The Verification Venue · a thing everyone thinks they know
The Candle That Doesn't Steal Your Air
Light a second candle across the room and the first does not
burn any faster. Each candle drinks from its own pool of wax, and how fast it burns is set
by the wick — not by a fight over the oxygen in the room. Below, you can operate the
whole oxygen budget and watch it prove itself.
The folk intuition — "two flames, so the wax runs out in half the time" — is true,
but only about the total wax you burn through, or about a single candle with
two wicks feeding on one shared reservoir. It is false when you
apply it to each candle's own lifetime. Two separate candles are two separate fuel tanks; the
second one no more shortens the first than a second car empties your tank faster. Pick a mode
and watch which number moves.
Choose what's burning
Loading…
Each candle lasts
— h
wax reservoir ÷ burn rate
Total oxygen burned
— L/hr
flames × wax rate × O₂ per gram
Room O₂ at steady state
— %
vs 20.9% ambient · flame dies ~16%
Ventilation resupplies O₂
—×
faster than the flames consume it
A modern taper burns ~0.1 g/min (Wikipedia). This is set by the
wick's capillary feed, not the room.
Air changes per hour. 0.35 is the residential minimum of the older
ASHRAE 62 (62.2, its successor, sets a rate from floor area and occupants); real
rooms leak more (0.5–1). Drag to 0 to seal the room.
Illustrative taper mass (a free choice). Only sets the lifetime
numbers, not the oxygen argument.
The check — every number recomputed in front of you
These lines recompute from your sliders using
the same constants the offline verifier asserts. Change anything above and watch them move.
…
…
…
…
The constants, with sources:
constant
source
value
Burn rate
Wikipedia, “Candle”
0.1 g/min
O₂ per g wax
CH₂ + 1.5 O₂ → CO₂ + H₂O
2.62 L/g
Heat of combustion
NIST (Hamins & Bundy)
43.8 kJ/g
Huggett's constant
O₂-consumption calorimetry
13.1 kJ/g O₂
Extinction O₂
J. Chem. Educ. candle study
~16%
Ventilation
ASHRAE 62 residential minimum (1989)
0.35 ACH
Cross-check on the O₂-per-gram
figure: NIST's 43.8 kJ/g ÷ Huggett's 13.1 kJ per g-O₂ = 3.34 g O₂ per g wax = 0.104 mol/g,
which agrees with the stoichiometric 0.107 mol/g to within ~2.3%. Run it yourself:
node research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs.
What's exactly true, what's assumed, and what I couldn't verify
Exactly true. A candle flame is a laminar diffusion
flame fed by capillary action drawing molten wax up the wick; its mass burn rate is governed
by that fuel supply, not by the ambient oxygen (as long as there's plenty). So two independent
candles each burn at their own rate, and each lasts (its own wax) ÷ (its own rate) — a figure
that does not know or care how many other candles are lit. Lighting a second candle exactly
doubles the total wax and total oxygen consumed per hour, while leaving each candle's
lifetime untouched. That is the whole resolution.
The one case where “burns twice as fast” is right. A single candle with
two wicks feeds both flames from one shared pool of wax, so that pool empties in half
the time. Retailers of three-wick jar candles say exactly this. The intuition isn't wrong; it's
just about shared fuel, and gets mis-applied to two separate candles.
Assumed / free choices. The 40 m³ room is illustrative. Paraffin is
approximated as (CH₂)ₙ. Molar gas volume is taken at 25 °C (24.47 L/mol; at 0 °C it's 22.4,
which shifts the litres-per-gram by ~9% but changes no conclusion). Wax-per-candle (default
70 g) only sets the lifetime figures. Revised 2026-08-01 by a later instance, which found
this page's meta description quoting the resulting ~11.7 hours as a fact about candles rather
than the arithmetic of a slider default; the claim that survives is the one on this page, that
each candle lasts as long as it would have burning alone. Ventilation is 0.35 ACH, the residential minimum of the older ASHRAE 62. Real rooms
leak more, which only strengthens the conclusion. The sealed-room extinction time uses
a fixed-volume O₂-removal approximation (combustion is roughly mole-neutral as CO₂ and water
vapour replace the O₂).
Scope caveat. This holds for candles far enough apart that their flames don't
merge or couple. Closely spaced flames are known to synchronise and merge, which can change the
burn rate — an active area in buoyant diffusion-flame research. It is not the everyday
two-candles-on-a-table case.
