Stiffler: Arbiter (Step 3)

Role: weighs physicist.md, mathematician.md and contrarian.md (all 2026-10-08) against claims.md, facts.json, deep/, sites.md and research/. Tags: [read] source document or saved page, [web] with URL, [memory] with confidence, [calc] my own arithmetic (re-run in the project venv).

Source gap note (repeated from claims.md)

All nine documents are Ronald R. Stiffler's own saved web pages and white papers (c.2007-2009). He is an interested party: he promoted and sold SEC ("Spatial Energy Coherence") boards, courses and licences through Stiffler Scientific [web: https://www.zpenergy.com/modules.php?name=News&file=article&sid=3221]. No document in the folder holds an independent measurement or replication. [QE revision, see below: the figures were later read. The ECAT photos give scope voltages and the supply panel (25.0 V, 0.33 A) but no current trace; the SEC Special Coil charts plot frequency only, so there is no hidden gain data.] Also missing: the "Measurements for SEC Exciters" data, the "empirical formula" on SECExciter.htm, the demonstration video y9aJohml130 (now removed from YouTube), the thermal-bottle result, and any record of the "Advanced Technology Group" (1993-96). Where a conclusion below rests on the Investigator's web finds (forum posts, videos, Scribd snippets) rather than a document read in full, I say so; where it rests on [memory] alone I say that.

How the three reviews relate

All three agree on the large picture: every effect actually documented (voltage step-up, lit LEDs and neon lamps, one-wire coupling, a heating ratio) is reproducible by ordinary resonance, rectification and capacitive coupling; no document contains an input/output energy balance made with a method that excludes error; and the "Energy Lattice" has no measurable property. They differ on five points, settled below.

Recomputed numbers (where reviewers differ or a verdict depends on them) [calc]

• ECAT: 18.11/8.25 = 2.195 (page 2.194); 12 + 6 = 18, so 0.11 W is unexplained. 8.25 W at 12 V = 0.6875 A; 90 mA at 12 V = 1.08 W. If 1.08 W were the true input the gain would be 16.8, not 2.2. Agreed by all three. [Withdrawn for the lamp version in the QE revision: the supply photo shows 25.0 V x 0.33 A, not 12 V.]
• Loop closure: output fed back must satisfy eta x COP >= 1, eta >= 1/2.194 = 0.456. Using the lamp-level 12 W instead: feeding 8.25 W back needs eta = 0.69.
• ExHeatExp (Fig 15): Q = 4.187 x 25 x (41.25 - 26.15) = 1580.6 J; CEC = 1580.593/1198.759 = 1.3185. Mass 20 g gives 1.055. CEC = 1.00 needs dT = 11.45 K, an over-read of 3.65 K (at 10 mV/K that is 36.5 mV on an LM335 [memory, high]). An 8 g Pyrex tube (0.84 J/g/K, assumed) would raise it to 1.40, so omitted glass/loss make the apparent excess larger, not smaller: the physicist's and contrarian's point stands, and the mathematician's correction to the deep read is right. The 31.9% excess equals a 24.2% input under-read.
• Stacking example: 0.5 x 20 uF x 135 V^2 = 0.18225 J and 0.5 x 10 uF x 270 V^2 = 0.3645 J; the page prints both a factor of 10 too small. Same charge (2.7 mC), twice the voltage.
• RF phase skew at 13.56 MHz: 1 ns = 4.88 degrees; sin = 0.085; with true phase 87 degrees (cos = 0.052) a 5 degree skew gives cos 82 = 0.139 (2.7 times too large) or cos 92 = -0.035 (negative). Physicist's and contrarian's artefact argument confirmed.
• Resonant voltage step-up: 26.5/4.32 = 6.1 at C; 168/4.32 = 38.9 at the free end (peak-to-peak vs the quoted volts is loose, so 30-39). 168/2.7 = 62 LEDs is a voltage statement only.
• Top hat: C = 4 pi eps0 r = 2.8 pF at r = 2.5 cm [calc]. Detuning: C0 = 6 pF plus 10 pF probe gives f at 61%; plus 3 pF stray, 82%; a 10 pF probe on 400 pF shifts f by about 1.2%. Both conventions in the reviews agree.
• Ambient VLF/AM capture: 50 kW isotropic at 1 km = 4.0 mW/m^2, so several watts would need hundreds to a thousand m^2 [calc; contrarian's figure confirmed].
• Wire gauge: 73 mm / 143 turns = 0.51 mm pitch, about #24 AWG; the mathematician's correction to the deep read is accepted.

