Arbiter (Step 3): Schuster and Hemsalech, "The Constitution of the Electric Spark" (1899)

Tags: [read] project document (6-page Proc. Roy. Soc. abstract, pp. 331-336); [web] URL; [calc] my own arithmetic (Python, .venv); [memory] with confidence. I read claims.md, facts.json, the three reviews, sites.md and the research notes. I did not re-read the PDF page images; I rely on the three reviewers' agreeing readings of the text layer and Table I.

Source gap note

The project holds only the 6-page abstract. The full paper (Phil. Trans. A 193 (1900) 189-213, DOI 10.1098/rsta.1900.0006) with all tables for Al, Mg, Bi, Cd, Hg and the photographs is NOT in the project and was read by no reviewer. Claims about metals other than zinc rest on one sentence each in the abstract. Everything said here about Royds 1908, Hemsalech 1901, Schaeffer 1908, Strutt 1913 and the referee reports rests on the Investigator's reading ([web], second-hand for the reviewers). Hull 1906 was read directly by the Contrarian from the archive.org text. Where a conclusion rests only on these, it is marked.

Plain-language setting (for the reader)

A spark between metal electrodes shows two kinds of light: from air (nitrogen, oxygen) and from the electrode metal. The authors photographed the spark's spectrum on film on a disc moving about 100 m/s. A line from a source that is fixed in place and brief draws a straight line on the film. A line from something moving along the gap is drawn slanted: slope = (source speed)/(film speed). So a metal line 5 times as steep as the film speed implies 500 m/s. Worked check [calc]: film speed at the mean radius 13.8 cm, 120 rev/s: 2 pi x 0.138 x 120 = 104 m/s; at the inner and outer edges (11.1 and 16.5 cm) 84 and 124 m/s. The one thing the slope cannot say by itself is WHAT is moving: the atoms, or a front of "being lit up" travelling through atoms that stay put. This is the real scientific question of the project.

Numbers where reviewers differ, recomputed

1. Gap-crossing speed: 5 mm (centre of a 1 cm gap) in 6 microseconds = 833 m/s; subtracting 0.5 microsecond of air luminosity gives 909 m/s [calc]. The quoted Cd mean is 560 m/s, which would need 8.9 microseconds. Physicist and mathematician say the abstract does not reconcile this; the Contrarian says 560 is "a different average". Arbiter: both are possible readings; the abstract does not say, so the mismatch (about 1.5x) stands as an unreconciled internal inconsistency of the abstract, not an error proven. Contrarian's point that facts.json misstated "crosses a 1 cm gap" is right: the text says "reach the centre" [read: p.6].
2. Sound-speed constant: c = sqrt(gamma R T/M), M = 0.002016 rho kg/mol gives k = 76.0 (gamma 1.4), 79.95 (gamma 1.55), 82.9 (gamma 5/3) [calc]. The authors' 80 equals gamma 1.55; monatomic Cd (M = 112 g/mol) is 5/3, so k = 83 and T from 560 m/s = 2544 K, from 80: 2744 K (printed 2700). All three reviewers agree; the Contrarian's 2530 K matches my 2544 K. At gamma 5/3, T = 2700 K, c = 577 m/s [calc]. Agreed: arithmetic fine.
3. Sensitivity of T to the 560 vs 830 m/s question: T goes as V squared, so with 830 m/s, T = (830/80)^2 x 56 = 6030 K [calc] (mathematician). Physicist did not give this; I adopt it. The "2700 K" is therefore uncertain by a factor of about 2 from the paper's own numbers, before the question of whether the slope is a gas speed at all.
4. Doppler scale: v/c x lambda gives 0.0095 A for 560 m/s at 5086 A, 0.0080 A for 500 m/s at 4800 A, 0.024 A for 1420 m/s at 5086 A; Hull's 0.002 A at 4800 A is 125 m/s [calc]. Hull's stated 150 m/s limit is consistent with his resolution. A coherent net drift of 500 m/s toward the observer would be 4x his limit. Reviewers agree; they differ on how much weight to give it (see claim 1).
5. [Corrected in the QE revision below: the figures in this item were wrong.] Diffusion: sqrt(2Dt) with D = 1 and 5 cm^2/s, t = 6 microseconds is 0.035 and 0.077 mm [calc, rechecked], i.e. 65 to 145 times too short for 5 mm. (Original text here said 0.35 and 0.77 mm and "7 to 15 times"; that was an arithmetic slip of a factor of 10.) The conclusion (pure diffusion cannot do it; pressure-driven flow is the right description) stands. The D value is [memory: medium], not checked.
6. Free expansion into vacuum: front speed up to 2c/(gamma-1) = 3c for gamma 5/3 [memory: high; standard gas dynamics], 1730 m/s for c = 577. So "speed of sound" is a lower-end estimate for the front; the authors' "not much difference between these views" is loose by up to a factor of 3 (mathematician). Physicist notes the same qualitatively.
7. Al/Zn: M^(-1/2) predicts 1.56x; observed "over 3x". Bi/Zn predicts 0.56x; observed up to about 2.8x on some lines [calc]. All three agree that a simple 1/sqrt(M) law fails. But the abstract also says Al and Mg (light) are fast [read: p.4], so the Contrarian is right that Royds' "lighter metals faster" is the first half of the abstract's statement, not a conflict. Both halves belong in claim 5.
8. Table I, zinc: 4811 exceeds 4925 in all nine cells by 10 to 67 percent (mathematician's ratios 1.10-1.67; I trust them, all three reviewers read the same table). The prose on p.3 says "uniformly higher velocity deduced from the double line 4925", which contradicts the table; the next sentence agrees with the table. A slip of wording. The inductance implied by a 2 microsecond period and 0.033 microfarad: 3.07 microhenry [calc].

