Smith-Don: Physicist review (Step 2)

Reviewer: PhD physicist. Project: Don (Donald L.) Smith, resonant air-core coil and capacitor "free energy" devices. Date: 2026-10-08. Tags: [read] from the documents (I re-read the key passages myself, see "What I checked first-hand"); [web] with URL; [memory] with confidence.

Source gap (repeated from claims.md)

The project folder holds no measurement, test report, witness statement, instrument reading or patent grant. The only evidence that hardware existed is seven uncaptioned photographs (binder pp.3-9) and Smith's statements. The Lagace document (don-smith.pdf) is a secondary source: a later author's reconstruction, not Smith's words. Every number in both documents is an illustrative calculation, not data. The single quantitative performance statement by Smith that the Investigator found (the 1996 "suitcase") is Smith's own account written about five years after the event. No verdict below rests on a measurement of a Smith device, because none is on record in the files. Verdicts that are "refuted by calculation" rest on the documents' own formulas and on accepted laws, which is the one thing the documents do supply.

What I checked first-hand

• Deep reads (all of deep/): Smith's worked examples A and B and the f vs f-squared formulas, the L-1/L-2 turn argument, the impedance-zero passages, the Patent Office passage, and Lagace's step-by-step energy tables [read: deep/ files, which quote binder pp.36-75 and Lagace pp.1-38].
• Magazine printing, ExtraOrdinary Science Apr/May/Jun 1996, p.37 (page image rendered 2026-10-08): example B reads "Joules = 0.5 x .000,010 x 100,000^2 x CPS^2 = 500 x One Million squared ... more than 500 MegaJoules (Watts) plus" [read: page image p.7 of research/files/1996-ExtraOrdinary-Science-Electrical-Energy-System.pdf]. The Investigator's reading is confirmed.
• Mexican application MX NL02000035 A (OCR of the 15-page PDF, Spanish): abstract, description and claims 1-4 read [read: research/files/MXNL02000035A.pdf, pp.1-9].
• I did not view the lecture videos, the Scribd papers or the Lagace 2026 videos; they are the deciding records for the demonstrations and are weighed below only as reported by the Investigator.

Corrections to Step 1 (and to the Investigator)

1. Example B has three different printed answers, and the magazine's is also an arithmetic slip. The formula Smith prints (J = 0.5 C V^2 x f^2) with his own numbers gives 0.5 x 1e-5 x (1e5)^2 = 50,000, then x (1e6)^2 = 5 x 10^16. The binder (p.49) states "greater than 50 Mega K-Joules" = 5 x 10^10. The 1996 magazine states "500 x One Million squared" and "more than 500 MegaJoules" = 5 x 10^8. The magazine's first intermediate "500" does not follow from the printed factors (the correct intermediate is 50,000), so the magazine version contains a further error beyond the binder's. Step 1 (facts.json, claims.md claim 1) says example B's answer is "50 Mega K-Joules"; that is true of the binder only. The Investigator's lead already notes the third figure; I add that it is not merely a different rounding. [read: page image p.37 of the magazine; deep read p.49; arithmetic mine]
2. claims.md claim 3 says "ambient potential is 200,000 V" as if Smith's invention. Smith's 200,000 V (500,000 V at the poles) is of the order of the real ionosphere-to-ground potential of about 250 kV [web: https://arxiv.org/pdf/physics/0506077 and https://centaur.reading.ac.uk/7311, both list V_I about 250 kV], but Smith cites no instrument and attaches it to the wrong object (a source of power near the ground). This is a clarification, not an error in Step 1: it is the one number in his source list that is of the right size, and it should be stated fairly. The later "16 trillion volts" and "multi billions of volts" figures are not.
3. Step 1 treats the 1 F / 1 kW claim as Smith's. It reaches us only through Lagace, with no source or date (claims.md already says so). The Investigator found no first-hand Smith source for it either. Not a Step 1 error; I keep it flagged as Lagace-reported.
4. claims.md claim 6 is described as "Lagace's reconstruction" but the energy bookkeeping defect can be stated more exactly. The deep read finds three sources of error: a x10 slip in the initial energy (0.5 mJ, not 5 mJ), a factor-of-2 inconsistency in the per-pulse capture, and an unquantified outside source. I confirm all three and add one the deep reader left implicit: the mechanism Lagace actually draws (Figure D) is an ordinary inductive charge-recovery circuit (see claim 6 below), so the honest ceiling is 100% recovery of the 5 mJ, which is not nothing, and not over-unity.
5. Investigator, Leads 4 ("Patent outcome") and the Corrections section: confirmed. The located MX filing is an application ("SOLICITUD de PATENTE", published 20 May 2004), not a grant; claims.md and any later document should not repeat Kelly's "granted". I add that the claims it contains are modest in wording (claims 1-4, below) and do not mention over-unity; only claim 4 ("self-sufficient once started") touches it.
6. Lagace p.47 "peaked over 35 years ago" conflicts with the dated record (1994-2005). Investigator's note stands; it is a statement about Lagace's text only.