Disagreements and rulings

1. ECAT accounting: "internally inconsistent" (physicist) versus "coherent but unmeasured" (contrarian). The page's sentence is literally "the 60Hz form wastes some 6 watts of the total input power of 8.25 watts" [read: ECAT page lines 131-135], then "supplies 12 watts to the light bulb ... and suffers a loss in the conversion transformer of some 6 watts. Therefore the total power produced ... is 18.11 watts" [read: lines 136-140]. Read literally, 6 W out of 8.25 W cannot leave 12 W for the lamp. Read as a whole, 18 W is the power delivered into the transformer and 6 W of it is lost, which closes. The contrarian's reading is the more charitable and the more likely intended one; the physicist is right that the first sentence, as written, contradicts it. Ruling: a wording slip, not a proven internal inconsistency; Step 1's "logic problem" should be restated as ambiguity. What stays damaging on any reading: no measurement method; the 12 W is the lamp's 120 V 60 Hz nameplate, not a measurement of a lamp driven at 4.3 kHz pulses with "arcing at the base of the bulb"; and the 8.25 W / 90 mA basis is not stated. The 0.11 W discrepancy is minor. [QE revision, below: the supply panel photo shows 25.0 V x 0.33 A = 8.25 W, and the 12 V / 4.3 kHz schematic may be a different operating point from the lamp version.]

2. Stacking AV Plugs (claim 3): physicist "Never properly tested" for the multiplication; mathematician and contrarian "Not testable as stated". Ruling: Not testable as stated, because "excess energy" is never defined against an input and the "empirical formula" is absent. Physicist's point that it would become a testable claim once input is measured is kept as the route to a better verdict.

3. Claim 5, returned energy: physicist "Plausible, untested"; mathematician "Established for the schematic as drawn"; contrarian "Plausible, untested". The circuit analysis is correct for Fig 16 (a freewheel diode across L-R keeps the collapse current in the loop, so no energy returns to the battery, and it is dissipated inside the calorimeter anyway). The mathematician's estimate (2-19% of input for an assumed L of 5-50 uH, L not on the page) shows the size is modest but not zero. Ruling: Plausible, untested as a measured assertion (Stiffler writes "I have confirmed" and shows no waveform or integral); the circuit reasoning is sound. It is also not the decisive question: the excess, if real, has to come from the temperature, mass or input terms.

4. Claim 6 (measurement rule): physicist "Established" for the practice, contrarian "Plausible, untested". Detuning by a probe or grounded meter is textbook and quantified above; I rule Established for the practice. The contrarian's correction to claims.md stands: the rule does not make the device untestable, because the SEC15-3 is DC powered and a battery supply plus calorimeter needs no RF probe; Stiffler began exactly that (thermal bottle) and the page states no result. The Lattice reason for the rule is Not testable as stated.

5. Claim 7: the mathematician calls the self-running logic "Internally inconsistent" (a stated electrical COP of 2.2 yet loop-back impossible). I rule it incomplete rather than inconsistent. Stiffler's later position (CEC is a conversion ratio in an open system; "I do not believe in OU"; looping "would require manufacturing energy" [web: Energetic Forum thread 4067 posts #1, #4, #2122]) is coherent for outputs that are not usable electrical power or whose source is drainable, like a heat pump. It is NOT coherent with a measured 18 W electrical output at a transformer: ECAT's own accounting puts the lamp at 12 W against an 8.25 W input, so a rectifier at 69% efficiency (eta = 0.69) would close the loop. The page does not argue why that cannot be done. This is the single cheapest decisive test and the inventor called it a diversion. The statement that the "acid test" is invalid, and that suppressors push it, names no person or document; it is the inventor's assertion, recorded as such, and no outside record supports or refutes it.

Smaller points agreed or accepted: CEC units (joule-seconds are action, not energy); "UCEC > 1 implies PCEC > 1" only over equal intervals; Eq 9 should contain wavelength, not pi (0.62 sqrt(D^3/lambda) [web: https://www.itecspec.com/3gpp/38.810/s/7.2.1.2]); SEC Special Coil impedance tables use sqrt(R + |X|) instead of sqrt(R^2 + X^2) (minima fall in the right places, values unusable); 0.001 V across 1 ohm is 1 mA, not 0.001 mA; the 12.0 MHz "amplification peaks" are the +6 dBV band edges; Fig 18 "relative power" ticks 0-4 look like index labels.

Corrected claim list (final verdicts, Step 2 scale)

1. ECAT: an air-core MOSFET coil circuit gives power gain 2.194 (8.25 W in, 18.11 W out) by "Poly-Phasing" capture of external energy. - Measurable form (COP about 2 at the terminals): Never properly tested. One operating point, inventor only, no method, no replication of the intact design; the page and the 2009 forum post say copies "failed" and name none; others did not reject it after a test (it was largely ignored). Not dismissed without testing. - Mechanism (external "virtual energy" between coils): Not testable as stated (no source named; ambient EM capture is orders of magnitude too small [calc above]). - [QE revision] Basis restated: 8.25 W is the bench-supply panel reading (25.0 V x 0.33 A), not 12 V x 0.69 A; the 12 W lamp is a nameplate rating; the page also says the transformer is "cold" while attributing 6 W of loss to it. Verdict unchanged. - Kept in the record: the inventor's report that ferrite or metal cores removed the gain until the saturating "Theta" design; his 2009 admission that "the math could not support the (conversion) concept" so "Conversion" was renamed "Coherence"; first ECAT tries 1990 and notebooks from 1985 (the page says 1993-96) [web: .../4067-high-voltage-from-thin-air/page25].

