Free Energy and D. A. Kelly’s Manual: Document, Devices, and Physics
Kelly’s 1980s manual is a real document collecting dozens of unconventional energy projects. What is historically verifiable, what is established physics, and where does the over-unity claim require an independent energy balance?
The phrase “free energy” can mean at least two very different things. In thermodynamics, Gibbs and Helmholtz energies are precisely defined quantities. In alternative-energy literature, the same phrase often refers to a device claimed to deliver more usable energy than the measured input, or to tap an environmental source not captured by the conventional description. Mixing these meanings creates more confusion than clarity.
D. A. Kelly’s The Manual of Free Energy Devices and Systems is a real document from the 1980s, not a later internet invention. Its first printing carries a 1986 copyright and ISBN 0-932298-59-5, followed by later printings. The manual gathers Tesla, T. Henry Moray, John Searl, Joseph Newman, homopolar or “N-machine” concepts, magnetic motors, pulsed systems, transformers, and many other projects.
The historical reality of the document is not the same thing as experimental confirmation of its conclusions. Kelly does not merely catalogue devices: he argues that practical systems with output exceeding input already exist. Each case therefore has to be separated into device description, patent, testimony, measurement, independent replication, and physical explanation.
This article neither rejects the manual because it is unconventional nor accepts it because it contains patents, diagrams, and real inventors. The stricter question is: what does the document actually establish, which phenomena are already part of standard physics, and what reproducible energy surplus would have to be shown before an over-unity claim could be considered confirmed?
First the document: what is Kelly’s manual?
Kelly’s manual has a traceable bibliographic history. Title pages in surviving scans identify The Manual of Free Energy Devices and Systems, D. A. Kelly, copyright 1986, ISBN 0-932298-59-5, with a first printing by Electrodyne Corporation in Clearwater, Florida. Later bibliographic records list a second printing in 1987 and a third printing in 1991 by Cadake Industries/Tri-State Press. The object under investigation is therefore not an anonymous PDF without provenance, but a concrete publication from the alternative-energy milieu of the late 1980s.
The manual presents itself as a survey of roughly fifty independent researchers and claims that various “free-energy” devices have already demonstrated practical viability. That is the author’s thesis, not neutral metadata. Statements about Kelly’s own patents, engineering experience, or the success of particular machines should therefore be treated as claims by the source until independently verified.
Its contents reveal both the breadth and the difficulty of the project. Historical figures such as Tesla sit beside standard electromagnetic machines, patented devices, experimental prototypes, and more speculative interpretations. All are placed under one “free energy” umbrella although the evidential quality is not uniform. The first audit step is therefore to disassemble the collection into checkable claims rather than judge the whole by its most famous name or its strangest assertion.
“Free energy” means several different things
In physics, free energy does not mean energy “from nothing.” IUPAC defines Gibbs energy as enthalpy minus temperature times entropy, and Helmholtz energy as internal energy minus the same entropic term. These are thermodynamic potentials used to describe how much energy is available for processes and work under specified conditions. “Free” concerns thermodynamic availability, not a costless unlimited source.
Kelly uses the phrase more broadly, often for devices alleged to show “over-unity”—useful output divided by the counted input greater than one. Such a result has two very different possible explanations. An external or stored energy source may have been omitted from the accounting; or the ordinary energy balance would genuinely have to fail. The first is routine engineering, the second would be an extraordinary discovery.
Hydroelectric power is a useful distinction. Water can drive turbines for long periods because the system is open: solar-driven weather, gravity, and geography continually replenish potential energy. The user may experience the resource as inexpensive or environmental, but the plant does not create energy. The same audit rule applies to every claimed free-energy machine: where is the system boundary, and have all energy flows across it been measured?
What is actually inside the manual?
The contents combine several device families: Tesla coils and wireless transmission, T. Henry Moray projects, Searl magnetic ideas, the Newman motor/generator, homopolar or “N-machine” systems, pulsed motors, variable-reluctance devices, battery-recharge systems, solid-state or transformer arrangements, and permanent-magnet motors. The manual is therefore a valuable map of a particular technical subculture.
A map is not a laboratory notebook. Chapters mix diagrams, patent excerpts, correspondence, short reports, prototype photographs, the compiler’s interpretations, and reported results from third parties. In some cases a primary patent or measurement report can be traced; in others the trail stops at testimony or a description without enough information to reconstruct a full energy balance.
