RankingsPrimerArbiterPhysicistMathematicianContrarianQESites logClaimsSourcesResearch notes

Kelly-Patrick: central claims

Project: Patrick J. Kelly, "The Practical Guide to Free-Energy Devices" (PJKbook.pdf, version 32.9, 16 February 2018, 2,481 pages, 79 OCR pages). One document; facts.json lists it once as a whole and then by chapter/appendix group.

Source note (not a source gap): the text layer is readable, but circuit diagrams, patent figures, graphs and some equations are images and were not rendered by the digesters; deep readers rendered a few pages only. The book is a compilation: Kelly says it is mainly a report of what other people say and that he has not built or proved most devices ("you are welcome to consider this a work of fiction", p.11). It is therefore the primary source for what Kelly and the texts he reproduces claim, not for independent measurement. Coverage (updated after QE, 2026-10-09): about 85 percent of the pages are now deep-read: pp.42-1006, 1060-1071, 1111-1150, 1151-1277, 1278-1374, 1375-1417, 1418-1435, and Appendix pp.1445-1960, 2003-2087, 2196-2252, 2280-2313, 2329-2432. Digest-only, so weaker: the introduction (pp.1-41), Chapter 12 electronics outside pp.1060-1071, and Appendix pp.1961-2002, 2038-2040, 2088-2195, 2253-2279, 2314-2328 and 2433-2481. The fact-finder's first pass sampled some digests; QE read the rest in full and ordered the deep reads that are now in. The contents list names Chapter 22 "Dealing With Chemtrails" and Chapter 23 "The Mandela Effect", but the words occur in the extracted text only in the contents list; in the body, Chapter 21 is followed by Chapter 22 "The Sabourin Generator" (listed as Chapter 24 in the contents) [read: QE].

Non-physics content: the book also makes medical claims (Chapters 11-12, 14, 20-21: Beck blood electrification as a treatment for HIV and other diseases, suramin and autism, Lakhovsky tumour settings, an Alzheimer audio recording, colloidal silver as a cure for cancer, malaria, AIDS and Ebola, Keshe GaNS water) and essays on finance, media and climate (pp.993-1005). They are outside this review (no medical reviewer; no trial is cited for any of them). They are inventoried in facts.json (Chapter 12 and Chapter 20 entries) so that the review's silence is not read as endorsement; the book's own safety warnings (Beck's precautions; pacemakers within 10 m of the Lakhovsky machine; lethal capacitor charge, p.1408) are recorded there too.

The central claims

  1. The "closed system" claim. Conservation of energy only forbids taking out more than is put in or is already in a closed system; real devices are open to an ambient / zero-point energy field, so a coefficient of performance (COP, output power / the operator's input power) above 1 does not break the law. (Kelly, pp.12-17, 422.)

  2. Sharp pulses, resonance and a tiny drive can tap that field. A short sharp DC pulse or spark (pulsed coils, spark gaps, Joule Thief circuits, Benitez and Bedini circuits, Tesla coils) draws extra "cold" energy from the environment and earth, giving COP above 1 in battery chargers, motors and generators. Don Smith's resonant Tesla devices (output rising as V squared and f squared, 160 kW from a 12 V front end), Kapanadze's self-running unit (40 W start, 5 kW out), Benitez GB 121,561 (400 W in, 2,400 W battery power; the stored energy per spark is 0.16 J, so at most 160 W at 1,000 sparks/s [calc: arbiter]) and Tseung's FLEET toroid are the book's chief examples. Numerical COP claims added after QE: Adams-motor builder "Maimariati" COP 1,223 (27.6 W in, 33.78 kW out, p.93; a second statement gives 27 W in and 32 kW out, PDF p.28); RotoVerter "at least 10 times" (p.118); coil shorting "about 100 times" (open-circuit pulse voltage 16 V to 440 V, pp.123-128); Hatem "20 times" (pp.128-129); Thane Heins' patent test 0.29 W in and 11.51 W out, "COP of 39.6" (p.150); Klingelhoefer "Gabriel Device" 60 W in and 480 W out, "COP=8" (p.170); the South African pulse generator 0.89 W in and about 40 W out, "COP of 45" (Chapter 19); and Kelly's own LM2587 DC-DC converter test, 2.664 W in and 4.01 W out, "COP=1.5" (p.1259; no meter or averaging stated). Kelly's own caveat on such readings: volts times amps is not power for non-sine or reactive loads, and a reading of "COP = 1.56" "may be wrong" (pp.1089-1091).