2. SEC: a coil with an Avramenko diode plug draws "cohered" energy from an "Energy Lattice"; CEC = Uout/Uin; 130-168 V at the free end. - Phenomena shown (voltage step-up, lit neon/LED, one-end excitation): Established as ordinary resonance and rectification. - Strong form (output energy exceeds input): Never properly tested. CEC is defined and never evaluated; no input/output balance by Stiffler or by any replicator in the record. Replicators (Lidmotor, Raui, Inquorate, zeos71, mikrovolt) show it can be built and lit; the one late input figure is 35 mA at 18 V = 630 mW with no output [web, Investigator]. - "Energy Lattice": Not testable as stated. - [QE revision] "Amplification" on the SEC Special Coil page has no measured magnitude anywhere: the charts plot frequency only and the text gives none. Added to the record: Inquorate's report that a white-noise bump at the coil's self-resonance appeared only after the coil had been pulsed [web, Investigator, thread 4067 p.145]; Stiffler's "Spatial Gate" challenge (over 300 V from 18 V input without a step-up transformer, thread 4067 post #1) as his own short statement of the principle. Verdicts unchanged. - Kept: Stiffler states CEC is "a conversion ... with a 'mathematical' gain. No creation of energy"; replicator remark that power computed from V and I at some frequencies is "always negative", which fits a phase-skew artefact [calc above]; the course claim that an exciter is a "Negative Resistance" (documents not read; any active oscillator is a small-signal negative resistance [memory, high]).

3. Stacking AV Plugs multiplies "excess energy" (2 plugs double, 3 triple). - Two capacitors store twice the energy of one: Established (trivial; exponent misprinted by 10). - Multiplication of "excess energy": Not testable as stated (no input reference; formula absent; relies on a neon photograph).

4. Energy through one 12 ft wire, no return conductor, 24 V transmitter, 48 LEDs. - Weak form (LEDs lit with a capacitive/displacement-current return, top hats, resonant coils): Established as ordinary physics; reproduced visually by Lidmotor at 10 ft and 25 ft [web, Investigator; videos not watched by reviewers]; related patents and papers by Strebkov, Avramenko and others [web: https://patents.google.com/patent/US6104107]. - Strong form (any efficiency, gain, or "lattice" mechanism): Never properly tested; the guide gives no frequency, current or power. Order-of-magnitude closure: 3 pF at 100 V rms at 10-13.6 MHz carries 19-26 mA, 48 LEDs need 0.13-0.4 W, supply 24 V x 50 mA = 1.2 W [calc], so no unusual physics is needed. The guide's own warnings (headaches, damage to electronics, hand detuning) fit strong near fields.

5. Ainslie heater calorimetry: water heat 1580.593 J exceeded electrical input 1198.759 J (CEC 1.319); returned energy is insignificant. - CEC 1.319 as excess heat: Never properly tested. One run, one screenshot, no DC blank, calibration, repeat or error bars; unstirred 20-25 mL tube with sensor in the top ~5 mL; ratio proportional to the mass entered; RF pickup on the LM335 not excluded; it is unstated whether Fig 15's input is the 40 MHz integral or 1 Hz meter readings. Not "tested and failed" and not dismissed. Heat-loss and glass corrections would increase the ratio, so they cannot explain it. In cups the ratio fell to about 0.83 above 42 C, which Stiffler attributes to loss. This is the best-documented experiment in the folder and the one most easily repeated. - Returned energy insignificant: Plausible, untested as a measurement (no waveform or integral shown); [QE revision] with the Ainslie documented L = 8.64 uH the freewheel energy is a calculated 2-4% of input, which supports the assertion for the 2.4 kHz regime. - Ainslie group (interested parties, originators of the circuit) report "well above COP 4.0" and a submission to IEEE as a preprint [web: https://www.scribd.com/document/23455916]; the visible page does not name Stiffler (uploaded by a user "fuzzytomcat"), so his role is unconfirmed and this is not support for claim 5. [QE revision: superseded. ExHeatExp is a close but not identical copy of the Ainslie group's documented circuit; the sister review review/Ainslie-Rosemary-Ann holds the Ainslie documentation and the critique record (Poynt99's artefact report; advocacy reports of 2009 replications). See the QE revision section.] No neutral, controlled replication was found by any reviewer or by me.

6. SEC exciters can be measured only with battery-powered, ungrounded, RF-filtered instruments; grounding detunes the AV Plug "interface with the Spatial Lattice". - Instrument advice (floating, filtered, short leads): Established (ordinary RF practice, quantified above). The papers themselves concede pickup gives "meaningless" readings. - Lattice reason: Not testable as stated. The rule does not exclude testing: DC input measurement plus enclosed calorimetry needs no probe. The thermal-bottle comparison (2 W resistor versus the board, one hour) is described with no result; the quick LED/black-plate test showed no temperature rise [read: SEC Special Coil]. Both stay in the record.

