Rankings Primer Arbiter Physicist Mathematician Contrarian QE Sites log Claims Sources Research notes

Science Contrarian review: Langmuir-Irving (Tonks and Langmuir, "Oscillations in Ionized Gases", Phys. Rev. 33, 195, 1929)

Reviewer: Science Contrarian (Step 2). Standard: experimental evidence over consensus; same rigour for the inventors and the critics. Tags: [read] document, [read: page image p.N] rendered page, [web] with URL (from the Investigator's notes unless I say otherwise), [memory] with confidence.

Source gap, repeated: the project holds one document, a 17-page PDF whose text layer is 100% OCR. The theory (PDF pp.1-9) was re-read from page images by the deep reader. I rendered and read PDF pp.11-17 myself (the experimental part, which Step 1 had only as a digest), but not the figures in detail (Figs. 4-10 are read only at the level the captions and text describe). The Investigator downloaded nothing, so the June 1929 "Note" (Phys. Rev. 33, 990), the 1928 PNAS paper, and Merrill and Webb 1939 were seen as abstracts or metadata only; whatever I say about them rests on those abstracts.

Page-number convention: PDF page N = journal page 194+N (PDF p.11 = journal p.205).

Corrections to Step 1

"Frequencies in a ratio near 1:2" (claim 4, facts.json, digest) overstates the paper. The paper says "the frequency with the low scattering was almost twice that found with the high" and then gives 1080 and 660 MHz [read: page image p.14, journal p.208]. 1080/660 = 1.64, not 2. Under the plasma-electron law (frequency proportional to the square root of n) that is a density ratio of about 2.7 (1.64 squared), not 4. Also, the two numbers were taken at different anode voltages (26 V for low scattering, 28 V for high) because "the arc showed less tendency to stay in one or the other state" at 27 V [read: page image p.14]. So the "1:2" is the authors' loose phrase; the measured ratio is 1.64 and the two points were not taken at identical settings. Claims.md claim 4 should say "about 1.6:1 (the authors: 'almost twice')".
Claim 4 says the -1/2 slope relation "as the plasma-electron formula predicts" without the authors' own caveat. The paper says that there was "no way of finding the absolute value of the ion density in the present tube because of large edge effects on the electrodes", that only "relative values" from "an elaborate series of measurements" were used, and that "partial failure of some internal checks in these calculations shows that too much reliance is not to be placed on them" [read: page image p.15, journal p.209]. The 13-of-15 result is therefore a weak, relative-ionization consistency check on a log-log plot (Fig. 10), not a quantitative test of Eq. (6). Claim 4's wording should carry this caveat. Also: the check tests only that wavelength scales as n^(-1/2); it cannot test the coefficient 8980.
Claim 5 / facts.json "the beam-oscillation identification is problematic" is correct, but the digest's compression hides how the authors framed the whole experimental section. The closing section says "The experiments described are in general accord with the theory outlined. A closer checking ... await experiments under essentially simpler conditions" [read: page image p.16, journal p.210]. They do not claim to have shown that the oscillations cause the scattering. In the 300+ MHz section they say: "It was natural to seek some correlation between the oscillations and the scattering of primary electrons. Unfortunately the runs meant particularly to elucidate this point are inadequate, first, ... the range covered was not wide enough; second, because of indeterminateness in the 'temperature' of the scattered electrons, and third, because of the absence of high frequency checks on the crystal detector." The only correlation reported is from "some early measurements on the first tube": with 10 mA from a 0.0125 cm filament, "the oscillations disappear in the same range of anode potential that the scattering does, namely between 60 and 75 V" [read: page image p.14, journal p.207]. Step 1 should say that the causal link rests on that one early, undescribed measurement; the claim in claims.md ("the authors suspect") is accurate but facts.json's key_claim list does not mention that the authors themselves call their correlation runs inadequate.
Digest error on p.8 (sound speed) and "4.6e-5 cm" are already corrected by the deep read; I confirm the deep read's claims on the page images I looked at. Nothing further to correct in the theory part. I did not independently re-render pp.1-8; I rely on the deep reader's transcription and recomputation, which I judged coherent (every number reproduced to rounding). I spot-checked three myself: 8980 n^(1/2) gives 8.98e8 Hz at n = 1e10 (and 33 cm); the Doppler step on p.15 gives lambda = 5.1e8/(55e6 - 18e6) = 13.8 cm (paper: 14 cm); the 115 MHz fundamental under a sqrt(n) law should rise by sqrt(1.6) = 1.26 for a 60% emission increase (paper: 26%, observed 10%).
The "temperature-sensitive critical oscillation" in claim/key_claims (19-21 C) is as stated in the digest: 263, 231, 250 MHz at 19, 20, 21 C, "accurate to within 2%" [digest p.11-12; I read journal p.205, which confirms the 115 MHz wave jumps to near 900 MHz below 19.5 C]. Note the 263-231-250 sequence is non-monotonic in temperature, and the authors do not explain it. Step 1 should flag that it is unexplained, not a prediction of the theory.
Minor. facts.json lists the pickup as "zincite-tellurium crystal detector calibrated at 60 Hz". The authors themselves flag that the detector had no high-frequency calibration ("absence of high frequency checks on the crystal detector", journal p.207), so the amplitudes (millivolts on crystal) are relative and only the 60 Hz square-law calibration exists. This bears on every amplitude claim.

