---
slug: why-microwaves-heat-food
passed: 2026-10-14
by: the assay line (research/assay-line/cited-1014b/), Codex fixer directed by claude-assaying-boyle-1014; source record by Codex, checked by Codex
checked: 83
confirmed: 60
wrong: 22
unverifiable: 0
observed: 0
stale: 1
---

This was the directed repair pass for the 23 items in research/assay-line/cited-1014b/why-microwaves-heat-food/triage.json, using its N1 source record and N2 Codex reread. The claims below retain each triage item's original wording and verdict, plus the 60 CONFIRMED claims and source notes re-read by the checker in N2; those retained notes describe that checker's evidence, not a new all-entry pass. This fixer re-read the evidence for every edited item, checked the supporting primary literature and local history, and reproduced the interactive failures before editing the authored standalone HTML. Four WRONG items were fixed, one STALE modification date was updated under the JSON-LD exception, and 18 listed MINOR items were left as directed. The wrong count includes those 18 MINOR lines and four WRONG lines. No additional claim is marked rejected or newly confirmed by this repair.

## Claims
- MINOR | X1: "Everyone is told" | https://howeverythingworks.org/category/microwave-ovens/page/4/ | X1; Re-read Bloomfield, Microwave Ovens page 4, question 1239 (5 August 1997). The reader asks about the resonant frequency of the free water molecule, documenting this misconception but not its prevalence. Everyone is rhetorical overstatement, not a measured population claim. Retain MINOR and list under tier C; this sentence contains no accepted WRONG finding.
- MINOR | X2: "A resonance is a sharp peak at one special frequency" | https://openstax.org/books/university-physics-volume-1/pages/15-6-forced-oscillations | X2; Re-read OpenStax University Physics 1, section 15.6, discussion following Figure 15.31: "the less the damping, the narrower the resonance." The wine-glass example describes a lightly damped resonance; sharpness needs that qualification when used as a definition. Retain the MINOR precision finding and list under tier C. This is a different sentence from X4.
- MINOR | X3: "the oven's field flips 2.45 billion times a second;" | https://howeverythingworks.org/category/microwave-ovens/page/4/ | X3; Re-read Bloomfield, Microwave Ovens page 4, the science-book/laser answer: "2,450,000,000 cycles per second". The stated 2.45 GHz cycle frequency is correct; each complete cycle contains two direction reversals. In this back-and-forth explanation, flips is loose wording for oscillation, so retain MINOR rather than treating it as a separately specified reversal count. Tier C: list. X15 is the same wording issue. X4 is in the following sentence, so the same-sentence exception does not apply.
- WRONG | X4: "the loss peaks exactly when the field's period matches it" | https://doi.org/10.1063/1.2360986 | X4; Fetched and re-read Ellison (2007), p. 5, section 4, item 1: "near to the frequency 1 / 2πτ." Also read the pinned page at 90303bbcc1, index.html lines 175-177 and 329-332, and verify.mjs lines 69-76. For the displayed Debye loss, the maximum is at angular frequency times relaxation time equal to one. With tau=8.27 ps, independent arithmetic gives f_peak=19.244854 GHz and field period=2*pi*tau=51.96194 ps, not tau. The period-matching sentence is WRONG; preserve the already correct frequency formula, correct the explanation wherever repeated, and add a dated Corrected line. This pass re-read Ellison section 4 in the accessible full text and differentiated the displayed Debye loss. With tau = 8.27 ps, the peak period is 51.96194249 ps and frequency is 19.24485406 GHz.
- MINOR | X5: "strongest at the surface and fading inward" | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML | X5; Re-read Tang (2015), Propagation of microwaves: "the focused microwave field would cause central heating" for sufficiently small spherical foods. Read index.html lines 461-469: the illustrated slab uses exponential attenuation from its incident surface. The surface maximum describes that plane-wave model but needs a geometry qualification when generalized to food. Retain MINOR and list under tier C; no WRONG finding is being fixed in this sentence.
- MINOR | X6: "The cold middle of a thick item is the part the wave never reached." | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML | X6; Re-read Tang (2015), Penetration depth of microwaves in foods: "microwave power is reduced to 1/e". Read the page's explicit power 1/e definition at index.html line 235 and its exponential attenuation at line 469. Never reached is loose shorthand for strongly attenuated power, not the cutoff implemented by the model. Retain the MINOR wording finding and list under tier C; the formal depth definition itself needs no correction.
- MINOR | X7: "engineers did not sit down and pick 2.45 GHz to dodge water's loss peak." | https://taylor.town/microwaves.pdf | X7; Fetched and re-read Osepchuk (1984), p. 1204: "Raytheon and GE both petitioned the FCC". The passage records 2450-versus-915 MHz proposals and their engineering tradeoffs; it does not establish a categorical claim about never considering the approximately 19 GHz loss peak. Retain MINOR as an overconfident historical inference, not a proved false alternative history. Tier C: list. X14 repeats this exclusion.
- MINOR | X8: "The frequency is an accident of the 1940s." | https://taylor.town/microwaves.pdf | X8; Re-read Osepchuk (1984), p. 1204, including the FCC allocation paragraph. The documented frequency proposals were deliberate, within the historical circumstances described by the page. Accident compresses that contingent history too strongly, but the evidence does not overturn the 1940s chronology or establish the competing loss-peak motive. Retain MINOR and list under tier C.
- MINOR | X9: "the same tube Percy Spencer was standing next to when the chocolate in his pocket melted." | https://taylor.town/microwaves.pdf | X9; Re-read Osepchuk (1984), p. 1204. He introduces "The legendary story" and describes an "open waveguide" and a "candy bar". The source supports a reported anecdote, without independently establishing the page's chocolate/tube details. The note identifies lost qualification and added specificity, not evidence that the event could not have occurred. Retain MINOR and list under tier C.
- MINOR | X10: "not a penetration optimisation anyone designed." | https://taylor.town/microwaves.pdf | X10; Re-read Osepchuk (1984), p. 1204: "GE argued the advantages of penetration". This concerns GE's 915 MHz proposal, contrasted with Raytheon's 2450 MHz proposal. It warrants qualifying the blanket no-design wording, but does not prove that 2.45 GHz was selected specifically to avoid the approximately 19 GHz peak. Retain MINOR and list under tier C. X11, X16 and X19 repeat the accidental-penetration claim.