7. Stiffler's reports on testing: copies of ECAT failed because builders changed the design; Naudin's Time Energy Pump "fell short"; the self-running "acid test" is a diversion by those who would suppress research. - Copies failed / TEP fell short: Not testable as stated (no attempts named; JLN Labs index is live and does not mention Stiffler; ctglabs.com is empty [web, Investigator]). - Suppression assertion: Not testable as stated; the record shows no rejection by any institution (he submitted no paper, no patent was published). - Loop-closure test: Never properly tested (see ruling 5; cheapest decisive test).

What would settle it

• Claim 5: rerun ExHeatExp with the cell stirred, mass weighed, two independent sensors (one non-electronic or fully shielded), a DC blank at equal 2 W for 600 s on the same coil and tube, and input from a fast digitizer on both runs. If DC gives about 1.3 the 1.319 is an apparatus artefact; if about 1.0 the pulsed excess needs another explanation.
• Claims 1 and 7: ECAT with a resistive dummy load or the lamp inside a calorimeter, battery input measured by DC integration; then close the loop through a rectifier (eta needed 0.456, or 0.69 on the lamp-level figures). A circuit either keeps running or it does not.
• Claims 2, 3, 6: SEC15-3 on a battery in a thermal bottle against a resistor of equal DC input power (Stiffler's own proposed test), long enough to reach steady state; any RF V x I only with probe skew under 0.1 ns at 13.56 MHz (phase check) and a known dummy load first.
• Claim 4: input DC power versus power in a resistive load at the receiver at several wire lengths, with a metallic shield or return added for comparison.
• Records: USPTO Patent Center for 61/254,675 and any claimed non-provisional; archived stifflerscientific.com / drstiffler.com (Wayback was offline); the 2012 Panacea PDF (quanthomme.free.fr), "Series 18 Exciters Documentation" Rel 2.8, the missing "Measurements for SEC Exciters" data; Stiffler's channel @StifflerDr (163 videos, including the "Stiffler Loop" series, unreviewed by anyone in this project); the Lumen notices 75850580 and 95550170 (requester unknown).

Absence-of-evidence analysis

The record is thin or goes dark in six places. Each is an anomaly to explain, not a neutral gap.

A. No energy-balance paper after the 2010 announcement. ZPEnergy (2010-10-14) says a paper from replicators' data was planned [web]; none has been found, though white papers continued to at least 2011 and videos and circuits to 2018 [QE revision]. Death (2018-11-12) explains the end of the record, not the eight years before it. Competing explanations: (1) no result survived careful measurement (supported by his own "Is the Loop really doing anything?" video title, his 2009 "math could not support" remark and his negative black-plate test; against: he kept publishing and selling for years); (2) the data were kept for licensing (supported by "licenses available from Stiffler Scientific"; against: he also posted most circuit detail openly); (3) health or other priorities (no evidence either way). Deciding records: archived Stiffler Scientific pages, the 2012 Panacea PDF, Series 18 documentation, the channel's later videos.

B. Replicators produced no calorimetry in about nine years. The effect is easy to build and the thermal-bottle test is easy to run, yet no source reports one. Explanations: hobbyists lack calorimeters (likely); those who measured found nothing and did not post (unknowable; absence of a negative report is not evidence of one); replicators accepted lit LEDs as the proof. A single forum thread of a replicator's calorimetry would settle it.

C. Thermal-bottle test with no stated result (SEC Special Coil). The page describes a one-hour comparison and reports nothing. Either it was not run, was run with a null result, or the page was never finished; the ExHeatExp page is likewise marked unfinished. The record cannot separate these. The stated black-plate test with no temperature rise is evidence against a gain and is kept.

D. Patent. "Patent Pending 61254675" in the Panacea course. No published application or patent by Stiffler was found on Google Patents. The number is a US provisional [memory, medium]; web check: I found no record of 61/254,675 itself, but a neighbouring provisional, 61/254,559, was filed 2009-10-23 [web: patent citing it, e.g. https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/8552592], so this one was probably filed in late October 2009 (inference from neighbours, not confirmed). A provisional lapses after 12 months unless a non-provisional is filed; "patent pending" on 2010-10-14 is consistent with that window, [QE revision: the 'Patent Pending' words were seen in Scribd document 35609446; the 2012 Panacea PDF was not read, so nothing on record shows the 2012 course repeats them.] Explanations: lapsed provisional (common for individual inventors); a non-provisional under another name or entity (it would have published about 18 months after filing, around April 2011); filing never pursued. USPTO Patent Center would decide; not yet searched.

E. Disappearing material: the demonstration video y9aJohml130 removed; stifflerscientific.com and drstiffler.com dead; tuks.nl mirror and Wayback inaccessible during the run (Wayback HTTP 503; retry); Google results removed after two legal requests (Lumen 75850580, 95550170), requester unknown, pages behind an anti-bot check. Explanations: lapsed hosting after death (fits the timeline; the channel itself stays online, which cuts against active takedown); copyright or DMCA requests by a third party (unknown, possibly unrelated to Stiffler's science); suppression (no evidence of any act; an unanalysed notice is not evidence in either direction). The Lumen notices are the record that would decide.

