What to boil, what to steep, and the rule that was never tested

There are twenty-three teas in this index. Nineteen of them tell you to simmer something. Thirteen of those nineteen do a very particular thing: they simmer the hard material first, then take the pot off the heat, and only then add the leaves and the flowers. Twelve say to keep the lid on.

Nobody wrote that rule into the recipes. It arrived by taste, one recipe at a time.

The European Medicines Agency did write it down. In a template it publishes to tell patients how to prepare a herbal tea, there is a line that reads: aromatic herbs like mint may be infused at the end of a decoction.¹ That is the deep sleep tea exactly: simmer the fennel, kill the heat, add the chamomile and lavender and mint.

Four operations, and one of them is not what you think

The same document defines all four, with times, and the definitions are more specific than the kitchen versions.

An infusion is made by pouring boiling water onto the herb and letting it steep, usually 5 to 15 minutes.¹ A macerate is made by soaking the herb in water at room temperature, usually for 30 minutes.¹ Between them sits a fourth method almost nobody names: when the soak is done with gentle heating above room temperature but short of boiling, the process is called digestion, and the template puts it at 30 to 50 degrees.¹

And a decoction starts cold. The herb goes into cold water, the whole thing is brought up to the boil together, and it simmers 15 to 30 minutes depending on the size of the material.¹ Most popular writing treats decoction as the same as an infusion with more heat, which is a real procedural difference thrown away. The template even suggests moistening tough plant material with cold water first, before any heat at all.¹

The traditional rule appears in that document as official guidance, and it reads: infusion is appropriate for leaves, flowers and delicate parts whereas decoction or maceration is appropriate for roots, rhizomes and barks.¹ Read that carefully. The official version puts decoction and maceration on the same side, both as answers to hard material. The folk version that treats cold soaking as the gentle opposite of boiling is a corruption of a rule that never opposed them.

The measurement I could not find

The rule's central claim is that hard, woody, lignified material will not give up its actives without boiling. It is stated everywhere. I went looking for the study that put it on a bench, comparing the same root prepared both ways and measuring what came out, and I did not find one. That does not mean it is wrong. It means a rule that governs how millions of people prepare medicine every day is resting on a great deal of practice and very little published measurement.

What the literature does contain is a set of experiments running the other way: cases where the method genuinely changes the outcome, and cases where it makes no difference at all. Those turn out to be more instructive than the rule.

Mistletoe, where it matters enormously

In 2011 a group led by Sebastian Jäger prepared mistletoe both ways and measured three things.² Syringin, a small phenylpropanoid glycoside, came out at 93 to 105 per cent by either route: the method was irrelevant to it. Flavonoids did mildly better with heat. And then there is mistletoe lectin, a large protein: 2.57 micrograms per millilitre in the cold macerate, and undetectable in the hot preparation.

The elegant part is what they did next. They took a finished cold macerate, one that already contained the lectin, and boiled it for sixty seconds. The lectin vanished.² That rules out a failure of extraction and leaves only thermal denaturation, which is what you would expect of a protein and would not expect of a small stable glycoside.

So the method matters exactly as much as the fragility of the thing you are after, and not at all otherwise. A rule stated at the level of roots and leaves is a rough proxy for a fact that lives at the level of molecules.

A leaf that does better boiled

Oregano is a leaf. The rule says infuse it. In 2014 a Portuguese group at Bragança prepared oregano three ways and measured the phenolics, and the decoction won: 98.05 milligrams per gram of extract against 92.40 for the infusion.³

That looks like a clean refutation until you read the rest of the table, and the honest reading is better than the slogan. The margin is six per cent. The phenolic acids did not move at all: 22.89 for the infusion, 22.80 for the decoction, statistically indistinguishable, and rosmarinic acid, the most abundant of them, sat at 15.91 against 15.42.³ The entire decoction advantage lives in the flavonoid glycosides, and mostly in one compound. On the antioxidant assays, two of the four showed no significant difference between steeping and boiling at all.³ The same group found the same direction in thyme a year later, though I have only their abstract and will not quote numbers from it.⁴

A leaf that rewards boiling, by six per cent, in some assays. The rule is not so much wrong as coarse.