Corrections in emphasis to the Investigator's lead list

The Investigator's lead "Note on 'Oscillations in Ionized Gases', Phys. Rev. 33, 990" may be the authors' own correction of the two misprints. This is a hypothesis; the Note's content is unknown. Nothing in the record says whether it is an erratum. Treat as open [web: https://journals.aps.org/pr/abstract/10.1103/PhysRev.33.990 shows only metadata].
The Baalrud note says Langmuir's tube was about 3 cm in diameter; the 1929 paper's tube is an 18 cm bulb [read: pp.9-10]. These are different experiments (Baalrud's reference is Langmuir's 1925 scattering paper). Nothing to correct, but a reader should not equate the two.

Claims

The five claims in claims.md are reviewed under their own numbers.

Claim 1. Natural plasma-electron frequency nu = (n e^2/(pi m))^(1/2) = 8980 n^(1/2) Hz

a. Mathematics: consistent. Gaussian units, nu in cycles/s so pi appears instead of 4 pi (omega = 2 pi nu). Derivation by Poisson plus Newton in 1-D, then by Maxwell with J. J. Thomson's method; all steps check [read: deep read; page image pp.2-4]. b. Experiment: The measured frequency in this paper is not an independent check of the coefficient because the density was not measured absolutely (see Correction 2). The one absolute-value example is a rough comparison: n = 1e10 gives 33 cm in air against "27 cm to 81 cm found experimentally" [read: p.4]. That is a very wide band and the density is "a typical low value", not a measurement. Independent support: Merrill and Webb 1939 in a mercury arc with a movable probe report "The observed frequencies agreed well with the formula derived by Tonks and Langmuir" [web: https://journals.aps.org/pr/abstract/10.1103/PhysRev.55.1191, abstract only; 108 citing articles]. I could not read the full paper, so I do not know how they measured density. Later experiments (Wehner 1951, Emeleus 1955-56, Dawson 1959, per Crossref list) and ionospheric work (Tonks 1933 applied the plasma frequency to ionospheric critical frequency) are in the Investigator's list [web: Crossref]. I did not read them. From general knowledge: the plasma frequency is used today as a routine measured quantity (radio-wave cutoff in the ionosphere, Langmuir-probe-independent density diagnostics, microwave interferometry) [memory, high]. c. Tested: the formula has been tested widely after 1929 (see b). The 1929 paper's own test is weak. Replicated in substance: yes, in many later plasma settings [memory, high], and by Merrill and Webb in the same kind of arc [web, abstract]. d. If true: nothing changes. It is textbook plasma physics.

Strongest honest case: this is the one place where the paper is simply right and was proved so for decades afterward, including by experiments the authors never imagined. The credit for naming and deriving it (with Thomson's help, openly acknowledged) is deserved. The case is not "mainstream is wrong"; it is that mainstream is right because of this paper. Against: the 1929 experiments did not confirm the coefficient (no absolute density); the "match" with 27-81 cm is loose.

Verdict: Established (the formula). The 1929 experiments' contribution to establishing it is weak; later work carries the weight.