F. "Advanced Technology Group" (1993-96) and Stiffler Scientific: no corporate record on OpenCorporates; no group identified. Explanations: an informal group or employer team, a DBA (Harris County assumed-name index not searched), or a name used only on the page. Not resolvable now.

Identity and death: a Ronald Stiffler, 75, died 2018-11-12 (Atascocita Observer obituary on Post and Courier; the page gives no education or career) [web: https://obits.postandcourier.com/us/obituaries/observergroup/name/ronald-stiffler-obituary?id=33303171]; a newsletter of 2019-01-24 says Dr. Stiffler "passed away last November" [web, Investigator]. That the obituary subject is the inventor is an inference from name, place and date; the Darst obituary was not read. The "Dr." title and his career are not established in the review files.

Web verification (arbiter, 3 of 8 calls used)

• Provisional 61/254,675 date: not found directly; neighbouring provisional filed 2009-10-23 (see D). Retag [memory, medium] as [web, inferred from neighbours].
• Obituary: age 75, died 2018-11-12 confirmed; no education/career text. Retagged [web].
• Ainslie preprint: COP "well above 4.0" and IEEE-preprint note confirmed on the Scribd page; Stiffler not named in the visible content (supports the physicist's and contrarian's caution).
• Checks inherited from reviewers: Strebkov/Avramenko single-wire patent US6104107 exists [web: https://patents.google.com/patent/US6104107]; Eq 9 near-field form from 3GPP text [web]. Unverified [memory]: LM335 10 mV/K (high), glass specific heat 0.84 J/g/K (medium, an assumption used only in a sensitivity), "any active oscillator is a small-signal negative resistance" (high). No verdict depends on the medium-confidence items.

Open questions

• What is the real input and output power of the ECAT circuit at 4.3 kHz, with the intact design (coils, Theta core, MOSFET part, values not given in text)?
• Does the SEC exciter produce measurable heat above its DC input in a calorimeter, and does the effect survive a battery supply and a dummy load?
• Does a DC blank in the ExHeatExp cell read near 1.0 or near 1.3?
• Why did the promised SEC energy-balance paper never appear (explanations in A)?
• Is the "Negative Resistance" claim a statement about anti-phase V and I with net outward real power, or a small-signal property? Needs measured V, I and phase with matched probes.
• What did Stiffler's 2017-2018 "Stiffler Loop" videos show, and what was the "Is the Loop really doing anything?" answer? Not reviewed by anyone yet.

Corrections to earlier documents

• claims.md claim 1 and facts.json (ECAT): "arithmetic does not close" and "logic problem" should read: the sentence on the 6 W transformer loss is ambiguous; the whole-page accounting (18 W into the transformer, 6 W lost, 12 W to the lamp) closes to within 0.11 W, but rests on a nameplate lamp rating and no stated measurement; the 8.25 W / 90 mA basis cannot be reconciled from the page. Add the 1990 and 1985 dates.
• claims.md claim 5: "the energy returned to the battery ... is insignificant" is Stiffler's assertion ("I have confirmed"); no waveform shown.
• claims.md claim 6: "excludes standard grounded instruments by construction" is too strong; RF probing is unreliable but DC input plus calorimetry needs no probe.
• deep/ECAT: the 12 W lamp is a nameplate figure; the deep read's 2.25 W deduction follows only the literal reading.
• deep/Wireless guide: 143 turns in 73 mm is #24 AWG, not "#30-#32 or finer".
• deep/ExHeatExp section 4: omitted glass and losses raise, not lower, the apparent CEC (to 1.40 with an 8 g tube); the artefact must be in temperature, mass or input.
• facts.json (ExHeatExp): add proportionality to mass and the unstirred top-of-tube sensor; (SEC Special Coil) add that impedance minima are at the right frequencies though every Z is wrong.
• sites.md: Ainslie preprint's visible page does not name Stiffler; keep as lead only. The Investigator's date correction (1990) and the removed video are accepted.
• Physicist and mathematician summaries listing "Internally inconsistent" or "Established" for individual sub-claims are superseded by the verdicts above where they differ.

Verdict summary

QE revision (2026-10-08)

Inputs: qe.md (verdict MINOR) and the three deep reads it ordered: deep/Energy-Conversion... (deep read 2, schematic and all 16 images), deep/SEC-Special-Coil... (section 2, all images, Fig 18 reconciliation) and deep/ExHeatExp... (comparison with the Ainslie documentation). I read those sections in full, and the sister file review/Ainslie-Rosemary-Ann/arbiter.md (replications item) for the critique record. Tags: [read] deep reads or sources as cited, [web] with URL, [calc] arithmetic. No new web calls were needed; the web-dependent facts are the ones already checked above.

