Arbiter: Frollov-Alexander (Alexander V. Frolov)

Inputs read: facts.json, claims.md, physicist.md, mathematician.md, contrarian.md; I also re-read the relevant passages of FreeEnergy.pdf myself (text layer). Tags: [read] from the document, [web] with URL, [memory] with confidence.

Source gap (repeated from claims.md)

The folder holds only two documents, both by Frolov: a 3-page 1994 note (single-wire transmission) and an 18-page 2003 magazine article (FreeEnergy.pdf). Both have a text layer and were read as text; figures were not viewed as images. Nothing else by Frolov (books, patents, later papers) and no primary document for any device he cites (Brown, Sweet, Searle, Avramenko, Strebkov, Godin-Roshin, Kozyrev, Bearden) is in the folder. Everything about those is Frolov's second-hand report. Where a conclusion below rests on [memory] alone, it says so.

Corrections to Step 1 that all reviewers raised (accepted)

1. (Superseded by the QE revision below: a deep read of PDF pp.10-16 has now been done.) FreeEnergy.pdf how_read should read "full text layer extracted; body read in full, patent/organisation appendices skimmed", not "text layer sampled".
2. Claim 5 in claims.md overstates Frolov. He writes "Ionization is not considered here!" [read, FreeEnergy.pdf ~line 525] and then "We have to exclude ideas on electrokinetic apparatus since it is just a reactive method and propulsion force is result of ionization flow" [read, ~line 545]. He also explains Searle's disk by Lorentz forces and treats unbalanced wheels as "two different systems" that classical theory can explain. So he is self-contradictory on ionization and only partly claims anomalies.
3. Missing sub-claim (contrarian, verified by me [read, ~lines 1180-1208]): Frolov says he personally tested a device in which switching on a load in a transformer secondary containing an arc made consumed power fall, not rise, at about 300 mA, and that to exclude phase-shift errors input was measured in a DC battery circuit. No numbers, waveforms or instruments given. I add this to Claim 1/3 as a sub-claim (1a below).
4. The same article says "in each receiver the power can't be more than power of this generator" [read, p.15], contradicting the 100%-plus claim and the 1994 remark that generator power "is not important".

Numbers recomputed (reviewers agreed, I checked)

• Spark energy: 0.5 x 1e-8 F x (2000 V)^2 = 0.02 J; x 5 Hz = 0.1 W; 0.1/3 = 3.3% of the 3 W DC input. Correct.
• Mean charging current: Q f = C V f = 1e-8 x 2000 x 5 = 1.0e-4 A (0.1 mA).
• Stray capacitance needed: C_s = I / (2 pi f V) = 1e-4 / (2 pi x 1e4 x 2000) = 7.96e-13 F, about 0.8 pF. Mathematician and contrarian agree. Caveat from me: this treats 2 kV as the voltage across the coupling capacitance and ignores waveform shape and rectifier action, so it is an order-of-magnitude statement only (real stray capacitances of a wire and nearby objects are of this order or larger, so the conventional explanation is adequate, but it does not prove that explanation is the right one).
• Resonance: the physicist (L = 1 mH, C = 1 nF, R = 10 ohm: Q = 100, 10 V drive gives 1000 V) and the contrarian (L = 100 uH, C = 1 nF, R = 5 ohm: Q = 63) used different values; both are right. Input power at resonance is V^2/(2R) (peak amplitude V) and equals the power dissipated in R, so power efficiency cannot exceed 100% in a passive linear circuit.
• Quarter-wave line: at 100 kHz, lambda = 3e8/1e5 = 3000 m, lambda/4 = 750 m (mathematician; correct).
• Air breakdown about 3 kV/mm: 0.5-1 mm gap breaks down near 1.5-3 kV, consistent with his 2 kV [memory, high].

No numerical disagreement between reviewers needed resolving; differences were in verdict wording only.

Web checks done by the arbiter (budget 8; search tool was exhausted, so WebFetch was used)

