Corrections: What Almost Everyone Gets Wrong

This dive found more repeated errors than most. They are collected here with the evidence that overturns each one, and with an honest note on how confident that evidence makes it.

The corrections are the product. Anything below that is contested is labelled as contested rather than smoothed over.

1. Tamari is not “the gluten-free soy sauce”

The claim: Tamari is soy sauce made without wheat, for people avoiding gluten.

What the record shows: Tamari is a different product with a different origin — the liquid drawn off fermenting miso. The 1603 Vocabulario da Lingoa de Iapam gives it a separate headword from shoyu, defined by its origin: “Hum licor muito saboroso que se tira do Miso” — “a very tasty liquor that is drawn from miso.” Shoyu is a different entry with a different definition.

And practically: the Japanese standard defines tamari as having “little or no wheat” — not no wheat. Commercial tamari commonly contains around 10% wheat or less. Not all tamari is gluten-free.

Confidence: high. The 1603 entries are primary documents and are reproduced on the tamari page.

2. Dark soy is not simply “aged light soy”

The claim: Chinese dark soy sauce is light soy that spent longer in the tank.

What the record shows: the ingredient panel of a mainstream Hong Kong dark soy reads “Water, Soybean, Salt, Colour (E150a), Wheat Flour, Sugar, Preservative (E202), Flavour Enhancer (E631 & E627).” E150a — plain caramel colour — is the fourth ingredient, ahead of the wheat and the sugar.

Sources genuinely disagree here and the page says so. Wikipedia states dark soy “is also aged longer” and is “often sweetened by adding molasses”; a leading practical source attributes the character to added sugar. Extended ageing does darken soy sauce through Maillard chemistry, so the ageing claim is not baseless.

The defensible conclusion: some dark soys are aged longer, essentially all are also coloured and sweetened, and darkness is not evidence of age.

Confidence: high on the label, medium on the reconciliation. The photograph is on the Chinese page.

3. “Naturally brewed” is unregulated — in the United States only

The claim, as usually stated: “naturally brewed” is a meaningless marketing term.

Half right, and the missing half matters. In the US it carries no legal definition and there is no FDA standard of identity for soy sauce. But Japan, China and Korea all regulate precisely this distinction: JAS 本醸造 / 混合醸造 / 混合, with added amino acid liquid capped at 80% on a nitrogen-equivalent basis; China’s GB/T 18186 requiring 釀造醬油 or 配制醬油 on the bottle; Korea’s MFDS separating 양조간장 from 산분해간장 and 혼합간장.

⭐ The blanket claim that the term is unregulated is an American parochialism. Three of the four major producing countries mandate the disclosure.

Confidence: high.

4. Soy sauce is not gluten-free by default

The claim: soy sauce is made from soybeans, so it contains no wheat.

What the record shows: Japanese koikuchi — the default bottle — is roughly half wheat. Chinese soy sauce uses wheat flour. Japanese shiro is about 90% wheat. The only traditional wheat-free soy sauce is Korean hansik ganjang.

The subtler finding: fermentation degrades gluten to the point that brewed soy sauce often tests below the limit of quantitation, with coeliac-active peptides measured at 0.2–0.5 mg/kg. Yet the FDA final rule of 13 August 2020 holds that sandwich ELISA cannot be trusted on fermented foods, and judges gluten-free claims on records proving the inputs were gluten-free before fermentation.

🔴 “Tests clean” and “may be labelled gluten-free” were deliberately decoupled — because of an analytical limitation, not a bureaucratic one. Details on the gluten question.

Confidence: high.

5. Colour runs backwards to salt

The claim: pale soy sauce is the mild one.

What the record shows, using Umami Information Center figures:

• Usukuchi, the pale everyday sauce: 18–19% salt — the saltiest.
• Koikuchi, the dark default: 16–17%.
• Saishikomi, the darkest and most intense: 12–14% — the least salty by a wide margin.

⭐ Across the Japanese range, colour and salt are close to inversely related. The English mistranslation of usukuchi as “light soy sauce” actively causes this error.

Confidence: medium-high. The koikuchi and usukuchi figures are widely corroborated; the saishikomi and shiro figures rest on a single source and were not independently confirmed.

6. Korean ganjang is not a variant of Japanese shoyu

The claim: ganjang is Korea’s version of soy sauce.