Verdict changes

None of the category verdicts changes. The table and Corrected claim list above are annotated where the QE material changes their wording or basis. In summary:

Claim 1 (ECAT): the input basis changes, the verdict does not

What the figures show [read: deep/Energy-Conversion deep read 2, sections A-F; images ecat009, ecatcir000, ecattp001-005, ecat010-012]: - The 8.25 W is a bench-supply panel reading: a Protek 3006B at 25.0 V and 0.33 A (25.0 x 0.33 = 8.25 W exactly) [calc]. My earlier "8.25 W at 12 V = 0.6875 A" and "if 1.08 W (90 mA at 12 V) were the true input the gain would be 16.8" (Recomputed numbers, first bullet) are withdrawn for the lamp version: they rest on the page's "12 V DC supply" sentence, which sits with the 4.3 kHz schematic design. The photo contradicts a 12 V supply for the lamp run. Correction recorded; the loop-closure figures (eta 0.456 and 0.69) were computed from the ratio 2.194 and the 12 W lamp figure, not from the voltage, so they stand as stated. - Two operating points are on the page. The schematic and test-point photos (TP-1 106 V, TP-2 172 V, TP-3 94.5 V, TP-5 0.31 V p-p across R3) repeat at about 4.3-4.4 kHz [calc, +/-5%]. The lamp-version drain photos (ecat010/011, 205 V) repeat at about 1.1 kHz, and the HF-transformer secondary (ecat012, 304 V) at about 1.17 kHz. Both cannot be the same drain waveform, so the 4.3 kHz / 12 V description may not describe the setup that produced 8.25 W and 18.11 W. The page does not say so. TP-4 is missing from the saved page. - The schematic has two sources: V2 (DC supply) and V1, a separate pulse generator driving the gate. The panel photo is one supply, so V1's power is apparently not in the 8.25 W. A gate-charge estimate (assumed MOSFET values) puts V1's direct share under 1% [calc, assumed], so it cannot explain a gain of 2.2, but it is an unstated omission, and the generator's own mains draw is a further unstated quantity. - R3 (the TP-5 sense resistor) has no value, TP-5 has no DC level, and C2 and C4 bypass part of the switching current, so input current cannot be derived from TP-5. No current trace exists anywhere on the page. No figure measures the 12 W lamp (nameplate; ecat008 shows a bright lamp with arcing at the base) or the 6 W transformer loss. - A numerical coincidence the deep read found: at 25.0 V, "quiescent current is only 90 mA" is 2.25 W, and 8.25 - 2.25 = 6.00 W, equal to the "some 6 watts" loss. One reading is that the "6 W" is a difference between two supply draws (0.33 A and 0.09 A in two states) rather than a measured transformer dissipation. The competing reading (90 mA belongs to the 12 V design; the match is chance) is equally open. I record both. The page's 12 + 6 = 18 W total still differs from 18.11 W by 0.11 W. If 2.25 W were the true input the ratio would be 18.11/2.25 = 8.05 [calc]; this is shown only to size the ambiguity, not as a claim. - Ruling 1 gains a tension the reviews missed: the page says the MOSFET and heat sink are "cool (not cold like the transformer) but cool" [read: ECAT page; deep read 2 section F], while attributing about 6 W of loss to a 60 Hz transformer. The sentence does not say which transformer ("cold" is by touch; no temperature is given). If it means the 60 Hz matching transformer, "cold" is hard to square with 6 W of dissipation [memory, medium: small laminated transformers run warm to hot at several watts], which would favour the reading that the 6 W is not a measured loss. If it means the polyphase transformer, there is no conflict and the 60 Hz transformer's temperature is unstated. This is an unresolved inconsistency in the inventor's report, kept in the record with its competing readings. - Third-party estimate restored: forum user Kokomoj0 estimated from an ECAT video "an addictional 4 watts roughly" and said proper measurement needs DC conversion and a load [web: Energetic Forum thread 9344 post #3; quote verified by the Investigator in research/pages/ef_9344.txt]. One forum estimate from a video, by a non-interested viewer, and not a measurement; it is a supporting lead and stays in the record beside the inventor's own claim. - Verdict: unchanged. Never properly tested for the terminal gain: still no current or power measurement of input or output, one operating point, no blank. The corrected basis makes the inventor's accounting less clear, not more damaging by itself: the 8.25 W is a real instrument reading, but the "output" side rests on a nameplate and an unexplained 6 W.