• Biefeld-Brown vacuum results: Wikipedia summary states Tajmar's 2004 test (electrodes in a box suspended on wires, excluding corona wind) found "No linear thrust", concluded the effect is corona wind; Talley (1988) found no thrust at 1e-6 torr under DC; the ARL report says vacuum studies failed to replicate Brown and attributes the force to corona wind [web: https://en.wikipedia.org/wiki/Biefeld%E2%80%93Brown_effect, via fetch summary; primary papers not opened]. This upgrades the contrarian's "[web, search summary]" and the physicist's "[memory, medium] later groups reported no thrust in vacuum" to [web]. Note a tension I could not resolve in budget: the physicist cites Bahder and Fazi (ARL-TR-3005) as saying ballistic ion wind is three orders of magnitude too small, mechanism not understood [web: https://arxiv.org/vc/physics/papers/0211/0211001v1.pdf, search summary only], while the Wikipedia summary attributes to the same ARL source the statement that follow-up studies blame corona wind. These can both be true (ballistic wind too small; ionic drift/charged-air momentum transfer the likelier mechanism), so I keep: force in air is real, mechanism incompletely settled but probably ionic, vacuum thrust not seen.
• Kozyrev: a 1960 Soviet Academy commission concluded his theory "is not based on accepted clearly formulated axiomatics" and the "quality and accuracy of conducted laboratory experiments do not allow drawing of specific conclusions" [web: https://en.wikipedia.org/wiki/Nikolai_Aleksandrovich_Kozyrev, via fetch summary]. The mathematician's "no conclusive reproducible result in independent replications" [web: researchgate 344839308, search summary] is consistent. Upgrade of the physicist's [memory, medium] to [web]. Replication details not checked.
• Tesla London 1892 (physicist, [web: https://teslauniverse.com/nikola-tesla/articles/mr-tesla-london-institution-electical-engineers-and-royal-institution]): lectures of Feb 1892 with single-terminal lamps and motors on one conductor at high frequency and voltage. I accept: single-terminal operation is historical; there is no evidence of net usable power without source cost. Not re-fetched by me.
• Avramenko plug as diode-capacitor rectifier block; VIESH 20 kW single-wire system reported by its authors; Strebkov group efficiencies 10-67% for small systems [web: https://viesh.ru/pre/development-of-resonance-methods-of-electric-power-transmission/ and https://link.springer.com/article/10.1007/s43236-022-00383-4, search summaries only, not re-opened]. Frolov's "20-100 kW" range: the 100 kW figure remains unconfirmed.
• Sweet VTA never independently certified [web: http://amasci.com/freenrg/sweetvta.html, search summary; advocate-adjacent site, so only the absence-of-certification point is relied on]. Not re-checked by me.
• Not checked: Searle/Godin-Roshin 7 kW and 100 kg force (no source found by any reviewer); Hubbard, Cooper, Hendershot. Conclusions there rest on Frolov's report alone and are labelled so.

Contrarian's distinction, kept visible

For every claim the record is one of three kinds: - Tested and failed: only Brown-type thrust in vacuum (Tajmar 2004, Talley 1988 [web]); Kozyrev's torsion experiments were reviewed negatively in 1960 [web]. - Never properly tested (the claim could be tested, nobody ran a neutral controlled test, and it was not dismissed after testing): Frolov's free-power claims, the arc-loading effect, Sweet's VTA, Searle/Godin-Roshin, seasonal output variation. - Dismissed without testing: I found no documented case for Frolov's own claims. The contrarian's finding, which I accept, is that Frolov never put a claim in testable form (no input-with-load versus input-without-load, no waveforms, no calorimetry), so there was nothing for critics to reanalyse, omit, or suppress. The one partial exception is Brown's air thrust, which was long dismissed as "ion wind" before a quantitative estimate (ARL) showed simple ballistic wind was too small; that is a Brown story, not Frolov's.

Claim-by-claim arbitration

Claim 1: potential field does net work around a closed cycle, without fuel

Agreement: all three say the weak form (time-varying fields do net work; someone pays for the switching) is textbook, and the strong form has no accounting of switching or screening cost. Curl E = 0 for statics, so the line integral around a loop is zero by Stokes. Disagreement: physicist and mathematician give the strong form "Refuted by calculation" plus "Never properly tested"; the contrarian gives strong form (static conservative field) "Refuted" and the weak "pulsed/gradient beats input" form "Never properly tested". Stronger argument: the contrarian's split is cleaner. There are really three statements: (i) static field, closed loop, net work: Refuted by calculation (zero by Stokes). (ii) time-varying field does net work: Established, with energy supplied by the switcher. (iii) a specific Frolov device gives output exceeding total input: Never properly tested (no numbers). The mathematician's screen-bookkeeping and the physicist's 10 pF, 1 kV tube example (0.5 C V^2 = 5 microjoules supplied when the capacitor was charged) both show where the energy comes from. Sub-claim 1a (arc loading lowers consumed power, DC-measured, ~300 mA): Never properly tested. It is the only place Frolov states he excluded a measurement artefact, which counts in his favour, but no data, circuit or instruments are given; a falling input in a battery-fed inverter with an arc load can come from changes in transformer loading or inverter behaviour, so input power must be shown with and without arc over a full cycle. What would settle it: log battery V and I (or calorimetry of all dissipation) with the arc gap open, closed and with load removed, same frequency. Final verdict: Weak form: Established. Strong form (static field, closed cycle, or any cycle with no accounted input): Refuted by calculation. Frolov's specific devices including the arc test: Never properly tested. (Sub-claim 1a was strengthened by the QE deep read: see "QE revision".)

