Steaming: Water Level, Condensation, and Stacking

Steam never passes 100°C and still cooks faster than water

This is the heart of steaming and the part that feels wrong.

When steam meets a cold surface it condenses, releasing its latent heat all at once. The heat given off by one gram of steam turning to water is more than five times what one gram of water gives off cooling from 100°C to 60°C.

So at the same temperature, steam delivers heat far faster. And because the food is not submerged, water-soluble compounds stay in it. The difference between boiling and steaming is not temperature; it is those two things (dry heat and moist heat).

A third difference: the surface does not dissolve

For anything that has to stay as separate grains, this is the decisive one. Boiled, the starch leaving each grain's surface has somewhere to go — it dissolves into the surrounding water and cements the grains to one another. Steamed, the only water a grain gets is what condensed on it, so the surfaces never dissolve into each other.

Which is why Dambou in Niger steams its coarsely cracked grain in two stages, and why Attiéké in Côte d'Ivoire presses and sieves its fermented cassava and then steams rather than boils it. If the goal is grains that stay separate, boiling was never an option.

That turns on where the water ends up

The paragraph above gets the mechanism right — the starch leaving each grain has somewhere to go — and then treats boiling as one thing. It is not. In the pot most rice is cooked in, every drop of the water is absorbed, so whatever dissolved into it is still there when the lid comes off, which is why Cooking rice puts the rinse before the pot as the moment to take surface starch off. Boil the same grain in a large excess of water and pour that water away, and the starch leaves with it.

Biryani and the chelow under Kabab Koobideh do exactly that, and then steam anyway. The biryani parboils basmati "to about 70% in boiling water with the whole spices and salt" and drains it; the kebab's rice is boiled "in salted water to al dente, drain, then steam covered on low for 15 minutes to make chelow" — and chelow is the rice whose whole identity is grains that fall apart on the plate.

So the boil and the steam are not alternatives here, they are stages, and the paragraph above is what makes the second one necessary. The drain does after cooking what a rinse does before it, and the steam then finishes the grain with no free water left to put anything back. Biryani is the extreme version: inside a sealed dum the only liquid there is is what condenses, which is why it stops at 70%.

Where the rule holds exactly as written is where there is nothing to pour off. Dambou's cracked grain goes into the steamer merely damp — four tablespoons of water rubbed through by hand — and Attiéké's pressed granules would come off the sieve as paste if they ever met a pot. For those two boiling really was never an option, and both pages say so in their own words.

Absorption is not a losing position either; it moves the work earlier. Plov wants the same separate grains, rinses "until the water runs clear", soaks, spreads the rice over the zirvak and never stirs it — the starch came off in advance, and after that the only rule is not to rub the grains against each other.

If the water touches the food, you are not steaming

Basic, and half of all failures. The water sits 2-3cm deep below the basket and must not reach the food.

And the food goes in after the steam is up. Start it in cold water and the food spends a long stretch merely warm while the water comes to a boil: doughs slump in that window, and gyeranjjim goes bubbly.

On long steams the real hazard is running dry. For anything over twenty minutes, keep a kettle of boiling water to top up with. Cold water breaks the steam.

The other end of the range is worth knowing too. Starch takes up water and sets between 60 and 75°C, so poured thin it is done in two minutes under 100°C steam. Cheung Fun is that, and longer does not make it softer — it makes it tougher. A thick cake like Lo Bak Go takes close to an hour by the same logic: the time is set by thickness, not by temperature.

The drip off the lid is the commonest failure

Condensation collects on the underside of the lid and falls onto the food. Dumplings burst where it lands, steamed buns come out blotched, fish sits in a puddle.

Three fixes, most effective first.

• Stretch a cloth under the lid to catch it. It is also why Chinese steamers are bamboo — the wood absorbs water.
• Set the lid very slightly ajar so the water runs down the wall instead.
• Cover the food with parchment.

Unless the water coming back is the point

All three treat the condensate as a contaminant, and "most effective first" ranks them by how completely it is taken away. There is a whole family of dishes where taking it away is the failure, and the deciding question is not the lid at all — it is whether the food is above the water or in it.

In a basket the food sits above the water, so anything that falls on it is dilution it never asked for and a wet patch where it lands. In a covered pot the food is in the liquid, and condensate returning to the pot is the dish's own cooking water going home. Run the three fixes above on the second case and you are pouring the sauce down the side of the pan.

Two dishes here take that as far as it goes, and they do it by opposite means.