What the record shows: traditional hansik ganjang uses no wheat and no Aspergillus koji. It is made from meju — boiled, pounded soybean blocks colonised by ambient wild moulds and Bacillus subtilis, then steeped in brine in onggi.

⭐⭐ And the structural point: after 50–60 days the liquid is drawn off and boiled to become ganjang, while the mashed solids become doenjang. One jar, two products, separated on one day. That is the same relationship the 1603 dictionary describes for tamari and miso — which makes Korean ganjang tamari’s true structural cousin, not Japanese koikuchi.

🔴 A practical corollary: the Korean bottle most households buy is jin ganjang, which is normally a blend containing acid-hydrolysed sauce.

Confidence: high. UNESCO inscribed jang-making on the Representative List on 3 December 2024.

7. “Aged” means almost nothing on a soy sauce label

The claim: an age statement indicates quality.

What the record shows: no jurisdiction regulates an age claim on soy sauce the way spirits ages are regulated. There is no minimum, no definition and no audit. Ageing also does not reduce salt — salt is fixed at the brine step — and does not improve a hydrolysate.

The honest proxy is nitrogen, printed in China as 氨基酸态氮 (≥0.80 g/100 mL for special grade) and compressed in Japan into 特級. See salt, time and aging.

Confidence: high.

8. Reduced-sodium soy sauce is not brewed differently

The claim: low-sodium soy sauce is fermented with less salt.

What the record shows: it is brewed identically and then has roughly 40% of its salt removed afterwards — about 590 mg sodium per tablespoon against 920 mg. A traditionally brewed low-salt soy sauce is not possible, because the 17–18% salt in the mash is what selects the microbiota.

Confidence: high.

9. 3-MCPD is a contaminant of a process, not a property of soy sauce

The claim, in its alarming form: soy sauce contains a carcinogen.

What the record shows: 3-MCPD and 1,3-DCP form during acid hydrolysis, from residual lipids reacting with chloride from the hydrochloric acid — identified as an acid-HVP by-product in 1978. The 2000 UK FSA survey found 3-MCPD in 32 of 100 samples, above 1 mg/kg in 16, with a worst case of 82.8 mg/kg.

And then regulation worked. Limits applied from 5 April 2002; by the 2002 follow-up survey about 6% exceeded, against nearly a quarter the year before.

🔴 Brewed soy sauce does not generate chloropropanols, because there is no hydrochloric acid in the process. The hazard tracks the production method, not the food. See brewed and chemical.

⚠ The toxicological thresholds are not settled: JECFA’s PMTDI is 4 µg/kg bw/day while EFSA’s TDI is 2.0 µg/kg bw — itself a 2017 revision upward from 0.8. A factor of five separates the expert bodies.

Confidence: high on the mechanism and the survey figures; deliberately hedged on the thresholds.

What could not be verified

Stated plainly, because a dive that lists only its wins is not trustworthy:

• The 1254 Yuasa origin story. The account of the monk Kakushin bringing Kinzanji miso from China and soy sauce being discovered as its run-off is presented as settled by Japan Heritage, the town of Yuasa and the breweries. No source consulted disputed it — but neither did any demonstrate it from thirteenth-century documents. What is solidly attested is that Yuasa was producing commercially by the mid-sixteenth century.
• The Kikkoman Institute journal article on the miso-to-soy-sauce derivation could not be read — the PDF returned unparseable binary. The miso-origin claim therefore rests on the 1603 dictionary plus secondary sources, which is still strong, but is not the primary industry account it could have been.
• The full Korean Food Code category list. A fifth category, enzyme-hydrolysed 효소분해간장, is referred to in places but could not be confirmed against the code itself.
• Saishikomi and shiro salt percentages come from one source.
• The resident microbiota of kioke barrels. Plausible and universally asserted by producers; no controlled study quantifying the contribution was located.
• The ketchup etymology. The derivation through Malay kicap from a Hokkien term is widely repeated; the exact source word is disputed and this dive did not adjudicate it.
• A specific 2007 Vietnamese 3-MCPD scandal was searched for and not confirmed. The Taiwanese figure used instead — 10 of 212 samples over the limit in 2002 — is documented.

---

In this dive: The story · How it is made · Tamari and shoyu · The Japanese types · Chinese soy sauces · Korean ganjang · Beyond the big three · Brewed and chemical · Salt, time and aging · The gluten question · Buying, storing and using · Corrections