Mathematician review: Malove-Eugene (Mallove, "Tesla and the Aether", Infinite Energy 48, 2003)

Reviewer role: PhD mathematician (Step 2). I check derivations, notation, units and hidden assumptions. Source tags: [read] from the project documents, [web] from an Internet source (URL given), [memory] recalled, with confidence.

Note on name: the folder is "Malove"; the author is Eugene F. Mallove. One 4-page article, full text layer, no source gap in the article. The experimental sources behind it (Correa monographs AS2-13, -14, -16; paid downloads) were not obtained, so conclusions about the Tesla-coil data rest on what the article reports. [read: claims.md, facts.json]

0. Headline for the mathematics

The article contains almost no mathematics: one number (a "10/1 power ratio"), one speed factor ("at least 1,000 times"), one time figure ("80 microseconds"), and verbal definitions ("massfree", "massbound", "cold electricity"). So the check is mostly on (i) the arithmetic and definitions behind the numbers, (ii) whether the quantities are measurable as defined, and (iii) hidden assumptions. The Investigator's findings are valuable here because they supply the numbers behind the 10/1 figure that the article omits.

1. Background a reader needs (plain language)

• Power P is energy per unit time, in watts (W). For a steady DC source, P = V x I (volts times amps). For a rapidly pulsing signal, the correct average power is the time average of the product, P_avg = (1/T) integral of v(t) i(t) dt over one period T. This is NOT in general (average of v) x (average of i). Example: a pulse train with v = 100 V and i = 1 A for 10% of each cycle and zero otherwise has P_avg = 0.1 x 100 W = 10 W, but (average v) x (average i) = 10 V x 0.1 A = 1 W. A factor of 10 can come purely from the averaging mistake. Whether Mallove or the Correas made it is unknown; the point is that "power ratio" must be defined.
• Mechanical power of a rotating shaft is P = torque x angular speed (tau x omega, omega in radians per second). It is not a direct electrical measurement and needs a dynamometer (a brake that measures torque).
• Energy balance (conservation of energy): average output energy over a cycle cannot exceed input energy plus any stored energy released. "Output/input = 10" over a complete steady cycle with no hidden store would need a new energy source. That is the physicist's question; mine is whether the ratio, as defined, is computed correctly.
• Speed from time: a signal speed is distance divided by delay, v = d/t. The smallest delay you can claim is limited by the time resolution of the measuring instrument.

2. Claim-by-claim review

Claim 1. PAGD tubes give about 10 times more electrical power out than in (Mallove's observations, 2000-2002)

a. Is the mathematics internally consistent? The article gives only "approximately 10/1 power ratio" with no definition. [read: p.3 of teslaandtheaether.pdf] The Investigator found the number behind it: Mallove's letter of June 14, 2001 says he observed "an input DC power to the PAGD reactor of 50 watts, with an output motor power (mechanical shaft power) of approximately 500 watts," measured with "conventional electric meters, a digital storage oscilloscope, and a computer data acquisition system"; he adds "no one else from Infinite Energy was with me". [read: research/files/eugenemallove.com-June_14_2001.html; I re-read the passage; also web: https://www.eugenemallove.com/Letters/June_14_2001.html]

Checks: 1. 500 W / 50 W = 10, so the arithmetic of the ratio is right (dimensionless). 2. The ratio compares different quantities: DC electrical input to the tube against mechanical shaft output of a motor. The article calls it a "power ratio" of "electric power" claims; the letter shows it is not electrical-out over electrical-in. Defining it consistently needs the motor's electrical input and efficiency. A small motor typically converts at best about 70-90% of its electrical input to shaft power [memory, high confidence], so 500 W of shaft power requires at least about 550-700 W of electrical power delivered to the motor. Where that electrical power came from (the 50 W tube, a battery in the inverter, or a separate supply) is the whole question, and the letter does not say. This is an unstated boundary of the system, i.e. a hidden assumption. 3. How the 500 W shaft figure was obtained is not stated (Investigator's note agrees). If it was computed from torque and speed estimates rather than a dynamometer, errors of several-fold are possible. The Investigator reports that the later Z-Prize rules quoted in the correspondence excluded torque-based claims under 10x because torque-to-wattage calculations "may be inaccurate by an order of magnitude". [web: https://www.eugenemallove.com/emwords_SEAS.html, via sites.md] That is the rules' author's statement, not a measurement of PAGD, but it is the same size as the claimed effect. 4. Pulsed waveform: the tube "self-oscillates in a particular negative resistance region" (Mallove). [read: p.3] Pulsed power must be averaged as the time integral of v x i, as in section 1. A DC meter on a supply carrying large current pulses reads the average current correctly only if its response is linear and fast enough; the product of average voltage and average current is wrong if v and i are correlated in time. The letter mentions a digital storage oscilloscope, which is the right tool, but gives no waveform data. Unknown whether the correct product was integrated.