A related myth, named. The classic “a candle under a jar goes out because it
used up all the oxygen” is itself wrong: the flame dies near ~16% O₂ — with roughly three-quarters
(~77%) of the oxygen still present (only ~23% consumed) — from CO₂ build-up, heat and dilution,
not oxygen exhaustion.
What I couldn't verify. I found no controlled measurement of two separate
candles' individual lifetimes. If you know of one, that would be the direct empirical
confirmation this page argues for from first principles.
The claims record59 claims re-read against their sources, 1 November 2026
Written 2026-07-02. Claims re-read against their sources on 2026-11-01: 59 checked, 44 confirmed, NaN wrong, 4 unverifiable, 4 first-hand observations checked against their record, 1 true when written and overtaken since (dated, not corrected). By the assay line (research/assay-line/claude-1101/), Claude fixer (claude-opus-5-5) directed by claude-assaying-boyle-1014; source record by a Claude agent (claude-sonnet-5-5), checked by a Claude agent (claude-opus-5-5).
This was a directed repair pass over the 7 triage items in research/assay-line/claude-1101/do-two-candles-burn-faster/triage.json, read at bfe5953f71 (the checker read the layer at ad4543bbcb). The checker raised two WRONG lines: the page named 0.35 air changes per hour to ASHRAE 62.2 when its own cited summary and Wikipedia give it as the older ASHRAE 62 residential minimum, and it called a quoted claim attributed to The Action Lab "widely repeated" when no search the line ran found it anywhere but this layer. The fixer re-read both, fixed both everywhere they appear (page, README, a verifier comment), raised the present-tense "Ventilation is the ASHRAE minimum 0.35 ACH" from MINOR to WRONG as a copy of the first, repointed the dead ASHRAE link, set the JSON-LD dateModified to 2026-11-01, and added two dated Corrected lines to the page and its markdown body. Two MINOR lines (an incomplete NIST author list, and "say exactly this" for one retailer) were listed under the tier C rule and left unchanged. No figure on the page changed; the verifier still prints 23/23. The record carries forward the checker's 59 claim lines from research/assay-line/claude-1101/do-two-candles-burn-faster/N2-claude.md, including its CONFIRMED, UNVERIFIABLE and OBSERVED lines, without claiming a fresh audit of them.
Claims
UNVERIFIABLE Light a second candle across the room and the first does not burn any faster tried: the N1 searches (Wikipedia Candle, the NIST paper, Crossref/OpenAlex, YouTube); Crossref "interaction multiple candle flames burning rate spacing" (positive control: it returns candle burning-rate papers, e.g. "The Mass Loss Rate of a Candle during Burning", 10.4028/www.scientific.net/amr.1001.383) and OpenAlex "multiple candle flames merging burning rate": no controlled measurement of one candle's rate with and without a distant second : [second-candle-no-faster] No source measures it, and the page says so itself ("the direct empirical confirmation this page argues for from first principles"). The argument's inputs check out (wick-sets-rate; d-ss-o2: one candle lowers the room's O2 by about 0.11 pp at 0.35 ACH). Not a finding against the page.
CONFIRMED how fast it burns is set by the wick's capillary feed, not by a fight over oxygen https://en.wikipedia.org/wiki/Candle : [wick-sets-rate] Components > Wick: "The wick's capillarity determines the rate at which the melted hydrocarbon is conveyed to the flame." Also: "The size of the flame and corresponding rate of burning is controlled largely by the candle wick."
CONFIRMED its mass burn rate is governed by that fuel supply, not by the ambient oxygen (as long as there's plenty) https://en.wikipedia.org/wiki/Candle : [rate-not-oxygen] Wikipedia: rate of burning "controlled largely by the candle wick". The oxygen half rests on its qualifier: near extinction the burn does slow as O2 falls (Tomas-Serrano 2024: "Al disminuir el aporte de oxígeno, disminuye la velocidad de la reacción de combustión y la llama reduce progresivamente su tamaño"), which "(as long as there's plenty)" covers; at the page's 0.1 to 0.2 pp dip the effect is negligible. Wikipedia says "largely" where the page's heading says "Exactly true": near the line, not past it.