Even its internal organization shows an editorial layer: promotional text speaks of fourteen types of projects, while surviving tables of contents extend through sixteen numbered sections. This is not a physics failure, but it is a reminder that the manual is not a standardized laboratory report. It should be read as a compilation of claims and documents, with each technical conclusion tested separately.
An open system is not a machine that creates energy
The first law of thermodynamics is nature’s accounting rule: a system’s energy change equals energy entering minus energy leaving through work and other transfers. If a device receives light, heat, fuel, mechanical torque, radio-frequency power, battery chemistry, or energy from a pre-charged capacitor, that belongs in the input balance even if no obvious power cable is visible.
A device can therefore legitimately show a coefficient of performance above one, or an output larger than a narrowly defined electrical input, if it draws additional energy from its environment. A heat pump is the familiar example: it transfers more heat than the electrical energy it consumes because electricity is not the only energy flow. This does not violate conservation and is not evidence of energy creation.
The decisive issue in over-unity testing is thus the system boundary. Are we measuring control electronics but not mechanical drive? Is a battery, capacitor, or magnetic system discharging? Is heat entering from the room? Are input and output integrated over different time windows? Until those questions are closed, a number above 100 percent is not enough for a revolutionary conclusion.
Maxwell’s equations are not a loophole in the energy balance
Some alternative models speak as though electromagnetic fields contain energy that ordinary electrical engineering fails to account for. In fact, energy flow is explicit in electromagnetic theory. Poynting’s theorem connects the change in field energy, the flow of electromagnetic energy through a system boundary, and work done on charges. MIT’s standard electromagnetics course teaches it precisely as an energy-conservation relation.
This does not make all electromagnetic systems intuitive. Resonance, back-EMF spikes, phase shifts, near fields, reactive power, and transients can produce spectacular voltages and currents. Tesla genuinely demonstrated resonant and wireless energy transfer; modern IEEE terminology still defines wireless power as energy transmitted from a source to a load through electromagnetic coupling or radiation.
It is therefore legitimate to investigate unusual field topologies or improved transmission. What is not sufficient is a high voltage, a resonant peak, or a striking magnetic effect. To establish a surplus, the integral of all energy flows across the chosen system boundary—after accounting for stored energy—must still show stable net output above input.
Why over-unity measurements become difficult very quickly
The experiments most dangerous to naive metering are exactly those common in alternative motors: short pulses, high voltages, high crest factors, strongly nonsinusoidal currents, inductive flyback, and phase shifts. NIST has specifically studied the influence of nonsinusoidal waveforms on voltmeters, ammeters, and phasemeters and the performance testing of true-RMS instruments.
Simply multiplying average voltage by average current can be wrong for such signals. Real power requires time-aligned voltage-current measurement or equivalent integration, sufficient bandwidth, proper phase handling, a known load, calibrated sensors, and accounting for energy in all storage elements. With pulsed signals, an instrument limitation can be larger than the claimed surplus.
A result also needs a measurement uncertainty. A claim of 103% is not evidence of excess if total uncertainty is ±10%. A strong test defines the protocol in advance, calibrates instruments, uses independent measurement methods, cross-checks by swapping channels where possible, and allows another group to reproduce the result. For small anomalies, metrology is not bureaucracy; it is the evidence itself.
Joseph Newman: a rare claim that received a metrology-grade test
Newman’s motor is especially useful for this audit because Kelly devotes a section to it while a detailed, court-ordered investigation by the U.S. National Bureau of Standards (NBS, now NIST) also exists. The dispute arose around Newman’s patent application and the claim that the machine delivered more energy than it received.
The history is not one-sided. Before the NBS tests, a special master had viewed part of the evidence more favorably to Newman. Newman later disputed aspects of the test conditions, especially grounding and the interpretation of output. Preserving that disagreement matters; a source audit should not erase documented objections simply because the later result was negative.
The 1986 NBS test was nevertheless technically specific: input and output were measured by multiple methods, pulsed waveforms and calibrations were addressed, and dozens of measurements were performed. The report concluded that under every condition tested, input power exceeded output power; the court summarized that no measurement showed efficiency above 100%. This does not prove that no conceivable device could ever tap an unknown source. It does mean that the Newman model in this independent test did not confirm the over-unity claim for which it became important in free-energy literature.
A patent is not laboratory validation
Kelly frequently emphasizes patents. That is understandable: a patent is a traceable legal document with descriptions, drawings, claims, and dates, and is far better evidence of what an inventor proposed than an anonymous post. But patent status answers different questions from independent physical validation. Patent offices do not routinely perform long-duration energy-balance tests on every patented device.
