A dosa can get as long as the griddle under it: heat only has to cross 2 to 4 mm of batter, so length never stops it cooking, and the Guinness record is 37.5 metres. What makes a dosa collapse is handling. By my estimate from rice-starch strength data, a crisp dosa can stick out unsupported only about 8 to 25 cm before it snaps, so a giant one survives only if it is lifted with a hand or spatula every 17 to 50 cm, which is almost exactly how the 75 chefs behind the record did it.
- 37.5 m is the Guinness record dosa, made by MTR Foods and Lorman in Bengaluru in March 2024.
- Length is free: heat only crosses 2–4 mm of batter, so a long dosa cooks like a small one.
- About 5 mm is the thickness ceiling, because heat-crossing time grows with thickness squared.
- 8–25 cm is the estimated overhang before a crisp dosa snaps under its own weight.
- One lifter every 17–50 cm keeps a giant dosa whole; 75 chefs over 37.5 m is one per 0.5 m.
Every dosa cook knows the two failure modes:
- Too thick: pour the batter heavy and the middle stays raw while the bottom burns.
- Too far: lift a crisp paper dosa by one corner and the far end cracks off.
Both are physics problems with clean answers. The numbers come from a rare thing: a peer-reviewed heat-transfer study of dosa itself, measured on a commercial dosa-baking machine. This post solves the problem the way an engineer would, then designs the dosa that beats the record.
What is the record for the biggest dosa?
The longest dosa on record is 37.5 metres (123 ft), made by MTR Foods and Lorman Kitchen Equipments in Bengaluru on 15 March 2024 to mark MTR's 100 years, according to Guinness World Records.
In March 2024 MTR Foods, the packaged-food company that grew out of Bengaluru's Mavalli Tiffin Rooms, set out to mark its centenary with a 100-foot dosa. The attempt ran at its Bommasandra factory, on an induction stove that Lorman Kitchen Equipments built for the job and called the biggest it had ever made.
75 chefs from MTR's culinary team worked the batter, the company's red-rice mix, along the cooktop. The finished dosa measured 123 feet, overshooting the target and more than doubling the previous record of 16.68 m. Guinness adjudicator Rishi Nath handed the certificate to MTR chief executive Sunay Bhasin, and the dosa was shared with employees, local schoolchildren and neighbours.
Newsvoir: MTR celebrates 100 years with a Guinness World Records title for the longest dosa (Mar 2024)Notice what the record is measured in: length. Nobody has a record for the widest or thickest dosa, and that is not an accident. A dosa is a thin sheet, and thin sheets scale in two directions freely and in the third not at all. To see why, we need the numbers.

What are the assumptions and variables?
All the material properties below come from one source, the dosa heat-transfer study by Venkateshmurthy and Raghavarao, except the strength value, which is my estimate from rice-noodle tests and is marked as such.
| Symbol | Meaning | Value | Source |
|---|---|---|---|
| \(t\) | Dosa thickness | 2–4 mm (3 mm used) | Venkateshmurthy & Raghavarao |
| \(D\) | Diameter of a standard dosa | 0.20–0.23 m | Same |
| \(m_b\), \(m_d\) | Batter in, baked dosa out | 0.100 kg, 0.072–0.076 kg | Same (Table 1) |
| \(k\) | Thermal conductivity of dosa | 0.42 W/m·K | Same |
| \(c_p\) | Specific heat | 1.83 kJ/kg·K | Same |
| \(T_p\), \(t_b\) | Plate temperature, bake time | 170–180°C, about 120 s | Same |
| \(h_{fg}\) | Latent heat of evaporating water | 2,257 kJ/kg | Standard steam tables |
| \(\sigma\) | Tensile strength of cooked rice-starch sheet | ~47–410 kPa (estimate) | From Kraithong & Rawdkuen breaking forces, assumed 3 mm² cross-section |
The authors measured temperatures through dosas baked on a 1.2 m rotating hot plate preheated to 170–180°C, making about 360 dosas an hour, each from 100 ml (0.1 kg) of batter at 65% moisture, spread 2 to 4 mm thick and 0.20 to 0.23 m across, baked about 120 s to 35% moisture. They measured thermal conductivity of 0.42 W/m·K and emissivity of 0.31, and found that about 98% of the heat arrives by conduction from the plate and about 1.5% by radiation, with convection negligible. Thermal efficiency was about 51.5%, and stainless steel and Teflon-coated aluminium plates gave the best surface. The limit: some energy figures in the tables do not add up cleanly, so I cross-check them below rather than relying on one number.