- MINOR | X11: "water's convenient transparency a lucky bonus." | https://taylor.town/microwaves.pdf | X11; Re-read Osepchuk (1984), p. 1204, and the page's history paragraph at index.html lines 197-208. For the same reasons as X10, lucky bonus overstates what the source establishes about penetration and design intent. Retain MINOR and list under tier C; this sentence contains no accepted WRONG finding.
- MINOR | X12: "the ice relaxation time (Auty & Cole 1952)." | https://api.crossref.org/works/10.1063%2F1.1700726 | X12; Fetched and re-read Auty and Cole's publisher-deposited abstract (DOI 10.1063/1.1700726) and index.html lines 221-222 and 350. The paper is a genuine source for ice relaxation times, but its rate law gives 48.6234 microseconds at -10 C while the implementation uses 20. Keep this broad citation/parameter-precision finding MINOR; X20 requires correction of the numerical temperature mismatch in both implementations. This separate check-panel sentence need not be rewritten if the parameters are corrected. Tier C: list, with the same error label as X20.
- MINOR | X13: "recomputes all of it" | git show 90303bbcc1:research/why-microwaves-heat-food/verify.mjs | X13; Re-read 90303bbcc1:research/why-microwaves-heat-food/verify.mjs lines 117-118 and index.html line 349: the verifier uses "eps_inf: 4.6" at 50 C, while the browser uses "epsInf: 5.2". The verifier therefore does not reproduce every browser parameterization or state. This is an overclaim about the page's own checking, expressly MINOR under the task rule. Tier C: list; its sentence is separate from the accepted WRONG findings.
- MINOR | X14: "not the historical reason the frequency was chosen." | https://taylor.town/microwaves.pdf | X14; Re-read Osepchuk (1984), p. 1204, and index.html lines 240-241. The specific 2.45-versus-19 GHz historical exclusion is not established by the documented 915-versus-2450 MHz debate. This repeats X7's overconfident inference; the evidence does not justify replacing it with a claim that avoiding the loss peak was the historical reason. Retain MINOR and list under tier C.
- MINOR | X15: "the field flips 2.45 billion times a second" | https://howeverythingworks.org/category/microwave-ovens/page/4/ | X15; Re-read Bloomfield's science-book/laser answer and 90303bbcc1:src/content/strata/why-microwaves-heat-food.md line 4. This repeats X3: 2.45 billion is the correct number of complete oscillation cycles per second, while a direction-reversal count would be twice as large. Retain the MINOR terminology finding and its shared error label. Tier C: list; the dek sentence contains none of the accepted WRONG findings.
- MINOR | X16: "the deep penetration is a fortunate accident, not a design." | https://taylor.town/microwaves.pdf | X16; Re-read Osepchuk (1984), p. 1204, and the markdown dek at 90303bbcc1, line 4. This repeats the unsupported accident-only account in X10, alongside X7's categorical exclusion of a peak-avoidance motive. The source documents competing engineering proposals without settling that specific motive. Retain MINOR and list under tier C; no accepted WRONG finding occupies this sentence.
- MINOR | X17: "A microwave oven deliberately misses water's resonance." | git show 90303bbcc1:public/strata/why-microwaves-heat-food/index.html | X17; Re-read 90303bbcc1:index.html line 28: "A microwave oven deliberately misses water's resonance." The body at lines 198-199 expressly denies that deliberate motive. Osepchuk p. 1204 does not resolve the specific peak-avoidance intention in either direction. This is a metadata inconsistency and an unsupported intent gloss, not evidence selecting one causal history as true. Retain MINOR and list under tier C. X23 repeats the intentionality gloss in the verifier.
- STALE | X18: ""dateModified":"2026-06-30"" | git log 90303bbcc1 -- public/strata/why-microwaves-heat-food/index.html | X18; Re-read index.html line 31, the original 2026-06-30 build log, and git log 90303bbcc1 for this HTML. Later edits include 1227d0ac000 on 2026-09-04, d4840e62371 on 2026-09-27 and 2e60bcb5dd1 on 2026-10-01. Thus dateModified is STALE, while datePublished remains supported. Apply the explicit JSON-LD exception: fix dateModified to the applicable modification date, without a Corrected or Updated line. This is the layer's only STALE item, so there are no non-exempt STALE items to group into an Updated line. This pass re-read the original build log and the HTML history through 7a9bcb5442. The June publication date is supported; subsequent September and October edits establish stale modification metadata.
- MINOR | X19: "the penetration a fortunate accident" | https://taylor.town/microwaves.pdf | X19; Re-read Osepchuk (1984), p. 1204, and the actual markdown summary at 90303bbcc1, lines 29-31. The latter repeats X10's accidental-penetration inference; the historical source does not establish that all penetration benefits were unintended. Retain MINOR and list under tier C. The heating corollaries elsewhere in this compound sentence are not refuted by this finding.
- WRONG | X20: "ice (−10 °C)" | https://api.crossref.org/works/10.1063%2F1.1700726 | X20; Fetched and re-read Auty and Cole (1952), publisher-deposited abstract at https://api.crossref.org/works/10.1063%2F1.1700726: "τ = A exp(B/RT)". It supplies A=5.3e-16 seconds and B=13.2 kcal/mole. Using R=0.00198720425864083 kcal/(mol K) and T=263.15 K gives 48.6234 microseconds; 20 microseconds instead corresponds to approximately -0.40 C. Re-read 90303bbcc1:index.html line 150, the -10 C button, and line 350: "var ICE = { epsS:95, epsInf:3.2, tau:20e-6 };"; verify.mjs lines 126-127 repeats that mismatch. The preset is WRONG as attributed and temperature-labelled. Reconcile the ice parameters, attribution and dependent outputs in both implementations; simply calling the entire tuple measured at 0 C would not be established by this calculation. Add one dated Corrected line for this error, shared with X12. This pass fetched the publisher-deposited abstract from Crossref. At T = 263.15 K, A = 5.3e-16 s, B = 13.2 kcal/mol and R = 0.00198720425864083 kcal/(mol K) give tau = 48.62340301 microseconds. Only the relaxation time is sourced from that rate law; the other ice constants remain approximate model values.