What boiling does not destroy

The other half of the folklore says heat wrecks the good things, and at kitchen timescales this is mostly false. It is worth doing the arithmetic rather than repeating the warning.

Hibiscus anthocyanins, the pigment in a roselle tea, have a half-life at 100 degrees of roughly ninety minutes.⁵ Boil the pot hard for half an hour, far longer than anyone does, and you lose about a fifth of them. Ascorbic acid in rosehips has a half-life of 46 minutes at 95 degrees, so a ten-minute brew costs around fourteen per cent.⁶

The same rosehip study measured three vitamins side by side, and vitamin C was the most thermally stable of them, holding better than thiamine and better than riboflavin.⁶ Its reputation as the delicate one comes from its sensitivity to oxygen, to metal ions and to pH, and not from heat as such.

What does leave, and why the boiling is innocent

There is one class where the caution is entirely justified, and the mechanism is not the one people give.

In 2010 Tschiggerl and Bucar at Graz poured boiling water over 60 grams of rosemary and left it fifteen minutes. A normal infusion. It never boiled. They then distilled the infusion and compared it with the oil in the leaves, and recovered only 19.6 per cent of it.⁷ Four fifths of the essential oil was simply not in the cup.

The composition tells you where it went. Total terpene hydrocarbons fell from 23.30 per cent of the oil to 0.67. Trans-caryophyllene, alpha-humulene and terpinolene were not detected at all. Meanwhile 1,8-cineole, an oxide, sat unmoved at 41.58 against 42.43, and camphor's share nearly doubled from 17.0 to 31.4 per cent purely because the compounds around it had left.⁷

The authors attribute it to higher volatility and lower water solubility, noting boiling points around 160 degrees and log P values of about 4.4 to 4.6.⁷ The hydrocarbons are too oily to dissolve and too volatile to stay. It is partitioning and evaporation, not thermal destruction, and it happened without the water ever reaching a boil.

Which is the honest form of the advice: keep aromatics out of long heat because they leave, not because they break. The EMA states it as regulation rather than opinion, writing that decoction is generally not applicable to herbal substances containing volatile active constituents.¹

One admission. Twelve recipes in this index tell you to cover the pot, and I believe that is right, and I could not find a controlled study measuring what a lid actually retains. The mechanism is sound and the effect is unquantified, and I would rather say so than invent a number for it.

Cold water, which has a ceiling

Now the question this note started from. Cold brewing gives a rounder, sweeter cup, and the usual explanation is that cold water is simply slower, so you steep for twelve hours instead of three minutes and arrive in the same place. That is not what happens.

In a review for the Japan Agricultural Research Quarterly, Manami Monobe sets out the comparison against a standard 80 degree, two-minute brew. At around 10 degrees, EGCG reaches only about 20 per cent of the hot figure. Caffeine reaches about 50. EGC about 70. Free amino acids come out best of all, about 80 per cent even at half a degree above freezing. And then the line that matters: further extraction did not affect these concentrations.⁸

Cold water is not slow. It stops. Leaving the jug overnight does not close the gap, because the gap is not about time.

The reason is in the leaf rather than the water. Michael Spiro's group at Imperial College, the same laboratory that spent the 1990s working out what the film on a cup of tea is made of, measured the activation energy of caffeine extraction: 61.6 kilojoules per mole from green tea, 40.0 from black. In plain water, with no leaf in the way, the figure is 17.3. Caffeine diffuses through the swollen leaf between 96 and 132 times more slowly than it moves through water.⁹ What the heat is fighting is not the solvent. It is the leaf.

Why cold tastes sweeter, in two different ways

So is a cold brew sweeter because it holds more of the sweet things, or because it holds less of the bitter ones and the sweetness is merely unmasked? The literature gives two answers, and which one is right depends on where on the thermometer you are standing.