Claim 2: single wire plus Avramenko plug, "wattless" displacement current, free power (2 kV, 10 nF, 5 Hz: 0.1 W out vs 3 W in)

Agreement: arithmetic right; output is 3.3% of input, so his own data show no gain; the missing control is generator input with and without load; "wattless" means reactive and carries no mean power, so "wattless becomes wattful for free" is a definitional move; single-wire capacitive coupling plus diodes is conventional (stray capacitance of about 0.8 pF suffices); Tesla's single-terminal lamps use the same principle; kW-scale systems exist (Strebkov, VIESH) with efficiencies below 100%. Disagreement: only on label for the strong form. Physicist: Never properly tested, with "wattless to wattful free" Refuted by calculation; mathematician: Never properly tested, plus "Refuted by his own data" and "Internally inconsistent" for the concept; contrarian: Never properly tested. Stronger argument: "Refuted by his own data" is not on the Step 2 scale and not quite right: output below input does not refute a claim of hidden free source power (3% could be both). What his data do is fail to support it. The fair split: (i) as a power-transfer technology: Established; (ii) "free" because the current is displacement: Never properly tested, and his numbers cannot distinguish free from hidden input; (iii) the concept that reactive (wattless) current can deliver net real power to a load without a source supplying that real power: Internally inconsistent as defined (mean power of a purely reactive current is zero; conversion needs a nonlinear element fed by the source), and at the level of energy bookkeeping, Refuted by calculation (real power at the load equals real power from the source minus losses). I take Internally inconsistent as the primary label for (iii) because the problem is his own definition. Also: "60 mA x 50 V = 3 W" for Avramenko's lamp is a product of values on a rectified, probably reactive waveform; P = V I cos(phi) <= V I, so it is an upper bound unless waveforms are shown. Not a power measurement. What would settle it: scope true power (mean of v(t) i(t)) at generator input with load connected, disconnected and diodes shorted/reversed, plus calorimetry of the load. Final verdict: Single-wire capacitive/resonant transfer: Established. "Free" power: Never properly tested (not dismissed without testing; never put in testable form). "Wattless becomes wattful at no source cost": Internally inconsistent.

Claim 3: resonance can raise output to efficiency above 100%

Agreement: all three: amplitude/voltage gain by resonance is Established; efficiency above 100% in a passive network is Refuted by calculation (Q example); Frolov's own text concedes receiver power cannot exceed generator power; "generator power is not important" (1994) confuses amplitude with power. Disagreement: none material. The mathematician adds Internally inconsistent; I accept that for the 2003 article (assertion of >100% against its own concession). Caveat kept from the contrarian: this refutes only passive linear networks. A system with a hidden active source is outside the calculation, but nothing in the documents offers one. Tellegen's theorem (power conservation in lumped circuits) is the precise statement. What would settle it: measure true input and output real power of any resonant transformer (done routinely; always below 100%). Final verdict: Voltage/amplitude gain: Established. Efficiency above 100%: Refuted by calculation; as written, Internally inconsistent with Frolov's own "can't be more than the generator" statement.

Claim 4: generalised 4-D conservation; reversed-time anti-entropy; vacuum/aether unlimited source; aether density sets rate of time (Kozyrev)

Agreement: no equations; "A + B = 0" is bookkeeping; undefined terms (reversed-time energy, aether density, rate of time); relabelling Kozyrev's "density of time" as "density of aether" adds nothing measurable; an extractable vacuum source needs a non-ground state; Kozyrev has no accepted replications. Disagreement: mathematician adds Internally inconsistent (conservation "preserved" yet "creation of energy" asserted; "unlimited source" vs "power can't exceed generator"); the physicist also lists the latter; the contrarian does not but notes that a season/day-modulated supply is not "unlimited". Mathematician rejects "Historically superseded but instructive" for Kozyrev; I agree, it was never an accepted position that was later replaced. Stronger argument: the three-part split is right and consistent: (i) the generalised-conservation statement, as written, cannot be evaluated: Not testable as stated; (ii) the specific consequence that device output varies with season/time of day because aether density does: Never properly tested (Frolov's only support is a remark that Sweet's output "was variable in different time of day and night" [read]; no logs); (iii) the pair of statements "unlimited source" and "power cannot exceed generator" and "conservation holds but energy is created": Internally inconsistent. What would settle it: a year-long logged output of any claimed device against a sealed control, with a pre-stated seasonal signature; and, for the theory, any operational definition of "reversed-time energy" with units. Final verdict: Reversed-time balance: Not testable as stated. Seasonal/aether-density variation: Never properly tested. The internal tension: Internally inconsistent.