  3. Permanent magnets can be switched or "conditioned" to give large gains. Small control power that switches or wobbles the flux of a permanent magnet yields far more output (Flynn, Tseung frame, Heins, Annis and Eberly COP 200-400 for the switches, Asaoka, Kunel, Kawai COP 3.18), culminating in Floyd Sweet's Vacuum Triode Amplifier: more than 500 W at 120 V 60 Hz from under 0.31 mW (COP about 1.6 million; 1.69 million from the 1987 video pair, 528.75 W and 0.312 mW), explained as "negative energy" from the Dirac Sea. Kelly knows of no replication of the VTA. Annis and Eberly's COP 200-400 is given for the switches "without the coupling circuit"; Flynn's claim is a factor-2 coil power claim [arbiter corrections].

Sub-class added after QE: motors driven by permanent magnets alone (Chapter 1). Kelly's premise: "the power of a magnet flows from the environment and not from the magnet" (p.42). Examples: ShenHe Wang's 5 kW colloidal-magnet generator ("two ... completed the Chinese government's mandatory six-month 'Reliability and Safety' testing programme in April 2008", pp.42-43; no tester, data or report named); Howard Johnson US 4,151,431 (1979); Ecklin-Brown, Tracy, Teal, Jines, Kundel, Hohl/Ascanius (a magnet stack that rolled on a V-track, "did work"); Perendev ("not less than 100 kilowatts"; Kelly says this is unconfirmed, with "a brief test by Sterling Allan" the only test, and by mid-2010 the backers were "most unhappy", pp.78-79); Mr Sung 20 kW and an unsourced 225 hp machine (pp.67-68). Kelly's own caveats: PM-only motors are "very difficult" to make work, and repulsion-mode magnets reportedly "lost their magnetisation" after about three months (p.82). Also Tewari (Kelly: COP 2-3; the patent says only "partial nullification" of back-torque) and Kawai (19.55 W in, 62.16 W out, one patent run).

  1. Impulse, gravity and buoyancy devices produce excess mechanical energy. Jerked or pulsed flywheels (Tseung COP 3.3, Chas Campbell 750 W in / 2,340 W out in a replication video, Hardy water jet), pendulum and lever machines (Milkovic, Dmitriev), Skinner's 1939 torque machine ("1200 percent") and buoyancy generators (Kwok "Hidro") gain energy "from the gravitational field"; self-powered running after start-up is asserted. On Skinner, the primary United Press text says "about 1,200 per cent" as Skinner's own estimate for a 1/8 hp motor (a factor of 12 [web: https://rexresearch.com/skinner/skinner.htm]); COP 1,200 is Kelly's inference (pp.405-406).