Patent data (Investigator, which I checked). US 5,449,989 reports "breakeven efficiencies" by comparing energy lost by a "drive" battery pack with energy gained by a "charge" battery pack. I read the saved patent text. [read: research/files/gp-US5449989A.html] - It states a surplus factor of 3.9x from 0.962 kWh gained and 0.246 kWh lost in 18 minutes. 0.962/0.246 = 3.91; the arithmetic checks. - It states 485.2% +/- 18% (n=64), "with projected 48.3Wh drive pack loss and 221.7Wh charge pack gain". But 221.7/48.3 = 4.59, i.e. 459%, not 485.2%. The patent says this "matches rather closely the observed 483% ... 37.7Wh loss as well as the 182.2 kWh gain" (182.2/37.7 = 4.83; "kWh" appears to be a typo for Wh). So the statistics quoted do not exactly reproduce from the two numbers given. There may be a scale factor (the text elsewhere mentions a 29/46 = 0.63 pack scaling) that is not shown. I flag this as an unexplained internal discrepancy in the patent, not as proof of error. - The patent itself warns that estimating battery charge from open-circuit voltage "overestimates net power consumed and produced by up to 10 fold" and that the lead-acid gel cells have "grossly diminished (50-60% of manufacturer's theoretical estimate) effective charge capacity". [read: gp-US5449989A.html] This is the central hidden assumption: the method infers stored energy from battery state of charge (voltage, "% of total capacity"), not from measured energy in and out. State-of-charge inference during pulsed charging is known to be unreliable because terminal voltage after charging includes a transient "surface charge" and depends on rest time [memory, medium confidence]. The runs include 15-minute relaxation periods (patent text), a choice that affects the result. A 10-fold uncertainty in the method is the same size as the claimed gain (3.9x to 9x). The inventors report this honestly; it also means the patent's figures do not establish the claim on their own terms.

Verdict for 1 (maths): Not internally inconsistent in arithmetic, but the "10/1 ratio" is undefined as published; the figure behind it compares shaft power with DC input, and the patent's 4-9x figures depend on a method whose stated error is 10x.

b. Consistent with established experiments? Conservation of energy (first law) has been verified to extreme precision in every closed-system test; a ratio of 10 at steady state with no external store would contradict it. A transient release of stored energy (battery, capacitance, inductance) can look like gain. The arithmetic of the claim does not itself show a violation, because the energy boundary is not defined. I defer the physics to the physicist.

c. Tested? Replicated? Tested by the inventors (patent), by Mallove alone (June 2001 letter, Aug 2000 and Mar 2001), and viewed by others reached only through Correa-run sites: Mike Carrell (an open-box demonstration in Dec 1999, wrote IE 8(3):10-15), Axelrad (an Aether Motor demo), Aspden, Soudak, Tiller. [web: research notes 2026-10-04-aetherometry.com.md, -eugenemallove.com.md] Not replicated by an independent party at an independent site. The Investigator records the 2002 SEAS/Z-Prize episode: the Correas declined off-site testing and NASA/Navy participants and required testing "at our premises" tied to $15-16M funding; a meeting took place on Nov 16, 2002 in Concord, Ontario; no validation report found. [web: https://www.eugenemallove.com/emwords_SEAS.html] Competing explanations for this record (inventor's trade-secret and funding conditions; distrust of state agencies; device not showable under independent conditions; lack of funds, as Axelrad wrote in April 2002 the Correas were shutting down the laboratory for lack of funds) cannot be decided from the record; the decisive documents would be SEAS/Loder evaluation notes, the "Berlin video" and the later J Aetherom Res PAGD papers. This is an absence-of-evidence anomaly, kept in the record, not a refutation. Also Mallove's own caveat: "reproduction by others is the only way in which your ... work will ultimately be accepted." [web, June 14, 2001 letter]