Mainstream physics in brief (so each verdict can be read against it)

• Energy stored in a capacitor: U = (1/2) C V^2 (joules). In an inductor: U = (1/2) L I^2. Multiplying U by a frequency f (cycles per second) gives a power in watts: it is the rate at which the full stored energy would be delivered if it were dumped and recharged f times a second. It is not an energy and not "energy per cycle squared".
• Resonance in a series L-C-R circuit: reactances cancel (X_L = X_C), so the impedance Z falls to R, the resistance, which is finite. Z does not become zero. The quality factor Q = omega_0 L / R is finite and is set by R. In a parallel L-C circuit the impedance at resonance is a maximum (about Q^2 R) [memory, high].
• Transformers: for tight coupling, volts per turn are equal on both windings (Faraday: V = N dPhi/dt), so voltage ratio = turns ratio, current ratio = inverse turns ratio, and power out cannot exceed power in. A Tesla coil is loosely coupled (k about 0.1-0.2); its large voltage gain comes from resonance, approximately sqrt(L2/L1) in the ideal lossless tuned case, and energy is conserved, with a well-designed coil moving over 85% of the primary energy to the secondary [web: https://richieburnett.co.uk/operation.html ; https://pupman.com/listarchives/2005/Jan/msg00389.html].
• Energy conservation is not suspended for open, non-linear or resonant systems. An open system exchanges energy with its surroundings, and that exchange must be measured and added to the budget. Resonance stores and returns energy; it does not create it.
• Skin effect: at radio frequency current is confined to a layer of depth delta = sqrt(2 rho / (omega mu)) (about 66 micrometres in copper at 1 MHz [memory, high]), so resistance rises, roughly as sqrt(f).

Claim 1. Delivered energy scales with the square of frequency: E = (1/2) C V^2 x f^2

a. Mathematics. The document is inconsistent with itself and with units. Smith's p.44 formula is (1/2) C E^2 x f (first power); p.47 and the worked examples use f^2. Example A (60 Hz, 120 V, 10 microfarad) follows the f^2 formula: 0.072 J x 3,600 = 259.2; the p.44 formula gives 4.32. Example B's printed answer follows the f formula (50,000 x 10^6 = 5 x 10^10), not f^2 (5 x 10^16); the 1996 magazine version prints a third figure (5 x 10^8, with a mistaken intermediate). Units: (1/2)CV^2 is joules; times f it is watts; times f^2 it is J/s^2, which has no standard meaning, yet Smith labels it "Joules [Watt Seconds]" and in B "[Watts]". p.47 defines C "in microfarads" but both examples use farads. p.44's "one henry and one farad equal one volt" is dimensionally false (H = V s/A, F = A s/V). The justification offered is an analogy: "E = (V x A) x CPS^2" is called "the electrical equivalent of E = mc^2" (p.38), and frequency is put in place of velocity in (1/2) m v^2 (p.43). There is no derivation. [read: deep read pp.38, 43-44, 47-49; magazine p.37]

b. Consistency with experiment. Where f^2 really does appear, it is for different quantities, which Smith does not invoke: the EMF induced in a pickup coil by a field B0 sin(omega t) is proportional to omega, so the power into a fixed resistive load goes as omega^2; the power radiated by a small dipole of fixed current goes as omega^2 (fixed dipole moment: omega^4) [memory, high]. These are rates and apply under stated conditions. The reactive volt-amperes of a capacitor on AC, Q = 2 pi f C V^2, is proportional to Smith's "(1/2)CV^2 x f", and reactive power is exactly the energy that sloshes back to the source each half cycle and does no net work. In example B the 10 microfarad capacitor at 1 MHz has reactance 1/(2 pi f C) = 0.0159 ohm; at 100 kV it would carry about 6.3 x 10^6 A, and resonating it needs L of about 2.5 nanohenry [computed]. No experiment is known to show delivered energy rising as f^2 for a fixed stored energy.

c. Tested? No test of the formula is on record in the files. The Investigator found that Smith claimed a 1996 briefcase demonstration (see claim 3); it is not a test of this formula. Not tested by others in any source found.

d. If true. Stored energy would have to be multiplied by f^2 per second with no source: a violation of energy conservation (a megahertz device would out-produce a 60 Hz one by 10^12 with the same stored energy). Circuit theory, Poynting's theorem and thermodynamics would all fail together.