Claim 5: others' anomalous devices are real examples, including reactionless thrust and gravity reduction

This is a bundle; verdicts are per device. Reviewers' disagreements: Searle/Godin-Roshin/Cooper/Hubbard: physicist and mathematician "Never properly tested", contrarian "Not testable as stated (no measurements given)". The 7 kW and 100 kg axial-force claims are quantitative and testable in principle, so "Not testable as stated" is the wrong label for the claim; what is missing is data in the article. I rule Never properly tested for the claims, noting that the evidence in the folder is Frolov's one-sentence reports only and that I found no independent test (neither success nor failure), so this rests on absence of sources. Brown: Agreement that a force on asymmetric high-voltage capacitors in air is real and that the mechanism is not fully settled (ARL measured it; ballistic ion wind appears too small [web, search summary], ionic drift probably responsible). Vacuum/reactionless thrust: Tested, not supported [web: Wikipedia summary of Tajmar 2004, Talley 1988]. "Efficiency million to one" has no definition (output/input with no measured input) and no data. Frolov himself concedes ionization flow for the electrokinetic apparatus. Mathematician's "Tested, not supported" for "no ionization" strong form agrees. Note one neutral test (Tajmar) plus one older (Talley) is thin but real; this was not "dismissed without testing". Sweet VTA: all agree Never properly tested; no independent certification, secret conditioning procedure [web, amasci]. Self-cooling 20 degrees while delivering electrical power would, taken at face value, need a cold reservoir or heat pump (second-kind perpetual motion by Frolov's own classification), and weight loss proportional to output would need a momentum sink; these conflict with thermodynamics and momentum conservation if real, but no data are given to test them. Searle disk: Frolov's own explanation is Lorentz force and ionization currents, which would be conventional: Plausible, untested for that conventional reading; free-energy/antigravity reading: Never properly tested. Unbalanced wheels and Archimedes-screw schemes: Refuted by calculation (uniform gravity is conservative; work around a closed cycle is zero, W = mg times net height change = 0); Frolov himself concedes a classical "two systems" reading. As historical claims (Bhaskar, Worcester) they are Historically superseded but instructive; the Worcester 1620 story has no source here [read; no web check]. Zaev dielectric cooling to electricity: electrocaloric cooling itself is Established [memory, high] but needs a field source; as single-reservoir heat-to-work it would be Refuted by calculation (second law); as claimed by Frolov, Never properly tested (no data). Capacitor with unequal plates pulling a dielectric ball (F proportional to grad E^2, dielectrophoresis): Established as a force [memory, high]; as a "no-fuel force against gravity", Refuted by calculation (the ball stops where the field is strongest; moving it back costs the energy). What would settle it: for Brown, thrust per watt of measured supply power (V x I including leakage) in air and in hard vacuum with a free-hanging box; for Sweet, an independent test with logged input/output power, temperature, mass on a balance and a sealed control. Final verdicts: Brown asymmetric capacitor force in air: Established (mechanism open, probably ionic). Reactionless/vacuum thrust: Tested, not supported. "Million to one" efficiency: Not testable as stated (undefined). Sweet VTA: Never properly tested. Searle (conventional reading): Plausible, untested; free-energy reading, Godin-Roshin, Cooper, Hubbard, Hendershot: Never properly tested (only second-hand reports). Unbalanced wheels, screws, unequal-plate ball: Refuted by calculation. Zaev: as above.

Corrected claim list (with final verdicts)