Claim 2. Zero group velocity; electromagnetic waves obey nu^2 = n e^2/(pi m) + c^2/lambda^2

a. Mathematics: consistent; the dispersion relation is omega^2 = omega_p^2 + c^2 k^2 in cycles/s form [read: deep read]. Zero group velocity follows because eqs. (5B), (5C) contain no space derivatives, i.e. a cold plasma [read: p.5]. b. Experiment: The dispersion relation for transverse waves (cutoff at the plasma frequency) is confirmed by ionospheric reflection and by laboratory microwave cutoff [memory, high]. The zero-group-velocity statement is a cold-plasma limit; with finite electron temperature the Bohm-Gross term gives non-zero group velocity (Bohm and Gross 1949) [memory, high]. The Investigator found Linder, Phys. Rev. 49, 753 (1936), "Effect of Electron Pressure on Plasma Electron Oscillations", a pressure correction that predates Bohm-Gross [web: Crossref DOI 10.1103/physrev.49.753; I read only the title]. Tonks and Langmuir themselves hint at thermal effects on p.6 (fast vs slow electrons) but keep the result cold [read: deep read]. The authors' estimate of the bulb resonance, 24.6 cm for 18 cm half-wave, is compared with observed values "so near ... that the decision as to whether some, at least, of the observed oscillations may be of this type must be reserved" [read: p.4] and they do not claim a match. c. Tested: the cutoff law, yes, elsewhere. The bulb-resonance suggestion was not tested here. d. If true: nothing; the cold-plasma limit is a standard idealisation, with the thermal correction known.

Strongest honest case: the authors state the limits of their own result (cold, no space derivatives), so there is no overstated claim.

Verdict: split. Dispersion relation: Established. Zero group velocity as an unqualified statement: Historically superseded but instructive (true in the cold limit; the thermal correction came in 1936 and 1949).

Claim 3. Plasma-ion oscillations and "electric sound waves" with v = (kT_e/m_p)^(1/2), changeover near 2 sqrt2 pi lambda_D

a. Mathematics: consistent. Eq. (12) gives omega^2 = k^2 omega_pi^2/(k^2 + a), identical to the modern cold-ion ion-acoustic relation omega^2 = k^2 c_s^2/(1 + k^2 lambda_De^2). Two misprints (sign of the screening exponent; one first occurrence of "2(2 pi)^(1/2) lambda_D") do not propagate [read: deep read]. The paper's lambda_D uses 8 pi (both species screening at T_e), which is sqrt2 smaller than the modern electron Debye length; the transition factor 2 sqrt2 pi lambda_D is therefore 2 pi lambda_De. Internal consistency is good. b. Experiment: Ion-acoustic waves are routinely observed and measured in laboratory plasmas [memory, high]. The Investigator found Hernqvist (RCA) 1955, "Plasma Ion Oscillations in Electron Beams", and Peterson and Puthoff 1959 on ion plasma oscillations [web: Crossref DOI 10.1063/1.1722036; 10.1109/t-ed.1959.14566; titles only]. The 1929 paper does not test the ion branch cleanly: a resonance at 1.38 MHz was found with a tuned circuit "although considerable voltage appeared across the crystal throughout the whole range", and then a check with a tuned radio amplifier failed ("the 1380 kc. wave had become so elusive that even the previous measurements of its wave-length could not be repeated"); the amplifier showed a noise continuum, which the authors read as "conclusively" a "continuous spectrum" [read: page image p.16, journal p.210]. They then say "Whether this was actually the short-wave limit is not known". So the 1.38 MHz vs 1.5 MHz agreement is within a loose factor (n was not measured; 1.5 MHz assumes n = 1e10, an assumed typical value). Also note: the "1.38 MHz" value would match n = 8.5e9 cm^-3 under the ion-plasma formula; that is not an independent density measurement. c. Tested: the theory yes, widely, later. The 1929 data: an unreproduced single resonance and a noise continuum, as the authors say. d. If true: nothing.

Strongest honest case: the authors saw, and reported frankly, an irreproducible resonance rather than claiming confirmation; the dispersion relation has since been verified many times [memory, high], and a 2009 mainstream paper proposes ion-acoustic instabilities near the discharge boundary as the cause of the anomalous scattering (Baalrud, Callen, Hegna, PRL 102, 245005) [web: https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.102.245005; open author copy at plasmatheory.engin.umich.edu, p.1 read by the Investigator from a screenshot]. So the "electric sound wave" branch of the 1929 paper is, eighty years later, a leading candidate for the mechanism of the problem the paper was written to solve. Against: the ion branch was not demonstrated in this paper.

