Wrapping and Sealing: Why Dumplings Burst
A dumpling bursts because of water in the filling, almost every time. How to deal with the filling first, when to seal with water and when with egg, why the pleats are structural, and how the cooking method decides the thickness of the wrapper.
A dumpling bursts because of water in the filling
It looks like a sealing problem and it usually is not. Heat turns the water in the filling to steam, the volume jumps, and the pressure pushes out through the weakest part of the seam. Thicker wrappers and harder pinching change nothing while the source of the pressure is untouched.
So the order is: fix the filling, then seal it.
A filling can also take water in, and the pleats do not answer that
Everything above is water the filling already has — salted and squeezed out of the cabbage, wrung out of brik's tuna, chilled into jelly in xiaolongbao, left on the prawns in har gow. A filling built on a dried starch runs the other way. Glass noodles, oatmeal and rice take water in, through the casing, out of the pot the parcel is cooking in. Sundae writes it in the very step that cites this guide — "the noodles and vegetables take up water and swell, so a tightly packed casing always bursts" — and answers it with a number: fill it seven-tenths. Ojingeo-sundae says 60%, samgyetang and haggis two-thirds.
Slack cannot cover this, and sundae's own method shows why: five minutes in, once the surface has firmed, it is pricked all over with a skewer. Steam is a pressure with somewhere to go, and a longer seam buys time for it to get there. A swollen noodle is not a pressure, it is volume — it arrived to stay and nothing vents it. So the room has to be set aside before the seal, as a fraction of the casing. The thickness table and the pleats both come after that.
- Soaking first reduces it and does not remove it. Ojingeo-sundae soaks its glass noodles and still chops them small; samgyetang stuffs soaked glutinous rice and still stops at two-thirds, because starch takes up water and swells from 60°C (thickening a sauce) and the pot is where that finishes. Soaking is the step that gets as much of it done as possible in a bowl, where the volume is in front of you.
- The container can move too. Ojingeo-sundae: "the filling swells and the body shrinks, so a full one bursts." Squid contracts with heat, so the fraction there is tighter than the filling alone would ask for. A gut, a stomach and a bird's cavity are all generous by comparison.
- One parcel can want both rules at once. Ojingeo-sundae squeezes the tofu and the bean sprouts, citing the section above while it does, and in the same filling leaves the noodles and rice room to drink. Water out of one ingredient and into another, under one seam.
Vegetable fillings get salted and squeezed
Cabbage, napa, chives and courgette are over 90% water. Chop them and add them raw and all of that water arrives during cooking.
Toss with a teaspoon of salt, leave 15-20 minutes, then squeeze hard by hand. You are drawing the water out in advance by osmosis — the same mechanism as salting, used in the opposite direction. Skip it in mandu or gyoza and the filling is already sloppy while you are still wrapping.
Meat fillings go cold, and get stirred one way
A meat filling does the reverse: you want it holding water.
- Keep it cold. Fat melted by the warmth of your hands renders straight out in the pan.
- Stir in one direction for three or four minutes. Muscle protein (myosin) unwinds and links up into a network that traps juice. The bouncy filling in siu mai and har gow is that network.
- Fillings built around stock or gelatin go in chilled and set. Added as liquid they cannot be wrapped at all.
Nothing sticks to a sealing edge with oil or filling on it
Obvious, and still the commonest mistake. Leave 5-7mm clear when you place the filling. If it is already smeared, wipe it before sealing — water brushed on top of grease does nothing.
Water or egg wash
The wrapper decides the glue.
- Wheat and rice wrappers take water. Surface starch dissolves into paste. Mandu, gyoza, buuz, mantu. Too much water and it slides; a fingertip's worth is right.
- Pastry takes egg wash. Butter-rich dough will not bond with water. Egg protein coagulates as it bakes and glues the seam, and brushed on top it colours the crust too. Empanadas and samosas.
- Rice paper needs nothing. In gỏi cuốn the wetting is the adhesive, and the real risk is soaking it too long — pull it out while still slightly stiff and let it finish softening on the board.
When the wrapper supplies no glue
Each of those works because the glue comes out of the wrapper. Raw dough has loose starch at its surface for water to dissolve; raw pastry has protein for egg to bond to; a rice-paper sheet is dried starch, and taking water back is what makes it tacky. A flatbread that was baked before you bought it is none of them. Its starch was gelatinised and set inside a crumb that fat and gluten are holding, so water on the edge of a wheat tortilla only makes it soggy. Burrito puts it plainly — a flour tortilla takes none of the three. The join has to come from somewhere else, and the dishes built on that wrapper each name where.