1. (Corrected) A time-varying potential field does net work per cycle, with the energy supplied by whoever varies it: Established. Frolov's strong form, that net work per cycle can come from the potential field itself with the "negative half" removed by pulsing, gradients, polarity change or screening and no accounted input: Refuted by calculation. His specific devices (tube with grid under cathode, pulsed-anode tube of Fig.2, unequal capacitor, arc-loading with falling DC input at ~300 mA): Never properly tested (Round 2: Figs 1-2 viewed; unlabelled schematics, no data; the pulsed-anode idea that the battery "is not discharging" ignores the energy already spent accelerating the electrons, so as a no-input-cost claim it is Refuted by calculation).
2. (Corrected) A single-wire line with a two-diode Avramenko plug delivers real power to a load through capacitive (displacement) coupling: Established as a conventional rectified capacitive-coupling technique. Frolov's home test gives 0.1 W out for 3 W in (3.3%), so it does not show gain. "Free" power: Never properly tested. "Wattless current becomes wattful at no source cost": Internally inconsistent.
3. (Corrected) Resonance raises voltage and amplitude (Established) but cannot give passive-network efficiency above 100% (Refuted by calculation; Frolov's own text says receiver power cannot exceed generator power).
4. (Corrected) "Generalised" four-dimensional conservation with reversed-time anti-entropy and an unlimited aether whose density sets the rate of time: Not testable as stated; its seasonal/daily output-variation prediction: Never properly tested; its simultaneous claims of unlimited source, conservation preserved, and power not above generator: Internally inconsistent. Kozyrev's underlying experiments: reviewed negatively in 1960 and without accepted replication [web]; not Historically superseded.
5. (Corrected; extended by QE revision, see items 6-8) Frolov selectively cites others' devices, conceding ionization for Brown's electrokinetic apparatus and Lorentz forces for Searle. Per device: Brown force in air Established (mechanism open); Brown reactionless/vacuum thrust Tested, not supported; Sweet VTA, Godin-Roshin, Cooper, Hubbard, Hendershot Never properly tested; gravity wheels and screws Refuted by calculation. Second-half devices (Methernitha, Johnson, Adams, Hide, Echlin, Moray, Koler, RQM, de Palma, MITI, Lambertson, Correa, Goriatchev, Kanarev, Latchinov, Filimonenko cell): Never properly tested (second-hand figures, no named measurement); Serogodsky-Scheffer and Buynov second-kind machines: Refuted by calculation (Kelvin-Planck) as free-power claims.
6. (New, QE) Frolov's own bench experiments. (a) Ferrite self-cooling alternator: Never properly tested; as heat-to-work from one reservoir, Refuted by calculation. (b) Rotor accelerated by back-EMF: Never properly tested as written (acceleration alone is not an energy gain); as net output with no accounted input, Refuted by calculation. (c) Arc "negative resistance" with DC input falling to zero and reversing, ~300 mA, shown once in 1996: Never properly tested (upgrades 1a; no data). (d) Electrolysis needs no primary power because the field moves the ions: Refuted by calculation (energy is EMF x charge; the source pays).
7. (New, QE) Suppression allegations (Filimonenko 1957-60, 1994 Moscow commission, Johnson patent delay, institutes): all about others' work; sourced only to Frolov plus unquoted Russian journals; status "unsourced beyond the document". Frolov's own claims remain "not dismissed without testing, never put in testable form".
8. (New, Round 2) Frolov's 1995 flat-spiral-coil experiment: a lamp on the secondary of a solenoid-primary / flat-spiral-secondary transformer is "free from input power in the primary winding" ("asymmetrical mutual induction"). As an observation: Never properly tested (no numbers, no instruments, not even whether the lamp lit; no publication). As a general statement that load power need not be drawn from the primary: Refuted by calculation for a passive transformer (reciprocity M12 = M21 and energy conservation: a real load on the secondary reflects into the primary) [read: PDF p.4; the reciprocity point is [memory, high]]. Not Historically superseded; it is a Frolov-specific assertion.

What would settle it (single measurements)

• Claims 1-3: true input power (mean of v(t) i(t), or DC battery V and I plus calorimetry of everything) of the generator with the load connected, disconnected, and with the diodes shorted or reversed. If input does not change with the load, the 3 W is idle loss and the load draws a small increment; if input falls with the load, that is the anomaly Frolov claims and worth a proper replication with waveforms.
• Claim 4: year-long logged output against a sealed control, with the seasonal signature specified in advance.
• Claim 5 (Brown): thrust per measured watt, in air and in hard vacuum, with the whole apparatus free to move.
• Claim 5 (Sweet): independent logged input/output power, magnet temperature and balance weight with a sealed control.

Open questions

• What does the generator draw with and without the load in Frolov's 1994 setup, and in the arc-loading experiment (does consumed power really fall at 300 mA)? Nobody has the data; it is cheap to measure.
• Is the Avramenko-plug lamp result (60 mA, 50 V, 100 kHz) a real-power measurement or a product of reactive quantities?
• What is the true mechanism of the Brown force in air (ionic drift versus something else), given that ARL called ballistic ion wind too small by three orders of magnitude? I did not open the ARL report; the Wikipedia summary and the search summary were not fully reconciled.
• Are the 20-100 kW single-wire lines real? 20 kW is reported by the system's authors [web]; 100 kW is unconfirmed; none is documented as over-unity (efficiencies reported are below 100%).
• Which Kozyrev experiments, if any, were replicated independently? I only confirmed the 1960 commission's negative finding.

