Science Contrarian review: Sykes-Chris (Step 2)

Reviewer role: Science Contrarian (judges by experiment, builds the strongest honest case, examines how claims were treated, applies the same rigour to the inventor). Source: Chris Sykes, "Guidelines to Bucking Coils: Lenz's Law Free Power Extraction", v2.7 (26.02.16), 49 pp. The text layer is available; no source gap. The Investigator's material (sites.md, research/ notes, research/files/) is weighed below.

Tags: [read] from the document or a saved file, [read: page image p.N], [web] with URL, [memory] with confidence.

0. Corrections to Step 1

1. Stale reading note. claims.md says "Read via digest only; no deep read was on file", and facts.json how_read says the same. A deep read of pp.14-19 exists (deep/Guidelines-to-Bucking-Coils-5ecaec.md, with page images) and is the basis of my claim-3 analysis. The pages cited in my verdicts that are outside pp.14-19 (8-10, 21-24, 31-32, 40-41, 46-49) I read myself in the text layer.
2. Claim 6 is misworded. claims.md says generators "lose 98-99% of their loss budget to Lenz drag and only 1-2% to friction". The document (p.23) says: the most efficient generators are 98-99% efficient, "100% of Rotational Shaft Energy, with a 1-2% loss ... and the other 98-99% are the effects of Lenz Law creating Drag". Sykes is not describing a loss budget. He calls the useful conversion work "drag". That is the mistake to examine (claim 6), and the Step 1 wording hides it.
3. Alternator test is incomplete in facts.json. facts.json says "an alternator test gave output with feedback and no extra field input". The document (p.48) says the alternator is driven by an electric motor (mechanical input), the field coil is shorted or put in series with the output, and adds "Not Free Energy" and "ONLY because of Lenz Law, Resistive Mechanical Force is applied on the Shaft". These qualifiers should travel with the claim [read: p.48].
4. "Input power falls as output rises" has a conventional mechanism in Sykes's own text. On p.46 he explains the falling input as a transformer whose load is "another Power Source" with matched phase, so power is returning to the primary side. That is ordinary power flow reversed through a transformer, not a new effect [read: p.46]. Step 1 lists the claim without this context.
5. Source for the p.17 Sweet quote is now found. The deep read found it uncited. The Investigator located it in Sweet's "Letter to Mark from Sparky" (research/files/hyiq-Letter-to-Mark-from-Sparky.pdf). I checked it: "lowering copper duty factor, at the same time opening the gates wider" is there; the Guidelines add "and" [read]. The letter concerns a 6-pole 7.5-10 kW machine with a "current feedback winding spirally traversing the total length of the coil conductor" [read].
6. "Bloch wall" in claim 4. The document uses "Bloch wall" for a macroscopic, moving boundary between the two output coils. In mainstream magnetism it is the roughly 100 nm thick transition region between magnetic domains inside the material [memory, high]. Step 1 does not say this is a change of meaning.
7. The "5000 W" and "6 W" figures are reported from others (p.13, p.19-20) and are not Sykes's measurements, as facts.json says. Confirmed. The IECEC 1991 paper (Sweet and Bearden) is the one conference record [web: https://www.rexresearch.com/sweet/1nothing.htm, via the Investigator's saved copy].

No other Step 1 errors found. The 650 us / 75 ms figures, coil data (55.6 mH, 0.231 H, 0.24 mH, 3.8 / 1.9 / 7.8 ohm) match p.15-16 [read].

1. What Sykes did that deserves credit

• He gives full coil data and photographs for the coil experiment and says plainly that Experiment 2 is "an expected behaviour" (p.16).
• He says the result is not new energy (p.47: "there was no new form of Energy"; p.48: "Not Free Energy") and agrees energy cannot be created (forum reply #7, 2015 [web: Wayback of overunity.com/15395]).
• He tells builders that perhaps 1 in 10 builds works and that "confusion easily sets in" (p.34, p.45).
• On aboveunity.com (2019) he set a rule that "Any COP > 1 Claims must be accompanied by Measurements" and told a user that scope mean, not RMS, gives power (one example: 59.5 W RMS-based versus 17.1 W mean on a 0.25 ohm shunt) [web: research/files/aboveunity-measurements.txt]. That is correct practice.
• On the same forum he writes "one can have Asymmetrical Regauging and still be Below-Unity", and "we have not succeeded, in general" [web: aboveunity.com threads]. He published a replicator's COP of 0.82 on his own forum without deleting it.
• The Investigator found and saved Sweet's primary documents on his site (lab notes, letter, "Nothing is Something"), which is real documentary work.

