Why does sake specify iron at 0.02 ppm when whisky does not? Miyamizu, soft water, and the drink that dilutes twice

Key takeaways

• Sake’s iron limit is a hard number, and it is stricter than drinking water. Japan’s National Research Institute of Brewing states that “the iron concentration in water used for sake making must be less than 0.02 ppm, which is lower than water quality standards for tap water (0.3 ppm or less)” (NRIB).
• The reason is a pigment carbon cannot remove. Sake’s five colour faults are “pigments derived from the raw materials (flavins), iron-related coloration (ferrichrysin), melanoidin compounds… sun exposure, and copper” — and “except for ferrichrysin, all of these can be removed with relatively small amounts of activated carbon” (Sake Street).
• Sake is proofed down with water, exactly like a spirit. “The stored sake contains approximately 20% alcohol. Before bottling, it will be diluted with water to reduce the alcohol concentration to approximately 15%” (NRIB). The industry has a word for that water: wari-mizu.
• The famous water story is a controlled experiment, not folklore. In 1840 the Nada brewer Yamamura Tazaemon carried well water seven kilometres from Nishinomiya to his Uozaki brewery; the sake improved, and the water was named miyamizu, “an abbreviated form of Nishinomiya no mizu” (Nippon.com).
• Hard water is not “better” — it is a different process. Hiroshima’s Senzaburō Miura spent “more than 20 years” developing a soft-water brewing method, published in 1898, for water of “less than 3 degrees hard” that the trade had written off as unbrewable (Sake Hiroshima).

Sake specifies iron at 0.02 ppm because iron in sake is not a taste problem that can be corrected downstream — it is a colour problem that survives the one treatment brewers rely on to fix colour. Everything else that darkens sake can be pulled out with activated carbon; the iron-derived pigment ferrichrysin largely cannot. Whisky does not carry the same published number because whisky’s chemistry and its treatment options differ, not because its water matters less — and when you look at where sake actually uses water, it uses it in exactly the two places a distillery does.

How much of a bottle of sake is water?

About four-fifths of it, before you count anything the brewery poured down a drain.

Japan’s National Research Institute of Brewing — the government body that researches and examines the country’s brewing industries — puts it plainly: “Sake contains approximately 80% of water,” and “sake making needs approximately 10 times as much water as rice” (NRIB). Other industry sources put the total far higher, at “more than 30 times the weight of the rice used” (Sushiya) or “as much as 50 times the total weight of rice” (Sake Street).

That spread is not a contradiction. It is a boundary question — whether you are counting the water in the mash, or the washing, soaking, steaming, cleaning and equipment water too. It is the same accounting problem we set out in how many different waters a distillery uses. For specification purposes only two of those waters touch the product: the brewing water in the mash, and the water added at the end.

NRIB gives the mash ratio directly: “Normal sake is made using 130 L of water for 100 kg of rice.”

What does sake’s water specification actually say?

Two trace metals, both at 0.02 ppm, both for reasons that are visible rather than gustatory.

NRIB is unambiguous about the first: “Iron is the most unfavorable mineral in water. If sake contains too much iron, it has a reddish brown color, and its taste and aroma are unpleasant. The iron concentration in water used for sake making must be less than 0.02 ppm, which is lower than water quality standards for tap water (0.3 ppm or less).”

Note what that comparison does. It places the product specification an order of magnitude below the potability floor — the same relationship between “legally drinkable” and “fit for the product” that we describe in the regulatory standards for water used in spirits. Potable is a public-health threshold. It was never a quality specification.

Manganese sits alongside it. Good brewing regions have “water with less than 0.02 mg/l iron and manganese,” and manganese “promotes a chemical reaction that changes or discolors the appearance of a sake” on exposure to ultraviolet light — visible, in direct sunlight, “in less than three hours” (Sushiya). Both metals are on our own blending-water control list for the same family of reasons, set out in iron, copper and manganese.

Why 0.02 and not 0.2?

Because of what happens after the fault appears.

Activated carbon is the brewer’s standard tool for stripping unwanted colour and aroma. Sake Street’s survey of sake fining lists the five principal causes of coloration — raw-material flavins, iron-related ferrichrysin, storage melanoidins, sun exposure and copper — and then delivers the operative sentence: “Except for ferrichrysin, all of these can be removed with relatively small amounts of activated carbon” (Sake Street).