Interested parties: Mallove edited Infinite Energy, promoted the Correas and was their business intermediary; the Correas hold the patents and run the sites that carry the witnesses' statements. Their descriptions are reported as theirs. Axelrad states he is "not trained in this field".

d. If true, what must change? The first law would need an energy source beyond the system boundary (zero-point or aether energy, per Mallove), or the accounting of stored energy would be shown to be wrong. A correct, boundary-closed accounting would be the first task.

Verdict: Never properly tested (a testable claim; the only measurements are by the inventors and the sympathetic editor, on-site, with an ill-defined ratio; an independent test was proposed in 2002 and not held; no neutral controlled test on record; not dismissed without testing, rather the tests that were possible were not carried out). The patent's battery-accounting method carries a stated 10x uncertainty, so the "Tested, not supported" verdict is not warranted either.

Claim 2. Tesla coils emit a "massfree" radiation that is not electromagnetic and not from ionic emissions; pulsed input converts local aether energy to electric form

a. Internal consistency. No equations, numbers or definitions are given in the article. [read: p.3] Terms to check: - "Massfree" and "massbound" are used as adjectives on electric fields and radiation. Charge is a conserved quantity attached to carriers; "mass" is a property of matter. A field has no mass in the sense used; electromagnetic waves are already massless in the standard view (photon). So "massfree non-electromagnetic" needs a definition of what the quantity is, its units and how it couples to the instruments. None is given. Without it, the claim cannot be checked for consistency. The Investigator found the Correa monographs list AS2-13, -14, -16 titles that contain "frequency functions" and a "resultant characteristic electromagnetic frequency", suggesting formulas exist, but they are paid downloads and were not obtained. [web: https://www.aetherometry.com/correa_publications.php] I cannot check them. - "Longitudinal waves": In a vacuum or ordinary gas without free charges, Maxwell's equations allow only transverse waves (div E = 0 forces E to be perpendicular to the wave direction for a plane wave). Longitudinal electric waves exist in plasmas and in guided structures (for example waves on a wire, which are TEM and still transverse in the field components), and sound is longitudinal. So the word is not inconsistent per se; the claim of a non-EM longitudinal wave in vacuum is a departure that requires new equations, which the article does not give. - "Ambipolar electric, longitudinal radiation" (Correa quote, p.4) is undefined. - The chain "triggered the ratemeter via a plate antenna at substantial distances" shows a signal, not its nature. Reasoning from "not blocked by X, not detected by Y" to "massfree non-EM" assumes the list of known detection channels (EM pickup, ion/charged-particle effects, ground and mains coupling, X-rays from spark gaps, electromagnetic pulses entering GM tube electronics) was exhausted. That is a hidden assumption: an exclusion argument is only as strong as its list of alternatives. Spark-excited Tesla coils are strong broadband EM emitters, and Geiger counters and ratemeters are well known to respond to EM interference [memory, medium-high confidence]. No data are given to show this was excluded.

b. Consistent with experiment? The standard analysis of Tesla coils (coupled resonant circuits with radiated EM and ionic/corona effects) is verified in radio engineering. The Correa claim needs measured data, which are not in the article. No count rates, distances or backgrounds are given. [read: p.3] Fine for the reader to know: no numerical value appears in the whole Tesla-coil passage.

c. Tested/replicated? By the Correas only (monographs; in-house, per the text not journal-peer-reviewed). No independent replication found by the Investigator. Mallove did not run these experiments ("the Correas were able to..."). [read: p.3] Ref 5 numbering (S2-15 vs S2-16): the Investigator's check of the Correa list shows the Tesla-coil part 3 is AS2-16 and AS2-15 is on magnetism, so the article's ref 5 as printed is probably a citation slip; this resolves a small Step 1 uncertainty. [web: aetherometry.com/correa_publications.php]

d. If true: Maxwell's equations would be incomplete, or the Tesla-coil near/far-field behaviour would contain a non-EM channel; relativity's treatment of the aether would need revisiting; the first law would need an external source if energy is "converted" from aether.