The USPTO explicitly explains that a working model is ordinarily not required, although one may be requested for alleged perpetual-motion devices. This is an illuminating boundary: the patent system requires enabling disclosure and other legal criteria, and may demand extra demonstration when a claim directly conflicts with established physical expectations. Yet a patent number is not itself a metrology certificate for an energy surplus.
The proper use of a patent is therefore genealogical and technical: it establishes what was claimed, when, with what construction, and how the inventor framed the idea. To establish over-unity, one still needs an independent measurement protocol, data, error bounds, and replication. Patent evidence and experimental confirmation are distinct evidential categories.
Tesla, magnets, and homopolar generators: real physics, stronger conclusions
Tesla gives the manual considerable rhetorical force, but precision matters most here. Tesla unquestionably belongs to the history of high-frequency systems, resonance, and wireless energy transfer. The Smithsonian preserves his material on “the transmission of electric energy without wires,” while modern IEEE sources describe wireless power as transporting energy from a source to a load. This is real technology—but it is transmission, not energy creation.
Likewise, a permanent magnet is not bottomless fuel. Magnetic systems can exert forces and store field energy, but a cyclic motor must restore its configuration over a full cycle; an attractive force on one part of the path does not by itself guarantee positive net work without corresponding costs elsewhere. Homopolar generators and Faraday-disc geometry are also genuine electromagnetic physics, not automatic proof that more electrical energy can be extracted indefinitely than the mechanical drive supplies.
The weakest argument on both sides is therefore symmetrical: “the machine looks strange, so it is impossible,” or “it uses a real strange effect, so over-unity is confirmed.” The useful audit lies between them: the phenomenon can be real, the explanation partly wrong, the device interesting, and the measured energy balance still below one.
Vacuum energy, zero-point energy, and tachyons
Quantum physics contains effects that popular literature often links to “vacuum energy.” The Casimir effect is a serious quantum phenomenon, but even its interpretation requires care: R. L. Jaffe showed that Casimir forces can be formulated without treating them as direct proof of a physically extractable reservoir of zero-point energy. The existence of quantum-vacuum effects is therefore not, by itself, a generator design.
There is also literature that directly explores zero-point-energy extraction. The review by Moddel and Dmitriyeva is more open to such proposals than a standard textbook and is useful for that reason: it still concludes that proposed methods have not been reliably demonstrated, while several classes face equilibrium or thermodynamic obstacles. An interesting hypothesis remains a hypothesis until reproducible net energy output is shown.
The manual offers another concrete audit point. Gerald Feinberg’s 1967 paper was titled Possibility of Faster-Than-Light Particles and constructed a theoretical framework for hypothetical particles later called tachyons. When a later source upgrades such work into a claim that tachyons had been “proved,” it raises the certainty beyond what the original paper established. The lesson generalizes: theoretical possibility, a patent, or a real unusual phenomenon is not the same as experimental confirmation of excess energy.
| Element | What Kelly or the tradition presents | Independent control | Evidence status |
|---|---|---|---|
| The manual | a compilation of many free-energy projects; practical viability claimed | traceable 1986/87/91 publication history | the document is real; its central physics claim needs separate validation |
| Tesla / wireless power | resonance and wireless energy inside a free-energy narrative | Smithsonian and IEEE document transfer from source to load | real phenomenon; not evidence of energy creation |
| Newman motor | claimed output greater than input | NBS 1986: input exceeded output under all tested conditions | over-unity not confirmed in the independent test |
| Patents | legal/technical documents supporting device history | USPTO: models usually unnecessary; may be requested for perpetual-motion claims | evidence of filing/disclosure, not by itself proof of efficiency |
| Nonsinusoidal/pulsed measurement | high-voltage and pulsed systems | NIST: waveform shape affects meter response and true-RMS requirements | poor metrology can create an apparent surplus |
| Vacuum / Casimir | possible environmental or vacuum source | Casimir physics is real; net vacuum-energy extraction has not been reliably demonstrated | real quantum physics ≠ demonstrated vacuum power plant |
| Tachyons | manual treats Feinberg’s work as validation of the concept | Feinberg 1967 develops a theoretical possibility | theoretical possibility is not experimental proof |
What is historically valuable in the manual—and what would have to be shown today
Kelly’s manual is worth preserving precisely because it captures a technical culture that often operated outside universities and major laboratories: individual inventors, alternative societies, patents, workshop prototypes, older electrical literature, and genuine dissatisfaction with accepted limits. Some machines employ entirely real phenomena; others combine real components with interpretations that the measurements do not establish.