2. Claim-by-claim review

Claim 1: bucking output coils cancel the magnetic component, the A component adds, Lenz drag on the input disappears

a. Mathematics. Sykes gives no derivation. The only equations are emf = dPhi/dt and emf = Bvl (p.8-9), both standard. The assertion that "A is the electric component of the magnetic field and Phi is the magnetic component" (p.6) uses A (vector potential) and Phi (flux) loosely; in standard theory E = -grad(V) - dA/dt and B = curl A, so A is not the "electric part of B" [memory, high]. His geometric condition (exactly half of the conductors seeing the opposite change, p.8) is a coherent engineering statement: two coils whose emfs add in series while each sees flux of opposite sign is how a centre-leg-driven two-outer-leg core already works. I find no internal contradiction in the geometry. I find no quantitative prediction at all (turns, flux, load), so nothing can be checked numerically.

What can be checked: the coil data in Experiment 2. Series-opposing two identical windings gives L = 2L1(1-k). With L1 = 55.6 mH and residual 0.24 mH, k = 1 - 0.24/111.2 = 0.998 [calculation, from p.15-16]. So the "non-inductive" coil is just two tightly coupled windings whose fields cancel. That is a textbook non-inductive (Ayrton-Perry style) winding [memory, high].

b. Established experiments. Where the two output mmfs (ampere-turns) cancel each other in the shared path and do not link the input winding, the input sees less reflected load. That is leakage-flux / loosely coupled transformer behaviour, and Sykes's p.31-33 observation that loose coupling "works" is consistent with it. It does not remove the energy requirement: the load power must still come out of the flux produced by the input. I find no established result in which a coil pair delivers power with zero reflected power at a fixed input. The mainstream argument that a delivered power P = emf x i has to be balanced by power entering through the flux linkage is, in my reading, a consequence of Maxwell plus energy conservation, so a successful test of the strong claim would be a violation of it.

Honest point for Sykes: the vector potential is not a bookkeeping fiction. The Aharonov-Bohm effect shows A shifts the phase of electrons in a region where B = 0 [web: https://www.hitachi.com/rd/research/materials/quantum/aharonov-bohm/index.html; Tonomura et al., 1986]. Bearden and Sykes cite it (p.18). But that effect is a quantum phase shift of a single electron beam; no one has derived from it a way to draw load current without a reaction on the source. Citing it supports "A is physical", not "Lenz drag can be removed".

c. Tests. - Sykes's own: the 2014 hyiq.org update reports "Input is 1 Watt (5V 200ma) ... Globe ... rated at 12V 300ma (3.6 Watt). I am getting slightly more Output than this" [web: research/files/hyiq-VTA-Replication-Project.txt]. The sentence is ambiguous (more than 1 W, or more than 3.6 W?), gives no instrument, no waveform and no load measurement. A lamp rated 3.6 W at 12 V does not reach rated output on 1 W; the statement cannot be tested as written. - Skeptic's request (MileHigh, 16 Jan 2015): refers to an "alleged COP 1.7 device" and asks for scope shots, meter readings and power in and out. Sykes's own reply stated no data in the portion read [web: Wayback overunity.com/15395]. I could not find the 1.7 figure in Sykes's own words, so I do not count it as his claim. - Replications on record, from Sykes's later three-coil design: dewetw (30 Oct 2019, aboveunity.com, N87 ferrite, UF5405 diodes, non-inductive load resistors) "never saw any result where I am getting more than about 0.82 COP", while reproducing the sawtooth waveform. patrick1 (23 Oct 2019): "successful of course, but thats because it isnt OU." [web: research/2026-10-04-aboveunity.com.md]. Itsu's replication (overunityresearch.com topic 3951, 2020): Sykes calls it "failed" because of load (6 V, 0.6 W), unknown coil length and "less than a basic Transformer Efficiency"; Itsu's own data are behind a login and I have only Sykes's account [web: Wayback overunity.com/18464]. Partzman (2020) argued the posted schematic has diodes reversed relative to Sykes's own scope traces and that L1's stored energy returns to the supply [same source]. Sykes replied "My Diagram is correct, change it, it wont work! Period!" and "Input can become Negative!". - Sykes's own 2019 three-coil circuit (12 V, 400 mA input, "some data off screen, my 10x probes are not enough") gives no complete in/out table [web: aboveunity.com non-inductive thread].