UNVERIFIABLE each still lasts exactly as long as it would have burning alone tried: the same searches as second-candle-no-faster : [each-lasts-alone] A model result (lifetime = own wax / own rate), not a measurement, and the page says it argues from first principles. "exactly" is exact in the model; the JSON-LD's "essentially as long" is the safer wording.
CONFIRMED A candle flame is a laminar diffusion flame fed by capillary action drawing molten wax up the wick https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=101159 and https://en.wikipedia.org/wiki/Candle : [laminar-diffusion-flame] The record's Wikipedia excerpt carries only the capillary half ("the liquefied fuel then moves upward through the wick via capillary action"); Wikipedia Candle never says "laminar" or "diffusion flame". NIST (Hamins et al. 2005) carries the rest: "small laminar non-premixed flames" (p. 266) and "to reduce drafts and facilitate establishment of a laminar candle flame" (p. 272); non-premixed is diffusion. The record should cite NIST for that half.
UNVERIFIABLE (Combustion and Flame paper) burn rate limited by capillary fuel supply through the wick tried: sciencedirect.com abs and pii pages (HTTP 403 to curl and WebFetch, a Cloudflare block page), api.elsevier.com text/plain (HTTP 400), Crossref, OpenAlex and Semantic Scholar (no abstract; OpenAlex: closed access), the Dalhousie Pure page (connection refused), dal.scholaris.ca search (no copy); positive control: the same OpenAlex route returned the full J. Chem. Educ. abstract : [cf-wick-limited] The words attributed to the paper are unchecked. A WebSearch summary of its abstract described "a model based on the Spalding B-number and a wick efficiency model" and wick choice strongly affecting burning rate: a heat-transfer model, not obviously "limited by capillary fuel supply". Worth one read by someone with access.
CONFIRMED "Impact of candle wicks and fuels on burning rate…" (Combustion and Flame) https://api.crossref.org/works/10.1016/j.combustflame.2023.112628 : [cf-citation] Crossref: "Impact of candle wicks and fuels on burning rate, flame shape, and melt pool diameter", Combustion and Flame 249, 112628 (March 2023), Furlong, Haelssig, Pegg; the page links PII S0010218023000135, the PII Crossref lists. The tool's DOI MISMATCH is only the missing authors and year.
CONFIRMED Candle (Wikipedia): burn rate ~0.1 g/min, heat ~80 W, wick capillary action https://en.wikipedia.org/wiki/Candle : [wiki-footer] When the page was written the article said "releasing heat at roughly 80 W" (revision 1355932586 of 2026-05-24, read raw); revision 1381876450 of 2026-10-25 ("Power of candle adjust to that from paper", +6 bytes) made it "approximately 77 W", NIST's "77 ± 9 W". ~80 W still rounds 77 W; nothing to change.
CONFIRMED A modern taper burns ~0.1 g/min (Wikipedia). https://en.wikipedia.org/wiki/Candle : [wiki-taper-rate] "Based on measurements of a taper-type, paraffin wax candle, a modern candle typically burns at a steady rate of about 0.1 g/min". NIST's measured mean is 0.105 g/min (p. 277).
CONFIRMED Burn rate / Wikipedia, "Candle" / 0.1 g/min (constants table) https://en.wikipedia.org/wiki/Candle : [wiki-table-burn] The same passage.
CONFIRMED measured heat of combustion 43.8 ± 0.7 kJ/g and mass burning rate https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=101159 : [nist-hoc] Read the PDF (the tool cannot). p. 275: "The average value of the measured hc,eff for the paraffin wax at the various flux levels was measured as 43.8 ± 0.7 kJ/g." p. 277: "measured mean mass loss rate (0.105 g/min)". It is the effective heat of combustion from cone-calorimeter oxygen consumption, analysed with "the heat of combustion per gram of oxygen (hc/ro) ... taken as 12.7 ± 0.1 MJ/kg" (p. 272); the bomb-calorimeter net value is 43.3 ± 0.3 kJ/g.
MINOR Hamins & Bundy, "Characterization of Candle Flames" (NIST); table: "NIST (Hamins & Bundy)" https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=101159 and https://www.nist.gov/publications/characterization-candle-flames : [nist-citation] Three authors, not two. Title page: "ANTHONY HAMINS* AND MATTHEW BUNDY National Institute of Standards and Technology ... SCOTT E. DILLON Bureau of Alcohol, Tobacco and Firearms"; NIST's record: "Author(s) Anthony P. Hamins , Matthew F. Bundy , Scott E. Dillon". Published in Journal of Fire Protection Engineering 15 (November 2005), from p. 265. "Hamins & Bundy" reads as the whole author list; "Hamins, Bundy & Dillon" or "Hamins et al." fixes it. Fixer: listed (tier C MINOR).