New point settled by a web check

The Contrarian identified the "double line 4925" (authors' 4924.8) with Zn II. I checked NIST: Zn II lines at 4923.889 and 4924.013 A, the latter the stronger (relative intensity 500), and no Zn I line in 4920-4930 A [web] https://physics.nist.gov/cgi-bin/ASD/lines1.pl (query Zn I, Zn II, 4920-4930 A; the tool's summary text called them "doubly ionized" but the species label is Zn II, i.e. singly ionised). So the identification is confirmed to the extent that the only zinc lines there are Zn II. The authors' value is 0.8 A off, plausible for a prism wavelength scale [memory: medium]. The 4810.7 line as neutral Zn I [memory: high; not rechecked]. If right, the "4925 versus 4811" difference compares an ion line to a neutral line, which is a physical explanation neither the authors (who offer different molecular mass or a measuring artefact) nor the physicist (who mentions neutral versus ion generally) specifically tested. Retagged [web] for the identification. The Hull 1906 content I did not re-fetch (a search did not locate it); I rely on the Contrarian's direct reading of the archive.org text [web] https://archive.org/stream/proceedingsroya02britgoog/proceedingsroya02britgoog_djvu.txt, quoted by two other reviewers via the Investigator. Hull's text is therefore verified by one reviewer, second-hand by the others.

Corrected claim list (final)

1. Metal vapour crosses the gap at measurable speed (about 6 microseconds to the centre for Cd; 400-1500 m/s)

Agreement: all three accept the time-ordered observation (air lines first, metal emission later and further into the gap) and note the 830 vs 560 m/s mismatch and the absence of error bars. All three give a split verdict. Disagreement: how strongly Hull 1906 counts. Physicist: thin, unresolved, never properly tested. Mathematician: the matter-velocity reading is a hidden assumption. Contrarian: not refuted, neglected. Stronger argument: the mathematician's and Contrarian's. The slope fixes the speed of ANY front of emission, matter or excitation; the data cannot tell them apart. Evidence for the front reading: Hull's Doppler limit (150 m/s vs 500 m/s needed), and Royds from Schuster's own laboratory writing that the 1899 workers "did not discuss ... propagation of luminosity through vapour which may be, as a whole, stationary" [web, via Investigator]. The abstract itself concedes the oscillation period was too short and the metal "was possibly affected in its motion by the subsequent oscillation" [read: p.6]. Evidence for matter motion: physically sensible magnitudes (sound speed in hot vapour about 580 m/s [calc]); diffusion alone is too slow by about 65-145x [calc, corrected in QE revision], so some pressure-driven transport is required; modern vacuum-arc and ablation work treats electrode vapour expansion as established [memory: medium, via Investigator's IEEE 2018 listing, unread]. Against Hull's weight: his conditions (small capacity, time-integrated, viewing along vs across the axis) differ from the six-jar spark, and a symmetric outflow from both poles could cancel a net centroid shift while broadening lines. Hull reports no broadening finding; unverified. Final verdict (split): - (a) Delayed arrival of metal emission across the gap, observed as slanted lines, with speeds of order 10^2-10^3 m/s as apparent propagation speeds: Established (repeated by Royds in the same laboratory, and the method became standard; not independent in the strict sense). - (b) These are speeds of bulk vapour (matter) rather than of a luminosity front: Never properly tested. A 1906 preliminary Doppler note under different conditions did not support it; it was not dismissed without testing, but the neutral test (gated Doppler on a six-jar spark) is not on record. - (c) The 6 microsecond / 560 m/s pair: Internally inconsistent as stated in the abstract by about 1.5x, unless two different averages are meant (not stated).