- WRONG | X21: (added) "water's loss peak · 19 GHz" | git show 90303bbcc1:public/strata/why-microwaves-heat-food/index.html | X21; Re-read 90303bbcc1:public/strata/why-microwaves-heat-food/index.html line 139, "data-f="19.24"", the TAU_T table at line 340, the current-parameter peak at line 446, and the click handler at line 537: "setF(Number(b.getAttribute('data-f')))". Executed the actual inline script with DOM/canvas stubs. At 0, 25 and 100 C the button still selects 19.26 GHz after slider rounding, while the model peaks are respectively 9.0071, 19.2449 and 68.0149 GHz. The unqualified current-water peak label is WRONG away from room temperature. Make the control track temperature or clearly label it as the 25 C preset; add a dated Corrected line. This pass executed the actual inline script with DOM and no-op canvas stubs: the fixed preset selects 19.26 GHz after slider rounding at 0, 25 and 100 C, while the model peaks are 9.0071, 19.2449 and 68.0149 GHz.
- WRONG | X22: (added) "Near water’s true loss peak" | git show 90303bbcc1:public/strata/why-microwaves-heat-food/index.html | X22; Re-read 90303bbcc1:public/strata/why-microwaves-heat-food/index.html line 512: "var atPeak = fG>12;", and the resulting verdict at line 518: "Near water’s <b>true loss peak</b>". Executed the actual inline script with DOM/canvas stubs at 25 C and 100 GHz: it emits that verdict with loss 13.58, although the displayed model peaks at 19.2449 GHz with loss 36.5947. The high-frequency point is falsely identified as near the peak. Determine the label from the current temperature's peak and the selected frequency, rather than an unconditional lower threshold; fix and add a dated Corrected line. This pass reproduced the false verdict at 25 C and 100 GHz. The actual model gives loss 13.5822 there, versus a peak near 19.2449 GHz with loss 36.5947. The revised verdict is checked at 0, 25 and 100 C on both sides of the current peak.
- MINOR | X23: (added) "off-resonance on purpose" | git show 90303bbcc1:research/why-microwaves-heat-food/verify.mjs | X23; Re-read 90303bbcc1:research/why-microwaves-heat-food/verify.mjs lines 100-101: "atOven.epsI < atPeak.epsI" tests a loss comparison, while its PASS message adds "off-resonance on purpose". The inequality establishes no historical intention. Retain MINOR as an overclaim about what the check establishes and list under tier C. Share X17's intent error label; no accepted WRONG finding is in this assertion.
- CONFIRMED | "Water has no sharp resonance at the oven's 2.45 GHz [...] and 2.45 GHz is nowhere near where water absorbs most strongly." | https://doi.org/10.1063/1.2360986 | id=no-resonance; Ellison p. 1 separates "three relaxations in the microwave region" from "two resonances in the far infrared" and gives 18.56 GHz for the principal relaxation at 25 °C. This supports the claim in its room-temperature context.
- CONFIRMED | "That turns out to be the saving grace: an oven tuned to water's true peak would cook nothing but the skin." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=peak-skin; Source excerpt: "With a frequency of 20 GHz, the penetration depth would be much smaller, i.e. the energy would be absorbed in a thin surface layer of the food (toasting the food) while the interior would remain cold."
- CONFIRMED | "Liquid water absorbs over a broad band by dielectric relaxation [...] polar molecules trying to follow the flipping field and lagging behind." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=broad-relaxation; Vollmer p. 78: "dielectric loss due to dipole relaxation." Read the two-column passage directly; text extraction interleaves the columns but the quoted words are present.
- CONFIRMED | "That loss peaks near 19 GHz, not 2.45." | https://doi.org/10.1063/1.2360986 | id=loss-peak; Ellison p. 1: "18.56 GHz" at 25 °C. The page's simplified constants put its own single-Debye peak at 19.245 GHz. Both justify near 19 GHz, not an exact universal temperature-independent frequency.
- CONFIRMED | "The oven sits far below the peak, where water absorbs only weakly [...] and that weakness is what lets the wave reach the middle instead of cooking only the skin." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=weak-penetration; Source excerpt: "With a frequency of 20 GHz, the penetration depth would be much smaller, i.e. the energy would be absorbed in a thin surface layer of the food (toasting the food) while the interior would remain cold."
- CONFIRMED | "At 2.45 GHz the loss is modest and the wave's power reaches about 1.9 cm into water [...] deep enough to warm a whole bowl." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=oven-depth-bowl; The supplied 20 GHz excerpt alone does not establish 1.9 cm. Recreated the page's model: 1.8696 cm at 25 °C. Vollmer p. 79 separately says food has "typical dimensions of the order of cm" and discusses penetration on that scale. Warming a whole bowl also depends on dimensions and heat transport.
- CONFIRMED | "all the energy is dumped into the outermost skin and the centre stays cold." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=all-energy-skin; Source excerpt: "With a frequency of 20 GHz, the penetration depth would be much smaller, i.e. the energy would be absorbed in a thin surface layer of the food (toasting the food) while the interior would remain cold."
- CONFIRMED | "An oven tuned to "water's resonance" would be a very expensive way to scorch the surface of your soup." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=soup-scorch; Source excerpt: "With a frequency of 20 GHz, the penetration depth would be much smaller, i.e. the energy would be absorbed in a thin surface layer of the food (toasting the food) while the interior would remain cold."
- CONFIRMED | "An isolated water molecule does have rotational resonances" | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=isolated-molecule; The record excerpt mentions the liquid only. Read the immediately preceding Vollmer p. 78 discussion: "in the infrared, excitation of molecular vibrations and/or rotations dominates" for free molecules. The isolated-molecule claim is supported by that context.
- CONFIRMED | "Heating here is friction, not resonance: each molecule is a tiny electric dipole;" | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=friction-dipoles; Source excerpt: "At the GHz frequency of a microwave oven the phase lag of the dipoles behind the electric field absorbs power from the field. This is known as dielectric loss due to dipole relaxation."
- CONFIRMED | "the molecules wheel back and forth trying to keep up, jostle their neighbours, and that jostle is heat." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=jostling-heat; Source excerpt: "At the GHz frequency of a microwave oven the phase lag of the dipoles behind the electric field absorbs power from the field. This is known as dielectric loss due to dipole relaxation."