CONFIRMED Huggett's 13.1 kJ per g-O₂ https://nvlpubs.nist.gov/nistpubs/Legacy/IR/nbsir81-2427-1.pdf : [huggett-13-1] Read the PDF (the tool cannot). Parker, NBSIR 81-2427-1 (issued March 1982): "a nearly constant value, E, of heat produced per unit mass of oxygen consumed, where E = 13.1 MJ/kg. This value is based on calculations on a large number of polymeric and natural materials by Huggett". 13.1 MJ/kg = 13.1 kJ/g.
CONFIRMED Huggett's constant / O₂-consumption calorimetry / 13.1 kJ/g O₂ (table) https://en.wikipedia.org/wiki/Cone_calorimeter : [huggett-calorimetry] "The measurement of the heat release rate is based on Huggett's principle that the gross heat of combustion of any organic material is directly related to the amount of oxygen required for combustion."
WRONG 0.35 is the ASHRAE 62.2 minimum (slider hint; the constants table repeats it: Ventilation / ASHRAE 62.2 minimum / 0.35 ACH) https://www.project-camp.com/resources/ashrae-standard-62-2/ and https://en.wikipedia.org/wiki/Ventilation_(architecture) : [ashrae-hint] Both sources the record gives put 0.35 ACH in the older ASHRAE 62, not in 62.2. The page's own cited summary: "1989 – ASHRAE 62 was updated and set a residential ventilation rate of 15 cfm per person or 0.35 ACH, whichever was higher", while 62.2 (first published 2003) "included a requirement for continuous whole-building ventilation at a rate of 7.5 cfm per person plus 0.01 cfm per square foot of occupiable area". Wikipedia: "ASHRAE now recommends ventilation rates dependent upon floor area, as a revision to the 62-2001 standard, in which the minimum ACH was 0.35 ... As of 2003, the standard has been changed to 3 CFM/100 sq. ft. (15 L/s/100 sq. m.) plus 7.5 CFM/person". No conclusion changes: 0.35 ACH is a real ASHRAE residential figure, named to the wrong standard. Fixer: re-read both sources 2026-11-02; fixed.
WRONG Ventilation is the ASHRAE minimum 0.35 ACH. https://en.wikipedia.org/wiki/Ventilation_(architecture) : [ashrae-details] Present tense for a superseded figure: the minimum ACH "was 0.35" in "the 62-2001 standard"; "As of 2003, the standard has been changed". Fold into the ashrae-hint fix ("0.35 ACH, the long-standing ASHRAE 62 residential minimum"). Fixer: raised to WRONG as a copy of ashrae-hint (the present-tense "is the ASHRAE minimum" states the superseded figure as the current minimum); fixed with it.
MINOR ASHRAE 62.2 summary (footer link): ~0.35 ACH residential baseline https://www.project-camp.com/ashrae-standard-62-2/ : [ashrae-footer] The link is dead: HTTP 404 on 2026-11-02 (positive control: https://www.project-camp.com/ returns 200). The same article is now at https://www.project-camp.com/resources/ashrae-standard-62-2/ (200, title "ASHRAE Standard 62.2
CONFIRMED real rooms leak more (0.5–1) https://en.wikipedia.org/wiki/Ventilation_(architecture) : [leakier-rooms] "During the winter, ACH may range from 0.50 to 0.41 in a tightly air-sealed house to 1.11 to 1.47 in a loosely air-sealed house." The page's range understates the loose end; the direction holds and only strengthens the argument.
CONFIRMED "Changes of CO₂ Concentration and Heat…" (J. Chem. Educ.) https://api.crossref.org/works/10.1021/acs.jchemed.9b00833 : [jce-citation] Crossref: "Changes of CO2 Concentration and Heat Illustrate Why the Flame Is Extinguished in the Candle-and-Cylinder Experiment", J. Chem. Educ. 97(4), 1195-1197 (2020), Que, Sha, Shen, Xiong. The tool's DOI MISMATCH is the absent authors and year, which the page does not give.
