A samosa the size of a two-storey house, about 8 m tall, would weigh roughly 80 tonnes, and in 150°C oil its potato centre would need about 35 days to heat through, because conduction time grows with the square of size. Long before that the outside would turn to charcoal, the inner layer of pastry would stay raw and gummy against a cold filling, and the walls, now a 10 cm slab of soft dough, would sag under a load 100 times larger than a normal samosa carries.
- 80 tonnes: an 8 m samosa, 100 times the height of a street one, weighs a million times more.
- About 35 days in 150°C oil for its centre to heat, because conduction time scales with size squared.
- Bi ≈ 540: hotter oil or more stirring barely helps; only a thinner heat path does.
- 0.2% oil instead of 18%, because oil enters through the surface and area grows slower than volume.
- 153.1 kg is the real Guinness record samosa, about 12 times a normal one's size each way.
A street samosa works because it is small. Sakhale and colleagues, in one of the few lab studies of the snack, rolled 20 g of maida dough to 1 mm, folded it into a pyramid and fried it at 150°C for 8 minutes. Every step of that recipe quietly depends on the heat path being a centimetre or two.
Here I scale the samosa by a factor of 100, keep the physics honest, and solve for:
- frying time;
- oil, in the pan and in the samosa;
- the raw middle and the burnt crust;
- whether it can stand up.
Then I design the version that could actually be built.
What are the assumptions?
I take an ordinary aloo samosa 8 cm tall and scale every length by \(k = 100\), giving an 8 m pyramid, about the height of a two-storey house. All the numbers below follow from that one choice and a few measured food properties.
| Symbol | Meaning | Value | Source |
|---|---|---|---|
| \(k\) | Scale factor | 100 (8 cm to 8 m) | Assumption |
| \(m_0\) | Mass of a normal samosa | 80 g | Assumption, typical street size |
| \(d_0\) | Pastry sheet thickness | 1.0 mm | Sakhale et al. 2011 |
| \(T_{oil}\), \(t_0\) | Frying temperature and time | 150°C, 8 min | Sakhale et al. 2011 |
| \(\alpha\) | Thermal diffusivity of potato | 1.49 × 10⁻⁷ m²/s | Moro Pisco and Salas-Valerio 2020 |
| \(h\) | Oil-to-food heat transfer coefficient | 90–200 W/m²·K | Şahin, Sastry and Bayindirli 1999 |
| \(L_0\) | Half-thickness of the filling | 1.5 cm | Assumption |
| \(\rho\), \(c\) | Density and specific heat of the filling | 1,070 kg/m³, 3.5 kJ/kg·K | Assumption, typical for cooked potato |
Mass scales with volume, so \( m = m_0 k^3 = 0.08 \times 10^6 = 80{,}000 \) kg. That is about the weight of 60 small cars, sitting in a kadai.
The biggest samosa anyone has actually fried was made in August 2017 at a mosque in East London. Twelve volunteers from the charity Muslim Aid spent 15 hours on it, using 100 kg of potatoes, 25 kg of onion, 15 kg of peas and 44 kg of flour. No commercial fryer could hold it, so they had a deep-fat fryer built specially.
Guinness adjudicator Pravin Patel set the rules: it had to be triangular, contain flour, potato, onion and peas, be fried, hold its shape and be eaten without waste. It weighed 153.1 kg, beating Bradford College's 110.8 kg from 2012, and the Salvation Army handed it out to homeless people in the area.
Guinness World Records: Muslim Aid makes world's largest samosa to feed the homeless (Aug 2017)Run the record through the same scaling. 153.1 kg is about 1,900 of our 80 g samosas, so its scale factor is \( k = (1914)^{1/3} \approx 12.4 \): roughly a 1 m samosa. By the \(k^2\) rule below, its centre needed about 150 times as long to heat as a street samosa's, around 13 hours at Fo = 0.2, which sits inside the 15 hours the team took from start to finish. The house is eight times bigger again.

Why does a bigger samosa take so much longer to fry?