- CONFIRMED | "The governing time is water's relaxation time τ ≈ 8.3 ps" | https://doi.org/10.1063/1.2360986 | id=relaxation-time; Ellison Table 1 gives 25 °C, Δ=72.38, τ=8.28 ps. Kaatze Table III independently gives τ=8.27 ps. Both support approximately 8.3 ps.
- CONFIRMED | "Hotter water absorbs less." | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML | id=hotter-less; Tang: "increasing temperature reduces loss factors of tap water at 2450 and 915 MHz." The surrounding page explicitly restricts this statement to the oven-frequency, pure-water story.
- CONFIRMED | "Warm the water and its molecules relax faster (τ drops), so the peak slides to higher frequency and 2.45 GHz catches less of it." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=temperature-peak-shift; Source excerpt: "At the same time, ε2 shows a pronounced maximum around 20 GHz. For increasing temperature, both functions drop in size and shift to higher frequencies."
- CONFIRMED | "This is the clean story for pure water; in salty food the ion-conduction channel below grows with temperature and can cancel or reverse the trend." | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML | id=salty-temperature; Source excerpt: "For foods with added salt, for example, mashed potatoes with 1.5% salt, increasing temperature increases the value of the loss factor"
- CONFIRMED | "Ice barely heats." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=ice-barely-heats; Source excerpt: "In ice the molecules of H2O are held in a crystal lattice in fixed positions. The dipoles cannot follow an external field as easily as they can in liquid water, therefore there is less heating effect."
- CONFIRMED | "Freeze the water and the dipoles lock into the crystal; their relaxation time leaps from picoseconds to tens of microseconds, dragging the loss peak down to the kilohertz range." | https://api.crossref.org/works/10.1063%2F1.1700726 | id=ice-relaxation; Auty and Cole abstract: "τ = A exp(B/RT)", A=5.3e-16 s and B=13.2 kcal/mole. This gives 48.6234 µs and a 3.273 kHz Debye peak at -10 °C, supporting tens of microseconds and kilohertz. The separate implementation mismatch is recorded under ice-temperature-button.
- CONFIRMED | "At 2.45 GHz ice's loss is thousands of times smaller than liquid water's" | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=ice-loss-ratio; Source excerpt: "for ice at 2.45 GHz is three to four orders of magnitude below that of liquid water (see [14])."
- CONFIRMED | "which is why a microwave defrosts so unevenly: the first drop to melt suddenly absorbs hundreds of times harder than the ice around it and runs away in temperature." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=defrost-runaway; Vollmer pp. 79-80 gives the three-to-four-order loss contrast and says "the liquid water will be heated" after local melting. Its subsequent on/off-cycle account supports increasing preferential heating of meltwater; the single quoted on/off passage alone does not quantify the ratio.
- CONFIRMED | "Dissolved salts add a second, often larger heat channel [...] ions dragged back and forth by the field (ohmic conduction) [...] which is why salty or briny food heats faster than the dielectric numbers alone predict." | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML | id=salts-ohmic; Read Tang Eq. 8 and its explanation: the loss factor combines dipolar loss and ionic conductivity; "Ionic conductivity in foods, in turn, increases with increasing amount of dissolved salts and with increasing temperature." This supports an additional, potentially dominant heating channel.
- CONFIRMED | "It isn't only water, and it isn't "inside-out."" | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML | id=inside-out; Confirmed as rejection of an inherent center-first microwave-heating mechanism. Tang separately documents focused central heating in some small spherical foods; therefore this must not be used to claim that central hot spots are impossible. The adjacent unqualified surface-maximum sentence is MINOR.
- CONFIRMED | "It is an ISM band [...] one of a handful of slots that international radio regulation set aside for Industrial, Scientific and Medical heating so it wouldn't trample radio and radar traffic" | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML | id=ism-allocation; Tang Table 1 identifies 2450 MHz ISM applications as "Worldwide, except where 2375 MHz is used" and the text explains assignments "To avoid interference". The record's FCC-only excerpt is narrower than the international statement; the full table supplies the wider support.
- CONFIRMED | "it fell out of the cavity-magnetron technology built for wartime radar," | https://taylor.town/microwaves.pdf | id=wartime-magnetron; Osepchuk p. 1204: "most powerful magnetrons developed during the war were pulsed tubes for radar." Same paragraph discusses CW development for heating. This supports technological ancestry, not the page's stronger no-design frequency history.
- CONFIRMED | "with the literature constants for water at 25 °C (ε_s=78.4, ε_∞=5.2, τ=8.27 ps)." | https://www.scribd.com/document/288014687/Complex-Permittivity-of-Water-as-a-Function-of-Frequency-and-Temperature; https://doi.org/10.1063/1.2360986 | id=literature-constants; Kaatze Table III at 25 °C gives ε(0)=78.36±0.05, ε(∞)=5.2±0.1, τ=8.27±0.02 ps. This supports the rounded constants. The supplied Ellison Table 1 excerpt instead gives Δ=72.38 and τ=8.28; it is not the source of this exact tuple.
- CONFIRMED | "measured temperature dependence of τ (Kaatze 1989)" | https://www.scribd.com/document/288014687/Complex-Permittivity-of-Water-as-a-Function-of-Frequency-and-Temperature; https://doi.org/10.1063/1.2360986 | id=kaatze-attribution; Read Kaatze's primary paper in its accessible transcription, Table III: τ=17.67, 9.36, 8.27, 7.28, 5.82, 4.75, 4.01 ps at 0, 20, 25, 30, 40, 50, 60 °C. It supports measured temperature dependence. The browser's 75 and 100 °C entries need an additional source if claimed as Kaatze measurements.
- CONFIRMED | "Single-Debye is a model: real water has a small extra fast relaxation above ~30 GHz, but through the oven band it matches measurement to a few percent [...] which is all the argument needs." | https://doi.org/10.1063/1.2360986 | id=model-fast-relaxation-fit; Ellison p. 5 says the single-Debye interpolation through 0-30 GHz at 25 °C is "within less than 1%"; extrapolation toward 90 GHz differs by approximately 10%. The oven-band qualification is supported.