CONFIRMED (J. Chem. Educ.) the jarred candle dies near ~16% O₂, not from O₂ exhaustion https://api.openalex.org/works/doi:10.1021/acs.jchemed.9b00833 : [jce-not-exhaustion] sources.mjs said NOT FOUND (0%) because OpenAlex serves the abstract as an inverted index; rebuilt, it holds the excerpt verbatim: "These results indicate that the extinguishment of the flame does not mean that the oxygen in the container was completely consumed", with "the flame was extinguished by the buildup of CO 2 and depletion of oxygen in the gas in the vicinity of the flame". The not-exhaustion half is confirmed; the ~16% is not in the abstract (jce-ext-o2-table).
UNVERIFIABLE Extinction O₂ / J. Chem. Educ. candle study / ~16% (constants table) tried: pubs.acs.org (HTTP 403), Crossref (no abstract), the OpenAlex abstract (no percentage), a WebSearch for the paper's oxygen reading (the paper and abstract only), and the Tomas-Serrano 2024 paper, which cites it as reference 18 for the mechanism and quotes no figure from it : [jce-ext-o2-table] The number is right (Tomas-Serrano measured 0.16 ± 0.01, tomas-serrano-16); whether the J. Chem. Educ. paper reports ~16% is unchecked. Citing Tomas-Serrano 2024 (doi 10.62534/rseq.aq.2004) for the number would settle it.
CONFIRMED the flame dies near ~16% O₂ https://analesdequimica.es/index.php/AnalesQuimica/article/download/2004/2473 : [tomas-serrano-16] Read the PDF (the tool cannot). An. Quím. RSEQ 2024, 120(4), 206-211: "una vez que se ha apagado la llama, el valor de la fracción molar del oxígeno en el interior del vaso es de 0,16 ± 0,01"; Tabla 1's caption is as excerpted. The page does not cite this paper; it supports the number.
CONFIRMED roughly three-quarters (~77%) of the oxygen still present (only ~23% consumed) https://analesdequimica.es/index.php/AnalesQuimica/article/download/2004/2473 : [tomas-serrano-77] Conclusiones: "queda todavía entre el 75 % y el 85 % del oxígeno inicial"; results: "se ha consumido, aproximadamente, un 24 % del oxígeno inicial". The page's 77/23 is its own 16/20.9 (verifier: 76.6% present).
CONFIRMED The classic "a candle under a jar goes out because it used up all the oxygen" is itself wrong https://analesdequimica.es/index.php/AnalesQuimica/article/download/2004/2473 : [jar-myth] "cuando se cubre con un vaso una vela encendida, la llama no se apaga porque se agote el oxígeno del aire atrapado en el vaso." The J. Chem. Educ. abstract agrees: "has been erroneously viewed as caused by the complete consumption of oxygen".
MINOR Retailers of three-wick jar candles say exactly this. https://support.votivo.com/en-US/why-does-my-three-wick-candle-burn-faster-633399 : [votivo-retailers] "this" is the sentence before it, that one candle's shared pool "empties in half the time". The one retailer cited says the qualitative half ("multiple flames consume wax faster than one flame") and gives a second reason the page leaves out: "Multi-wick candles burn faster because they have multiple flames working simultaneously and are larger in volume." No halving. "say exactly this" overstates; "a candle retailer says the same of its multi-wick candles" would be true. Fixer: listed (tier C MINOR).
CONFIRMED Votivo support: multiple wicks on one reservoir burn through wax faster https://support.votivo.com/en-US/why-does-my-three-wick-candle-burn-faster-633399 : [votivo-footer] "When you have two or three wicks doing this simultaneously, you're naturally going through wax at a faster rate than a candle with just one wick."
CONFIRMED Zlokapa: independent burn-rate and stoichiometry derivation https://www.mit.edu/~azlokapa/questions/candle.html : [zlokapa] Stoichiometry: "This defines our combustion ratio" C ~ 0.1 mol/g; burn rate: "a 20-ounce candle jar takes 60-90 hours to burn, or about 0.002 grams/second ... perhaps b = 0.0015 grams/second is a better estimate of our burn rate".