Verdict: Internally inconsistent (two formulas, f and f^2; units mismatch; example B answers disagree across the binder, the magazine and the formula) and, in its f^2 form, Refuted by calculation: it is not an energy, and what the first-power form delivers is a power that is mostly reactive and returned to the source. The weak true core is that coupled power can rise with frequency under the conditions above (Established, but not what Smith claims).

Claim 2. At resonance impedance is zero: room-temperature superconductor; Ohm's law does not apply

a. Mathematics. Z = R + j(X_L - X_C). At series resonance Z = R, not zero; at parallel resonance Z is large. Smith equates ohms with "wasted energy" and amperes with "the rate of wasting" (pp.44, 71), but current flows in a lossless LC tank with no dissipation (dissipation is I^2 R). His "V.A.R. system" (volt-amperes reactive) reading treats high VAR as "near unity energy output"; in circuit theory Q-power averages to zero net work per cycle. His version of the power triangle (p.46) puts apparent power on the base, uses an energy ("V x A x Time") as the hypotenuse of a power triangle, and treats "Reactive" as a word instead of sin(phi). [read: deep read pp.39, 44-48, 50, 46] Smith's own physical picture ("at radio frequency the electrons ... encircle the conductor and are free of the conductor's resistance", p.48) is the skin effect, which raises resistance with frequency, the opposite of what he claims. Smith's own appendix (pp.91-94) converts 1 gauss to 1 x 10^-4 weber per square metre, in conflict with his "1 gauss = 100,000 volts" (pp.55, 68): the document contradicts itself. [read: deep read pp.91-94 per facts.json]

b. Experiment. Resonant-circuit measurements (Q-factors, Tesla-coil and RF-coil data) are among the best-documented in electronics; Q is finite and falls with R and radiation. Room-temperature superconductivity has not been observed in a resonant copper circuit; a true superconductor has zero DC resistance and expels field below T_c, neither of which a tuned copper coil does.

c. Tested? Every RF engineer who has measured a tank circuit has tested the finite-R prediction; the zero-impedance claim has never been supported. Smith offers no measurement, and none of the sources found.

d. If true. Resonant circuits would be lossless, ending Q-factor and skin-effect theory and implying a new superconducting state at room temperature in copper.

Verdict: Refuted by calculation (series resonance gives Z = R; a finite-R tank, which Smith's own photographed coils would be, cannot be a superconductor) and Internally inconsistent (a textbook conversion table bound into Smith's binder [corrected 2026-10-08 after QE: was "his own conversion table"; pp.91-94 are a standard table, not Smith's writing, see arbiter.md claim 2 and qe.md correction 2]; skin effect). Tested, not supported applies in the weak sense that the opposite (finite Q) is the experimental norm.

Claim 3. The extra energy comes from the Earth's and air's ambient electrical and magnetic background; conservation laws "excluded" for open non-linear systems; devices become "self sustaining once started"

a. Mathematics. The budget numbers do not close. The Earth is said to hold 17.6 x 10^18 EJ, "accumulated at 3.9 EJ per year for 4.5 billion years". The product of those two inputs is 3.9 x 4.5 x 10^9 = 1.76 x 10^10 EJ, so the printed total is 10^9 times larger than its own inputs give (printed 1.76 x 10^37 J; from inputs 1.76 x 10^28 J). The 1,600 gamma at 1,000 ft worked example (p.55) gives, by its own arithmetic, 1,000^2 x 1,600 = 1.6 x 10^9 gamma (not "1.6 trillion"), times 10 V = 1.6 x 10^10 V, not "16 trillion volts"; and the unit step is wrong: the gamma (1 nanotesla), gauss and oersted are units of magnetic flux density or field strength, not of voltage, so "1 gamma = 10 volts" has no meaning. Dividing a remote reading by distance-squared is also not how an aeromagnetic anomaly is continued down to the ground (a dipole field falls as 1/r^3, and the correction uses potential-field continuation, not an inverse-square multiplication) [memory, medium]. [read: deep read pp.54-55, 67-69] Smith's "Earth is an energy sink" and his "electrons pulled in from earth and air grounds" amount to an assertion of an unaccounted source; no energy balance is attempted. The real ambient figure he does cite (about 200 kV) is the right size for the ionosphere-to-ground potential.