Overall

Frolov's documents contain one conventional, correctly computed experiment (rectified capacitive coupling, 3% of input) and one theoretical frame (potential field as free fuel, time-reversed balance) that does not connect to it. His field is a literature survey of others' devices with no data. The criticism that the mainstream "refused to test" does not apply to anything in these two documents, because nothing testable was offered. The conclusions about other people's devices rest on Frolov's second-hand reports plus a handful of web summaries that were not read in full.

QE revision (after qe.md, verdict MINOR)

Inputs added: qe.md; the new deep read deep/FreeEnergy-db73c1.md (PDF pp.10-16 = magazine pp.20-26, Figs 8-12 viewed as images; the file notes it overwrote an earlier deep read of other pages, so pp.1-9 and Figs 1-7 still rest on digest and text reading). Page convention: PDF page = magazine page minus 10. Corrections of fact: my earlier "~line" cites in the corrections list do not match _text/; the correct places are PDF p.5 (ionization conceded), PDF p.13 (arc test), PDF p.5 / mag. p.15 for "can't be more than the generator" and PDF p.4 / mag. p.14 for "efficiency can be more than 100%". The 1994-note spark rate is 5 Hz [read]; 5-10 kHz is the generator frequency. The earlier headline verdicts for Claims 1-5 do not change.

New evidence and what it changes

Frolov's own experiments are each one to four sentences with no instruments, waveforms or power values [read: PDF pp.11-13]. Photographs (Fig.11) show a prototype and an unlabelled multimeter [read: page image p.12].