2. A film on a spinning disc records metal lines slanted, air lines straight; slope measures the motion of the emitting particles

[Attribution of the calcium H and K observation below corrected in the QE revision: it is the 1899/1900 authors' own result as summarised by Royds, not Royds's own finding.] Agreement: kinematics trivially correct; method sound; interpretation is the issue. Claims.md wording should read "was taken by the authors to measure" the motion of the emitting particles (physicist correction; I adopt it). Additional points kept: V1 (pole to 2 mm) and V2 are two different quantities mixed in the text; no velocity-versus-distance profile; film speed varies about 40 percent across the band (84-124 m/s) and the abstract does not say it was corrected for [calc]; calcium H and K lines (impurity in silver electrodes) lit first at the gap centre and propagated TOWARD the electrodes (see QE revision: this is listed by Royds 1908 p.333 among the results of the 1899/1900 work) [web, Royds 1908 p.333, via Investigator; documented in one secondary summary by a colleague in Schuster's laboratory; the full paper and plate not seen]. Calcium coming from the electrodes yet lighting from the middle outward is awkward for a pure "vapour streams out of the poles" picture and fits a front reading better; kept in the record whatever the verdict. Final verdict (split): the method (timing the appearance of emission on a moving film): Established. Slope = velocity of emitting particles: Not testable as stated from the film image alone (the two readings produce the same image), and Never properly tested by an independent probe at the 1899 conditions.

3. The discharge begins through air; metal vapour then carries the later oscillations, against Feddersen 1862; evidence: self-induction makes air lines nearly vanish

Agreement: the self-induction effect is real and became standard in spark analysis (physicist: Established; Contrarian and mathematician: replicated in qualitative form). Circuit physics is sound: period 2 pi sqrt(LC), peak current falls as sqrt(C/L). Worked: L 3 to 30 microhenry with 0.033 microfarad: period 2.0 to 6.3 microseconds, peak current down 3.2x [calc, mathematician]; the 6 microsecond scale is comparable to the Cd transit time, which makes the authors' argument sensible. Disagreement: physicist rates the whole claim Established; mathematician and Contrarian separate the mechanism ("all later oscillations run through vapour") as untested. The mathematician and Contrarian are right: the abstract measures spectra, not the current path; no energy budget is given; Hemsalech 1901 (interested party, co-author) found the effect is not universal (Fe, Mn, Co, Ni lines strengthened, Cd, Zn, Mg, Al, Sb, Sn, Bi, Pb, Cu, Ag weakened) [web, via Investigator, https://archive.org/details/recherchesexpri01hemsgoog]. The electrode origin of the vapour is stated three different ways: both poles (1899 abstract), positive electrode only (Hemsalech 1906, as summarised in Nature 1908), negative pole (Schaeffer 1908), none of the three primary papers read by any reviewer. Feddersen's view is called "partly true, especially with magnesium" by the authors themselves [read: p.5]. Final verdict (split): the air-first sequence and the suppression of air lines by self-induction: Established (replicated by others; Schaeffer 1908 credits the discovery [web, via Investigator]). The mechanism that the later oscillations are carried by the vapour: Plausible, untested (no direct current-path measurement). The energy statement ("a great part of the energy may do useful work heating metal molecules"): Not testable as stated. Electrode origin: unresolved between sources; Never properly tested in a neutral comparison.