b. Experiment. What the Earth's ambient electric field can supply has been measured. [web: https://arxiv.org/pdf/physics/0506077 ; https://centaur.reading.ac.uk/7311] give the ionospheric potential as about 250 kV, the fair-weather surface field as about 120-130 V per metre and the fair-weather current density as about 2.7 pA per square metre. Multiplying the two local figures, the power available from the fair-weather field is of the order J x E = 2.7 x 10^-12 x 120 = 3 x 10^-10 W per square metre; a 100 square metre collector would intercept about 3 x 10^-8 W [computed from the web values]. The whole global circuit carries about 1-2 kA at 250 kV [memory, medium], that is a few hundred megawatts for the entire planet, and the capacitor-like energy is of order 10^10-10^11 J [memory, medium]. The fair-weather ambient source is real, large in voltage and tiny in current: it is a poor energy source for the kilowatt to megawatt outputs Smith claims for breakfast-table devices. A device could in principle rest on the earth's field only as a high-impedance, microwatt-scale collector (as atmospheric electricity "lightning-rod" collectors, Kelvin's water-dropper and Tesla's own proposals were).

c. Tested? There are the following records, all weighed here as the sources describe them. - Smith's 1996 suitcase. Smith's own account (NuEnergy group post, about 2001; a single published source, written about five years after the event, with no second account found): "Input was 12 V DC at 300 milliamperes and the Output measured by an Electrical Engineer, picked by the audience, was a continuous 120 V at 80 Amperes as Documented in the VCR Tape" [web: https://energyevo.org/2015/01/18/updated-compilation-of-important-info-don-smith-shared-to-nuenergy-yahoo-group/]. That is 3.6 W in and 9.6 kW out by his figures, a ratio of about 2,700. The workshop is independently documented in the symposium programme (18 July 1996, Colorado Springs) [read: research/files 1996 ExtraOrdinary Science]. The measurement itself is documented only by Smith's description; the engineer is unnamed, the meters, the load and the duration are not given. The VCR tape is the deciding record. It has been listed by Rick Friedrich (an interested party, he sells a Smith-derived kit) but I have not viewed it. For scale: 9.6 kW at 120 V is 80 A, which needs a heavy conductor and a load such as a large heater; a 12 V, 7 Ah battery (about 300 kJ) would be drained by 9.6 kW in about half a minute if it were the source [computed]. So the claim stands or falls on whether the measured "continuous" figure was a true RMS real power measured over a stated time, and a clamp meter or reactive power could produce the number (see claim 1: Smith's "power" is reactive VA). That is a reading of what a flawed measurement could be, not evidence that it was. - 1994 demonstration, as described by Rick Friedrich (interested party). Friedrich says the flat-coil setup ran "to the point of the neon bulbs"; the four-secondary coil was not loaded; part of the demonstration failed, the audience reacted negatively; Friedrich attributes the failure to a 12 V battery on a 24 V module. Smith said then that a battery held its charge for a month (later "a year", "2 years") [web: https://www.youtube.com/watch?v=BHr3eDELyHk]. A battery that "stays charged" under no load is not evidence of anything: any battery does. - Bench reproduction (energeticforum, tswift, 24 Jul 2016): flyback at 30 kHz, battery, kickback diode, earth ground; about 1 W continuous; reports "self-charging" and flags a possible meter artefact himself [web: https://www.energeticforum.com/forum/energetic-forum-discussion/renewable-energy/13932-reproducing-the-don-smith-effect]. Not an independent test; one amateur report, uncontrolled. A flyback-plus-diode battery circuit is a standard energy-recovery arrangement and can make a battery meter read as if charging, especially with a high-frequency spiky waveform that cheap meters mis-average. - Planned validation that did not happen (absence of evidence): for Inventor's Weekend 2002 organiser Perreault announced Smith would bring "good test equipment to validate his claims"; Smith then had a stroke and did not present [web: https://energyevo.org/2016/08/05/nuenergy-and-freeenergy-missing-posts/]. Competing explanations for the lack of later test data, none decided by the record: illness (documented in that source); commercial or contractual secrecy (Smith's "contract obligations" appear in the Kelly part of the archive.org compilation); no working device. The 2001 recording was edited by the organiser, who says he cut about five minutes where Monroe Fischer questioned Smith, to protect a speaker weakened by a stroke; a critic website called the event report misleading [same page]. Both are the organiser's account (interested party: event organiser and video seller). The uncut tape and the critic report would decide. - Joel Lagace's 2026 videos (author of the secondary document): he states a working "modified Don Smith replication" in which "the input inverter stays completely cold" while the output "melts standard consumer power supplies" [web: https://www.youtube.com/watch?v=Tiji0JqY90o]. No input and output power figures were found in the part the Investigator read. A cold input stage and a hot output is what an unmetered input would look like if the input power is small, and equally what a poorly measured system looks like if energy comes from a source not being metered (for example the supply in the inverter's own housing, a charged bulk capacitor, or an unmeasured ground path). No independent measurement exists. No controlled, neutral test with calibrated input and output power measurement is on record. There is no record that anyone tested and failed either, because no neutral test appears to have been run.

d. If true. Energy conservation would fail for a class of circuits, or a new energy reservoir (a vacuum or earth-electron "background" thousands of times richer in current than anything measured) would have to exist. Atmospheric electricity, circuit theory and the first law would all need a new term. Mainstream physics already allows open systems to take energy from outside (solar cells, antennas, wind) provided the source is measured; Smith's claim is the specific one that the source is unmetered and large.