None of this is a neutral, controlled test. The replications on record are enthusiasts' bench builds on the inventor's own forum, and none was done by an independent laboratory.

d. If true. Conservation of energy would need an input from somewhere not on the meter, or Lenz's law would need a loophole under non-sinusoidal, diode-switched, gapped-core conditions. Sykes himself rules out the first (no new energy). If a loophole existed it would imply the load power can exceed the metered source power in a closed circuit, and that Poynting's theorem is incomplete. The more modest outcome, which his own p.46 text describes, is a transformer returning power, a known effect.

Verdict (split). - Weak form (opposed, loosely coupled output coils reduce the reaction seen by the input; leakage-type behaviour): Plausible, untested as a quantified claim; no measurement of the reduction exists in the document. - Strong form (load power delivered with no corresponding input, or COP > 1): Never properly tested (no neutral, controlled, complete power-in/power-out measurement on record), with the available replications, which found COP <= 0.82 or "isn't OU", pointing against it: Tested, not supported at the level of informal bench tests only. It was not dismissed without testing; it was tested by enthusiasts and the inventor rejected the tests that failed (see section 3).

Claim 2: Lenz's law is the result of drawing current from already-separated charge, not its cause

a. Qualitatively consistent: with the secondary open there is no secondary current and no reflected load (p.21), which is textbook. Sykes's "if we were able to turn off the Magnetic, but at the same time keep the Electric components, there would be no Reflected Power" (p.21) is a hypothetical, not a derivation. The statement "Simply put, if Lenz's Law could be by passed, then every Generator on the planet could be a perpetual motion machine" (p.22) is correct as a logical statement: Lenz's law is how energy conservation shows up in induction.

b. Established: induced current opposes the change (Lenz); no load, no current, no reaction. Not established: that a load current can exist with no magnetic field attached to it. Any current produces a B field by Ampere's law [memory, high]; Sykes's own text concedes this ("A Moving Charge constitutes a Magnetic Field", p.10) and p.46: "With no Flow of Charge Carriers, there can be no Magnetic Field". Cancelling net B between two coils leaves each coil's field in the local space (he says so, p.45-46).

c. No test is described. The p.14-16 experiment concerns inductance (claim 3), not Lenz drag.

d. Nothing in mainstream physics would have to change for the weak form. The strong form would require current without field, or field without reaction.

Verdict (split). "Lenz opposition appears only when current is drawn": Established. "Charge separation can occur without a reflected magnetic field while delivering power": Not testable as stated (no measurable prediction attached), and if the prediction is supplied as "input power does not rise with load", Never properly tested.

Claim 3: the magnetic field itself slows charge flow; 650 us versus 75 ms at 7.8 ohm "proves" it

a. Mathematics, checked by me from the deep read's page-image numbers. Series R = 7.8 ohm coil + 1 ohm sense resistor = 8.8 ohm [read: page image pp.14-15]. tau = L/R: - Exp. 1: 0.231 H / 8.8 ohm = 26 ms (3 tau = 79 ms). Sykes's "75 ms" is about 3 tau. The trace shows an exponential with tau about 22-27 ms and a plateau of about 1.6 A (V/R = 15.4/8.8 = 1.75 A) [read: page image p.14]. - Exp. 2: 0.24 mH / 8.8 ohm = 27 us. The trace shows a rounded rise of about 0.4 ms to a peak of about 0.7 A (V/R = 7/8.8 = 0.80 A), and the pulse width is about 640 us [read: page image p.15]. The "650 us" is close to the pulse width set by the generator (650.4 Hz) and is not a time constant. The rise implies an effective inductance of about V/(di/dt) = 7 V / (0.7 A / 0.4 ms) = 4 mH, about 17 times the quoted 0.24 mH. The inductance ratio is 231/0.24 = 960 but the claimed time ratio is only 75/0.65 = 115.