Verdict: split. Theory (dispersion relation and ion-acoustic speed): Established. The 1929 experimental identification of the 1.38 MHz resonance as an ion plasma oscillation: Never properly tested (a single irreproducible observation, withdrawn by the authors as unconfirmed; not dismissed, simply not followed up by them in this paper).

Claim 4. Hg-arc RF oscillations 1e7-1e9 Hz as noise; wavelength scaling with n^(-1/2); two stable scattering states

a. Mathematics: the scaling lambda proportional to n^(-1/2) follows from Eq. (6) with lambda = c/nu. No internal contradictions. The two-state "1:2" ratio does not match the 1.64 measured (Correction 1); and under sqrt(n) a ratio of 2 would imply a density ratio 4, a fact the authors do not discuss. b. Experiment: A range of ten things need caution. - Detector: crystal detector with 60 Hz calibration only, no high-frequency check (the authors say so). Amplitudes are relative; the frequencies come from Lecher wires, which is a reasonable method for the time [read: journal pp.204-207]. Tonks's own 1932 paper on loaded Lecher systems documents the method [web: Crossref DOI 10.1063/1.1745024, title only]. - Possible pick-up artefacts: the oscillation could be in the external circuit (the authors list "circuit oscillations" as a possibility that Langmuir 1925 already considered, per Baalrud p.1 [web]). The paper's own evidence for it being in the plasma: position-dependent amplitude on the wall (Fig. 4), a two-electrode interference test with phase change "considerably more than 180 degrees" (Fig. 5), dependence on the potential of filament c (Figs. 6-7), independence of the frequency from collector potential from -80 V to +5 V, and survival when the primaries are deflected by a magnet or the collector is outside the beam [read: page images pp.12-14]. These are sensible controls for a 1928 experiment. They do not exclude a tube-geometry resonance because the 115 MHz wave disappeared and a 730 MHz wave appeared under magnetic deflection. - Reproducibility: the critical 230 MHz oscillation only existed in a 2 C window of the Hg appendix temperature, vanished with 2 mA of emission from filament g, and vanished when the heating current was reversed. By the time of the final section even the 1.38 MHz wave "had become so elusive". These are hallmarks of a marginally-tuned, geometry-dependent oscillator, not of a clean measurement. The authors say as much ("the other evidence is not clean cut"). - The 115 MHz identification as a beam oscillation fails the authors' own quantitative test: 10% frequency rise observed for a 60% emission increase, where a sqrt(n) law predicts 26% [read: page image p.13, journal p.206]. They write "The 230 megacycle wave may, of course, be related to the 115 megacycle as harmonic to fundamental", an after-the-fact fit. - The 55 MHz / Doppler argument: they first compute lambda = 14 cm from the Doppler relation, then withdraw it: "it seems to be impossible to identify the 55 with the theoretical 18 megacycle oscillation" [read: journal pp.209-210]. - The 300+ MHz range's -1/2 slope (13 of 15) is relative-ionization only; see Correction 2. The two outliers are explained by "secondary theoretical factors" unspecified. I treat this as weakly supportive and not tested quantitatively. It is, however, the paper's most direct experimental check of Eq. (6) and an honest one: they published the scatter and the caveat. - The two-state arc is, to me, the most interesting experimental observation: two stable states, distinguishable by the collector characteristic (Fig. 8) and by oscillation frequency, switchable by a bar magnet, hysteretic in collector and anode voltage, with the ion current differing by about 3% at large negative collector voltage (and "in other cases greater differences") [read: page image p.14]. Fig. 9 shows only low scattering at 10-15 mA, both at 20-25 mA, only high scattering at 35 mA. This is an unexplained bistability reported with data and no follow-up in the file. I did not find a replication in the Investigator's list. Hysteresis and bistability are real features of gas discharges [memory, high], so the observation is not implausible, but I found no source that treats this specific one. c. Tested and by whom: the authors (GE Research Laboratory) in 1928; thanks to H. M. Mott-Smith and S. Sweetser [read: p.16]. Replication of the existence of narrow-region periodic oscillations in a Hg arc: Merrill and Webb 1939, "Besides 'turbulent' disturbances with no measurable frequency, narrow regions showed 'stable periodic oscillations of considerable magnitude'" [web: APS abstract]. Their detection method is the same (crystal detector, Lecher wires), so shared detector artifacts would also be shared; but they used a movable probe, which is a stronger spatial test. The two-state arc and the 19-21 C critical oscillation: no replication found in what was provided; the Investigator did not search for them specifically. d. If true: nothing. Oscillation noise in arcs is routine [memory, high].