- Heat, because heat is what that starch still answers to. Leche frita states the property in one line — "a starch gel is reversible, firm when cold and loose when hot" — and that is the half water cannot reach. So chimichanga goes into the oil seam-side down and stays there a full minute: "the cocktail stick is a backup; this minute is what actually holds it." A burrito does the same thing in a dry pan as its last step. The same page spends the other end of the gel too, refrigerating the rolls twenty to thirty minutes, because "only a cold, firm roll holds its shape in the oil."
- A paste mixed up separately. Chimichanga also brushes flour paste onto the last overlap before frying. That is the wheat-wrapper route above, made in a bowl: the starch a mandu wrapper hands over for nothing, because this one will not.
- The filling. Quesadilla has no seam at all — one disc folded — and its page says the recipe states it outright: the cheese is the glue. That moves the setting into the cooling rather than the cooking (dairy and heat), which is why the minute before it is cut is a set and not a rest.
Or the join is given up. Enchiladas are rolled and simply laid seam-side down, and what holds them shut is the dish, the sauce poured over and the bake. What that recipe protects instead is the fold, since a corn tortilla has no gluten and the same reversible gel is all it has — which is why each one is warmed in oil before it will roll at all (deep frying).
The pleats are structural
Pleating lengthens the seam and builds in slack, and that slack absorbs the expansion during cooking. It is why a drum-tight seal is the first one to split.
Gyoza's one-sided pleat does a second job: it flattens the base so there is a face to sit on the pan. The Korean habit of joining both ends of a mandu into a ring keeps the seam from concentrating at a single point as it rolls in the broth.
The cooking method sets the wrapper thickness
- Steamed — thinnest. Steam applies no mechanical force, so a thin wrapper survives. The translucent skin of har gow is the extreme case (steaming).
- Pan-fried — medium. The base sticks and crisps while the top cooks in steam, so it has to withstand both. Gyoza.
- Boiled or fried — thickest. Convection in boiling water and thermal shock in oil hit the seam directly. The heavy wrappers on pierogi and samosa are not a stylistic choice.
But the table is half the pressure, and the other half is this guide's first line
What the seam has to hold is the water in the filling turning to steam. The method decides how violently that happens; the filling decides whether there is anything to turn. Change the second term and the table inverts.
- Brik sits in the harshest row — deep-fried — and uses pastry thin enough to belong in the steamed one. The potato is cooled, the onion salted and squeezed, the tuna wrung out, so there is nothing left to flash into steam. Deep-frying sets out the pair it makes with samosa: ninety seconds at 180°C against nearly twenty minutes in barely-warm oil, because a samosa's filling still has to cook, which is why its thick wrapper, low oil and long time all travel together.
- Xiaolongbao is the opposite corner, and the table cannot explain it at all: the thinnest row, with a filling that is deliberately all liquid. The answer is the thing this guide already says further up — the stock goes in chilled and set. At the moment of wrapping there is no water, only jelly; by the time there is, the parcel is in a steamer, where steam applies no mechanical force. A line filed above as advice about handling the filling is what sets the thickness.
- Har gow is the control inside that same row. Any water left on the prawns "turns to liquid in the filling as it steams and bursts the wrapper from the inside — the usual way har gow fail", and oversteaming waterlogs the skin until it sags. The thinnest row in the table, and the filling still decides.
So read the table as a default, not a rule: it tells you the thickness a wet filling needs. Dry the filling and you can spend the saving on thinness; chicken Kyiv answers the same question the other way, building a crust thick enough to contain melted butter on purpose.
The list the method is chosen from has no dry oven in it
The section above changes the second term, the filling, and ends on: dry the filling and you can spend the saving on thinness. Empanadas do the reverse and get away with it — the filling is deliberately jugoso and the pastry is thin and butter-rich. What lets them is the row they are not in. Steam applies no mechanical force, boiling water and hot oil hit the seam directly, a pan splits the difference: every row sorts by what is outside the parcel, pushing on the join or wetting it. A dry oven does neither.
The pressure inside still rises and the seam is still the only thing holding it, but nothing outside is working on it, so the method stops setting the thickness. What sets it instead is how long the parcel has to hold while its inside is wet — and inside this one row that runs from fifteen minutes to six hours.
- Short: the filling arrives cooked, and the oven is baking pastry. An empanada is 15-20 minutes at 200°C over beef cooked in a pan and cooled, which is the samosa pair above read backwards — thin pastry, high heat and a short time travel together for the same reason a thick wrapper, low oil and a long time do. Karjalanpiirakka is the far end: a 1mm rye shell at the oven's top setting, with its rice porridge simmered forty minutes and cooled first, because, as its page puts it, "Fifteen minutes at 300°C is nowhere near enough to cook rice and exactly right to bake a 1mm shell."