Verdict (split): - Weak form (a small amount of energy can be collected from the Earth's fair-weather field or from ambient radio fields): Established in principle (atmospheric-electricity collectors, crystal radios), at power levels of nanowatts to microwatts per square metre. The calculation above shows this cannot supply a kilowatt. - Strong form (kilowatt to megawatt outputs from ambient background, "self sustaining", conservation "excluded"): Never properly tested as a claim about Smith's device (no neutral controlled measurement on record; Smith's own suitcase claim is a single account and the video record has not been reviewed here; the 1994 demonstration is reported by an interested party as partly failing); and, as an energy budget with the ambient source stated in his own documents, Refuted by calculation, since the fair-weather source has about 10^-10 W per square metre of power density. - The printed Earth-energy figures are Internally inconsistent (10^9 error; unit conversion of gamma to volts). - "Conservation laws are excluded for open non-linear systems": Refuted (open systems are accounted for by flows across the boundary, not excluded from the law).

Claim 4. An air-core two-coil system gains energy through turn ratio and by "squaring" flux

a. Mathematics. The arithmetic is right (8,000 V / 4 turns = 2,000 V per turn; 2,400 V over 10 turns = 240 V per turn, 100 turns gives 24,000 V, pp.45, 76-77). The reading of it is not. With tight coupling the volts per turn are equal, so N2 turns give N2 times the volts per turn and N2 times less current, not a net power gain (power conserved). That "flux lines square with each unit of voltage" is wrong for flux: field between plates goes as V, stored energy as V^2 (p.43 picture is right for energy and wrong for flux). "Each time the Hz are doubled the effectiveness of induction is squared" (p.61) is not true: induced EMF is proportional to f. "Energy input in L-1 is divided by the number of turns" (p.72) divides energy where the p.61 text divides volts; the dominoes analogy counts the stored potential energy of every domino, which is real energy already in the system, not energy gained. In his own p.45 example, the number of L-2 turns, the frequency, the coupling and the input current are not given, so the output cannot be computed. The turn-ratio step-up described on pp.76-79 (the Builder's Guide, including the 247/MHz quarter-wave rule, about 246.5/MHz in free space [memory, high]) is ordinary Tesla-coil behaviour and consistent with mainstream physics. [read: deep read pp.43, 45, 61-63, 72; facts.json pp.76-79]

b. Experiment. Step-up of voltage by resonant coil pairs (Tesla coils, flyback and ignition coils, the "high-voltage module" and neon bulbs touched to the L-2 top) is routine and well measured: a 12 V battery into a laser-type module gives about 8 kV, then the coil pair raises it by resonance. Such sources have high voltage and small current, as the p.45 text itself concedes ("lots of volts but no amperes"). Output current at that voltage cannot raise the output power above input.

c. Tested? No power measurement of the coil-pair step in the files. The step-up itself is tested daily by hobbyists and engineers; the net-gain claim has not been tested in any source found, and the 1994 demonstration (above) reportedly lit neon bulbs but did not load the output.

d. If true. Power out would exceed power in for a passive linear transformer-like network, contradicting the reciprocity and Poynting-theorem accounting used throughout RF engineering.

Verdict (split): voltage step-up from turn ratio and resonance: Established. "Net gain" in energy or power, and "flux squaring" as its source: Refuted by calculation (V_out I_out cannot exceed V_in I_in plus supplied energy) and Internally inconsistent (energy divided in the speech, volts divided in the paper; "squared" flux). No test of Smith's actual outputs is on record, so the weak experimental claim "his device gave net gain" stays Never properly tested.