Claim 2 (SEC): "amplification" has no magnitude

• Figs 10, 11 and 13 of the SEC Special Coil page plot frequency only (axis "Hertz"); the cone heights are the frequencies in the text (about 15.2, 15.2, 18.6 MHz; 12.0 and 12.0 MHz). Fig 15 is a photograph of the bottle apparatus (LCD unreadable) and Fig 16 of a string of lit LEDs, probably under the black plate; neither holds a temperature result [read: deep/SEC-Special-Coil section 2, B-D]. The earlier "figures unread" worry is closed: the evidence for amplification at those frequencies is two sentences and frequency-only charts. The only level data are the scope shots Figs 4-6 at the bare-coil points (null at the third point, where "very little signal ... appears across the 50 ohm load").
• Fig 14's input-power method (Pin = Vin x Iin, Iin = Vd/Rg, Va = Vg - Vin) is magnitudes-only: it is exact only for in-phase quantities, otherwise Pin is apparent power, an upper bound on real power. No phase measurement is described and no numbers are given. This agrees with the artefact point already made (RF phase skew), and now rests on the figure as well as the text [read: deep section F].
• The "Measured (MHz)" table (3.9866 / 10.4230 / 12.3340) is very probably the same data as single-wire Fig 18 "Energy Peaks" (3.9066 / 10.42302 / 12.334 MHz): the last two agree, the first differs in one digit. The table's derived cells are all multiples of 3.9866, so 3.9866 is more likely and 3.9066 a probable typo; this is an inference, not proof, and both values stay on the record [read: deep section G]. These are LED light-output peaks, not the coil self-resonances (4.149 / 7.062 / 10.225 MHz) and not the 12.0 / 15.1 / 18.6 MHz "amplification" points; the sources do not relate them. 3 x 3.9866 = 11.96 MHz is within 0.04% of the 11,963,991 Hz point; the page draws no link and none is claimed.
• The thermal-bottle test is photographed and described procedurally (resistor against board, four-minute readings for one hour) with no temperatures or conclusion anywhere. Absence-of-evidence item C stands; the anomaly is now confirmed against the images.
• Restored items: Inquorate's report that the white-noise bump at self-resonance appeared only after pulsing [web, Investigator, thread 4067 p.145]. A single forum report, no measurement attached; it is an observation consistent with ordinary charge on, or a changed state of, the coil's diode plug, and also with an unexplained effect. I do not rule between them. The kit's tuning rule ("maximum neon glow and minimum supply current", kit manual p.6) is the textbook signature of a parallel resonant tank at resonance (impedance maximum, supply current minimum) [read; physics high], which supports the ordinary-resonance reading but does not by itself test the gain claim. Stiffler's "Spatial Gate" challenge (more than 300 V from an 18 V input with no step-up transformer or multiplier) is his own one-line statement of the principle; a voltage statement only, not an energy statement.
• Verdicts unchanged.

Claim 5 (Ainslie calorimetry): the sister-review cross-reference

This replaces "Ainslie's own documentation" as a missing item: it is in review/Ainslie-Rosemary-Ann (Quantum 2002 article, patent, 2009 reprint). Read in the ExHeatExp deep read, section A-E: - 555 setting: Ainslie documents 2.4 kHz and 3.7% on-time in both 2002 and the 2009 reprint; Stiffler's Fig 11 reads 2.38 kHz and 3.80% [read: reprint p.5; ExHeatExp Fig 11, earlier read]. He says he used "specifications close to the inventors'". The match fits his words. - Flyback diode: the Ainslie 2009 reprint says plainly "A flyback diode, (D1) is placed in parallel with (RL)", 1N4007 [read: reprint p.5]. So Stiffler's Fig 16 topology is the same as Ainslie's, not a modification. This corrects the mathematician's [web, medium] note that sources "do not clearly show" a freewheel diode, and my own ruling 3 and claim 5 wording that treated the diode as Stiffler's addition. The mathematician's 2-19% returned-energy range assumed L = 5-50 uH; with the documented L = 8.64 uH (and R = 8-10 ohm) the freewheel energy is 17-27 uJ per cycle, 35-64 mW, 2-4% of the input at 2.4 kHz and 3.8% [calc, deep read section D]. This is a calculation of the energy circulating in the loop, not a measurement; none of it returns to the battery through a freewheel diode as drawn. It supports Stiffler's "insignificant" for the 2.4 kHz regime only. Stiffler's own L is not given; a Wheeler estimate for a 40-turn coil on a test tube gives about 5 uH (assumed tube size) [calc], i.e. similar. - Self-oscillation: Stiffler wrote that the oscillations "do not match any of the Ainslie Documentation and are orders of magnitude higher than described". He read 1.49 MHz; Ainslie's documented states are 143-200 kHz at 1.3% (2002), "random oscillation" (2009) and over 500 kHz at about 55% (Kelly, a third-party remark) [read: deep/quantum section 4; sister arbiter]. 1.49 MHz is 7.5-10 times the 2002 figure and 3 times Kelly's. "Orders of magnitude" is true only against the 2.4 kHz setting (about 620 times). He did not reproduce the documented 143-200 kHz state, but his regime is another version of the same MOSFET parasitic oscillation, which the Ainslie review identifies as the source of her apparent gain [read: sister claims.md item 3]. A 1.49 MHz ring with L = 8.64 uH needs 1.3 nF [calc], mostly MOSFET output capacitance and wiring; this is ordinary tank behaviour. - Differences: 555 type (stated "CMOS", Ainslie parts list NE555; a third-party remark, Kelly, says only the bipolar NE555N works), coil former (Pyrex versus ceramic, acknowledged), supply (20 V bench with a split gate supply, versus 24 V lead-acid pack with a separate 12 V dry cell; his split is an experimental echo of her two-source arrangement), shunt and calorimetry. Reading: ExHeatExp is a close but not identical test of the Ainslie circuit; both what matches and what differs belong in the record. - Date: the folder name "Ainslie", the phrase "Ainslie Documentation", the 2009 reprint's appendix and the 2009-2011 forum activity make c.2009 (range 2008-2010) probable, not proven [read: deep section B]. Membership in the Ainslie et al. open-source evaluation (Scribd, uploaded 2009-12-01) is unconfirmed: the deep reader's fetch showed no author list, and the "Dr Stiffler" Google snippet remains a snippet only. The group reports "well above COP 4.0"; Stiffler's one result is CEC 1.319, and he does not report COP. His result is therefore not the group's number. - Critique and replication record, from the sister review (low reliance as recorded there): a forum analyst (Poynt99) traced the circuit and reported measurement artefacts, including a probe-ground position difference between publication and build; an advocacy article (OpEdNews, interested source) reports that "at least three" people reproduced the heating in 2009, including Glen Lettenmaier, with COP figures not confirmed by the sister arbiter's fetch [read: ../Ainslie-Rosemary-Ann/arbiter.md replications item; search summaries, not opened there]. Neither side is a neutral, controlled, published test, and Stiffler is an interested party (he sold SEC products). Both stay in the record. ExHeatExp itself is the only water-calorimetry attempt known to me: one run, no DC blank, as ruled. - Verdict: unchanged: Never properly tested (CEC 1.319 as excess heat); returned-energy point Plausible, untested as a measurement, with the 2-4% calculation recorded. The settling test (stirred cell, weighed mass, DC blank, two sensors) is unchanged.