Verdict: Not testable as stated in its present form (undefined terms, no numbers), with the experimental part Never properly tested by neutral parties (no independent replication on record; the monograph data were not available to me). Only the vocabulary can be assessed as it stands.

Claim 3. Two forms of electricity: ordinary "massbound" and "massfree" ("cold electricity", Mallove's name), flowing in and around wires and as Tesla waves through gas and vacuum

a. "Cold electricity" is Mallove's suggested label; "massfree/massbound" are Correa terms. [read: p.3] No definition in terms of charge, current or field, no equation. "Cold" has no measurement tied to it in the article (thermometers are only listed as instruments). So the statement is a naming, not a quantitative claim. In standard electrodynamics, current I is charge per unit time; a "massfree" current would have to be charge flow without carriers of mass, or a field effect with no charge flow (displacement current already covers the latter). That is a distinction without a stated difference in measurable output. b. No experiment is cited that distinguishes it from displacement current, skin-effect surface currents, or capacitive coupling, all of which are standard and flow "in and around wires". c. Same as claim 2. d. The classification of electric phenomena would need a new category.

Verdict: Not testable as stated (an interpretation/naming with no measurable difference given).

Claim 4. The aether is real and was dropped from physics by fashion (relativity), not experiment; Tesla's global longitudinal-wave power was therefore possible

a. The article quotes Tesla (1891) that "electricity and ether phenomena are identical" and attributes the aether's removal to Einstein. [read: p.1-3] The argument is historical and rhetorical; there is no formal derivation. Two checks: - A "dynamic, non-luminiferous aether" (Reich, per the Correas) is introduced without equations. Whether it is Lorentz-covariant is not discussed. The standard objection to a luminiferous aether is that it must be undetectable by any velocity-dependent test to the accuracy of Michelson-Morley type experiments; the article does not engage this. Hidden assumption: that a "dynamic non-EM aether" escapes those tests because it is not the luminiferous one. That needs a stated mechanism. - Tesla's global scheme: "longitudinal waves ... in the resonating cavity surrounding the body of the Earth". The Earth-ionosphere cavity does resonate; Schumann resonances are at about 7.8 Hz (and 14, 20 Hz...) from the cavity circumference [memory, high confidence]. A check: circumference of Earth about 40,000 km, c = 3 x 10^8 m/s, so c/circumference about 7.5 Hz, close to 7.8 Hz (the small difference arising from the ionosphere's finite conductivity). So the cavity is real; it hosts ordinary transverse-field EM modes, and the energy is dissipated, not usable as a lossless power bus. Whether power at "millions of receivers" is deliverable depends on losses (Q factor of Schumann modes is only about 4-6 [memory, medium confidence]) and is a quantitative question the article does not touch. I leave the conclusion on feasibility to the physicist; the point for the reader is that the cavity's existence does not itself supply the numbers. - Tesla "illuminated electric bulbs at good distances" is unsourced. [read: p.3] Lighting a bulb at distance is consistent with near-field inductive and capacitive coupling, which falls off steeply with distance. The unstated distance and power make the claim non-quantitative.

b. Michelson-Morley and later experiments, and relativity's successes, are the experiments to confront (physicist's domain). The article cites dark energy and the cosmological constant as the mainstream's own "patching". [read: p.1-2] That is an argument by analogy, not a result. c. The article reports no test of the specific aether claim. d. Relativity's role would need re-examination; ZPE/aether energy extraction would conflict with energy conservation unless a source is specified.

Verdict (split): the historical statement that Tesla and Maxwell-era physicists discussed an aether is established history (not rated by the scale beyond "Historically superseded but instructive"); the physical claim that a dynamic aether supplies usable power is Not testable as stated for lack of equations or measurable predictions. The strong form (Tesla's global scheme works as Mallove describes) is Never properly tested in the article's sense: no quantitative test is reported.