Claim 5. Dipole / capacitor-plate collection; "mirror" plate effect

Sources. MX NL02000035 A (Smith's own filing, 15 Nov 2002; published 20 May 2004; "Application Discontinuation" per Google Patents; family of one) and binder pp.10-19; Lagace p.2 for the mirror effect. I read the application: claim 1 is "an equipment for generating electric energy by magnetic means" comprising a potential generator (perpendicular to the magnetic flux) with capacitor plates perpendicular to it, connections collecting the voltage generated; claim 2 copper/dielectric/aluminium plates; claim 3 the generator is a coil, a plasma system, or plasma plus coils; claim 4 the equipment can become self-sufficient by feeding back part of its output. The description says the plates change the dipole's magnetic waves into "useful electrical energy (Coulombs)", that waves passing through the plates "do not degrade", that "each group produces an exact copy of the whole force and effect" and that the original source "is not reduced as in conventional transformers"; it says prototypes were "built, designed and functionally tested by the inventor" (4 ft plasma tube) with no results [read: MX NL02000035 A, description pp.2-6, claims p.9]. The Investigator's lead (application only, no grant) is confirmed. The prior-art statement says an international patent database search "did not reveal any prior Art" [read: p.2]; capacitive and near-field coupling to resonant antennas is long-established (Tesla's wireless-power patents, capacitive power transfer, antenna loading).

a. Mathematics. The application contains no equations or numbers beyond the 4 ft by 6 in plasma tube. The term "Heaviside current component" is not defined. "Coulombs" is named as the unit of useful energy; a coulomb is charge, not energy (the same unit slip as binder p.36). So there is nothing to check for consistency except the units, which fail.

b. Experiment. Mainstream facts [memory, high]: a resonant dipole has a reactive near field that stores energy, and a conducting plate placed in it picks up voltage by capacitive (electric-field) coupling; this is how capacitive wireless power transfer, antenna probes and Tesla's own demonstrations work. The source power draw rises when a load is attached (reflected impedance); the claim that "the original source is not reduced" is true only to the extent that the load takes negligible power, and fails at the point where significant power is collected. A small electric dipole radiates very little (radiation resistance goes as (length/wavelength)^2), so most of the input to such an antenna is not "radiated and wasted energy" waiting to be caught; it is mostly reactive energy returning to the source, which is why the claimed recoverable resource is smaller than the input. The "mirror effect" (a charged plate seems to charge a distant plate, which when grounded gives current) is ordinary electrostatic induction: a finite charge flows once, then stops; a steady current needs a steady source of work. [Lagace p.2; mechanism memory, high]

c. Tested? The application says prototypes were tested and lists no measurement. The Investigator found no neutral test. The photographs in binder pp.3-9 show hardware (coils, high-voltage modules, capacitors, a plasma globe, oscilloscope and meters) but without captions, values or readings. The 2005 TeslaTech talk is recorded (and the 1994-2001 talks), but the files do not contain a measurement of plate-collected power; the recordings have not been reviewed here. Competing explanation for the missing record, as in claim 3.

d. If true. If the plates produced more power than the dipole's source consumed, that would break energy conservation (Poynting's theorem). If the claim is only that capacitive coupling collects near-field energy, nothing changes: that is accepted physics and is already used.

Verdict (split): - Capacitor plates near a resonant coil or dipole pick up voltage and can deliver a small power to a load, with the source supplying that power: Established (capacitive and near-field wireless power transfer). - "Collecting energy that would otherwise be radiated and wasted" as a net generator, with the source "not reduced": Never properly tested (no neutral input-output measurement on record; the application is discontinued, not granted) and, if the claim is gain over the source's own input, Refuted by calculation (Poynting theorem; the radiated fraction of a small resonant dipole is small). - Self-sustaining feedback (application claim 4): Never properly tested (no measurement; a loop gain above 1 is the over-unity claim). - "Mirror" plate effect as a steady source: Refuted by calculation (electrostatic induction moves finite charge only); as a one-time charge transfer: Established.

Claim 6. Lagace's displacement-current capacitor scheme (1 F / 1 kW)

Source status. Secondary (Lagace, undated; no measurement). The 1 F / 1 kW statement is reported by Lagace with no source or date. For scale: 1 kW for 1 s takes 1 kJ, the whole energy of 1 F at 44.7 V; Lagace's own 4.75 kW from 500 J per second at 1 F and 100 V (950%) is a per-second claim from a 5 kJ capacitor [read: deep read]. Investigator lead: Utkin (c. 2012) described a Smith arrangement in which "the capacitor is charged by sparks and powerful displacement current is produced. The transformer with the ferromagnetic core is collecting this current" [corrected 2026-10-08 after QE: the PDF reads "is produced", not "is around"; it is Utkin's caption for a 2006 rendering titled "Don Smith Device" (credited to Zero Point Films), not Smith's words] [web: http://www.free-energy-info.tuks.nl/VladimirUtkin.pdf; research/files]. So the displacement-current idea is not unique to Lagace; it is attributed by Utkin to Smith, in a secondary compilation.

