rotation" the lecture lacks. Their absence is a limit on what I can say for claims 2, 3 and 6.
• Crookes's notebooks (Science Museum MS/0410), unavailable because "the volumes register slight radiation" [web: https://collection.sciencemuseumgroup.org.uk/documents/aa110067393/laboratory-notebooks-of-william-crookes]. The reason is stated in the catalogue (radioactive contamination); I see no sign of suppression, but the working record for the lecture's numbers is not currently reachable.
• Gimingham and the pump: Gimingham's obituary (The Electrician, 1890, as quoted by the IET) says he assisted Crookes for 12 years on "the radiometer and radiant matter" and invented a multiple-tube Sprengel pump in 1876. Wikipedia gives 10^-6 to 5 x 10^-8 atm for Crookes's range. My calculation shows 5 x 10^-8 atm is 0.005 Pa, which a Sprengel pump with chemical absorbents could plausibly reach, but I have no independent measurement. The pressures in the lecture are Crookes's own report (single source); I accept them as order-of-magnitude. The 1870-1882 versus about 1869-1881 date discrepancy is unresolved and does not matter for any verdict.
• Stokes's referee reports (RR/8/151, RR/8/152): contemporary expert endorsement; the "clarifying experiment" suggested on the second paper is unknown and would show what remained unsettled in 1879.
• Hittorf (1869), Goldstein (1876), Puluj (1883), Varley (1871): rival or earlier work on the same phenomena. Priority for straight-line travel, shadows and normal emission is shared. Puluj's electrode-matter theory is a closer ancestor of the electron picture than the gas-molecule picture; his characterisation of Crookes as "transcendental" is the opinion of a party with a rival theory.
• Crookes's own later view ("now passes as electrons"): the inventor's reinterpretation; the date and place of the quote are unconfirmed.
• Absence of evidence: none found. The radiant-matter work continued in print to 1887 and the technique fed into rare-earth spectroscopy.

Summary table

Limits of this review: web checks done: Wikipedia "Cathode ray" (Schuster, Perrin, Thomson, Hertz; fetched), and the Royal Society page RR/8/152 (failed: certificate error). The Starke and Thomson 1900-1903 results are taken from the Investigator's Harvard note and not from the primary texts. Ruby wavelength, glow-discharge dark-space theory and the Loschmidt number are [memory] and I flagged confidence. All calculations were done with numpy in the project venv; assumptions (beam 10 kV and 1 mA, field 0.01 T, air at 20 C) are mine because the lecture gives no coil, voltage or field values [read].