Claim 5. A foot acupuncture point drove visual-cortex activation at least 1,000 times faster than known nerve speed (So, Jones, Kidney, Saito, PNAS 1998, as described by Mallove); Galvani detected massless "Tesla radiation"

a. Mathematics. Here the article gives enough numbers to test internal consistency. Quantities as printed: speed "at least 1,000 times any known nerve transmission speed"; fMRI "minimum time localization of measurement is about 80-microseconds"; Jones said the "actual transmission speed to the brain might be faster than even this astounding 1,000-fold figure". [read: p.3-4, deep read section 6]

Worked check. Take the path from foot to visual cortex as d about 1.2 m (an adult; assumption, [memory]). Fast myelinated nerve fibres conduct at up to about 100 m/s [memory, high confidence]. A 1,000x speed is 10^5 m/s. The delay is then t = d/v = 1.2 m / (10^5 m/s) = 1.2 x 10^-5 s = 12 microseconds. Now the instrument limit: if the shortest time that can be resolved is 80 microseconds, the largest speed that could be demonstrated is v = d/t = 1.2 m / (80 x 10^-6 s) = 15,000 m/s, i.e. about 150 times fast nerve speed, not 1,000. To show 1,000x you need a time resolution of about 12 microseconds or better. So "at least 1,000x" and "80 microsecond resolution" cannot both support a speed claim: the resolution is 6.7 times too coarse. At the other end, the physical limit is light speed: 1.2 m / (3 x 10^8 m/s) = 4 nanoseconds, so there is a bound but it is irrelevant here. (Python check: 1.2/80e-6 = 15,000 m/s; 1.2/(1000 x 100) = 1.2e-5 s.) Further: standard BOLD fMRI (blood-oxygen-level-dependent imaging) measures a haemodynamic response, a change in blood oxygenation that begins about 1-2 s after neural activity and peaks at about 5-6 s; typical sampling times (TR) are 1-3 s [memory, high confidence]. Latency differences down to tens of milliseconds can in special designs be inferred statistically, but 80 microseconds is not a timing resolution of the method as I know it. The "80 microseconds" figure is reported from a lecture (SSE meeting, La Jolla, June 2001) and I could not confirm it from the paper. It may refer to something else (for example the gradient-echo readout time of a single scan line, which is of the order of tens of microseconds per sample [memory, low-medium confidence]); read as such, it would be a sampling interval inside one image, not a neural timing resolution. This is my reading, tagged low-medium.

So the speed claim is internally inconsistent as printed (resolution value cannot support the speed bound), and a "speed" requires a start time (stimulus) and an arrival time with known latency, which a slow haemodynamic signal does not provide. I mark this as the strongest mathematical defect in the article.

Citation check (Investigator, which I confirm from the saved note; the PNAS pages were blocked for me, so this stays [web] via the Investigator and I could not re-fetch: two of my own fetches, PubMed and the PNAS DOI, returned a cookie page and HTTP 403): the paper at PNAS 95:2670-2673 (1998) is "New findings of the correlation between acupoints and corresponding brain cortices using functional MRI" by Cho, Chung, Jones, Park, Park, Lee, Wong, Min. Mallove's reference names "So, Jones, Kidney, Saito" and a different title, with the same volume and pages. So the reference as printed does not match the paper at those pages. Whether a separate So/Jones/Kidney/Saito work exists was not established. [web: https://www.pnas.org/doi/10.1073/pnas.0602520103, via research/2026-10-04-pnas.md]. Retraction (PNAS 103(27):10527, 5 July 2006): Cho, Chung, Lee, Wong and Min retract, citing "no point specificity, at least for pain and analgesic effects"; "J. P. Jones, J. B. Park, and H. J. Park have not approved this retraction." The notice does not mention a speed-of-transmission result. This is part of the record: the retraction was by five of eight authors, and Jones, whom Mallove quotes, dissented. I report both sides: a retraction by the majority of authors on the grounds of no point specificity, and a non-signing co-author. Whether the 1998 paper itself contains the speed statement was not checked (paper not opened).

Mallove's additional statement that other journals rejected the paper "without review" has no source in the article. [read: p.3] Unknown.