a. Mathematics. The printed balances do not close (deep read, confirmed by me against the formulas): (1) initial energy is 0.5 mJ for 10 microfarad at 10 V, not 5 mJ (the figures draw 10 mF, which gives 0.5 J); (2) the capture of "1.25 mJ ... the same amount that would normally be lost" conflicts with the 2.5 mJ loss printed on pp.7 and 12; (3) the total "2.5 + 2.5 = 5 mJ" on p.15 conflicts with the 1.25 per pulse on p.14 (recovery 50%, not 100%); (4) chapter 6 prints 2.5 mJ for each capacitor at 5 V where the formula gives 1.25, so after step 1 the books already total 6.25 mJ for 5 mJ in; the 10 mJ "output" counts the recharged capacitor A, which is the next cycle's starting state, as output; the ceiling is 5 mJ; (5) a capacitor at 0 V "holds all the energy of B" (p.10) contradicts (1/2) C V^2 = 0. The Bearden quotation ("no rule in physics that says you can't transfer a joule from one capacitor to another with nearly perfect efficiency") is attributed to the "Energy from the Vacuum" series with no locator [read: p.23]; it is true as stated for inductive transfer (below), and does not support gain.

b. Experiment / mainstream. (i) Connecting two equal capacitors directly always dissipates half the energy, whatever the resistance (the loss (1/2)(C/2)(delta V)^2 is independent of R); that part is correct. (ii) Putting an inductor in the path changes this: charge oscillates, and with a diode or switch to stop the reverse swing the energy is transferred to the second capacitor with efficiency limited only by resistance. This is the adiabatic or LC charge-transfer and buck/boost-converter principle (efficiencies of 90-95% are routine) [memory, high]. Figure D draws exactly such a loop (C1, transformer primary, C2, SW2). So part of Lagace's recovery is real: with an inductive path you can recover most of the "lost half" instead of heating a resistor. (iii) "The primary does not feel the secondary load because there is no closed current loop" is wrong for the circuit he draws: inside a capacitor the displacement current equals the conduction current in the leads (Ampere-Maxwell continuity), so the current in the primary is ordinary conduction current of a series loop, and a loaded secondary reflects back to the primary (the standard back-reaction; Lagace himself states it at p.17: "This coupling ... ensures energy conservation" and p.38: "the primary must supply all the energy transferred to the secondary") [memory, high; read: deep read pp.13, 17, 32, 38]. The primary and the secondary both draw on the same 5 mJ. (iv) No external energy source is entered into any arithmetic: the "environmental energy" mentioned in words is not quantified.

c. Tested? By Lagace: no measurement printed; the circuit figures look like screenshots from an (unnamed) circuit simulator, with one voltmeter reading (5.0 V). Lagace's 2026 videos claim a working "modified" replication; no input/output measurements were found by the Investigator (above); that is a single author's report. No independent replication. Lagace's claim is explicitly about his own reconstruction and cannot be held against Smith's records.

d. If true. If a transformer primary in a capacitor discharge path could be isolated from its secondary's load, a passive network would deliver more energy than the source stored. That breaks energy conservation and the Poynting-theorem accounting; it also contradicts the basic transformer relations (Lenz's law).

Verdict (split): - Recovering most of the half of the energy lost in a capacitor-to-capacitor transfer by routing it through an inductor and rectifier (up to 100%, in practice below it): Established (standard energy-recovery/switched-mode practice; not a new effect and not gain). - "Primary does not feel the secondary load" because of "no closed current loop": Refuted by calculation (displacement current continues conduction current in a series loop; the energy-conservation reason for back-reaction is given by the source itself). - Output above 100% of input (200%, 950%, the 1 F / 1 kW): Refuted by calculation from the document's own corrected arithmetic (ceiling 5 mJ per 5 mJ in), and Internally inconsistent as printed (x10 slip, factors of 2, A at 0 V "holding" energy). - Lagace's recent bench claims: Never properly tested (single-author video claims; no input/output figures seen).

Claim 7. Smith's account of suppression

Smith asserts (binder pp.37, 50-51) that a Morgan, Edison and Tesla meeting rejected "Free Energy", that the U.S. Patent Office and academia "bastardized" the idea, and that routing over-unity applications to the Office's "metering group" is "direct and absolute evidence" the Office knows over-unity exists. Smith's stake: he held a pending application (08/100,074, 4 February 1992) and a business. The cover letter adds that "selfish special interest has made sure the technology remains discredited" (p.21). These are his assertions; he gives no date, place, art unit, examiner or office action [read: deep read pp.37, 50-51].

a. Mathematics. Not a mathematical claim. The inference as written is a chain (routed to metering group, therefore accepts log meters, therefore accepts the square law, therefore knows over-unity exists) in which each step does not follow from the previous one (the log scale on AC instruments is a convenience for wide dynamic range, not evidence of non-linearity).