1. Arc / self-generated discharge (SGD) test, revises 1a. The text says more than I recorded: "consumption current is decreasing until zero and then it can change its direction i.e. this system begins to generate the power"; a larger spark gap gives "more surplus power"; one effect was "demonstrated in 1996 during the conferences 'New Ideas in Natural Science', St.-Petersburg"; the only number is "about 300 mA"; input was measured on a DC battery circuit "to avoid" phase-shift errors [read: p.13]. He sets it beside Chernetsky and Galkin's SGD work (a 1971 Moscow Aviation Institute transformer destroyed by a reversed-current pulse "10 times" the consumed power [read: p.13]) and notes pinch effect needs hundreds of amperes, so he proposes electron acceleration across the gap. Verdict: still Never properly tested. In his favour: a testable dependence (gap length) and a DC measurement. Against: a DC-side average does not exclude pulsed or non-sinusoidal averaging error, stored inductor/capacitor energy or battery-sag effects, and nothing says what fed the transformer or what instrument read the battery current. Current reversal would be a real anomaly if shown; a public demonstration with no recorded data is not a test. Settling experiment: log battery V and I with a fast sampling scope (mean of v(t)i(t)) at several gap lengths, with the load on and off, and calorimetry of arc plus load.
2. Ferrite self-cooling alternator. "it was demonstrated that ferrite core of the generator coil is self-cooling"; a motor-driven iron plate passes through the gap between magnet and coil; no temperature, no thermometer, no control core, no power [read: p.11]. A separate pair of 1983 Russian patent applications on converting ferrite heat to electricity is named [read: p.10]. Self-cooling alone is not a power claim: eddy-current and hysteresis heating in a core varies with flux cycling, and a core that cycles through a magnetocaloric-type change can cool locally while the apparatus as a whole warms. Verdict: Never properly tested. If it means net electrical work drawn from a single heat reservoir with nothing else changing, it is Refuted by calculation (second law, Kelvin-Planck). Cheap test: thermocouples on core and a non-ferrite dummy, with the coil loaded and open, motor power logged.
3. Rotor accelerated by back-EMF (patent, Fig.11, "several experiments" 1994-2003). No speed, torque or power; patent not identified [read: p.12]. A rotor that accelerates is normal if the motor still supplies torque. In a generator, back-torque normally opposes the drive (Lenz). A scheme where the induced field aids rotation is possible in the sense of a motor/reluctance effect, but then the ledger is input electrical (or mechanical) work = increase in kinetic energy + coil dissipation + stored field energy + any output; net output with no accounted input would violate energy conservation. Verdict: Never properly tested as written; Refuted by calculation if it means net output beyond the accounted input.
4. Electrolysis "field does the work, so the battery need not." An argument only, no experiment [read: p.14]. Energy delivered by a source is EMF x charge passed. The minimum voltage to split water is about 1.23 V [web: https://en.wikipedia.org/wiki/Electrolysis_of_water, "the minimum voltage necessary for electrolysis is about 1.23 V"; thermoneutral voltage about 1.48 V, same page], so the electrical work to make 1 mol of H2 is at least 2 x 96,485 C x 1.23 V, about 237 kJ [memory, high; Gibbs energy]; burning that hydrogen returns 286 kJ/mol (higher heating value) [memory, high], where the extra 49 kJ is the TdeltaS heat that the cell takes from its surroundings, not extra energy from the field. "Burning the gases gives more heat than input electricity" is possible only by counting that ambient heat, and never gives surplus after all losses. Goriatchev's claim (0.2 V pulses, "1500%", no definition of output) would require the source to supply less than the thermodynamic minimum on average, so either the output is mis-defined (for example gas heat of combustion against only part of the input, or counting stored charge) or energy enters some other way. Verdict: Refuted by calculation for "no primary power required"; Goriatchev's 1500% and Kanarev's "output > input": Never properly tested (no numbers or method given, no independent test found).
5. Second-half devices (Frolov relays them; see the deep read's table of 25 items). All numbers are second-hand with no named instrument; Frolov calls Johnson's 5 kW "reported" and Adams's 10 kW a future plan, and doubts RQM's technical emphasis [read: pp.11-13]. Methernitha 750 kW total (20 cm disk 200 W, 3 m disk 30 kW): web check found no independent test; the Wikipedia pages for Methernitha and Testatika say nothing of verified output [web: https://en.wikipedia.org/wiki/Methernitha, fetch summary, so absence only]. Correa PAGD 483%: not found in budget; the figure comes from a conference report relayed by a magazine [read: p.14] with "efficiency" undefined. Verdicts: Methernitha, Johnson, Adams, Hide, Echlin, Reed, Moray (more than 50 kW, Yakovlev visit 1929, no result), Koler (60 kW), RQM, de Palma (unipolar, "published results" without numbers; Frolov's argument that the Lorentz force is radial so there is no back-torque ignores that the circuit current still loads the rotor through the return path; I give this [memory, medium] and recommend the Faraday-disc literature), MITI (a 40 kW generator size, not a measured gain), Lambertson (1.6 kW per unit), Correa: all Never properly tested. None is on record as tested and failed that I found; I did not search each one, so this is absence of sources, not proof of absence. Serogodsky-Scheffer (benzene-water retro-condensation at 154 C, German patent 4,244,016) and Buynov (N2O4 "mono-thermal" cycle): second-kind claims, Refuted by calculation (Kelvin-Planck) as free-power machines; Buynov's idea that entropy is "undetermined" in a reversible reaction is the usual confusion between a state function and its path integral [memory, medium]. Kanarev, Latchinov (1888 high-pressure cell that "freezes", i.e. endothermic electrolysis with no net gain claimed in the text), Zaev: as above.
6. Suppression allegations, assessed. (i) Filimonenko: extra power in heavy-water electrolysis (1957), 1960 support from Kurchatov, Korolev and Zhukov, a secret resolution, later suppression, and a 1994 Moscow City Council commission urging renewal [read: p.15]. Sources are an unnamed declassified 1960 document, a 1995 article in "Inventor and Rationalizator" and the journal "Russian Ideas" 1994; none quoted, no numbers; Frolov himself calls the military-interest explanation "an assumption" [read: p.15]. I did not find or open any independent source. Status: unsourced beyond Frolov; I do not know whether the effect was ever tested under controlled conditions. (ii) Johnson's patent delay 1973-1979: the dates match the patent data in the text (filed 6 Dec 1973, granted 24 Apr 1979), but "it is perpetuum mobile" as the reason is Frolov's inference [read: p.11]. (iii) Institutes unable to take new ideas, authors plagiarised, "two worlds" of hidden knowledge: assertions with no named case [read: pp.15-16]. These bear on the contrarian's "how was it rejected" test only for third parties. The arbiter's earlier statement that "refused to test" does not apply to Frolov's own claims stands; it must not be read as "no suppression is alleged in the article".

Changed verdicts

• 1a: wording strengthened (current reversal, 1996 demonstration, gap-length dependence), verdict unchanged: Never properly tested.
• New claims 6 and 7 (above) added to the Corrected list; Claim 5 extended to the second-half devices.
• No Claim 1-5 verdict changed. No verdict moved to Tested, not supported: I found no test record for any second-half item.

Remaining limits

PDF pp.1-9 (including Figs 1-7 and Frolov's 1995 flat-spiral-coil experiment) were not deep-read in this revision; the earlier pp.1-9 deep read was overwritten. Web checks done: 3 fetches (Methernitha, Testatika, electrolysis voltage); Correa, de Palma, Filimonenko, Moray, Koler and Lambertson were not checked and rest on Frolov's report [read] alone.