Heat crawls into food by conduction, and conduction time grows with the square of the distance, so a samosa 100 times bigger needs 10,000 times longer for its middle to get hot. This is Fourier's law of heat conduction, the same rule that makes a whole pumpkin take hours in an oven when cubes take minutes.
Fourier's law says heat flows down a temperature gradient at a rate set by the material's conductivity. Combine it with energy conservation and you get the heat equation, whose natural clock is the Fourier number:
For the centre of a compact shape to get most of the way to the surface temperature, Fo needs to reach roughly 0.2 (a standard rule of thumb from transient-conduction charts; treat it as an estimate). The full working:
- Normal samosa. \( t = 0.2 \times (0.015)^2 / 1.49\times10^{-7} \approx 302 \) s, about 5 minutes. That fits inside the 8-minute fry Sakhale used, which is a good check that the model is sane.
- House samosa. The half-thickness becomes 1.5 m: \( t = 0.2 \times (1.5)^2 / 1.49\times10^{-7} \approx 3.0\times10^{6} \) s.
- Convert. \( 3.0\times10^6 / 86{,}400 \approx 35 \) days. Equivalently, \( 302\ \text{s} \times k^2 = 302 \times 10^4 \) s.
Why can't hotter oil or stronger stirring fix it? That is what the Biot number answers. It compares how fast heat arrives at the surface with how fast it can move inside:
With \(h\) = 200 W/m²·K, the normal samosa has Bi ≈ 5 (200 × 0.015/0.56), and the house samosa has Bi ≈ 540 (200 × 1.5/0.56).
When Bi is far above 1, the surface is already at oil temperature and the inside is the bottleneck. Doubling \(h\) changes almost nothing; only shortening \(L\) helps.
See the data as a table
| Hours at Fo = 0.2 (hours) | |
|---|---|
| 8 cm | 0.08 |
| 50 cm | 3.3 |
| 1 m | 13.1 |
| 2 m | 52.4 |
| 4 m | 210 |
| 8 m | 839 |
Source: Calculated with the potato diffusivity of Moro Pisco and Salas-Valerio (2020) · t = 0.2 L² / α with L = 1.5 cm × (height / 8 cm). An estimate: it ignores steam and convection in the filling.
The authors heated whole Yungay potatoes (equivalent diameter 7.65 cm) in a 70°C liquid bath, logged temperatures at several depths and fitted the heating curves to get a thermal diffusivity of 1.48 to 1.49 × 10⁻⁷ m²/s, close to textbook values of 1.23 to 1.70 × 10⁻⁷. A finite-difference model built on that number predicted the centre temperature within 1.4 to 7.8%. The limit: whole raw potato is not spiced, mashed and oily samosa filling, so I use it as a close stand-in, not an exact value.
Read the paper (free)How much oil would you need, and how much would soak in?
By the lab recipe's ratio you would need about 2.4 lakh litres of oil, and to actually submerge an 8 m pyramid closer to 9 lakh litres, yet the finished samosa would be only about 0.2% oil instead of 18%.
- Recipe ratio. Sakhale fried at 1 kg of samosa to 3 L of oil. For 80,000 kg that is 240,000 L, or 240 m³.
- Submerging it. A pyramid 8 m tall needs a tank roughly 10 m × 10 m × 9 m, about 900 m³ (my estimate). At an oil density of 920 kg/m³ that is about 830 tonnes of oil.
- Heating the oil. Taking oil's specific heat as about 2.0 kJ/kg·K (a typical value, my assumption):
\[ Q_{oil} = 830{,}000 \times 2000 \times (150 - 30) \approx 2.0\times10^{11}\ \text{J} \approx 55\ \text{MWh} \]
- Heating the samosa. Filling from 30°C to 75°C: \( 80{,}000 \times 3500 \times 45 \approx 1.3\times10^{10} \) J, about 3.5 MWh. Most of the energy goes into the oil bath, not the food.
Oil uptake is the pleasant surprise. Sakhale's control samosas held 18.34% oil, about 14.7 g in an 80 g samosa. Oil enters through the crust, which is a surface, and surfaces scale as \(k^2\), not \(k^3\):
This is the square-cube law at work in your favour.