  2. Water can be split, and engines run on it, for far less than the textbook energy. Bob Boyce's electrolyser (211-216 percent of "Faraday" in DC, 1,000-1,200 percent pulsed), Meyer's water capacitor, Cramton (2,340 percent) and Zigouras's HHO cell yield many times the gas that 2.34 Wh per litre allows (the percentages are against the 2.34 Wh/L energy figure, not the charge law [arbiter correction]), with the surplus drawn from the environment; ordinary generators and cars run on the gas (three UK men: a 5.5 kW generator on about 3 litres of gas a minute). Added after QE, with source figures and measurement methods: - Puharich (article and US 4,394,230): 25-38 mA rms at 4.0-2.6 V rms, 98.8-100 mW for 30 minutes (180 J), no wattmeter; 10.80 cc H2 + 5.40 cc O2 by gas chromatography and mass spectrometry; claimed 91.3 percent efficiency and a 14.9 percent "surplus" (26.85 J). Recomputed by QE and the deep reader: 0.482 mmol, Gibbs efficiency 63.5 percent, burn-back 137.7 J against 180 J; the printed figures carry a factor-1.5 slip and use the enthalpy ΔH where the work is the Gibbs energy ΔG; and the reported gas is 1.36-2.07 times the Faraday maximum for the stated current. Kelly adds, without source, that Puharich drove a motor home "hundreds of thousands of miles" on water (p.1902). - Hasebe US 4,105,528: 2.8 V, 30 A (84 W) for 116 cc/s, "2.6 cc hydrogen per coulomb", "over 20 times" Faraday (22.4 times at STP); no measurement method stated; 0.72 J/cc against a Gibbs floor of 7.05 J/cc. - Chambers (Xogen) US 6,126,794: 12 V at 300 mA (3.6 W) pulsed; "1 p.s.i. per minute"; "successfully run such engine without any gasoline" (no run time, engine size or input power for the run). - Spiros WO 9528510: a claimed self-sustaining loop (electrolysis, self-compression, expander, fuel cell); measured test to 2,450 psi (about 1.85 kW, voltage about 29-30 V, 88-100 L/h); calculated figures to 2,000 C and 150,000 atm (760 W, 1,235 Wh from 1,000 Wh, break-even "at approximately 66 kAtmospheres"); no energy recovery was measured. - High-mpg carburettors and boosters: Shelton "approximately eight times the mileage", Schwartz "over 200%", Ogle "in excess of one hundred miles per US gallon" (patentees' one-sentence statements, no baseline, speed or method); Hotsabi 18 to 27 mpg; Pearson / Saturn SL 26 to 44-46 mpg and one 88 mpg run (a seller's report, leaner ECU maps). Kelly's note that "the oil industry does not want functional high mpg carburettors" is not in the patent texts and is unsourced.

  3. Aerial, passive and fuel-less devices deliver power. Moray's radiant-energy device (635-675 W of lamps and appliances in witnessed demonstrations; these are nameplate sums, and 635 W conflicts with "0.878 hp" [arbiter]), Tesla/Plauson aerial plates, Papp-style inert-gas / air-plasma engines, Coler's magnet-and-coil apparatus, pyramids, the Joe Cell and the Barbosa and Leal "electron trap" all draw on environmental energy; Kelly accepts many on testimony and says only that a failed replication does not prove a claim false. - Papp, per his patent US 4,428,193 (pp.2196-2228, deep-read after QE): a mixture of 35.6 percent helium, 26.3 neon, 16.9 argon, 12.7 krypton and 8.5 xenon; thorium-232 (about 4 g) and other radioactive material in the electrodes; "It is calculated that ... there is a minute fusion reaction in the helium consisting of the energy conversion of a single helium atom" at "approximately 100,000,000 degrees F"; "a cylinder can run at 1200 rpm approximately 1000 hours, if not more, on a single charge of gas"; input a 24 V battery plus 27.12 MHz at 3.4-8.4 A (RF voltage and power not stated). The patent reports no measured output; the 300 hp, 1968 and 1983 episodes rest on Kelly's chapter (p.643). The demonstration is "300 HP for 40 min" in Kelly's p.37 table and 35 minutes on p.642. - Barbosa and Leal (WO 2013/104041-43): patent "at least 1 to 100" (COP 100); "22 watts producing 6 kilowatts" in a video (p.182); "169 kilowatts of power flowing from the ground ... 104 times" (p.211; p.37 table "COP=102.4"; no date, witness or instrument); forum builder "Clarence" 3 kW with many ground rods, a battery, charger and inverter (pp.198-204). Matchett / Ascanius mains-rod circuits: 61 V under a 7 W LED load; Kelly doubts that they give net power (p.211).

  4. Suppression. Inventors have been silenced, bought out or classified by vested interests (more than 10,000 US builders, "40,000" confiscated patents, Moray, Meyer's death, Sweet's death), which Kelly offers as the reason none of this is commercial (pp.17, 22, 993-1005; mostly uncited). "More than 10,000 US builders" and the TC 2800 "subject matter of special interest" list image are both on p.17 [read: page image p.17; arbiter correction]. Further episodes, each from one source: Adams "was told that if he shared the information, he would be killed" (p.28; Kelly's introduction); the Chinese government barring Wang's generator from the Shanghai Expo (pp.42-43, no source cited); Maurice's account of Sweet (an associate with a stake): a VTA said to have gone to "Senator John Glen", then the Department of Energy and General Motors, a planned $200 million suit, and people "claiming to be from the FBI" removing equipment the morning after Sweet's death (digest 94, p.2243).