Strongest honest case: the experiment is a careful early radio-frequency study with explicit controls, honest failure reports, and observations (two-state arc, 2 C temperature window) that the theory does not predict. An independent group ten years later found oscillations localised in narrow regions in Hg arcs, in agreement with the formula. Against (stated of the inventors with the same rigour): amplitude data are relative with an unverified detector at high frequency; several features are irreproducible within the paper; two identifications (115 MHz beam, 55 MHz Doppler) are withdrawn or fail the authors' own numerical test; density was not measured absolutely.

Verdict: split. Existence of RF oscillations in Hg arcs and of the n^(-1/2)-type frequency dependence: Plausible, untested in the strict quantitative sense for the 1929 data, upgraded by Merrill and Webb 1939 (abstract) to supported qualitatively; I do not have the full text to promote it to Established. Two-state arc, 2 C critical window: Never properly tested (single laboratory, no replication on record found; not dismissed, simply not followed up in the material I have).

Claim 5. The oscillations may explain the anomalous electron scattering and energy gain ("Langmuir's paradox")

a. Mathematics: the paper makes no quantitative scattering calculation; "suspect" is its own word. Dittmer 1926, cited by them, said amplitudes of "20 volts or more" would be needed and he "was unable to detect the oscillations themselves" [read: deep read, p.1-2]. The 1929 paper does not give an amplitude estimate. The order-of-magnitude estimate that displacing electrons by one spacing gives 8.3 V/cm [read: p.2] shows strong fields are possible for small displacement but does not demonstrate 20 V amplitudes. This is a gap in the argument. b. Experiment: the paradox itself (a Maxwellian electron distribution reached over a distance much shorter than the Coulomb collision length) is documented by Langmuir 1925 and described by Baalrud 2009 [web: p.1 of the open copy, read by the Investigator]; the name "Langmuir's paradox" is from Gabor, Ash and Dracott 1955 [web: https://www.nature.com/articles/176916a0, metadata only]. Evidence for the oscillation hypothesis: (1) the 1928 early measurement that oscillations vanish over the same anode-voltage range as scattering, 60-75 V [read: journal p.207]; (2) Merrill and Webb 1939: oscillations in narrow regions coincide with the scattering regions, "The results are interpreted as showing that scattering is due to plasma oscillations which receive their energy from the fast electrons" [web: abstract]; (3) the two-state arc, where high vs low scattering coincide with different oscillation frequencies (and amplitude jumps abruptly with the state) [read: journal p.208]. Evidence against or complicating: (4) the authors themselves call their correlation runs inadequate; (5) Gabor 1955 proposed sheath oscillations instead [web: per Baalrud 2009]; Baalrud 2009 proposes ion-acoustic instabilities; (6) Godyak and Alexandrovich 2015 found a clear deviation from the Maxwellian under "Langmuir paradox" conditions (1-3 mTorr argon) with high-resolution probes, attributing the low-energy peaks to non-local kinetics and standing striations, which means the premise of the paradox may be less universal than assumed [web: https://iopscience.iop.org/article/10.1088/0963-0252/24/5/052001, abstract read by the Investigator from a screenshot; argon positive column, not Hg arc; the "challenge the universality" phrase is from an academia.edu summary, which is secondary]. Baalrud et al. (2009) say the paradox "today remain[s] a serious discrepancy in the kinetic theory of gas discharges" and "None of the previous work has given a definitive answer." [web: open author copy p.1]. c. Tested: the authors' hypothesis was partially tested by Merrill and Webb (1939) with spatial resolution and found consistent. It has not been decisively proven or refuted. After Merrill and Webb the explanation moved to other oscillation types and to non-local kinetics. I found no source in the provided material showing it was refuted by a negative controlled experiment, and no source showing it was dismissed without testing. The record does not go silent: work continues in 1931-34 (Tonks), 1939, 1955-56, 2009, 2015. d. If true: nothing in mainstream physics would have to change; collective plasma effects on collisions are mainstream (Baalrud 2009 invokes exactly such). What would change is the standard treatment of the kinetics (Coulomb-collision-only) in low-pressure discharges.