Volume grows faster than area, so the house samosa is, by fat fraction, the leanest samosa ever made.
The team made samosas from refined wheat flour, water and oil at 10:4:1, rolled 20 g portions to 1.0 mm, and fried them at 150 ± 5°C for 8 minutes with an oil ratio of 1:3. Thirteen treatments were run in triplicate. Control samosas held 18.34% oil; adding 1.5% xanthan gum to the casing cut that to 8.56%, a 53% drop, with CMC and guar gum close behind. The limit: one frying condition, an unspecified filling, and a small sensory panel.
Read the paper (free)Why would the middle stay raw while the outside burns?
The pastry wall would sit between 150°C oil and a filling still near room temperature, so its inner face would stay soft, wet dough for many hours while the outer face spent five weeks above the temperature where sugars and proteins char.
If the 1 mm sheet scales too, the wall is 10 cm thick. Heating through a slab from one side takes Fo of about 0.5:
For a normal 1 mm sheet the same sum gives about 3 seconds, which is why ordinary pastry crisps almost at once and the 8 minutes go into drying and browning.
In the giant, the cold filling acts as a heat sink on the inside face, so a band of gummy, raw dough survives against the potato for days. Meanwhile the outer crust keeps losing water and the Maillard browning that makes a samosa golden (explained in the Maillard reaction in Indian cooking) runs on into scorching.
Frying also does something a giant cannot copy: it relies on steam. The 2025 synchrotron study by Verma and colleagues showed that steam expanding the dough is what keeps oil out during the fry, and that most oil enters during cooling. A wall that takes 9 hours to heat would be venting steam unevenly for days.
Using fast synchrotron X-ray tomography at the ESRF, the authors filmed wheat flour dough in 3D while it fried and then cooled. Hotter oil formed a distinct crust with surface openings that drew oil into small pores by capillary action. Final oil content was 14.4% at 180°C, 12.2% at 150°C and only 1.3% at 120°C. The limit: a plain dough model, not a filled, folded samosa, and small samples sized for the X-ray beam.
Read the paper (free)Would the samosa even stand up?
Probably not for long: the stress at the base of the filling rises from under 1 kPa to about 82 kPa, a hundredfold jump, while the material holding it is the same soft potato and dough.
This is the square-cube law that Galileo used to explain why giants can't exist. Weight grows as \(k^3\) while the area that carries it grows as \(k^2\), so stress grows as \(k\):
A crisp 1 mm crust carries the first number with ease; you can pick up a samosa by one corner. The second is roughly the pressure under a car tyre, applied across the whole base of a wet, unset paste.
In the first hours, before any crust forms, the walls are only raw dough and would bulge outward under the filling's weight, much as a tall heap of mashed potato spreads.
I have not found a measured yield strength for samosa filling, so treat this section as a scaling argument rather than a calculation of failure. The same logic, tested on a thinner food, is worked out in how big a dosa can get before it collapses.
What changes if the assumptions move?
The answer is most sensitive to size and almost insensitive to oil temperature, which is exactly what Bi ≫ 1 predicts.
| Change | Centre heating time | Why |
|---|---|---|
| Baseline, 8 m | ~35 days | \(t \propto L^2\) |
| Smaller, 4 m (a hut) | ~9 days | Half the size, a quarter of the time |
| Bigger, 12 m (a four-storey block) | ~79 days | 1.5× the size, 2.25× the time |
| Diffusivity 20% higher | ~29 days | \(t \propto 1/\alpha\) |
| Diffusivity 20% lower | ~44 days | \(t \propto 1/\alpha\) |
| \(h\) of 90 instead of 200 W/m²·K | Almost unchanged | Bi is still in the hundreds |
| Fo of 0.4 instead of 0.2 (centre fully hot) | ~70 days | \(t \propto \text{Fo}\) |
Size your own giant samosa
Could anyone actually make one?
Yes, if you stop frying it as one piece. The version that works treats the samosa as architecture: fry thin parts at normal size, then assemble.