b. Experiment / record. I could not verify a joint Morgan-Edison-Tesla meeting; the standard account of Morgan's withdrawal concerns Tesla's Wardenclyffe project (1901-1906) [memory, medium; not web-checked, the check was beyond the budget]. On Patent Office practice: it is documented that applications claiming perpetual motion or over-unity are examined under the utility requirement as "inoperative", with special handling at examination (MPEP 2107.01, 707.07(g)), per secondary summaries [web: https://blueironip.com/ufaqs/how-does-the-uspto-handle-incredible-or-wholly-inoperative-inventions/ and https://en.wikipedia.org/wiki/Utility_(patentability_requirement); I could not confirm the official MPEP text]. That is a policy of treating such claims as lacking credible utility, not evidence that the Office believes they work, but it is evidence that over-unity applications are singled out, and Smith's reading (that routing means recognition) is his inference. The Office's own explanation would be the record in the application's file wrapper. The Investigator's lead stands: the file wrapper of 08/100,074 (likely unpublished, probably abandoned) has not been obtained. The Mexican application was discontinued; whether by IMPI action or abandonment is not known (IMPI file history not reached). Smith's account of the later period (Perreault's 2002 post: his "patents in Mexico have been approved") does not match the application record (discontinued; not a grant). Kelly repeats "granted". That discrepancy is real and not resolved by the files.

c. Tested? Not applicable as a physics claim. The record of how the claim was handled is thin: no examiner correspondence on file; the demonstrations were public (1994-2005) but the neutral measurement that would settle the physics was apparently never run (planned for 2002, cancelled by Smith's stroke). Absence of the later record has competing explanations (illness, contractual secrecy, no working device); the files do not decide among them.

d. If true. The Patent Office and academia would have to be knowingly suppressing a working device. That is a claim about institutions, not about physics, and no independent evidence for it is in the files other than the discontinued application and the unfinished validation.

Verdict: Not testable as stated (an account of motives and private meetings, with no dates or documents), with the following kept on record: the Office's treatment of over-unity applications is a documented practice; the file wrapper, the IMPI file and the uncut 2001 tape would shed light on Smith's own account.

Other material kept in the record (pertinent whatever the verdict)

• Smith's photographs (binder pp.3-9): seven pages of real apparatus: a flat spiral coil on a board with a "DANGER HIGH VOLTAGE" module; a plasma globe with a coil ring; a PVC tube winding with capacitor-like cases, with an oscilloscope, multimeter and two analogue meters nearby; copper straps and blue cased units. They show that Smith built and ran high-voltage RF hardware and had instruments; they show no readings. [read: facts.json from page images]
• The suitcase claim is the most specific numerical claim by Smith found and is worth a direct test if the VCR tape (1996, 2 h 39 min, Friedrich index) is reviewed: look for the load, the meter type (true RMS or clamp meter) and the duration. A continuous 9.6 kW needs about 10 kWh per hour from a source; a 12 V battery could not deliver this for more than seconds.
• Binder pp.76-79 (Air Core Induction Coil Builders Guide) is a valid, usable recipe for a resonant air-core step-up coil (quarter-wave wire length rule, volts per turn, neon-bulb probe for the voltage node): consistent with mainstream physics, makes no gain claim [read: facts.json]. The reader can build the step-up part and measure it.
• Historical passages (d'Arsonval list and European high-frequency medical coil work, p.38): the link between Tesla's work and European high-frequency medicine has a real basis; several dates in Smith's list are too early (d'Arsonval was born in 1851, so "as early as 1860" cannot be literal [memory, high]). Not verified on the web; low weight for any verdict.
• Related work (Investigator): Bearden's motionless electromagnetic generator line (CIPO 2172240, cited by Smith in a 9 Nov 2001 e-mail: "a close relative of Tom Bearden's Device. However I used a different Coupling System"); Utkin (above); Dalton, "Radiant Energy Induction Using RF Pulsed Plasma" (2014, not opened); Kelly Chapter 31 (secondary, favourable; "at least forty eight different devices"; a reader's note that an output choke's wire looks too thin for the stated currents). The label "over-unity" attached to Bearden's work is the mainstream's characterisation; Smith's closeness to it shows only that he regarded his device as in that family. None of them supplies a measurement either.
• Recordings and Smith originals not in the folder (Investigator): lectures 1994-2005, about 170 pages of Smith originals held by Rick Friedrich (interested party: sells a kit and a course), a "CAPACITOR FUNCTIONS" paper (14 Feb 2004). They are the deciding first-hand records and have not been reviewed here.

Summary of verdicts

Web checks done: 4 searches (global-circuit values; Tesla-coil gain and coupling; Patent Office practice). Not checked: the Morgan-Edison-Tesla meeting; the global-circuit current and capacitance (tagged [memory, medium]); gamma-to-field continuation.