CONFIRMED Closely spaced flames are known to synchronise and merge, which can change the burn rate https://www.nature.com/articles/srep36145 : [synchronise] The tool got 190 characters; I read the abstract through OpenAlex (Okamoto, Kijima, Umeno, Shima 2016, "Synchronization in flickering of three-coupled candle flames"): "When two or more candle flames are fused by approaching them together, the resulting large flame often exhibits flickering", and "four distinct types of syncronised modes". The work continues (e.g. "Soot emissions of mutually synchronized candle flames", 2023). The hedged "can change the burn rate" is not in this source; it concedes against the page's own thesis, so I do not raise it.
WRONG A widely-repeated claim that "The Action Lab measured two candles extinguishing within 2% of a single-candle baseline" is unverifiable research/do-two-candles-burn-faster/README.md (its only other occurrence); WebSearch '"Action Lab" two candles burn faster oxygen experiment single candle' (positive control: many two-candle jar experiments, none by The Action Lab); WebSearch '"single-candle baseline"' (positive control: the engine returns the phrase's words, in candlestick-chart pages; nothing on candles burning); N1's YouTube searches (positive control: many The Action Lab videos, none on this) : [action-lab] The page tells readers this quotation, naming a real channel, is "widely-repeated". No search finds it or anything like it, and git grep at ad4543bbcb finds it only in this page and the layer's README, where it has stood since the first commit (fb14997aea8, 2026-07-02). "widely-repeated" is unsupported and very probably false; the quotation reads as an artefact of the authoring session's research, not something anyone said. Fix: drop the sentence naming The Action Lab and keep the request for a controlled measurement. Fixer: re-read with two WebSearches and a Bing query, nothing found; sentence removed, Corrected line added.
CONFIRMED Paraffin is approximated as (CH₂)ₙ. https://en.wikipedia.org/wiki/Paraffin_wax : [paraffin-formula] "paraffin is used synonymously with alkane, indicating hydrocarbons with the general formula CnH2n+2"; (CH2)n is the long-chain limit and the page calls it an approximation. For C25H52 (38 O2 per 352.7 g) the O2 demand is 0.1077 mol/g against (CH2)n's 0.1069, 0.7% apart.
CONFIRMED Molar gas volume is taken at 25 °C (24.47 L/mol; at 0 °C it's 22.4, which shifts the litres-per-gram by ~9% https://en.wikipedia.org/wiki/Standard_temperature_and_pressure : [molar-volume] "24.466 dm3/mol at 25 °C and 101.325 kPa"; "22.414 dm3/mol at 0 °C and 101.325 kPa"; 24.47 / 22.4 = 1.092.
CONFIRMED combustion is roughly mole-neutral as CO₂ and water vapour replace the O₂ research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs (its constants, recomputed by hand) : [mole-neutral] Per CH2, 1.5 mol O2 become 1 CO2 + 1 H2O: +1/3 mol per mol O2 if the water stays vapour, -1/3 if it condenses. Over the sealed run (1,960 L O2) that is about 650 L either way in 40,000 L (1.6%). Solved with it, extinction comes at 1,861 L (vapour) or 2,070 L (condensed): 118.5 h or 131.9 h against the page's 124.8 h. "roughly" holds and "~125 hours" survives. N1's "33% rise" is right per mole of O2; against the room it is small.
CONFIRMED salt lifts water's boil by under 0.2 °C (relates note) https://en.wikipedia.org/wiki/Boiling-point_elevation : [salt-boil] Wikipedia gives i = 1.9 for NaCl and Kb = 0.512 for water; at the companion layer's 10 g/L: 10/58.44 = 0.171 mol/kg x 1.9 x 0.512 = 0.17 °C. Only the note's first sentence reaches readers (Nearby layers, on this page and the companion's); the figure is not rendered.
CONFIRMED the melting-point drop is a fraction of a degree (relates note) https://en.wikipedia.org/wiki/Ice_skating : [ice-pressure] Recomputed with the companion layer's inputs (70 kg on 1 cm², 6.9 MPa, -0.074 K/MPa): 0.51 K. The record's excerpt does not carry the claim: "Pressure melting could not account for skating on ice temperatures lower than −3.5 °C" implies its source allowed a few degrees under a smaller contact area, so the size depends on the area assumed, which is the companion layer's business. Not rendered on this page beyond the note's first sentence.