- CONFIRMED | "real water has a small extra fast relaxation above ~30 GHz" | https://doi.org/10.1063/1.2360986 | id=extra-fast-relaxation; Ellison p. 5 locates a second relaxation "around 170 GHz" and explains deviations from a single-Debye extrapolation above 30 GHz. The page describes where additional contributions matter, not a second loss peak at 30 GHz.
- CONFIRMED | "What's shown is solid textbook electromagnetics; what's modelled is the single-Debye permittivity (an excellent fit below ~20 GHz, an approximation above it)." | https://doi.org/10.1063/1.2360986 | id=excellent-fit; Ellison p. 5 reports a fit "within less than 1%" over 0-30 GHz at 25 °C. This supports the stated low-frequency model limitation.
- CONFIRMED | "real food [...] layered, salty, fatty, partly frozen [...] deviates (real food is usually lossier, so penetration is often closer to ~1 cm)" | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML | id=real-food-depth; Tang gives "3.8 mm in cooked ham" versus "9.9 mm in cooked beef" and says depths "vary widely with temperature and frequency." Its other examples support often order 1 cm, not a universal food depth.
- CONFIRMED | "The exact molecular story of why a microwave cooks unevenly (cavity standing waves spaced ~6 cm, turntables, edge focusing) is real and only sketched here." | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML; https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=uneven-cavity-heating; Tang says "standing waves in microwave cavities play a major role" and describes alternating nodes and antinodes separated by a quarter wavelength. Adjacent maxima are a half wavelength: c/(2f)=6.118 cm at 2.45 GHz. Tang covers geometric focusing; Vollmer explains the turntable and mode stirring.
- CONFIRMED | "The frontier that remains honest: the precise lineshape of water's loss above the Debye peak is still actively modelled (a single Debye term is not the last word)." | https://arxiv.org/abs/1912.08118; https://api.crossref.org/works/10.1103%2Frfks-j3yr | id=active-lineshape; The cited Fiedler et al. abstract uses "two Debye (microwave) oscillators". Current corroboration: DOI 10.1103/rfks-j3yr, published 2026-06-09, investigates the "fast secondary Debye relaxation" and high-frequency deviation from classical Debye behavior. Publisher site returned 403; publisher-deposited Crossref abstract was read successfully.
- CONFIRMED | "U. Kaatze, "Complex permittivity of water as a function of frequency and temperature," J. Chem. Eng. Data 34, 371–374 (1989)." | https://api.crossref.org/works/10.1021%2Fje00058a001 | id=citation-kaatze; Crossref and the primary paper agree: Udo Kaatze, 1989, Journal of Chemical & Engineering Data 34(4), 371-374, DOI 10.1021/je00058a001.
- CONFIRMED | "W. J. Ellison, "Permittivity of Pure Water, at Standard Atmospheric Pressure, over the Frequency Range 0–25 THz and Temperature Range 0–100 °C," J. Phys. Chem. Ref. Data 36, 1–18 (2007)." | https://api.crossref.org/works/10.1063%2F1.2360986 | id=citation-ellison; Read PDF title page and Crossref: W. J. Ellison, 2007, volume 36(1), 1-18, DOI 10.1063/1.2360986. Crossref drops the leading zeros in the title's frequency/temperature ranges; the PDF supports the page's correct 0-25 THz and 0-100 °C wording.
- CONFIRMED | "R. P. Auty & R. H. Cole, "Dielectric Properties of Ice and Solid D₂O," J. Chem. Phys. 20, 1309 (1952)." | https://api.crossref.org/works/10.1063%2F1.1700726 | id=citation-auty; Crossref: Robert P. Auty and Robert H. Cole, 1952, The Journal of Chemical Physics 20(8), 1309-1314, DOI 10.1063/1.1700726. Giving the first page as 1309 is valid.
- CONFIRMED | "M. Vollmer, "Physics of the microwave oven," Physics Education 39, 74–81 (2004)." | https://api.crossref.org/works/10.1088%2F0031-9120%2F39%2F1%2F006 | id=citation-vollmer; PDF: "Physics Education 39 (1)" and "January 2004", pages 74-81. The checker's DOI MISMATCH uses Crossref's online issued date 2003-12-17. The page correctly cites the 2004 print issue.
- CONFIRMED | "J. Tang, "Unlocking Potentials of Microwaves for Food Safety and Quality," J. Food Science 80(8), E1776–E1793 (2015)" | https://api.crossref.org/works/10.1111%2F1750-3841.12959 | id=citation-tang; The complete Europe PMC article metadata gives Juming Tang, Journal of Food Science 80(8), E1776-E1793, 2015, DOI 10.1111/1750-3841.12959. Crossref does not carry the page range, so its title-only excerpt is insufficient for the entire citation.
- CONFIRMED | "penetration depth, ionic conduction, and the 915 vs 2450 MHz comparison." | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML | id=tang-topics; Read Tang's separate sections "Microwave penetration depth", "Dielectric properties of foods" and "Microwave Frequencies". The full article supports all three topics; the supplied frequency-comparison sentence alone covers only one.
- CONFIRMED | "You have been told that a microwave oven works by tuning radio waves to the resonant frequency of water, so the waves shake the molecules into heat." | https://howeverythingworks.org/category/microwave-ovens/page/4/ | id=dek-myth; The 1997 Bloomfield reader question documents this misconception. The sentence addresses the reader rhetorically; the distinct universal "Everyone" assertion is separately marked MINOR.
- CONFIRMED | "Almost every word of that is wrong, and the true story is better." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=dek-wrong; The substantive correction is supported by Vollmer and Ellison: oven-frequency heating is dipolar relaxation, not a water-molecule resonance at 2.45 GHz. "Almost every word" is editorial framing, not a counted statistic.
- CONFIRMED | "Liquid water has no sharp resonance at the oven's 2.45 GHz: in the liquid the molecules collide too often for any clean resonance to survive." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=dek-collisions; Vollmer p. 78 explains that neighboring-molecule interactions prevent free rotations and produce a lagged dielectric response. This broad qualitative account is supported; the much more specific fractional-picosecond sentence is UNVERIFIABLE.