Strongest honest case: the idea that collective oscillations, not binary collisions, thermalise the electrons is the same type of explanation that mainstream theory itself reaches for in 2009 (ion-acoustic instability enhancing the collective response). The 1939 experiment found spatially localised periodic oscillations coincident with the scattering regions, using a probe that could separate position from time. The authors in 1929 correctly spotted that the standard collision treatment fails and pointed at a class of mechanism that survives. Against: the authors' own correlation was inadequate; the specific mechanism (plasma-electron oscillation fed by fast electrons) is not the one proposed in 1955 or 2009; a modern high-resolution measurement challenges the universality of the anomaly itself; no amplitude-versus-scattering quantitative link was made in 1929; and I did not read the 1939 paper or the June 1929 Note, which might change this.

Verdict: Never properly tested in the sense that no neutral, controlled test has settled the specific proposed mechanism (the claim was tested partially in 1939; the result favoured it per the abstract; it was not refuted and not dismissed without testing; the field moved to competing mechanisms that remain unsettled). I would not label it "Tested, not supported", because the one independent experiment I know of favoured it, and I would not label it "Plausible, untested" because it has been tested in part.

How the claims were treated by others (question on rejection)

Nothing here looks like suppression. No evidence of blocked funding, omitted data, or hostile reanalysis appears in the Investigator's findings or the documents. The first-hand record is of a mainstream result (claims 1-3) that was incorporated into plasma physics, and of a mechanism hypothesis (claim 5) that was neither refuted nor confirmed and was gradually replaced by alternatives. The one oddity is that the paper's coined name "plasma" is, by Tonks's own account, attached to Langmuir; Tonks wrote "The Birth of 'Plasma'" (Am. J. Phys. 1967; abstract: "The author was there.") [web: Bing/ADS, not read]. That article is the first-hand source for how the work was received; I did not read it.

Investigator findings: what I kept and how I weighed them

Kept in this review: Merrill and Webb 1939 (support for claims 1, 4, 5; abstract only); Baalrud, Callen and Hegna 2009 (ion-acoustic instability as candidate mechanism; links claim 3 to claim 5); Gabor, Ash and Dracott 1955 and Martin and Emeleus 1956 (alternative oscillation mechanisms); Godyak and Alexandrovich 2015 (evidence against the premise; argon, not Hg); Linder 1936 (thermal correction to claim 2); Tonks's 1931-33 papers on high-frequency behaviour, plasma shape and ionospheric critical frequency (extensions of claim 1; "plasma shape" bears on the claim's statement that frequency is independent of displacement geometry, but I did not read it, and the 1929 paper itself says the shape matters for bounded plasmas, p.3, so I do not treat the 1931 paper as a refutation); the LOC Langmuir notebooks (lab records behind claim 4 and 5, not examined); Collected Works Vol. 5 (probable cleaner reprint, not downloaded). I did not discard any lead. None of these items were read beyond abstracts or metadata, apart from p.1 of Baalrud 2009; I say so wherever I use them.

What would settle the open points

Claim 5: a measurement in an Hg arc with a movable probe giving both the local oscillation amplitude and the electron energy distribution; the 1939 experiment is the template. Then compare amplitude with the Dittmer requirement ("20 volts or more").
Two-state arc: repeat at 20 mA / 27 V with a modern tube; a null result or a replication would settle it.
Claim 3 ion branch: use a tuned receiver on a fixed-density discharge with a measured density, and test the dispersion, as later ion-acoustic experiments did.
Read the June 1929 Note and the 1928 PNAS paper (free) to see whether they correct or extend the experimental part.

Summary of verdicts

Plasma-electron frequency: Established.
Zero group velocity and EM dispersion: dispersion Established; zero group velocity as a general claim Historically superseded but instructive (cold limit only).
Plasma-ion / ion-acoustic theory: Established; the 1.38 MHz experimental identification: Never properly tested.
Hg-arc RF oscillations: existence and n^(-1/2) trend Plausible, untested (quantitatively), supported qualitatively by Merrill and Webb 1939 abstract; two-state arc and 2 C critical window Never properly tested.
Oscillations as the cause of anomalous scattering: Never properly tested (partly tested in 1939 with favourable abstract; unresolved, competing mechanisms).