So the experiment in Exp. 1 matches the ordinary L/R law, and Exp. 2 does not match it either, being 15 times slower than the 27 us it would predict. Both of these facts cut against a simple reading. The Exp. 2 shortfall may be a drive limit (the schematic shows a high-side N-MOSFET driven from a ground-referenced generator, which would not behave as drawn), core effects at 0.7 A, or the generator pulse width; none is tested. Also: supply (15.4 V vs 7.0 V), drive (9 V vs 5 V) and frequency (7 Hz vs 650 Hz) all change between the runs [read: page images]. The document does not say how 75 ms and 650 us were taken: no cursors, no 63% point.

b. Consistent with the textbook: an inductor opposes a change of current; in plain words, its magnetic field slows the rise of current. That is established and Sykes states it is "expected". What is not established is anything beyond it. "Slows the charge flow rates" confuses current rise time (di/dt) with charge speed (the drift velocity of conduction electrons in copper is of order mm per second or less and is not measured here) [memory, high].

c. Tested only by Sykes, once, with no repeats or error bars, on one coil. The experiment cannot separate his interpretation from L/R, because both predict the same curve. A distinguishing test would be a change in steady-state resistance or heating (Sweet's "10 times more current at the same temperature", p.17). The plateau currents equal V/R for the same 7.8 ohm, so the data show no change of that kind [deep read s.4-5].

d. If it meant more than inductance, the circuit equation V = iR + L di/dt would need a new term. As presented it needs no change.

Verdict (split). Observation (aiding versus opposing windings give slow versus fast current rise): Established (textbook L/R, run on his own coil). Interpretation ("proves" an extra friction on charge or drag that wastes energy): Not testable as stated by this experiment, because it predicts the same curve as ordinary inductance; the Sweet "duty factor" extension (lower I^2R at the same current) is Never properly tested: nothing in the Guidelines measures it. The quoted times themselves are not reproducible from the document (how taken is not stated).

Claim 4: conditions for the effect (loose coupling, Bloch wall, diode plus square wave at the BH knee, enough load); input falls as output rises; alternator output with feedback

a. The BH-curve claims ("this technology does not move through hysteresis ... best operates in and around the knee", p.41) come with no equations; the author adds "this section only appears to be valid under some configurations". The p.46 power formula is written V^2/R and then "I x V" with a phase-angle caveat; correct as far as it goes. "Bore of one coil will OPPOSE the magnetic field, gating this space into an off position" (p.40) has no derivation.

b. Loose coupling and diodes are ordinary transformer-engineering variables. The alternator test (p.48) is the ordinary self-excited generator: field winding in series with, or shorted to, the output, relying on residual magnetism and positive feedback until saturation [web: https://www.electrical4u.com/self-excited-generators/]. It keeps a motor turning the shaft and Sykes states that Lenz's drag acts on the shaft there, which confirms the standard picture. It therefore does not demonstrate "no extra input".

c. The Guidelines contain no table of input and output power for any device Sykes built. The 1-in-10 success rate (p.34, p.45), the load requirement ("Your Coils must be Loaded at all times", 2020 recipe) and the list of required conditions (loose coupling, diode, square wave, correct winding direction, right frequency, "uniform magnetic fields") make the claim hard to falsify: a failed build can always be assigned to a missing condition. The 2020 recipe says to "benchmark ... in Transformer mode ... around 80% efficiency" and treat anything above as gain [web: Wayback overunity.com/18464]. That is a weak baseline, because an efficiency of the same circuit in transformer mode does not bound what the real circuit's measured numbers should be. Replications: section 2, claim 1.

d. As claim 1, plus: the BH knee claim would require the conversion to happen at a point where the material's behaviour is already well characterised, so any loophole would be visible in a hysteresis-loop measurement. That is a cheap test no one has done.

Verdict. Alternator part: Established (self-excitation with mechanical input; Sykes does not claim more). Conditions and "input power falls as output rises": Never properly tested (no neutral measurement; informal replications gave COP <= 0.82 or "not OU"); several elements (Bloch wall between coils, knee operation) are Not testable as stated.