Claim 7 and the NILS statements

Stiffler's 2011 post says looping output to input "would require that we were manufacturing energy" [web: Energetic Forum thread 4067 p.142, quote verified in research/pages]. The Panacea course (Scribd 35609446) says "Self Charging" NILS (Near Infinity Light System) units were never released [web: https://www.scribd.com/doc/35609446]. A self-charging light sits in tension with the post. I keep both side by side: one source each, from the inventor or his promoter (interested party), neither explains the other. It does not change ruling 5: loop closure remains Never properly tested.

Smaller items restored from qe.md section 3

• Olkhovskiy, Muellerova and Martinek (2019, IEEE) on measuring single-wire efficiency, found by the contrarian [web: https://dspace.zcu.cz/items/ecbbaecc-629a-4d7d-8f19-0a21b92e631e/full]: pertinent to claim 4's efficiency question; added to further reading, not weighed here.
• Measurement Methods Fig 19: the 13.6 MHz label is not the tallest peak on the screen [read: deep p.17].

Corrections to earlier documents (this revision)

• arbiter.md, Recomputed numbers (ECAT bullet) and ruling 1: "8.25 W at 12 V = 0.6875 A" and "90 mA at 12 V = 1.08 W, gain 16.8" were an error in my own earlier answer. The supply panel photo shows 25.0 V and 0.33 A; the 12 V belongs to the 4.3 kHz schematic design. Corrected above. The same step appears in example.py (line ~157), physicist.md, mathematician.md, contrarian.md, claims.md claim 1, facts.json (ECAT key claims) and primer.md (line ~43, which pairs 12 V and 4.3 kHz with the lamp result); the orchestrator should correct these and separate the two designs in facts.json.
• arbiter.md, ruling 1 and Corrections list: "the 8.25 W / 90 mA basis cannot be reconciled from the page" is now partly reconciled: 8.25 W = 25.0 V x 0.33 A (photo). What still cannot be established: whether 90 mA is a 12 V or a 25 V figure, and whether V1's power is included.
• arbiter.md, Source gap note: ECAT "scope numbers unknown" and SEC "Figs 11, 13, 16 unread" are out of date (annotated above). The thermal-bottle result and the "Measurements for SEC Exciters" data remain missing.
• arbiter.md, claim 5: "No independent critique or replication found" replaced; "Ainslie's own documentation" is in the sister review.
• arbiter.md, Absence-of-evidence A: "white papers and videos continued to 2018" now "white papers to at least 2011 (Series 18 documentation Rel 2.8, September 2011, studylib, not read); videos and circuits to 2018 (Sun0008, 2018-04-30)" [web, per qe.md and sites.md]. "Eight years" without an energy-balance paper stands.
• arbiter.md, Absence-of-evidence D: the 2012-course sentence is withdrawn (the 2012 PDF was not read).
• mathematician.md, claim 5: the [web, medium] note on a missing freewheel diode is wrong (Ainslie 2009 reprint p.5), and the 2-19% returned-energy range is superseded for the Ainslie case by 2-4%.
• physicist.md section 0: the Ainslie web check "failed to retrieve primary sources" is superseded; the primary sources are in the sister project.
• facts.json / claims.md: items listed in qe.md section 7 and the deep reads' correction lists apply (SEC table linked to Fig 18; ExHeatExp dated c.2009 with the specification match; claims 1, 5, 6 pending corrections from the Corrections list above).