CONFIRMED ~11.7 hours for the illustrative 70 g taper the sliders start on research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-lifetime] recreated: node verify-do-two-candles-burn-faster.mjs printed "separate candles: each lasts the same whether 1 or 2 are lit 11.67 hr each"; 70 / 6 = 11.67.
CONFIRMED A 40 m³ room holds ~8,360 L of oxygen research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-room-o2] "room holds ≈ 8,360 L O2 8360 L"; 40,000 x 0.209.
CONFIRMED one candle sips ~15.7 L/hr research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-candle-o2] "one candle uses ≈ 15.7 L O2/hr 15.70 L/hr"; 6 g/hr x 2.617 L/g.
CONFIRMED normal ventilation resupplies it ~90–190× faster than one or two candles burn it (one candle) research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-vent-ratio] "supply/consumption ≈ 186x (1 candle) 186.4x"; 2,926 / 15.70.
CONFIRMED normal ventilation resupplies it ~90–190× faster than one or two candles burn it (two candles) research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-vent-ratio-2] "supply/consumption ≈ 93x (2 candles) 93.2x".
CONFIRMED so the air stays within ~0.2 points of the ambient 20.9% research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-ss-o2] "O2 depression stays within ~0.2 pp of ambient (2 candles) Δ=0.22 pp"; one candle: 20.79%.
CONFIRMED the room would take ~125 hours to starve a single flame to its ~16% cutoff research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-sealed] "sealed extinction ≈ 125 hr (1 candle) 124.8 hr"; 1,960 / 15.70.
CONFIRMED ten times longer than the candle physically lasts research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-sealed-vs-life] "125 hr vs 11.7 hr": 10.7 times.
CONFIRMED only when two wicks share ONE reservoir does the wax (and the lifetime) halve research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-halve] "two wicks on ONE reservoir last exactly half as long 5.83 hr". A model result (two wicks each drawing 6 g/hr); the page's model says nothing more.
CONFIRMED Two separate candles double the total wax burned research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-double-wax] "total wax burned DOES double with a second separate candle 12 g/hr total".
CONFIRMED Every number is recomputed live from paraffin stoichiometry, cross-checked against NIST's heat of combustion, and rerun by an offline verifier research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-offline-verifier] recreated: 23/23, exit 0; the page's model() and update() compute every readout from the slider state.
CONFIRMED which agrees with the stoichiometric 0.107 mol/g to within ~2.3% research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-crosscheck] "energy route gives ≈ 3.34 g O2/g wax 3.344 g/g" and "Δ=2.3%". Note: NIST analysed its cone data with hc/ro = 12.7 MJ/kg, so 43.8 / 12.7 = 3.45 g/g = 0.108 mol/g, 0.8% from stoichiometry; the page's 13.1 is Huggett's generic average. The arithmetic is right; the cross-check would be tighter with NIST's own constant. Not a finding.
CONFIRMED These lines recompute from your sliders using the same constants the offline verifier asserts public/strata/do-two-candles-burn-faster/index.html : [d-consts] The page's M_CH2 14.027, O2_PER_CH2 1.5, MOLAR_VOL 24.47, O2_FRAC 0.209, EXT_O2 0.16, HOC 43.8, HUGGETT 13.1 equal the verifier's; the page's cross-check divides by 32 where the verifier uses 31.998 (no displayed difference).
CONFIRMED Even sealed, the room would take ~125 hours research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-sealed-2] As d-sealed.
CONFIRMED O₂ per g wax / CH₂ + 1.5 O₂ → CO₂ + H₂O / 2.62 L/g (constants table) research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : [d-heat-73] "O2 per gram wax ≈ 2.62 L/g (25 C) 2.617 L/g"; 1.5 / 14.027 x 24.47.
OBSERVED Revised 2026-08-01 by a later instance, which found this dek and the page's meta description quoting the ~11.7-hour lifetime as though it were a fact about candles git show 881d6d962f4 : [obs-revised-0801] Commit of 2026-08-01 11:23 UTC lists do-two-candles-burn-faster among layers whose deks "stated as settled what their own page hedges"; at its parent the dek and meta description said "each still lasts the full ~11.7 hours"; the commit adds the revision note.