Cross-cutting facts the later steps must keep in view (reported, not verdicts)

  • The deep readers found no input and output measured together by a named independent party for most numerical COP claims in the book. Where measurements appear they are: voltage x current products without power-factor statements (Gray 111 W from 1 mW), bulb brightness (Nick Giannopoulos, Salty Citrus, Gegene), battery terminal voltages after rest (UFOpolitics, Kelly's FLEET swap), or patent examples by the patentee (Kawai, Newman, Alexander, Ainslie WO 99/38247 at COP 1.086, 2.45, 0.976-1.176).
  • Kelly's own caveats are in the record: the preface hearsay disclaimer; bulbs light brightly at low power (p.314); Sucahyo's warning of surface charge (p.556); Kelly's lux test where a Joule Thief was less efficient than direct LED drive (p.476); he doubts the Ribero patent and the Tseung YouTube prototype; volts times amps is not power for non-sine or reactive loads, and a "COP = 1.56" reading "may be wrong" (pp.1089-1091); Perendev's 100 kW "has not been confirmed" (p.78); Tewari "does not mention anything about 20 kilowatts of self-powered output" (p.102); of the Matchett mains-rod circuits, "the mains meter charges you for power assessed by multiplying the average voltage by the average current", and he does not recommend them (p.211); Kim's HHO heater needs 4.2 kW at Faraday efficiency by his own arithmetic (p.1158); the ram pump runs on sunlight-driven water flow (p.1217).
  • Numbers conflict inside the book: Sweet input 0.31 mW / 1 mW / 10 mW and output 500 W / 1 kW / 5 kW; Tseung COP 3.3 (text) vs 3.4 (heading); MEG "no replication" (p.1145) vs "replicated by others, COP>1" (p.1147); Williams' Joe Cell truck ran in "shandy" mode with the fuel line connected; Papp "300 HP for 40 min" (p.37) vs 35 minutes (p.642); Heins "thirty times" (p.149) vs "COP of 39.6" (p.150); Barbosa and Leal COP 100 (p.179), 273 (22 W to 6 kW, p.182), 102.4 (p.37) and 104 (p.211); Adams "27 W / 32 kW" (PDF p.28) vs "27.6 W / 33.78 kW" (p.93), probably one report at two roundings (Kelly does not say); the lux table (p.476) misprints one row ("101 lux per watt" should be 214; the range 139-221 is unaffected); Puharich's own Faraday table against his reported gas (claim 5); the Spiros patent's "88 litres per kWh" is litres per hour at about 1.85 kW (48-54 L/kWh); Kelly's carburettor patent dates (Pogue "US 642,434, 1932" against the later patents' 1,938,497 of 1933; Shelton "2 May 1940" for a number in the early-1960s range [memory, medium]) look like transcription errors in the book.
  • Interested parties: patentees, sellers (Kwok licensing, PlasmERG, Magnetic Energy), the inventors themselves, and Kelly as the operator of a free-energy information site. Their descriptions are reported as their characterizations.

Corrections to earlier documents (from the deep reads, for the orchestrator)

  • digest/PJKbook-9cc3a5.24.md (p.596): "His germanium diode ... is said to have been tested many times" misreads the text; the page says the equipment was examined and tested many times and, separately, that it is said to have contained a germanium diode Moray built.
  • Page labels in several deep-read briefs did not match the extraction: Annis/Eberly, Kunel, Ivanov, Asaoka and Leben come before the Sweet section (Sweet is pp.244-249); back-to-back and solid-state Tesla coils are at pp.312-317, not within pp.326-361; Joe Cell runs pp.690-721; Garrett and Blue are pp.913-915, Zigouras pp.916-918; Kelly's "chapter 17" reference to the pulsed wheel is chapter 15. Fact entries use the deep-read ranges.
  • Mohamed's cross-reference "A-1209" for the Sweet quotation does not match this edition; the passage is at A-817 (Mohamed cited another version of the appendix).