4. The vapour speed approximates the speed of sound, V = 80 sqrt(T/rho); about 2700 K for Cd; the spark molecule is not much lighter than the normal vapour molecule

Agreement: all three: arithmetic correct and constant 80 equals gamma about 1.55 (monatomic 5/3 gives 83). Disagreement: how much the temperature and mass inferences carry. Mathematician: strong form not derivable, because speed alone cannot give both T and mass. Physicist: temperature plausible; mass inference weak because Cd vapour is monatomic. Contrarian: correct at the level the authors claim ("seems a possible value"). Stronger: the mathematician's. T is an output of the assumed premise (slope = gas speed = sound speed), not a measurement; it changes by 2.2x if 830 rather than 560 m/s is used [calc], and the front of an expansion can run to 3c [memory: high]. Against 2700 K: spark channel electron temperatures are of order 10^4 K or more [memory: medium; checked by no reviewer]; for: Hemsalech 1901 argues vapour temperature does not greatly exceed arc temperature, partisan source. These are different temperatures (gas kinetic vs electron excitation), so they need not conflict (physicist). Retag: the spark-channel-temperature figure remains [memory: medium], unchecked; the "Historically superseded" half of the verdict below rests on it plus the premise problem (see QE revision, item 3). Final verdict (split): the sound-speed relation applied as an order-of-magnitude estimate (a few thousand kelvin): Plausible, untested. The specific 2700 K and the conclusion about molecular mass: Not testable as stated (one speed cannot fix both), and Historically superseded but instructive (a reasonable 1899 inference).

5. Velocity has no simple relation to atomic weight; different lines of one metal differ; velocity falls with capacity at short gaps (flagged unsettled by the authors)

Agreement: table trends are as described: at 0.51 cm, 4925 gives 814, 556, 416 m/s for 2, 4, 6 jars and 4811 gives 1014, 668, 529; at 1.03 cm flat; at 1.54 cm erratic (435 to 1526, with "?" marks); the authors say "very curious" and draw no conclusion [read: p.3-4]. A factor of 2 in V implies a factor of 4 in T under their own formula, which they did not remark on [calc, mathematician]. All reviewers: honest, tentative, flagged by the authors. Disagreement: verdict labels (physicist: Never properly tested; mathematician: Plausible, untested; Contrarian: Not testable as stated for the atomic-weight statement). Arbiter: the atomic-weight statement is a null statement about scattered data and so is Not testable as stated; the line-to-line difference and the capacity trend are testable and Never properly tested. Explanations on the table, all untested: (i) doublet measuring artefact (authors); (ii) the 4925 line is Zn II and 4811 Zn I, so ion and neutral emitters differ (Contrarian; line identity now supported by NIST, see above); (iii) the period lengthens as sqrt(C) with more jars, so a luminosity pattern timed by successive half-cycles mimics a velocity change (physicist and Contrarian; the authors' own "periodic time was rather too small" [read: p.6]). I consider (iii) the most testable, (ii) the most interesting, neither established. Final verdict (split): "no simple law with atomic weight": Not testable as stated; line-dependence and capacity trend: Never properly tested; "lighter metals faster" as a rough trend (Al, Mg fast; Bi anomalous): Plausible, untested.

Absence-of-evidence analysis

The record does NOT go dark for the work as a whole: it was refereed (Crookes 13 Feb 1899: "The paper marks an important step in understanding the electric spark"; Abney recommended Phil. Trans. [web, https://makingscience.royalsociety.org/items/rr_14_230, archivists' summaries; Crookes also asked about metal purity per the Contrarian]), published, continued by Hemsalech (1901 thesis), Royds (1908, 1910), Strutt (1913), and cited in the 1911 Britannica and 2018-2022 plasma papers [web, via Investigator]. Referees are fellows of the same society, so their praise is a contemporary assessment, not a test. No suppression is indicated. Where it does go quiet is one specific question: after Hull 1906's Doppler note and Royds 1908's remark, no Doppler or other discriminating test of "matter vs luminosity front" at the 1899 conditions has been found. Competing explanations: (1) the question was considered settled or beside the point as spark spectroscopy turned to analytical use (the self-induction technique of de Gramont and others) and, after 1913, to atomic structure (Bohr; Stark effect) and ionisation; (2) the tools to test it (Fabry-Perot interferometers, gated detectors) arrived later and the topic moved to plasma physics under other vocabulary (vacuum-arc cathode jets, laser ablation plumes) where the answer is "both: plasma jets move at 10^3-10^4 m/s and ionisation fronts also propagate" [memory: medium]; (3) the answer was published but not found by the Investigator (Strutt 1913 and Royds 1910 were not opened, Hull's follow-up was not searched). Evidence for (1)-(2): the work's later citations are methodological and historical. Evidence for (3): the Investigator ran 3.5 minutes of a 10-minute budget and left 8 sites unreached. What would decide: Royds 1910 Phil. Mag.; Strutt 1913 (Proc. R. Soc. A, DOI 10.1098/rspa.1913.0013, page failed to load); Hull's later papers; modern streak-plus-Fabry-Perot studies of spark plumes. The Hull note itself is preliminary, one page, and the Doppler-vs-line-broadening question was not examined in it (Contrarian).