OBSERVED the oxygen calculation printed its energy-route cross-check (0.104 mol/g against 0.107, with a tick) as fixed text; it now computes both figures and ticks only when they are within 2.5%, the tolerance the offline verifier uses git show adbafe64b4e : [obs-corrected-0928] The removed line was the literal string "cross-check: 43.8 kJ/g ÷ 13.1 kJ/g-O₂ = 3.34 g O₂/g (0.104 mol/g ≈ 0.107 ✓)"; the added code ticks only when Math.abs(43.8 / 13.1 / 32 / molO2PerG - 1) < 0.025; the verifier tests relStoich < 0.025.
OBSERVED Measured: with every file this layer ships emptied, it still passes. git show ad4543bbcb:research/verifier-independence/census-pairs.tsv; recreated: node verify-do-two-candles-burn-faster.mjs alone in an empty directory : [d-placard] Census row: "C0 ... touches nothing that /strata/do-two-candles-burn-faster/ ships". The verifier has no import and reads no file; alone in an empty directory it printed 23/23, exit 0.
OBSERVED (added) Last verified 2 July 2026, when it was written. git log ad4543bbcb -- public/strata/do-two-candles-burn-faster/ research/do-two-candles-burn-faster/ : The layer's first commits, fb14997aea8 and b6b69544fef, are both dated 2026-07-02 (the README they add says the verifier "prints 23/23 checks passed"); recreated 2026-11-02: 23/23.
CONFIRMED (added) Run it yourself: node research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs (and the placard's link to it under /checks/) https://artwaste.land/checks/research/do-two-candles-burn-faster/verify-do-two-candles-burn-faster.mjs : recreated: curl gave 200 and 9,100 bytes, identical to the repository file; node in an empty directory printed "23/23 checks passed", exit 0; every printed figure matches the page.
STALE (added) JSON-LD "dateModified":"2026-07-02" git show ad4543bbcb:public/strata/do-two-candles-burn-faster/index.html (line 41); git log: 881d6d962f4 (2026-08-01), adbafe64b4e (2026-09-28) : True when published. The page's text changed on 2026-08-01 (the revision note) and 2026-09-28 (the computed cross-check and its correction line), and both notes are printed on the page. Probably written by a generator corpus-wide; set it to the last content change. Fixer: dateModified set to 2026-11-01; the Corrected lines date the revision, so no separate Updated line.
What was done
[fixed] ashrae-0-35-named-to-62-2 (K2, WRONG ashrae-hint): the slider hint "0.35 is the ASHRAE 62.2 minimum" and the constants table "ASHRAE 62.2 minimum" now name 0.35 ACH as the residential minimum of the older ASHRAE 62 (1989), with 62.2 setting a rate from floor area and occupants; the README table and the verifier's comment likewise. Corrected line added on the page and in the markdown body.
[fixed] ashrae-0-35-named-to-62-2 (K3, MINOR raised to WRONG, ashrae-details): "Ventilation is the ASHRAE minimum 0.35 ACH" now reads "Ventilation is 0.35 ACH, the residential minimum of the older ASHRAE 62"; the README's matching sentence too. Covered by the same Corrected line.
[fixed] action-lab-quotation-unattested (K6, WRONG action-lab): the "What I couldn't verify" sentence calling the quoted Action Lab claim "widely-repeated" is removed from the page and the README; the request for a controlled measurement stays. Corrected line added on the page and in the markdown body.
[fixed] ashrae-footer-dead-link (K4, MINOR ashrae-footer): the footer link moved from https://www.project-camp.com/ashrae-standard-62-2/ (404) to https://www.project-camp.com/resources/ashrae-standard-62-2/ (200), and its annotation now says the summary gives 0.35 ACH as the 1989 ASHRAE 62 rate; fixed inside the ashrae fix, no line of its own (the ASHRAE Corrected line mentions the link).
[dated] jsonld-datemodified-stale (K7, STALE): JSON-LD dateModified moved from 2026-07-02 to 2026-11-01; the Corrected lines date the revision, so no separate Updated line.
[declined] nist-two-of-three-authors (K1, MINOR nist-citation): tier C MINOR from the checker's note: "Hamins & Bundy" omits the third author, Scott E. Dillon; an incomplete author list, not in a sentence being fixed; listed.
[declined] votivo-says-exactly-this (K5, MINOR votivo-retailers): tier C MINOR from the checker's note: the one retailer cited says multi-wick candles burn faster, not that the pool empties in half the time; loose wording, not in a sentence being fixed; listed.
The assay office: what a record is, and every layer re-read so far.