Iron reaching the fermentation does not stay as iron. It binds to a colourless siderophore produced by the kōji mould and becomes a pigment with its own stability. The specification is tight because the correction is weak. That is the entire logic of a water specification compressed into one compound, and it generalises: every number on a certificate of analysis exists because the alternative — fixing it later — is either expensive, incomplete, or impossible. We walk through how those numbers are actually generated in how blending water is tested.

What is miyamizu, and what did it actually prove?

It is the groundwater of Nishinomiya — and in 1840 it became the subject of what is, in effect, the drinks industry’s first controlled water trial.

Yamamura Tazaemon ran two breweries: one in Nishinomiya, one in Uozaki. The Nishinomiya sake was consistently better. Rather than attribute it to the site, the rice or the brewers, he changed one variable — he carried the Nishinomiya well water over seven kilometres to Uozaki and brewed with it there. The Uozaki sake improved. The water was named miyamizu, “an abbreviated form of Nishinomiya no mizu” (Nippon.com).

The reported mechanism is the one a modern specification would predict. The water, “although rich in minerals, contained very little iron,” and sake made with it “kept its taste past summer” — the aki bare or “fine autumn” phenomenon that made Nada’s sake commercially dominant. By the late Edo period “around 80% of the sake consumed in Edo was from Nada,” and the Nada-Gogō districts still account for “just over one quarter of the sake production in the entire country” (Wikipedia).

Published analyses of miyamizu give potassium at around 20 mg/L and phosphate at 5.2 mg/L, at a hardness of about 8 °dH (Sushiya). Potassium, magnesium and phosphorus are yeast and kōji nutrients; iron is not required and is actively harmful. Miyamizu’s reputation rests on carrying a lot of the first group and almost none of the second.

One honest caveat. The published hardness figures for these waters do not agree. Miyamizu is given as “around 6 degrees hard” by one Japanese industry source (Sake Hiroshima) and 8 °dH by another, while a third expresses it as “around 100” on the mg/L CaCO₃ scale (Sake Street) — a range of roughly 107 to 143 mg/L once converted. Regional water figures drift with the aquifer, the season and the unit; spec sheet figures do not. If the two scales are unfamiliar, our guide to hardness units covers the conversions.

So is hard water simply better for brewing?

No — and the proof is that the softest of the three regions produced the style the modern market prizes most.

Nada’s mineral-rich water drives vigorous fermentation and a dry, firm result: otoko-zake, “men’s sake”. Fushimi’s softer gokōsui ferments more gently and yields onna-zake, “women’s sake”, with a finer, mellower profile (Sake Street). NRIB notes that Fushimi’s springs are designated among the 100 best water resources by Japan’s Ministry of the Environment.

Hiroshima had neither. Its water is “less than 3 degrees hard,” and soft water was held to be unsuitable for brewing — low fermentation power, spoilage risk, unstable quality. Senzaburō Miura worked at the problem for “more than 20 years” and published a soft-water brewing method in 1898: mature the kōji fully so its enzymes compensate for the missing mineral nutrition, then ferment slowly and cold to suppress wild organisms. Some accounts position that method as the origin of ginjō sake (Sake Hiroshima).

That is the whole argument against water romanticism in one case study. Three waters, three hardness levels, three successful styles — and the difference between them is a process built around the water, not a ranking of waters. It is the same conclusion we reached about Speyside, Islay and the myth of the whisky region and about the terroir of whisky water. Mineral content shapes fermentation, which is biology. It does not shape dilution, which is chemistry.

Where does sake’s blending water come in?

At the end, after pasteurisation, in a step with its own name — and this is the part that is directly comparable to proofing a spirit.

“The addition of water after pressing is called ‘wari-mizu’. In most cases, wari-mizu is added after filtration and hi-ire pasteurization” (Sake Street). The arithmetic: “sake immediately after pressing contains about 18–20% alcohol, from which water is watered down to about 15–16% alcohol by volume.” NRIB describes the same step as taking roughly 20% down to approximately 15%.

Two details are worth a bottler’s attention.

First, Japan’s tax authority permits a sake to be labelled genshu — undiluted — “as long as the alcohol content adjustment is less than 1%.” That tolerance exists partly to accommodate mizu-oshi, the “water pushing” operation that uses water to clear what remains in the hoses (Sake Street). Line water ends up in the product whether or not anyone specified it — the same blind spot we examine in bottle rinse water on the spirits bottling line.

Second, the stated purpose of wari-mizu is not only strength. It is “to adjust the balance of aroma and flavor,” because sake comes off the press concentrated. Any brewer or distiller who has watched a tasting panel open up a sample will recognise that — it is the same effect we discuss in why panels dilute to 20% ABV.