What would settle it

1. Matter vs front (claims 1, 2): a time-gated high-resolution spectral measurement (Fabry-Perot or modern spectrometer with streak camera or gated camera) of a 6-jar (0.033 microfarad) spark in cadmium or zinc, viewed along and across the gap, looking for centroid shift and line shape at the same time the film shows the slope. Matter at 500 m/s gives 0.008 A shift at 4800 A [calc]; a pure front gives none. Also resolves Hull's geometry issue.
2. The 830 vs 560 m/s mismatch (claim 1): read the full 1900 paper to see whether the 6 microsecond figure and 560 m/s average are from different definitions, and read the velocity-vs-distance data.
3. The 4925 vs 4811 difference (claim 5): the line list in the full paper (confirm Zn II vs Zn I), then repeat with a Zn I line pair in the same multiplet. A same-ion pair showing the same velocity would exclude (ii) and (iii) splits.
4. Capacity trend (claim 5): vary C and L independently at constant period (add inductance when removing jars). If velocity follows period and not C, the trend is a timing effect.
5. Electrode origin (claim 3): film records with the polarity reversed or the anode and cathode of different metals (e.g. Zn and Cd), reading which metal's line moves first.
6. Temperature (claim 4): an independent gas temperature, for example from Doppler width of the Cd line with the front reading ruled out, or Boltzmann plots, compared with 2700 K.

Open questions

• Why do calcium H and K lines (from an impurity in the electrodes) start at the gap centre and spread outward (reported in Royds 1908's summary of the 1899/1900 results; full paper not read)? Documented in one summary; plate not seen by any reviewer.
• Are the Zn lines compared by the authors really ion versus neutral, and does that account for the difference?
• Which electrode supplies the vapour: both, positive only, or negative?
• Do the 1899 velocities survive recomputation with the film speed gradient (84-124 m/s across the band)?
• Is there any later gated Doppler test following Hull?

Corrections to Step 1 and to the reviews

• claims.md item 2: "measures" -> "was taken by the authors to measure"; mention the Hull and Royds challenge.
• claims.md item 5: add the abstract's first half ("Al and Mg ... higher molecular velocities", then "no simple law"); Royds's "lighter metals greater velocity" is compatible, not conflicting.
• facts.json: "crosses a 1 cm gap" -> "reaches the centre (5 mm)"; mainstream_conflict "none evident" -> "interpretation contested (Hull 1906)"; how_read note that no deep read exists is wrong (deep/TheConstoftheElectSpark-4e82a8.md exists and read the page images for p.3 and p.5); add the p.3 wording slip.
• Physicist's diffusion estimate ("about 0.1 mm, fifty times too short") is roughly right; the earlier correction to "7 to 15 times" was itself wrong and is withdrawn. Correct range: 65 to 145 times for D of 1 to 5 cm^2/s [calc]. No change to the physicist is needed.
• The physicist's statement that cadmium vapour is monatomic and the mathematician's implication that Cd2 might appear: Cd vapour is monatomic [memory: medium], not rechecked on the web; verdicts do not hinge on it (claim 4's weak inference holds either way).

Final verdicts in brief

QE revision (2026-10-04)

Source: qe.md (verdict MINOR). No deep reads were ordered; the one manifest document was already read in full with page images. I rechecked the QE items against the extracted text (_text/TheConstoftheElectSpark-4e82a8.txt, p.6) and research/2026-10-04-royalsocietypublishing.md. Tags as above. No verdict changes; two verdicts are re-annotated (below).

1. Diffusion arithmetic (A1). The QE is right and my earlier figure was wrong. sqrt(2Dt) with t = 6 x 10^-6 s: D = 1 cm^2/s gives 0.035 mm; D = 5 gives 0.077 mm [calc, rerun in Python]. Against 5 mm that is 144 and 65 times too short. The physicist's "about 0.1 mm, fifty times" was roughly right, and my "correction" of it is withdrawn. The conclusion (pure diffusion cannot carry vapour 5 mm in 6 microseconds; pressure-driven flow is needed) is stronger, not weaker. D = 1 to 5 cm^2/s is still [memory: medium], unchecked on the web, but no verdict turns on it: to cover 5 mm in 6 microseconds by diffusion alone, D would need to be x^2/(2t) = (0.5 cm)^2/(2 x 6e-6 s) = about 2 x 10^4 cm^2/s [calc], four orders of magnitude above ordinary gas values.