Corrections to earlier documents (QE and deep reads, 2026-10-09; applied to facts.json and this file)

  • facts.json #ch1-2 said the digest "records one replication with 27 W electrical output" for Adams. Wrong: 27 W is the input; the claim is 27 W in and 32 kW out (PDF p.28), and the p.93 builder "Maimariati" gives 27.6 W in and 33.78 kW out, COP 1,223. The error came from the fact-finder's sampling of the digests. Corrected; the "27 W / 32 kW" passage is on PDF p.28 (printed I-18), not p.29 [read: deep read of pp.42-146].
  • facts.json #ch3a (Heins) lacked the patent test figure "0.29 watts ... 11.51 watts, which is a COP of 39.6" (p.150); added, with Kelly's "thirty times" lead-in noted as unreconciled.
  • Arbiter corrections 1-6 and 11 (claims 7, 4, 2, 3, 5 and 6, Benitez patents) applied above; correction 12 (digest 24, p.596) was already recorded in the earlier section of this file.
  • Puharich: qe.md quoted "A wattmeter was not used"; the source reads "it was decided not to use the watt meter" (p.1909). The gas is 10.80 cc H2 + 5.40 cc O2, not only "16.20 cc of gas"; the 63 percent figure assumes STP volumes (58 percent at 25 C). New point, in no earlier document: by the article's own Faraday table the reported 10.80 cc H2 is 1.36-2.07 times the maximum for the stated 25-38 mA over 30 minutes.
  • Spiros: "88 to 100 litres per kWh" misreads the patent's own table (litres per hour); "1.49 kW from 1 kW" is an energy ratio at an assumed 2,000 C with the heat not counted as input.
  • Hasebe: the Gibbs floor is 7.05 J/cc of mixed gas (QE wrote 7.06; immaterial), 6.40 J/cc at the table's own 25 C and 1003 mbar.
  • Chambers: the patent excerpt gives no gas volume, only 1 psi/min, so the energy test cannot be done on the cell figure; the engine-run objection stands as stated because no input power for the run is given. Kelly's date "16th July 2002" (p.1938) disagrees with the grant date the deep reader recalls (3 October 2000) [memory, medium]; to be web-checked.
  • Papp patent: Kelly's reproduction (pp.2196-2228) has the specification and drawings but not the numbered claims; the earlier description of it as reproduced in full (p.643) should be qualified.
  • Book misprints to carry, not silently fix: Pogue's first patent number and date, Shelton's date, Hatem "1955" against FR 2,826,800 of January 2003, Flynn's 51.1 / 25.5 degrees (should be 51.4 / 25.7).

Open requests for later steps

  • The QE deep reads (Papp patent, Puharich / Hasebe / Chambers, Chapters 1-3a, Spiros, medical claims, Chapter 14, carburettors and boosters) are done and folded in above. Still digest-only: the introduction (pp.1-41) and Appendix pp.1961-2002, 2038-2040, 2088-2195, 2253-2279, 2314-2328, 2433-2481; also the Meyl source for 0.5 mW in / 6 mW out.
  • Leads for the Investigator that QE and the deep reads name: the 2008 Chinese "Reliability and Safety" test of Wang's generator; Perendev / Mike Brady's later record; Sweet's MIT MSEE 1969 (Maurice's statement; MIT DSpace thesis index); the 1902 Daily Mail report on Figuera; Gollob Analytical Services and the Puharich measurements; the original Papp, Chambers, Shelton and Pogue patent records (claims, grant dates); FOIA records for the Glenn / DOE / GM / FBI episodes in Maurice's account; a comparison of V34.8 with V32.9 for changed figures; the original Hotsabi and Pearson mpg reports.
  • Records the Investigator can chase (named in the deep reads): Sweet's lab notes, Aesop Institute / Mark Goldes funding records, Bearden's 1987 "Phase Conjugate Vacuum Triode" paper, Michael Watson (intalek.com) coil data, the Ainslie / Panacea test documents, the 1968 Papp demonstration (TRW, NUWC observers), Moray witness records, Dr Roger Hastings's Newman measurement, and Beck's printed sources (Science News 30 Mar 1991, Longevity News Dec 1992, US 5,188,738, LA Times 10 Dec 1996).