The Galvani statement (massless, electrically neutral radiation) is a Correa interpretation of a historical experiment, a quote from monograph AS2-28. [read: p.4] "Massless" and "electrically neutral" together describe the photon in standard physics; the claim that it is "different" from electromagnetism has no equation attached.

b. Standard nerve physiology (conduction velocity v about 100 m/s maximum; for a 1.2 m path, 12 ms minimum) is consistent with decades of measurements (evoked potentials, e.g. somatosensory evoked potentials from foot stimulation arriving at cortex at about 40 ms [memory, medium confidence]). A 12-microsecond arrival would be inconsistent with those, if it were established. c. Tested once (the 1998 paper, in humans), then retracted by five of eight authors; no independent replication of any speed result is known. Reproducibility ("toggled off and on") is asserted. No follow-up of the speed claim found. The wider acupuncture-fMRI literature has not been surveyed by me. d. If true, signal conduction would need a channel outside nerve electrophysiology, and fMRI timing would need to be re-understood. A cheaper fix is that the speed figure is an artefact of the time axis.

Verdict (split): the speed claim (1,000x nerve speed from an 80 microsecond resolution) is Internally inconsistent as printed (resolution too coarse by about 6.7x for the bound, and the measure is haemodynamic); the underlying paper's finding of point-specific activation was Tested, not supported by its own authors' later retraction (majority), with one dissent; the Galvani "Tesla radiation" statement is Not testable as stated.

Claim 6. Establishment science suppresses experiment-based challenges and favours gigantic projects; the dispute should be settled in the laboratory

a. Not a mathematical claim. The statement that "table-top experiments can show the foundations of high-energy physics are as weak as mud" is an assertion without data. [read: p.4] No quantity is defined. Whether suppression occurred is a historical and sociological question; the Investigator's record bears on it in two directions: the Skeptical Inquirer 1994 cover story (Jeff Johnson, for CSICOP) mocked the International Tesla Society meeting (that is the characterization of an interested organization; Mallove's rebuttal is also an interested party's) [web: https://cdn.centerforinquiry.org/wp-content/uploads/sites/29/1994/07/22165128/p32.pdf]; and, on the other side, the record shows the Correas declined independent testing in 2002 and tied testing to funding. Both are in the record. The "laboratory should decide" principle is correct and compatible with the mathematics; it also applies to the Correas' own device. b-d. Not applicable as a physical claim.

Verdict: Not testable as stated (a sociological opinion, with the specific "table-top experiments" assertion lacking data). The mainstream-funding comparison could be checked against budgets but is not done in the article.

3. Other quantitative items in the Investigator's findings (kept in the record)

• Senate record S. Hrg. 107-564 (field hearing, May 30, 2001, Durham NH): carries a Mallove statement and a Valone briefing citing PAGD; advocacy entered into a government record, not a test. [web: https://www.govinfo.gov/content/pkg/CHRG-107shrg80649/html/CHRG-107shrg80649.htm via sites.md; the file is saved in research/files]
• Graneau arcs "approaching 24%" excess electrical energy (Mallove's ICCF-10 framing, grouped with PAGD, BlackLight Power, Wallman/Dammann). Percent figures of that size are the kind that depend on calorimetry and electrical accounting; I did not read these papers and make no judgement. [web: https://www.eugenemallove.com/Writings/lenr_excess_heat.html, via sites.md]
• Aether Motor and ORAC observations (June 14, 2001 and Feb 20, 2002 letters): "70% reduction" in apparent weight of a ~45 mg gold foil; a 27,000 microfarad capacitor charged overnight to about 2 V with no input; hand-estimated outputs of "several watt" and "few tens of watts". Check of the capacitor figure: stored energy E = (1/2) C V^2 = 0.5 x 0.027 F x (2 V)^2 = 0.054 J, about 54 millijoules. A charge of that size over 8-12 hours corresponds to an average power of about 0.054 J / 36,000 s = 1.5 microwatts (about 1.9 microwatts over 8 h), which is within the range of ambient pickup, thermoelectric or electrochemical effects between electrodes [memory, medium confidence], so the figure does not by itself indicate any exotic source. (This is my arithmetic, [read: the quantities from the Feb 20, 2002 letter as reported in research notes; the calculation is mine].) Mallove did not examine the closed electronics box ("I do not represent to anyone that I have examined its innards"). [web: June 14, 2001 letter]
• Mallove's claim of a Reich-Einstein thermal anomaly observed "independently under careful conditions" at his lab, with results to be published: no publication found by the Investigator. [web: June 14, 2001 letter] Keep as an open lead.
• Witnesses (Carrell, Soudak, Axelrad, Aspden, Tiller, Bahman, Sapogin, Minchrowski, King): each reaches us through Correa-run sites; they are statements by named people, documented in published sources; no instrument data in what was collected, and no independent off-site measurement found. I do not dismiss them; they are accounts of observations whose quantitative content I could not see. [web: research notes]
• Patent family: US 5,416,391; 5,449,989; 5,502,354; WO 94/03918; WO 94/09560; GB 2,284,093; CA 2,141,309; CA 2,147,153; AU 5147693; IL 107285. [web: https://patents.google.com/patent/US5449989A/en] A patent grant is not a validation of operation; I note only that it contains the most numerical PAGD data on record, and that I found one unexplained arithmetic discrepancy (459% vs 485.2%, above).