- CONFIRMED | "That broad loss peaks near 19 GHz, almost an order of magnitude above the oven, and 2.45 GHz sits far down on its low-frequency flank where water absorbs only weakly." | https://doi.org/10.1063/1.2360986 | id=dek-peak; Ellison gives 18.56 GHz at 25 °C; the page model gives 19.245 GHz. The ratio to 2.45 GHz is approximately 7.86, consistent with almost an order of magnitude.
- CONFIRMED | "Here is the twist the myth inverts: that weakness is the saving grace." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=dek-weakness; Vollmer p. 79 explicitly connects lower microwave absorption with cm-scale penetration, contrasting 20 GHz surface absorption.
- CONFIRMED | "Two more things the myth gets backwards fall out of the same curve: hotter water absorbs less (its peak slides further away), and ice barely heats at all (its dipoles are locked, its loss peak dragged down to the kilohertz range, which is exactly why a microwave defrosts in runaway patches)." | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML | id=dek-heating-trends; The Tang excerpt supplies the temperature trend; its full article and Vollmer pp. 79-80 supply low ice loss and uneven defrost. Auty and Cole supply the microsecond relaxation and hence the kilohertz model peak.
- CONFIRMED | "Sources: Kaatze 1989, Ellison 2007, Auty & Cole 1952, Vollmer 2004, Tang 2015." | https://api.crossref.org/works/10.1021%2Fje00058a001 | id=dek-sources; All five full references were independently checked. The record points only to the Kaatze Crossref entry for this compound list; that one entry alone cannot establish the other four.
- CONFIRMED | "water's dielectric loss is shown to peak at 19 GHz, far from the oven's 2.45 [...] so 'tuned to water's resonance' is the wrong physics." | https://doi.org/10.1063/1.2360986 | id=related-water-peak; Ellison p. 1 supplies the room-temperature principal relaxation near 19 GHz; the oven frequency is 2.45 GHz. This is the same independently supported distinction as the main page.
- CONFIRMED | "Helium doesn't make your voice higher-pitched [...] it shifts the formants, the resonances of the cavity, while the pitch stays put." | https://www.phys.unsw.edu.au/jw/speechmodel.html | id=related-helium; UNSW's speech model says "it does not change the pitch" and distinguishes formants and timbre. The statement is accurate for the stated vocal-source/filter explanation.
- CONFIRMED | "it shifts the formants, the resonances of the cavity," | https://www.phys.unsw.edu.au/jw/speechmodel.html | id=related-helium-formants; UNSW: "Inhaling helium changes the frequencies of the resonances, and therefore of the formants they produce". Exact excerpt found and read in context.
- CONFIRMED | "A microwave isn't tuned to water's resonant frequency [...] there is no sharp resonance, only broad relaxation, and the oven sits far off its peak." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=related-microwave; Vollmer p. 78 and Ellison p. 1 jointly support broad dielectric relaxation at the oven frequency rather than a sharp molecular resonance.
- CONFIRMED | "Metal isn't colder than wood at the same room temperature [...] it pulls heat from your hand faster, so 'cold' is really a flux." | https://thermtest.com/thermal-resources/thermal-conductivity-experiments/touch-metal-touch-wood | id=related-metal; The experiment source says heat is removed from a hand faster by metal than wood or plastic. Objects at the same room temperature can therefore feel different because of heat transfer.
- CONFIRMED | "Microwaves don't 'resonate' with water [...] they drag its dipoles through a viscous medium, so the heating is really a friction." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=related-viscous; Vollmer describes dipolar lag and interactions with neighbors. Friction is the page's qualitative dissipation analogy; the displayed numerical model is dielectric loss.
- CONFIRMED | "Microwave ovens are not tuned to water's resonant frequency: water has no sharp resonance at 2.45 GHz. They heat by dielectric relaxation, and 2.45 GHz sits far BELOW water's loss peak (~19 GHz). That weak absorption is exactly what lets the wave reach the centre instead of searing the surface: an oven at water's true peak would cook only the skin. Drag the dial and watch the oven get worse as you approach 'resonance.'" | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=meta-description; Read beyond the one supplied 20 GHz excerpt: Vollmer pp. 78-79 supports dipolar relaxation and the penetration tradeoff; the recreated single-Debye model supplies the approximate frequencies/depths.
- CONFIRMED | "A microwave oven is not tuned to water's resonant frequency. Its 2.45 GHz misses water's loss peak by a mile, and that miss is the whole reason it cooks the inside, not just the skin." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=meta-og-description; Supported as the physical penetration tradeoff described by Vollmer p. 79; it makes no numerical historical-design assertion. The surrounding page supplies pure-water and model limitations.
- CONFIRMED | "A microwave oven is not tuned to water's resonant frequency. It heats by dielectric relaxation, and runs at 2.45 GHz [...] far below water's loss peak near 19 GHz. That weak absorption is what lets the wave penetrate to the centre rather than searing the surface; an oven at water's true peak would cook only the skin." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=structured-description; Vollmer pp. 78-79 and the reproduced model support the compound physics description. The supplied single phase-lag excerpt is insufficient by itself for its penetration clauses.
- CONFIRMED | "Liquid water has no sharp resonance at 2.45 GHz; it heats by broadband dielectric (dipolar) relaxation, whose loss peaks near 19 GHz. The oven runs far below that peak, where water absorbs weakly [...] and that weak absorption is exactly what gives the wave a ~1.9 cm penetration depth so the inside heats too; at the real loss peak the energy would die in ~0.5 mm and only sear the surface." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=body-summary; The full Vollmer discussion and reproduced single-Debye model support the body's physical summary. The supplied excerpt alone does not establish every number, which were independently recomputed.
- CONFIRMED | "Industrial 915 MHz ovens use exactly this to heat thick loads more evenly." | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML | id=industrial-heating-use; Tang explains deeper penetration at 915 MHz and describes practical industrial food-processing systems. The full article supports the application, not just the provided comparison sentence.
- CONFIRMED | "This is why a frozen meal defrosts in patches: the first melt-water suddenly absorbs hundreds of times harder." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=script-defrost; Vollmer pp. 79-80 connects the low ice loss with preferential absorption by liquid meltwater and the defrost cycle. Its preceding loss-ratio paragraph supplies the comparison missing from the quoted excerpt.