2. Attribution of the calcium H and K observation (A2). Royds 1908 p.333 lists it as point (5) of his summary of the 1899/1900 results (calcium impurity in silver electrodes "became luminous first in the centre of the gap, and were propagated towards the electrodes"), and the Investigator notes it is "not in the abstract; in the full paper" [web, via Investigator, page image seen by the Investigator only]. So it is most probably Schuster and Hemsalech's own observation, recorded in the full 1900 paper and reported by Royds. I earlier credited it to Royds; corrected above. Caveat: I rely on Royds's summary and the Investigator's note; the full 1900 paper has been read by no one here, so whether the observation was theirs or Royds's own addition to a list of their results is not directly checked. It is documented in one secondary source from a colleague in the same laboratory; no second source found yet. Weight: it is the observation in the authors' own work that fits a luminosity front (excitation travelling through vapour already present, or vapour first heated at the centre) better than a simple outflow from the poles; the authors' abstract did not draw that conclusion.

3. The authors' own statement that "the metallic vapours remain luminous in the centre of the spark for a longer period than near the poles", with luminosity traceable for about 1.5 x 10^-5 s with six jars [read: p.6]. Added to the record for claims 1b and 2. It is compatible with matter that has arrived and is staying hot at the centre, and with a front that left the centre emitting longest; it does not separate the two readings. It also supports the point that the observed picture was not a simple stream from the poles. Effect on verdicts: none for the labels; it adds weight to "Never properly tested" for the matter reading rather than to "Established".

4. Fuller p.6 timeline, for claim 1. [read: p.6] The air glow lasts about 5 x 10^-7 s, "the metallic vapours then begin to diffuse and reach the centre ... in a time which in the case of cadmium was about 6 x 10^-6 second"; the period with six jars and minimal self-induction was about 2 x 10^-6 s; luminosity traceable for about 1.5 x 10^-5 s. The wording "then begin" allows the 6 microseconds to be counted from the start of the vapour motion (909 m/s as computed above) or from the discharge (833 m/s); neither equals 560 m/s, so verdict 1c stands. The 5 x 10^-7 s is below the film's time resolution of about 1 microsecond at roughly 100 m/s [calc, physicist's figure, not rechecked], so the air-glow duration is an inference, not a resolved measurement.

5. Bismuth and mercury (claim 5). [read: p.4-5] Bismuth was measured with 2 jars, not 6, and its line 3793 was slower than the zinc and cadmium lines; mercury was tried with amalgamated zinc and cadmium poles; magnesium was too scattered to measure, and Feddersen's view is called "partly true, especially with magnesium". The bismuth 1420 m/s figure is therefore not directly comparable with the six-jar values and the "anomaly" is partly a change of condition. Verdict labels for claim 5 unchanged.

6. Spark temperature (A3). Claim 4's counter-figure (spark-channel electron temperatures of order 10^4 K) remains [memory: medium] and was not web-checked by any role. I did not run a further web check in this revision. The "Historically superseded" half of verdict 4 therefore rests on that memory figure together with the premise problem (one speed cannot give both T and M; the T output scales as V^2 and shifts 2.2x between 560 and 830 m/s [calc]); the "Not testable as stated" half rests on the premise problem alone. Also unchecked: cadmium vapour monatomic [memory: medium].

7. Referee note. Crookes asked about metal purity [web, makingscience.royalsociety.org item, archivists' summary]. That bears on the calcium result, which is an impurity line, and on line-to-line differences in claim 5.

8. Still open (not resolvable from the project folder). The full 1900 paper (Phil. Trans. A 193, 189-213, DOI 10.1098/rsta.1900.0006) and Royds 1908 pp. 334-347 remain unread because the Investigator did not download them. Verdict 1c (6 microseconds vs 560 m/s), the electrode-origin question, the Zn II identification of 4925 and the calcium passage all depend on them. Recommended: an Investigator recheck that obtains them, then a second arbiter revision.

Verdict changes

None of the final labels change. Annotated: claim 1 (b) and claim 2 (added the authors' own p.6 observation and the calcium result, re-attributed); claim 4 (superseded half flagged as resting on [memory: medium]). Numerical corrections: diffusion 65 to 145 times too short.