A modern proof standard would be straightforward in principle. A device should run long enough to exclude batteries, capacitors, thermal storage, and transients; its system boundary should include electrical, mechanical, thermal, and radiative flows; instruments should be calibrated and fast enough; the uncertainty should be much smaller than the claimed excess; and independent groups should reproduce the result. If a stable net surplus survived that process, it would be scientifically important and should not be dismissed merely because of the label “free energy.”
On the currently checkable record, however, Kelly’s manual does not reach that combined evidential threshold. The strongest conclusion is therefore neither “it is all fraud” nor “the manual proves new physics.” It is more precise to say: the document is real and historically valuable as a catalogue of unconventional energy experiments, while its central claim of validated over-unity systems remains unproven. The most useful question for every next device is simple: where is the independent, complete, reproducible energy balance?
Sources and further reading
- D. A. Kelly, *The Manual of Free Energy Devices and Systems* (first printing 1986; later printings 1987 and 1991) — primary source for the manual’s own claims, contents, device descriptions, and framing of “over-unity.”
- Open Library — *The Manual of Free Energy Devices and Systems* by D. A. Kelly — independent bibliographic record for the 1987 edition and publication metadata.
- IUPAC Gold Book — “Gibbs energy” — standard definition distinguishing thermodynamic free energy from popular “free energy” usage.
- IUPAC Gold Book — “Helmholtz energy” — standard definition of Helmholtz energy and the historical term “free energy.”
- OpenStax, *Physics*, “First Law of Thermodynamics: Thermal Energy and Work” — conservation-of-energy accounting for thermodynamic systems.
- OpenStax, *College Physics 2e*, “Introduction to the Second Law of Thermodynamics: Heat Engines and Their Efficiency” — cyclic engines, efficiency, and second-law constraints.
- MIT OpenCourseWare, *Electromagnetics and Applications*, Lecture 4 / course notes — Poynting theorem and electromagnetic energy conservation.
- Richard P. Feynman, Robert B. Leighton, and Matthew Sands, *The Feynman Lectures on Physics*, Vol. II, ch. 27 — electromagnetic field energy and conservation.
- NIST — Nile M. Oldham and Thomas L. Nelson (1991), “Influence of Nonsinusoidal Waveforms on Voltmeters, Ammeters, and Phasemeters” — measurement problems with non-sinusoidal signals.
- NIST / NBS IR 88-3736 — R. S. Turgel, O. B. Laug, T. E. Leedy, *Electrical Performance Tests for True-RMS Voltmeters* (1988) — instrument response and true-RMS testing.
- NIST Statistical Engineering Division — “Measurement Uncertainty” — why quantitative claims require an uncertainty budget rather than a single displayed number.
- Robert E. Hebner, Gerard N. Stenbakken, and David L. Hillhouse, NBSIR 86-3405, *Report of Tests on Joseph Newman’s Device* (1986) — court-requested independent test concluding input exceeded output under all tested conditions.
- *Newman v. Quigg*, 681 F. Supp. 16 (D.D.C. 1988) — court record summarizing the special-master history, Newman’s objections, NBS protocol, and test results.
- USPTO, “Applying for Patents,” section on models, exhibits, and specimens — a working model is normally not required, but may be requested for alleged perpetual-motion devices.
- IEEE Technology Navigator — “Wireless power transfer” — contemporary engineering definition of wireless transmission from a source to a load and major loss mechanisms.
- Smithsonian Institution — “Nikola Tesla papers” — includes Tesla’s 1904 material on the transmission of electric energy without wires and Wardenclyffe.
- R. L. Jaffe, “Casimir Effect and the Quantum Vacuum,” *Physical Review D* 72 (2005), 021301(R) — cautions that Casimir forces need not be interpreted as direct evidence for extractable zero-point-energy reservoirs.
- Garret Moddel and Olga Dmitriyeva, “Extraction of Zero-Point Energy from the Vacuum,” *Atoms* 7.2 (2019), 51 — review of proposed extraction mechanisms and their experimental/thermodynamic limitations.
- Gerald Feinberg, “Possibility of Faster-Than-Light Particles,” *Physical Review* 159 (1967), 1089–1105 — original theoretical tachyon paper; a possibility framework, not experimental discovery.
- American Physical Society / Physics — overview of precision Casimir-effect work and vacuum-fluctuation forces — evidence that the effect is a real quantum phenomenon while not establishing a usable over-unity energy source.