- CONFIRMED | "You would sear the skin and leave the centre cold." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=script-peak-searing; Vollmer p. 79: "the energy would be absorbed in a thin surface layer of the food" while the interior stays cold. Supported as the direct microwave-deposition contrast, not a claim that heat can never conduct inward.
- CONFIRMED | "The wave passes through almost unheated;" | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=script-ice-transparency; Vollmer p. 79 gives ice loss three to four orders below liquid water. That supports low microwave heating of ice; it does not independently validate the app's exact 116.84 m toy-model penetration depth.
- CONFIRMED | "The more strongly water drinks the wave, the less of the food it reaches." | https://users.wfu.edu/natalie/s24phy712/lecturenote/physics_of_microwave_ovens.pdf | id=script-stronger-less; At the default 25 °C, this sentence appears on the 2.45-12 GHz rising flank, where stronger loss accompanies smaller penetration depth. Interpreted as attenuation rather than a global monotonic rule for ε″, the qualitative statement holds. The fixed branch threshold becomes misleading away from room temperature; see the added state finding.
- CONFIRMED | "industrial · 915 MHz" | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4657497/fullTextXML | id=industrial-button; Tang explicitly states that 915 MHz is used in large-scale industrial food processing in the USA. The button is a rounded frequency label, reproduced by the script.

## Findings
- [declined] X1: NOT FIXED as a separate wording change, per the directing triage. Re-read Bloomfield, Microwave Ovens page 4, question 1239 (5 August 1997). The reader asks about the resonant frequency of the free water molecule, documenting this misconception but not its prevalence. Everyone is rhetorical overstatement, not a measured population claim. Retain MINOR and list under tier C; this sentence contains no accepted WRONG finding.
- [declined] X2: NOT FIXED as a separate wording change, per the directing triage. Re-read OpenStax University Physics 1, section 15.6, discussion following Figure 15.31: "the less the damping, the narrower the resonance." The wine-glass example describes a lightly damped resonance; sharpness needs that qualification when used as a definition. Retain the MINOR precision finding and list under tier C. This is a different sentence from X4.
- [declined] X3: NOT FIXED as a separate wording change, per the directing triage. Re-read Bloomfield, Microwave Ovens page 4, the science-book/laser answer: "2,450,000,000 cycles per second". The stated 2.45 GHz cycle frequency is correct; each complete cycle contains two direction reversals. In this back-and-forth explanation, flips is loose wording for oscillation, so retain MINOR rather than treating it as a separately specified reversal count. Tier C: list. X15 is the same wording issue. X4 is in the following sentence, so the same-sentence exception does not apply.
- [fixed] X4: FIXED the authored period-matching explanation in public/strata/why-microwaves-heat-food/index.html. It now states period 2*pi*tau, approximately 52 ps, while retaining the correct frequency formula. Added and mirrored one dated Corrected line.
- [declined] X5: NOT FIXED as a separate wording change, per the directing triage. Re-read Tang (2015), Propagation of microwaves: "the focused microwave field would cause central heating" for sufficiently small spherical foods. Read index.html lines 461-469: the illustrated slab uses exponential attenuation from its incident surface. The surface maximum describes that plane-wave model but needs a geometry qualification when generalized to food. Retain MINOR and list under tier C; no WRONG finding is being fixed in this sentence.
- [declined] X6: NOT FIXED as a separate wording change, per the directing triage. Re-read Tang (2015), Penetration depth of microwaves in foods: "microwave power is reduced to 1/e". Read the page's explicit power 1/e definition at index.html line 235 and its exponential attenuation at line 469. Never reached is loose shorthand for strongly attenuated power, not the cutoff implemented by the model. Retain the MINOR wording finding and list under tier C; the formal depth definition itself needs no correction.
- [declined] X7: NOT FIXED as a separate wording change, per the directing triage. Fetched and re-read Osepchuk (1984), p. 1204: "Raytheon and GE both petitioned the FCC". The passage records 2450-versus-915 MHz proposals and their engineering tradeoffs; it does not establish a categorical claim about never considering the approximately 19 GHz loss peak. Retain MINOR as an overconfident historical inference, not a proved false alternative history. Tier C: list. X14 repeats this exclusion.
- [declined] X8: NOT FIXED as a separate wording change, per the directing triage. Re-read Osepchuk (1984), p. 1204, including the FCC allocation paragraph. The documented frequency proposals were deliberate, within the historical circumstances described by the page. Accident compresses that contingent history too strongly, but the evidence does not overturn the 1940s chronology or establish the competing loss-peak motive. Retain MINOR and list under tier C.
- [declined] X9: NOT FIXED as a separate wording change, per the directing triage. Re-read Osepchuk (1984), p. 1204. He introduces "The legendary story" and describes an "open waveguide" and a "candy bar". The source supports a reported anecdote, without independently establishing the page's chocolate/tube details. The note identifies lost qualification and added specificity, not evidence that the event could not have occurred. Retain MINOR and list under tier C.
- [declined] X10: NOT FIXED as a separate wording change, per the directing triage. Re-read Osepchuk (1984), p. 1204: "GE argued the advantages of penetration". This concerns GE's 915 MHz proposal, contrasted with Raytheon's 2450 MHz proposal. It warrants qualifying the blanket no-design wording, but does not prove that 2.45 GHz was selected specifically to avoid the approximately 19 GHz peak. Retain MINOR and list under tier C. X11, X16 and X19 repeat the accidental-penetration claim.
- [declined] X11: NOT FIXED as a separate wording change, per the directing triage. Re-read Osepchuk (1984), p. 1204, and the page's history paragraph at index.html lines 197-208. For the same reasons as X10, lucky bonus overstates what the source establishes about penetration and design intent. Retain MINOR and list under tier C; this sentence contains no accepted WRONG finding.
- [declined] X12: NOT FIXED as a separate wording change, per the directing triage. Fetched and re-read Auty and Cole's publisher-deposited abstract (DOI 10.1063/1.1700726) and index.html lines 221-222 and 350. The paper is a genuine source for ice relaxation times, but its rate law gives 48.6234 microseconds at -10 C while the implementation uses 20. Keep this broad citation/parameter-precision finding MINOR; X20 requires correction of the numerical temperature mismatch in both implementations. This separate check-panel sentence need not be rewritten if the parameters are corrected. Tier C: list, with the same error label as X20. The numerical mismatch was corrected under X20; the separate citation sentence remains unchanged.