And here is the point that matters commercially. Fermentation water gets the famous name, the shrine, the 1840 experiment and the tourist signage. The water that makes up the last 20–25% of the bottle, added after every corrective process in the brewery has already finished, typically gets no name at all. By then the carbon treatment is done. Whatever the water brings in, the bottle keeps.

What should a spirits bottler take from sake?

Three transferable lessons, none of them Japanese.

1. Specify against the failure mode you cannot fix, not the one you can. Sake’s 0.02 ppm iron limit exists because ferrichrysin resists activated carbon. Work backwards from your own irreversible faults — haze, colour shift, metal-catalysed oxidation — and let those set your ceilings, as argued in can water be too pure?.
2. Separate your process water from your dilution water, conceptually and physically. Minerals that feed a fermentation are a liability in a proofing tank, where nothing downstream will consume them. That split is the foundation of what blending water is.
3. Potable is a floor, not a target. Sake sets iron fifteen times below the tap-water limit. Spirits regulations in the EU, the US and India largely defer to drinking-water law for added water — a floor most serious producers clear by a wide margin, per our demineralised water specification guide.

Glossary

• Miyamizu (宮水) — the mineral-rich groundwater of Nishinomiya, Hyōgo, used by the Nada-Gogō breweries; a contraction of Nishinomiya no mizu.
• Wari-mizu (割り水) — the water added to sake after pressing, filtration and pasteurisation to bring it from roughly 18–20% ABV down to 15–16%. Sake’s equivalent of blending water.
• Shikomi-mizu (仕込み水) — the brewing water used in the fermentation mash, as distinct from wari-mizu.
• Genshu (原酒) — undiluted sake. Japan’s tax authority permits the label where the alcohol adjustment is under 1%.
• Mizu-oshi (水押し) — “water pushing”: using water to clear residual product from transfer lines at the end of a run.
• Ferrichrysin — the iron-bound pigment formed when iron complexes with a kōji-derived siderophore; the one sake colour fault activated carbon does not readily remove.
• Otoko-zake / onna-zake — “men’s” and “women’s” sake: the dry, firm hard-water style of Nada and the mellow, delicate soft-water style of Fushimi.
• Aki bare (秋晴れ) — “fine autumn”: sake that improves rather than degrades over the summer following production.

Frequently asked questions

Is sake really 80% water? Yes. Japan’s National Research Institute of Brewing states that “sake contains approximately 80% of water” (NRIB). That is the finished liquid; total water consumed per unit of rice is far higher, with industry estimates ranging from roughly 10 times the rice weight for brewing up to 30–50 times once washing, cleaning and equipment water are counted.

Why is iron so tightly limited in sake brewing water? Because iron produces a reddish-brown discolouration that standard treatment cannot undo. Of the five main causes of sake coloration, “except for ferrichrysin, all of these can be removed with relatively small amounts of activated carbon” (Sake Street). NRIB sets the limit at “less than 0.02 ppm” against a Japanese tap-water standard of 0.3 ppm or less.

Does sake get diluted with water like whisky does? Yes, and it has a dedicated term for it. Wari-mizu is added after filtration and pasteurisation, taking sake from about 18–20% ABV to about 15–16% (Sake Street). Shochu uses warimizu the same way, as we cover in how Asia’s biggest spirits build a bottle out of water.

Is hard water better than soft water for brewing? Neither is better; they demand different processes. Nada’s harder miyamizu gives dry, firm otoko-zake, Fushimi’s softer gokōsui gives mellow onna-zake, and Hiroshima’s very soft water — “less than 3 degrees hard” — required Senzaburō Miura’s low-temperature soft-water method, published in 1898, which some accounts credit as the origin of ginjō sake (Sake Hiroshima).

Does a whisky distillery need a sake-grade iron limit? The failure mode differs, but the direction of travel is the same. Iron, copper and manganese are controlled in blending water because they catalyse oxidation, can drive colour and haze, and have no downstream removal step after the proofing tank — the reasoning set out in trace metals in blending water. Sake simply publishes its number.

Why does the water get a shrine and the dilution water gets nothing? Because the brewing water’s effect is dramatic and biological — it changes a fermentation — while the dilution water’s effect is silent and subtractive. It can only add faults. That asymmetry is why miyamizu has an origin story dated to 1840 and wari-mizu has a one-line definition, and why we argue the second deserves a specification at least as serious as the first.

Blenders International · Water Science series