Claim 5: Sweet's VTA and the devices of Moray, Meyer, Jensen, Koenig, Frolov, Melnichenko, Hendershot and others work on this one principle

a/d. No derivation links these devices to a single mechanism. The Bearden material quoted (p.18-22) uses wave/antiwave pairs summing to "vector zero", "negative time" and gravitational effects; as quoted these are statements without measurable predictions. The general Whittaker 1903 decomposition of the potential is mathematics; the physical claims added by Bearden are not derived from it in the quoted text [memory, medium; not rechecked].

b. Sweet's own primary document set, which the Investigator located and saved, is the best evidence on file, and I read it. The 1986 Roberts/Sweet lab notes (transcribed by Sykes from scans "from an anonymous source", with pages missing) record: input measured with a Beckman multimeter on the 200 uA scale, "7.5v (3.1 on 200 uA scale)"; output "10.41 (20 V scale) x 1.84 A (20 A scale) = 19.1544 W"; "19.1544 W / 0.023 mW = 83,260,000% gains" [read: hyiq-Lab-Notes-Transcribed-Floyd-Sweet.pdf]. I checked the arithmetic: 7.5 V x 3.1 uA = 23.3 uW (0.023 mW), and 19.1544 / 23.3 uW = 8.2 x 10^5, so the gain of about 83 million percent follows from the figures. Equally, 10.41 x 1.84 = 19.1544. Later: "4.6 A x 24.2 V = 111.32 W", "results of 6/4/86 were 105,000 x - 111,000 x gain" (this gain would need an input of about 1 mW; the forum analysis by "Anor" gives 7 V x 143 uA, about 1.0 mW, which matches [web: aboveunity.com Sweet VTA thread]; I could not check the scans). These are strong claims with the following specific, checkable gaps in the same notes: - The notes say power was "from a Wavetek Signal Generator, with line power" [read]. Input power was metered at the generator output; the generator's own mains draw is not metered, and a lab signal generator typically draws watts to tens of watts from the wall [memory, high]. If the stated 23 uW were all the generator delivered, the gain figure would still not account for the generator's own supply unless that was a purely passive device; nothing says it was. This one is the biggest hole in the "input" side. - The drive coil's impedance "goes way up to hundreds of thousands of ohms" in the field [read], i.e. an almost purely reactive load; a multimeter reading of V and I multiplies to volt-amperes, not watts, unless the phase is measured. Real power could be lower or higher than VA, so "input power" is undefined from these entries. - A hand-held multimeter on the 200 uA AC scale at a few hundred Hz with a high-reactance load has limited accuracy that nobody quotes [memory, medium]. - Output was read on 26 W lamps as 104 W "rated" and measured by V and A meters; waveform and power factor not recorded. - These are the figures of an interested party's source; the witnesses named (Mark Goldes and Darryl Roberts; Goldes was Sweet's backer [memory, medium]) were also interested. Single-source evidence is still evidence: the notes are signed and dated and record the instruments used. I do not dismiss them. I note that they contain the data needed to check the claim and that no one has re-run them. The Sweet and Bearden paper at IECEC 1991 (Boston) is a conference publication [web: https://www.rexresearch.com/sweet/1nothing.htm]. Bearden (interested party, promoter and co-author) says Sweet feared independent testing: "He feared that his life would be snuffed out immediately" [web, via search result: sources on Sweet's refusal of certification]. Sweet died in July 1995 [web: https://www.hyiq.org/Reference/Profile?Name=Floyd+'Sparky'+Sweet; search summary]. The circumstances of death were ruled non-criminal per one aggregated summary [web: search summary, unchecked primary]. I did not find any independent laboratory test of the VTA. Sykes himself rejects Bearden's magnet "conditioning" as "nonsense" (2025 forum [web: aboveunity.com Sweet VTA thread]) and the Guidelines (p.17) says it is "simply wrong". So his account of the VTA principle differs from the account of the person who tested it with Sweet.

For Moray, Meyer, Jensen, Frolov, Melnichenko and others: Sykes presents diagrams and patents; I found no neutral test of any of them in the files. Koenig US4806834A is a granted patent whose abstract describes two mirror-image coils on a magnetizable bar (filed 1987, granted 1989) [web: https://patents.google.com/patent/US4806834A/en]. A patent grant is not a performance test. The patent's claims were not read by the Investigator, and I did not read them. Whether these devices share Sykes's principle, as opposed to his reading of their drawings, is untested.

c. Tested by: Sweet and Roberts (1986, informal); Bearden's account of a weight-loss test; Sykes' forum replicators (0 positive on Sykes's own later design). No independent laboratory. No replication of the VTA found; the Roberts notes say the VTA's magnets were "(2) 4 x 6 x 1 in ceramic ... 500 + Gauss".

d. If the VTA worked as claimed (111 W out from a milliwatt-order input), conservation of energy would fail in a table-top device, or a measurement error exists in the 1986 numbers. Evidence for a metering explanation is concrete (the generator's mains draw; reactive load), which is what a neutral test would settle first.