Round 2 (new deep reads appended to deep/FreeEnergy-db73c1.md)

Inputs added: the new section of deep/FreeEnergy-db73c1.md covering PDF pp.1-5 (magazine pp.11-15): the closed-trajectory argument, Figs 1-3 (page images of PDF p.2 and p.5 viewed), the 1995 flat-spiral-coil experiment and the Strebkov sentence. The pp.10-16 section used in Round 1 is unchanged. I did no new web checks in this round; everything below is [read] from that deep read unless tagged.

What the new reads found

1. Claim 1 has no mathematics at all [read: PDF pp.1-2, 4]. The only formula-like items are the bookkeeping "A + B = 0" and "A + B + C + ... = 0". The standard theorem (work of a conservative force around a closed path is zero, equivalently curl F = 0 for a static field) is never written, only stated in words. Frolov concedes it ("it is right for only case: if one part of work is positive and another part is negative") and proposes to remove the negative part by screening, pulsing or polarity change. He asks himself what the source of the power is and answers with "aether density"/space-time and a reversed-time anti-process, not an energy account [read: PDF p.2]. This confirms the three-way split already made (static closed loop: Refuted by calculation; time-varying field: Established, switcher pays; his devices: Never properly tested). No verdict changes.
2. Figs 1 and 2 viewed [read: page image p.2]. Fig.1 (grid under the cathode) and Fig.2 (battery disconnected while electrons are in flight) carry no labels, values, waveforms or meters. In Fig.2 the battery was connected long enough to accelerate the electrons; that energy is the input, so "the primary source is not discharging" does not mean no energy was spent. Verdict for both: Never properly tested; the zero-input reading is Refuted by calculation (energy conservation, with the electrode capacitance charging energy 0.5 C V^2 supplied by the battery). Settling test: battery V and I sampled over the whole pulse, with and without the tube load.
3. 1995 flat-spiral-coil experiment [read: PDF p.4]. The whole description is one paragraph; nothing measured, no instrument, no numbers. Added as Claim 8 (above). Never properly tested as an observation; Refuted by calculation as a general statement about passive transformers. Cheap test: primary V, I and phase with and without the lamp on the secondary.
4. Strebkov "lines from 20 to 100 kW" [read: PDF p.5, ref. [8] on p.17]. The sentence is Frolov's, ends with an exclamation mark, gives no voltage, length, efficiency or tester, and is supported only by the same magazine's issue #5(8), 2002 (reference [8] is linked by name only, not cited by number). The paper is not in the folder and I did not open it. This leaves my Round-1 status unchanged: 20 kW reported by the system's authors [web, earlier fetch], 100 kW unconfirmed. Even if true, a 20-100 kW line does not support Claim 1, since Frolov himself writes "in each receiver the power can't be more than power of this generator" [read: PDF p.5].
5. Fig.3 (Avramenko plug) viewed [read: page image p.5]. The two-diode plug at the end of a lambda/4 line is exactly the conventional rectified capacitive-coupling circuit assumed in Claim 2. Frolov states "high efficiency mode is resonance mode" with no efficiency number or frequency; "It is not difficult to repeat" is not data. No change to Claim 2.
6. Brown tension confirmed [read: PDF pp.5-6]. Frolov excludes "electrokinetic apparatus" as ionization flow, yet the cited Brown patent figure (Fig.4) is itself titled "Electrokinetic apparatus by T.T. Brown". This supports Correction 2 (self-contradiction on ionization). No Brown verdict changes.
7. Other unsourced assertions found in pp.1-5 [read]. Tesla's resonance transformer "efficiency can be more than 100%" (p.4, supports Claim 3 as an assertion only); a Tesla "fuel-less car" in 1934 powered by a 12-tube generator (p.5; no source; Tesla died in 1943 [memory, high]); the "perpetuum mobile of the third kind" taxonomy (p.4); references [1]-[10] are theory or same-journal items, none an independent measurement of closed-trajectory work (pp.16-17).
8. Correction to my earlier source-gap note. I wrote that figures were not viewed as images. Figs 1-3 and 8-12 have now been viewed; Figs 4-7 (PDF pp.6-9) and the 1994 note's figure have not.

Verdict table (after Round 2)

Changed verdicts

No verdict for Claims 1-7 changes. One claim (8) is added to the Corrected claim list, and the wording of Claim 1 is extended to include Fig.2. Nothing moved to Tested, not supported, since no test record turned up for any new item.

Remaining limits (supersedes the Round 1 note)

PDF pp.6-9 (Figs 4-7: Brown and Searle-related figures, perpetual-motion material) were still not deep-read; they rest on digest and my own text reading. No new web checks were done in Round 2, so the web status of Strebkov ref. [8], Filimonenko, Correa, de Palma, Moray, Koler and Lambertson is unchanged and rests on Frolov [read] alone.