At the cold end, both effects run together. A 2025 study comparing 4 degrees with 80 found caffeine down by more than 40 per cent and L-theanine up by 9.2 per cent.¹⁰ That is an absolute gain in the sweet, umami-tasting amino acid alongside a large loss of the bitter alkaloid. Less of one and more of the other.

In the hot range the relationship inverts, which I did not expect. Working between 70 and 100 degrees, another 2025 group found that total free amino acids increased with brewing temperature while the umami and mellow taste scores fell, from 8.10 down to 5.60. Their sentence is worth having exactly: the concentration of free amino acids increased significantly with BT, but the umami/umami-mellow taste actually weakened.¹¹

More of the sweet compounds, less of the sweet taste. Composition and flavour came apart, because everything else came out too. No study I found isolates the unmasking cleanly enough to put a number on it, so the fair summary is that cold brewing wins on both counts at the cold end, and that above 70 degrees you can add sweet molecules to a cup and make it taste less sweet.

Cold brew coffee, where the story is simply wrong

The same enthusiasm has produced two claims about coffee that do not survive measurement, and both were tested by the same pair of chemists, Niny Rao and Megan Fuller.

The first is that cold brew is less acidic. At a matched ratio of 35 grams to 350 millilitres, they measured pH from 4.96 to 5.13 for cold brew and 4.85 to 5.10 for hot, and reported that varying the temperature of the extraction water did not result in distinguishable pH values.¹² Where a difference did appear it ran the other way, with hot brew carrying more titratable acid.¹² Their earlier paper adds the deeper problem: pH does not track perceived acidity anyway, so the popular claim is built on a measurement that would not predict the sensation even if it had come out as advertised.¹³

The second is that cold brew is lower in caffeine. At matched dose, cold brew came out equal or higher in all four of their pairings, reaching 1230 milligrams per litre against 970 for the hot equivalent.¹³ The claim is true for tea, where cold water genuinely leaves half the caffeine behind, and false for coffee. Most of the confusion comes from comparing a cold brew concentrate with a hot cup at a different dose ratio entirely.

Two edges

Marshmallow root is the classic cold-soak herb, and the EMA authorises only a macerate for the comminuted root.¹⁴ I went looking for the reason and read the 42-page assessment report that accompanies the monograph. It does not give one.¹⁵ Worse for the folklore, the one optimisation study of marshmallow polysaccharide extraction puts the optimum at 60.9 degrees, which is warm.¹⁶ A well-founded traditional practice, an official monograph, and no stated rationale anywhere in the file.

At the other edge, boiling is occasionally not optional. Raw kidney beans carry phytohaemagglutinin at 20,000 to 70,000 haemagglutinating units, which fully cooked beans bring down to 200 to 400, and the FDA specifically warns that slow cookers may reach only 75 degrees or less.¹⁷ A 1990 outbreak investigation in England traced cases to exactly that.¹⁸ There is an older report that 80 degrees leaves beans more toxic than raw, though a later group failed to reproduce it, so the safe reading is the one both agree on: get them properly boiling.¹⁹

Back to the pot

Set the library against the guidance and it holds up better than I expected. Thirteen of the twenty-three teas do the staged thing the EMA describes, hard material simmered, heat off, delicate material in afterwards. Six simmer throughout, and one of those, the spikenard and star anise, is root and spice alone with nothing delicate in it to protect, so it is right to. The other five contain flowers or herbs and could reasonably be staged.

One recipe in the whole index names a water temperature. The jasmine and lavender says to warm the water to about 80 degrees, because anything hotter scorches the jasmine. I wanted to tell you whether 80 is the correct number and I cannot: the tidy brewing-temperature charts that circulate are vendor pages, and the only published optimum I could verify is 70 degrees for a Longjing at four minutes, which is a different leaf and a different question.

So the honest state of it is this. Four operations, defined precisely by a regulator, distinguishing themselves mostly by how fragile your target molecule is and how badly it wants to leave the water. A rule of thumb that is coarse, occasionally wrong in the direction of boiling more, and never directly measured. And a great many cooks who arrived at the right procedure anyway, by putting the mint in last because putting the mint in first tasted worse.