4. Corrections to Step 1

1. Claim 1 definition. Step 1 (claims.md, facts.json) says the PAGD tubes produce "about 10 times more electrical power out than in". The source behind the figure (Mallove's June 14, 2001 letter) says 50 W DC input versus about 500 W mechanical shaft power of a motor. So it is not electrical-out over electrical-in. Step 1 correctly noted that no definition was given in the article; the Investigator's letter now supplies it. claims.md should be updated accordingly. [read: June_14_2001.html]
2. Claim 5 citation. Step 1 flagged that the reference was unchecked; it is now checked: volume and pages 95:2670-2673 belong to Cho et al., "New findings of the correlation between acupoints and corresponding brain cortices using functional MRI", not to "So, Jones, Kidney, Saito", and the paper was retracted in 2006 by five of the eight authors, with Jones among those who did not approve. [web, via Investigator, PNAS retraction notice 103(27):10527]. Step 1 and the deep read should not be read as accepting the printed reference. Also facts.json "mainstream_conflict" says fMRI temporal resolution is seconds; I agree for the BOLD signal [memory, high confidence], but the more decisive point is the arithmetic in claim 5a (80 microseconds cannot bound a 1,000x speed).
3. Reference numbering. Step 1's note that ref 5 is "S2-15 or S2-16" is resolved: the Tesla-coil part (3) is AS2-16; AS2-15 is on magnetism. [web: aetherometry.com/correa_publications.php, via Investigator]
4. "Observed himself" nuance. The deep read says it is unknown whether Mallove operated the instruments. The letter says the validation used meters, a storage oscilloscope and a data-acquisition computer and that he was alone from Infinite Energy; the Correas' narrative says the demonstrations were on their premises (and the Aether Motor ones were black-box). So "who held the probes" is still not stated. [web]
5. Scope of "no source gap". The article is fully readable, but the key experimental sources (Correa monographs) were not available to any step; the Tesla-coil claims therefore rest on the article's summary, which Step 1 states but claims.md could say more plainly.
6. Minor. Step 1's "approximately 10/1" row in the table of numbers is correct as a dimensionless ratio; the Reich dates (1897-1957) check out [memory, high confidence].

5. Web checks done

• PNAS retraction and the Cho et al. paper: my two fetches (PubMed, PNAS DOI) failed (cookie page; HTTP 403), so those facts are taken from the Investigator's notes of a successful Chrome read and the saved notice file (saved HTML is a bot-check page, content not verified by me). Tagged [web, via Investigator].
• Patent numbers read directly from the saved patent text [read].
• Medium-confidence [memory] items I used and could not check on the web in this run: lead-acid surface charge behaviour; Schumann resonance Q; haemodynamic response timing (high confidence); the interpretation of "80 microseconds" (low-medium, labelled as my reading only); motor efficiency range. None of these changes a verdict: the claim 5 inconsistency rests only on the article's own two numbers plus a body-length distance.

6. Verdict summary