- Panels, not a sheet. Fry 1 to 2 mm pastry panels in an ordinary commercial fryer at 150 to 170°C, the conditions the lab studies used. Each panel crisps in minutes because its heat path is a millimetre.
- A frame. Hang the panels on a light steel or bamboo pyramid 8 m tall, so the crust carries only its own weight, not the filling's.
- Filling in compartments. Cook the aloo masala hot in normal batches and load it into insulated sections, so no part sits in the bacterial danger band.
- Rough cost. About 80 t of potato filling at an assumed ₹40 per kg is about ₹32 lakh, before the frame, fryers and labour. That is my order-of-magnitude estimate, not a quote.
Frying a house-sized samosa in one piece is physically hopeless: by Fourier's law the centre needs about 35 days in 150°C oil (the square of a 100-fold size increase), by which time the outside is carbon, the inside pastry is raw and the filling has spent weeks at bacteria-friendly temperatures. Built as fried panels on a frame with hot filling in compartments, an 8 m samosa is buildable, and at about 0.2% oil it would even be lean.
The samosa's own story, from the Central Asian sambusa to the Indian street stall, is told in the history of the samosa in India, and the potato inside it only arrived a few centuries ago, as the history of the potato and tomato in India explains. What happens to spices in hot oil is covered in tadka chemistry. For the rest of the Taste Lab, see the Rasa world page.
What's new
What if next?
- Could you boil chai with body heat? The samosa needs too much time; chai with body heat runs into a law that no amount of time can beat.
- What if you ate only chillies for a day? Swap the samosa for its green chutney and follow the burn all the way down.
- What if all the rice India eats in a day were one pile? Another square-cube problem, this time 54 m tall.
Questions people ask
How long would it take to fry a giant samosa?
Frying time is set by conduction, which scales with the square of size. A normal samosa's filling heats in about 5 minutes; one 100 times bigger needs about 10,000 times longer, roughly 35 days for an 8 m samosa at 150°C.
Why do big samosas stay cold in the middle?
Because the surface reaches oil temperature quickly but the inside can only warm by slow conduction through potato and dough, which have a thermal diffusivity of about 1.5 × 10⁻⁷ m²/s. Doubling the thickness quadruples the time.
How much oil does a samosa absorb?
In a lab study, control samosas fried at 150°C for 8 minutes held 18.34% oil. Adding 1.5% xanthan gum to the casing cut that to 8.56%.
Does frying temperature change how much oil food absorbs?
Yes. In a 2025 synchrotron study of wheat dough, oil content was 14.4% at 180°C, 12.2% at 150°C and 1.3% at 120°C, because hotter oil forms a crust with pores that draw oil in as the food cools.
Sources
- Sakhale BK et al.: Effect of hydrocolloids incorporation in casing of samosa on reduction of oil uptake, Journal of Food Science and Technology (2011) (20 g dough, 1 mm sheet, 150°C for 8 min, 1:3 oil ratio, 18.34% oil)
- Moro Pisco JF, Salas-Valerio WF: Thermal diffusivity and temperature profiles of whole potato during heating in a liquid, Revista Colombiana de Investigaciones Agroindustriales (2020) (α = 1.49 × 10⁻⁷ m²/s)
- Şahin S, Sastry SK, Bayindirli L: The determination of convective heat transfer coefficient during frying, Journal of Food Engineering 39(3):307–311 (1999) (h = 90–200 W/m²·°C at 150–190°C)
- Verma U et al.: High-speed computed tomography to visualise the 3D microstructural dynamics of oil uptake in deep-fried foods, Nature Communications (2025) (oil content at 120, 150 and 180°C; uptake during cooling)
- ESRF: Scientists find ways to reduce oil intake in deep fried foods (2025)
- The Tribune (PTI): Warning labels for samosas, jalebis? Health ministry says claim misleading (2025)
- Guinness World Records: Video, Muslim Aid makes world's largest samosa to feed the homeless (2017) (153.1 kg, 12 volunteers, 15 hours, custom fryer, ingredients, rules, 110.8 kg previous record; the story)