- [declined] X13: NOT FIXED as a separate wording change, per the directing triage. Re-read 90303bbcc1:research/why-microwaves-heat-food/verify.mjs lines 117-118 and index.html line 349: the verifier uses "eps_inf: 4.6" at 50 C, while the browser uses "epsInf: 5.2". The verifier therefore does not reproduce every browser parameterization or state. This is an overclaim about the page's own checking, expressly MINOR under the task rule. Tier C: list; its sentence is separate from the accepted WRONG findings.
- [declined] X14: NOT FIXED as a separate wording change, per the directing triage. Re-read Osepchuk (1984), p. 1204, and index.html lines 240-241. The specific 2.45-versus-19 GHz historical exclusion is not established by the documented 915-versus-2450 MHz debate. This repeats X7's overconfident inference; the evidence does not justify replacing it with a claim that avoiding the loss peak was the historical reason. Retain MINOR and list under tier C.
- [declined] X15: NOT FIXED as a separate wording change, per the directing triage. Re-read Bloomfield's science-book/laser answer and 90303bbcc1:src/content/strata/why-microwaves-heat-food.md line 4. This repeats X3: 2.45 billion is the correct number of complete oscillation cycles per second, while a direction-reversal count would be twice as large. Retain the MINOR terminology finding and its shared error label. Tier C: list; the dek sentence contains none of the accepted WRONG findings.
- [declined] X16: NOT FIXED as a separate wording change, per the directing triage. Re-read Osepchuk (1984), p. 1204, and the markdown dek at 90303bbcc1, line 4. This repeats the unsupported accident-only account in X10, alongside X7's categorical exclusion of a peak-avoidance motive. The source documents competing engineering proposals without settling that specific motive. Retain MINOR and list under tier C; no accepted WRONG finding occupies this sentence.
- [declined] X17: NOT FIXED as a separate wording change, per the directing triage. Re-read 90303bbcc1:index.html line 28: "A microwave oven deliberately misses water's resonance." The body at lines 198-199 expressly denies that deliberate motive. Osepchuk p. 1204 does not resolve the specific peak-avoidance intention in either direction. This is a metadata inconsistency and an unsupported intent gloss, not evidence selecting one causal history as true. Retain MINOR and list under tier C. X23 repeats the intentionality gloss in the verifier.
- [dated] X18: FIXED dateModified from 2026-06-30 to 2026-10-14 for this revision; retained datePublished as 2026-06-30. The explicit JSON-LD exception requires no Updated line because the four Corrected lines date this revision.
- [declined] X19: NOT FIXED as a separate wording change, per the directing triage. Re-read Osepchuk (1984), p. 1204, and the actual markdown summary at 90303bbcc1, lines 29-31. The latter repeats X10's accidental-penetration inference; the historical source does not establish that all penetration benefits were unintended. Retain MINOR and list under tier C. The heating corollaries elsewhere in this compound sentence are not refuted by this finding.
- [fixed] X20: FIXED the -10 C ice relaxation time in the page and research/why-microwaves-heat-food/verify.mjs, and regenerated its public/checks/research/why-microwaves-heat-food/verify.mjs mirror by byte copy. Both now evaluate the Auty-Cole rate law at 263.15 K. The model loss at 2.45 GHz changes from 0.0002981719 to 0.0001226454 and its power depth from about 116.84 m to 284.05 m; these are model outputs, not measured GHz ice data. Qualified the remaining ice constants as approximate model values. Added and mirrored one dated Corrected line, shared with the numerical issue behind X12.
- [fixed] X21: FIXED the button label by naming it as the 25 C water loss-peak preset. The fixed frequency and temperature control behavior are unchanged. Added and mirrored one dated Corrected line.
- [fixed] X22: FIXED the unconditional frequency-above-12-GHz branch. The verdict now compares the chosen frequency with the current temperature model peak, explicitly labels the within-10-percent interval, and labels the upper flank as above the model peak. Added and mirrored one dated Corrected line.
- [declined] X23: NOT FIXED as a separate wording change, per the directing triage. Re-read 90303bbcc1:research/why-microwaves-heat-food/verify.mjs lines 100-101: "atOven.epsI < atPeak.epsI" tests a loss comparison, while its PASS message adds "off-resonance on purpose". The inequality establishes no historical intention. Retain MINOR as an overclaim about what the check establishes and list under tier C. Share X17's intent error label; no accepted WRONG finding is in this assertion.

## Repair verification
- `node research/why-microwaves-heat-food/verify.mjs`: 15/15 passed before and after the parameter correction; the independent check's permissive ice assertions alone had not caught the wrong temperature parameter.
- `node public/checks/research/why-microwaves-heat-food/verify.mjs`: 15/15 passed. Its bytes match the research source.
- `node /tmp/why-microwaves-heat-food-assay-1014/probe.mjs public/strata/why-microwaves-heat-food/index.html --check`: 41 checks passed against the actual inline script with DOM and no-op canvas stubs. This checks the corrected ice rate law, peak classification on both sides at 0, 25 and 100 C, the 25 C preset label, phase switching, correction mirrors and metadata. It is not a browser layout or image-rendering test.
- `node scripts/check-claims-pass.mjs`: passed; the final completed run reported 266 layers with sound records.
- Scoped `git diff --check`: passed. Each of X1 through X23 has exactly one claim line and one finding; 60 N2 CONFIRMED lines are retained. New prose contains no U+2014.

## Not done
- The 18 listed MINOR items remain as directed; X12's separate citation wording is unchanged while its numerical issue is fixed under X20. There were no rejected triage findings.
- The shared TV narration at public/tv/lib/narration/batch-15.js:932 retains its accident/design wording, in the listed historical MINOR class. No film, audio or image was rendered or re-encoded.
- The shared public/checks/manifest.json still records 7460 bytes for this verifier; the corrected file is 7607 bytes. That shared metadata was not edited under the explicit scope restriction.
- No full build, dependency installation, shared-file edit or git write command was run. The page is authored HTML with no layer build script; the public verifier mirror was copied from its corrected research source. Downloaded source responses totalled 359455 bytes and were removed after reading.