- Long: the parcel is the pot. A Cornish pasty says in as many words that it contradicts this guide. Everything goes in raw: "In a pasty that water is the cooking medium, so it stays — and the pastry and the seam are built to withstand it instead." Pour stock in as well and it "boils over inside and splits the crimp". Kalakukko is six hours of the same thing, and names the state its crust is in — a pot made of dough, holding a fish's liquid in and then eaten along with it.
- And the failure is not a burst. With nothing outside pushing on the seam, what goes wrong is underneath. Beef Wellington names it "The enemy is water, and only water", and says what that looks like: "A wet base that never bakes is what failure looks like here". Aussie meat pie gives the geometry — a thin gravy "soaks into the shortcrust from the inside, and no oven dries a base from below" — which is why both cool their fillings and neither is judged at the seam (browning on why a wet surface goes nowhere).
So on this row the fix moves off the filling and onto the wrapper, and the two long dishes name two levers no other row needs. The flour: a pasty takes strong flour and lard, because "A soft pie pastry splits under fifty minutes of steam." A kalakukko has to be rye, whose pentosans let a crust sit wet for six hours without going to paste, where wheat "sets and cracks in the oven" (choosing flour). And where the seam sits: "Crimped over the top, the seam sits at the highest point and takes the weight and the pressure", so a pasty is crimped along the side instead.
Dry heat and moist heat owns the family those two belong to — a seal, low heat and the ingredients' own water — and observes that dishes in it "often seal the rim with dough or a leaf". There the vessel is equipment and the only question is whether it leaks. Here the vessel is the food, so it has to survive its own steam for hours and still be worth eating, which is what makes the flour a sealing decision. Kalakukko puts the consequence in one line: patch a crack the moment you see one, because "One split at hour two and all the liquid is gone; then the filling is baked fish and the crust is dry bread."
When leaving it open is correct
Siu mai is left open at the top. The filling has already been stirred into a mass so it needs no seal, and steam entering from above cooks the centre evenly. Forcing it closed gives you an undercooked middle.
Gimbap and onigiri also hold together on rice starch rather than a glue. Rubbing a few grains along the far edge of the nori is the entire mechanism.
When nothing is cooked after it is wrapped
Every mechanism above assumes the parcel is going on to be cooked. The burst is steam; the filling's water "arrives during cooking"; cold fat matters because it renders in the pan; egg wash glues because it coagulates as it bakes; the slack in the pleats absorbs the expansion; and the thickness table is a list of cooking methods. Take the heat away and none of it fires. Gỏi cuốn, gimbap, onigiri and spam musubi are assembled out of components that are already finished, and nothing happens to them afterwards.
What starts instead is slower and never stops. The wrapper is the thin dry part and the filling is the wet one, so water crosses from one to the other from the moment they touch, and the wrapper is always the loser. Gimbap states the material fact — nori "is a thin polysaccharide sheet and it goes limp quickly on contact with water" — and onigiri states the direction: "nori is thin and dry and pulls moisture out of the rice fast." Which makes the rice-starch glue just above the same process as the failure: the nori sticks because it takes the rice's water.
So sealing turns into a scheduling question, and there are three answers, each written down by a dish.
- Hand it over now. Gỏi cuốn keeps the rolls apart under a damp cloth and serves them at once. Gimbap is "a same-day food", and its page traces that to the vinegar it does not have — sushi drops the pH and helps the rice keep, and sesame oil does nothing of the sort.
- Keep the two apart until the last second. Onigiri: "add the nori just before eating for crispness; pack it separately for lunch." The parcel is assembled twice, and the second time is at the table.
- Let it happen on purpose. Spam musubi says to wrap the nori dry — "the nori bonds to the rice and a little of the glaze soaks in." There the nori is a tie rather than a crisp surface, so the exchange is what holds the block together.
Gimbap uses a barrier as well: sesame oil coats each grain, which slows the rice drying out and stops the roll welding itself to the knife.
And two of the instructions above invert. Roll it tight, where the pleats section builds slack in on purpose — there is no expansion to absorb here, and a loose roll takes in air and comes apart. And the filling does not have to be wrung out: gỏi cuốn squeezes nothing where chả giò squeezes everything, because frying the same wrapper is what turns the same water into steam.
Reviving leftovers names this physics moisture migration and files it as a diagnosis — something that happened to a cooked dish while it cooled, and a question of which kind of heat reverses it. Here nothing was cooked and nothing is being reversed. It begins at assembly, at room temperature, and the only controls are the order you put it together in and when you hand it over.
Freezing what you wrapped
You do not have to cook the whole batch.
Spread them on a tray not touching, freeze hard for an hour or two, then bag them. Frozen while touching, they tear the moment you pull them apart.
And do not thaw them. Straight from the freezer into the steamer or the boiling water, with two or three minutes added. Thawed, the wrapper absorbs water and will burst without fail (freezing and thawing).