• Biryani seals the pot shut. Its page says outright that the dum "deliberately inverts that guide's commonest failure": the drip is the whole of the water the rice has left to finish on, which is why the rim is closed with dough and why lifting the lid to look ends the dish rather than interrupting it.
• Tagine seals nothing. The tall cone leaves vapour free to circulate and simply changes where it lands: the tip is the coolest surface over the pot, so condensate runs down the slope to the rim and back into the sauce at the edge instead of onto the meat. That is the second fix above — "set the lid ajar so the water runs down the wall" — turned into the shape of the vessel and made permanent. Which is why the liquid line barely falls over an hour and why the pot is not stirred.

Aloo Gobi is the same case without a drop of added water: "the lid is the whole cooking method", the liquid being what the cauliflower itself gave up. Dry heat and moist heat carries the rest of that family, the ones that seal the rim with dough or a leaf.

Steam has to travel through, so how tightly you pack is itself a variable

Steam applies no mechanical force. Unlike hot water it does not push its way in; it flows along whatever gaps exist and condenses on the surfaces it reaches. So density decides the cooking time.

Baekseolgi is the extreme case: the rice flour goes into the mould and is never pressed down. The gaps between the particles are the steam's route, and packed tight the outside sets while the middle stays powder. It is a cake that sets from loose flour rather than a dough, so there is no alternative path.

Jeungpyeon uses the same fact from the other end. Rice has no network to trap gas, and what lets it rise anyway is that the batter is loose enough to pour. Made stiff, it sets with a few big holes and nothing else.

So what goes into the basket, and how, is part of the method.

Stacking does not work the way people assume

"The top tier cooks less" is a common belief. Steam is the same temperature top and bottom, so there is very little difference in heat.

What actually differs is other things.

• The bottom tier catches more condensation — including what drips from the tier above.
• Aromas travel upward. Steam fish under rice cakes and the rice cakes will smell of fish.

So put strongly flavoured things on the bottom, not the top, and put whatever must stay dry above. And rather than swapping tiers midway, choose a combination that finishes together.

What stops sticking is a liner, not oil

Oiling the basket does not last — the steam washes it away.

That does not make oil useless, though; sometimes it is doing a different job. Cheung Fun pours a rice batter 1-2mm thick and sets it in two minutes, and the tray is lightly oiled. There the oil is not stopping anything sticking during the steam — two minutes is over before that could matter — it is there so a film that has fully set will release from smooth metal. Same oil, different purpose.

• Punctured parchment lets steam through and nothing sticks.
• Napa or cabbage leaves add aroma as well. Common under buuz and mandu.
• Pine needles — this is where the name songpyeon comes from. The aroma is the point; not sticking is a side effect (songpyeon).
• Thin slices of carrot for wrappers thin enough to stick even to parchment, like har gow.

Wrapping the food instead of lining the basket

A liner puts one layer between the food and the basket. Wrapping the food completely changes the conditions: no condensation reaches it at all, so the cloth under the lid becomes unnecessary, and nothing aromatic gets out either.

• Fish Amok folds banana leaf into a bowl. The leaf is the vessel.
• Ilish Paturi uses the leaf as a lid, because mustard's pungency is volatile and an open dish loses it.
• Ikan Pepes goes one step further: after steaming it goes onto coals still wrapped, which drives the surface moisture off and makes the scorching leaf part of the flavour. Reverse the order and the leaf is ash before the fish is cooked.

Banana leaf is passed briefly over a flame to soften before folding. Folded cold it splits along the grain.

The steamer you want depends on what you are steaming

• Bamboo — absorbs condensation. Best for dim sum (har gow, siu mai).
• Metal — condensation falls straight down. A cloth is effectively mandatory.
• A covered pan — fine for small amounts. A plate resting on three small bowls is a steamer.

When it is done — open at once, or leave it

This one splits.

• Dumplings: open immediately. Trapped steam condenses as it cools, wets the wrappers and glues them to the base. Even in pan-fried-then-steamed gyoza the lid comes off at the end to drive moisture away.
• Leavened dough: wait three minutes. Steamed buns and mantou collapse when the temperature drops abruptly.
• Gyeranjjim: heat off, lid on, two or three minutes. Residual heat sets the centre; opened at once, the middle is underdone.

A dish called "jjim" is not always steamed

A naming trap in Korean. Galbijjim is not steamed at all — it is simmered covered in a soy braising liquid, which is braising. Turning collagen into gelatin needs liquid and time, not steam (braising).

Telling the genuinely steamed dishes — mantu — apart from the ones that only carry the name saves you from choosing the wrong method.