Verdict (split). Sweet VTA: Never properly tested (a documented 1986 informal measurement exists, with specific unmetered inputs; no independent test; the inventor reportedly declined one, on Bearden's word). Moray / Meyer / Jensen / Koenig / Frolov / Melnichenko etc. sharing one principle: Not testable as stated until the principle is given a measurable prediction; each device individually Never properly tested in the files provided. The Bearden wave/antiwave and "negative time" language: Not testable as stated.

Claim 6: generators lose 98-99% of their efficiency to Lenz drag, so removing drag is the route to higher efficiency; the system is open and connected to the Aether; "Free Energy Future is here NOW"

a. Calculation (refutation of the efficiency argument). For an electric generator, shaft power is P_mech = T x omega (torque times angular speed). Electrical output is P_elec = emf x I. Take a 1 MW generator at 98.5% efficiency (typical of large machines [memory, medium]): P_mech = 1.015 MW, losses 15 kW (1.5%: copper, iron, friction, windage). The reaction torque from the load current that Sykes calls "Lenz drag" carries 985 kW, which is the output itself. Remove it and torque falls to the 15 kW of friction and loss; the output falls to zero in the same stroke, since the load current that causes the drag also is the power. The only improvement left is in the 1.5% of losses, a gain of at most 1.5 percentage points. So the argument "remove drag, raise efficiency" fails unless something else supplies the energy. This is Sykes's own premise in reverse: if Lenz drag is the energy conversion, bypassing it bypasses the output.

b. Established: efficiency figures for large generators (about 98-99%) [memory, medium; not rechecked]. Sykes's text (p.8) goes further: "Mechanical Energy is NOT the Source of the Energy Generation" because rotating a shaft with no correct conductor and iron configuration gives no output. That confirms a configuration is needed, which is no evidence against mechanical input as the source: a car with the engine disconnected also produces no motion.

c. No test attached.

d. If true, Lenz's law would have to fail in the Poynting-theorem sense for time-varying fields.

The claims "open system connected to the Aether" (p.47) and "Free Energy Future is here NOW" (p.49) are not attached to measurements; on p.47 he writes that "there was no new form of Energy". The document therefore contradicts itself on whether a new energy source is claimed. In the same document he says both "Not Free Energy" (p.48) and "Free Energy Future is here NOW" (p.49). I mark this as internal inconsistency in the claim as a whole, and not in any single equation.

Verdict (split). Efficiency argument (remove drag, get higher efficiency): Refuted by calculation (the "drag" is the conversion work; see 1 MW example). "Open system / Aether / free energy": Not testable as stated; combined with the "not new energy" statements, Internally inconsistent as to what the claim is.

3. How these claims were treated

• Tested and failed, never tested, or dismissed without testing. Not "dismissed without testing". The thread on overunity.com reached 514 pages and 1.14 million reads by Jan 2017 and carries builds by experienced bench people (TK, Itsu, MarkE, Picowatt were named by MileHigh as capable replicators) [web: Wayback overunity.com/15395]. The skeptic's early criticism (MileHigh) was a request for data, and that is a fair request that the Guidelines did not meet. The negative results on record are by volunteer replicators; none was run under a pre-agreed protocol.
• Were the negative experiments sound? Itsu: Sykes's critique that the load was 6 V 0.6 W and the coil length unknown is a point about missing information; but "he has failed" is the inventor's judgement, and his own required conditions (1 in 10 builds work, loading, loose coupling) mean any failure can be attributed to the build. That is the structure of an unfalsifiable claim: the effect is real when it appears and a build error when it does not. dewetw's 0.82 was not disputed by Sykes. Partzman's diode-polarity critique is a technical objection that I cannot resolve without the schematic and traces; Sykes's reply was an assertion.
• Was data reanalysed, omitted, funding or publication blocked? I found no evidence of this for Sykes. He publishes openly, says his work "Can not be Patented!" and is shared as public domain. aboveunity.com's home page says "We have been DEPLATFORMED by everyone" and "PayPal De-Platformed me! They REFUSE to tell me why!" [web: per sites.md]. That is the inventor's own statement; no second source was checked, and I treat it as his account, not as established fact, and not as dismissed.
• Absence of evidence (Investigator item). overunity.com, where the 514+ page thread lived, now redirects to a domain-for-sale page; its archive mirror is paywalled; overunityresearch.com holding the Itsu data requires login [web: sites.md]. The record of the 2015-2021 replication debate is therefore going dark; the Wayback Machine holds partial snapshots. Competing explanations: (1) ordinary commercial or hosting lapse (paywall of the archive site is documented; the domain sale is documented); (2) deliberate removal (no evidence found); (3) forum decline. Evidence best fits (1). A full Wayback CDX listing of topics 15395 and 18464 and an account of Itsu's own posts would decide how much of the debate can still be read; that is the record to target.
• Sweet. Sweet reportedly never had the VTA independently certified, and the explanation comes from Bearden, an interested party. Bearden's own claims were also queried as "HEARSAY ... via telephone conversations with Sweet" on the rexresearch page, and Bearden replied that he "did lots of experiments with Sweet, in Sweet's lab" [web: https://www.rexresearch.com/sweet/1nothing.htm]. Both are interested parties.

4. Investigator findings: kept in the record

(Whatever my verdicts, the pertinent ones stay.) 1. Current Guidelines v2.7 is still the latest on hyiq.org; version 2.5 exists on academia.edu / docslib / rodscontracts (not compared) [web per sites.md]. 2. overunity.com thread 15395: started 16 Jan 2015, 514 pages at Jan 2017, at least page 1772 by March 2021; "3 configurations to work"; "Loose Coupling is a requirement". 3. 2014 update: "Input is 1 Watt (5V 200ma) ... slightly more Output". 4. aboveunity.com three-coil build, "Asymmetrical Regauging" at 10% duty, 12 V 400 mA input with output data "off screen"; 7 videos; forum tags it ABOVE-UNITY (interested party's label). dewetw: COP <= 0.82. patrick1: "isn't OU". 5. 2020 Builders Group recipe, "Input can become Negative!", benchmark at 80% transformer efficiency. 6. Itsu replication (login-walled), Partzman critique, MileHigh's data request, Tinman's second-hand quote ("a bucking coil arrangement dose indeed add energy to a system", mechanical devices). 7. Sweet primary texts (lab notes, Letter to Mark, "Nothing is Something" 24 Jun 1988 with E = (B x V) + (-B x -V) = 2(B x V)), IECEC 1991 paper and Rosenthal notes. 8. Koenig US4806834A (mirror-image coils), and related names (Melnichenko, Kapanadze, Don Smith, Figuera 1908, Wiegand wire). 9. Anor's 2026 analysis of the Sweet numbers (1986 versus 1991 figures, "What does not add up"). 10. The overunity.com going-dark anomaly and Sykes's deplatforming statement, as above.

Not reached by the Investigator and still open: the YouTube device videos, Itsu's posts, academia v2.5, Melnichenko's Russian PDF, Koenig's patent claims, the scans of Sweet's notes.

5. Where I do not accept the mainstream reflex

• The mainstream is right that the 650 us / 75 ms comparison shows ordinary inductance. But the A-B effect and the leakage-flux behaviour of gapped, opposed-coil cores are real physics, and the strong "nothing to see" reading goes past the evidence in at least one place: nobody has published a calibrated power-in/power-out measurement (battery-powered input, true mean V x I, load calorimetry) of a Sykes three-coil circuit. The tests on record are informal.
• The 1986 Sweet notes list specific instruments and figures. They merit a re-run, not a wave of the hand. The weak points I name above (unmetered generator mains power, reactive input, accuracy) are testable.

6. What would settle the disagreements

1. Claims 1 and 4. A battery-powered (or separately metered) three-coil build, true mean V x I on all ports with a wide-bandwidth differential probe and an equal-time-base average, plus calorimetry of the whole assembly (core, coils, load) over several minutes, with the inventor present and signing off on the circuit beforehand. Pass/fail: energy in = energy out within error, or not.
2. Claim 3. Repeat Exp. 1 and 2 at the same supply, drive and frequency, with cursor-measured tau, and add a steady-state heating test of the plateau I^2R. Prediction of L/R: 26 ms and 27 us; anything else is a measurement or drive problem.
3. Claim 5. Rebuild the 1986 Sweet device as described in the notes (2 ceramic magnets 4 x 6 x 1 in, #18 250-turn excitation coil, #12 12-turn output coil), power it from a battery-driven oscillator, and meter true power at the oscillator's DC supply as well as at its output.
4. Claim 6. Already settled by calculation; nothing to test.

7. Verdict summary