Read the paper (free)Why can't a dosa be thicker? The heat-diffusion limit
A dosa cannot be much thicker than about 5 mm because the time heat takes to diffuse through a layer grows with the square of its thickness: double the thickness and the middle needs four times as long, while the face on the plate keeps getting hotter. This is the heat equation, Fourier's law of conduction applied over time.
- Density of a finished dosa. Mass over volume:
\[ \rho = \frac{m_d}{\tfrac{\pi}{4}D^2 t} = \frac{0.074}{\tfrac{\pi}{4}(0.215)^2(0.003)} \approx 680\ \text{kg/m}^3 \]For 2 mm it is about 1,020 kg/m³ and for 4 mm about 510. The dosa is porous, full of fermentation bubbles.
- Thermal diffusivity. How fast temperature spreads:
\[ \alpha = \frac{k}{\rho c_p} = \frac{0.42}{680 \times 1830} \approx 3.4 \times 10^{-7}\ \text{m}^2/\text{s} \]
- Diffusion time. The characteristic time for heat to cross thickness \(t\):
\[ \tau \approx \frac{t^2}{\alpha} = \frac{(0.003)^2}{3.4\times10^{-7}} \approx 27\ \text{s} \]
- Fourier number. Compare with the 120 s bake: \( \text{Fo} = \alpha t_b / t^2 \approx 4.5 \). A Fourier number well above 1 means the whole thickness has time to heat through. That is why a 3 mm dosa cooks evenly in two minutes.
Sensitivity check: what thicker batter does
| Thickness | Diffusion time τ | Fourier number in 120 s | Result |
|---|---|---|---|
| 3 mm | 27 s | 4.5 | Cooks through evenly |
| 6 mm | 107 s | 1.1 | Borderline, soft centre |
| 10 mm | 296 s | 0.4 | Raw top, burnt bottom |
At 6 mm, the centre ends up soft and uttapam-like. At 10 mm, \(\tau\) is two and a half times the bake time, so the top is still batter when the bottom burns. You can lower the plate temperature and bake longer, but then you have invented the uttapam.
See the data as a table
| Diffusion time t²/α (s) | Typical bake time (s) | |
|---|---|---|
| 1 mm | 3.0 | 120 |
| 2 mm | 11.9 | 120 |
| 3 mm | 26.7 | 120 |
| 4 mm | 47.4 | 120 |
| 6 mm | 106.7 | 120 |
| 8 mm | 189.6 | 120 |
| 10 mm | 296.3 | 120 |
Source: Author's calculation using k, cp and bake time from Venkateshmurthy & Raghavarao (J Food Sci Technol) · Uses α = 3.4 × 10⁻⁷ m²/s for every thickness; real batter changes properties as it dries.
Why the square-cube law doesn't apply
This is where the square-cube law, which says area grows with the square of size and volume with the cube, fails as a description of dosas. A dosa scaled up geometrically, ten times wider and ten times thicker, would be a 30 mm slab that never cooks.
Keep the thickness fixed, and size only changes area, so a giant dosa cooks exactly like a small one, square metre by square metre.
How much heat does a giant dosa need?
Each square metre of dosa needs about 2 MJ of heat, mostly to boil off water, delivered in about two minutes, which is a heat flux of roughly 17 to 20 kW per square metre.
- Water to evaporate. \( m_w = 0.100 - 0.074 = 0.026 \) kg per dosa, so
\[ Q_{lat} = m_w h_{fg} = 0.026 \times 2257 \approx 58.7\ \text{kJ} \]
- Sensible heat. Warming 0.1 kg of batter by about 75 K:
\[ Q_{sen} = m_b c_p \Delta T = 0.1 \times 1.83 \times 75 \approx 13.7\ \text{kJ} \]Using 3.5 kJ/kg·K, closer to wet batter, gives 26 kJ instead.
- Per square metre. One dosa covers \( \tfrac{\pi}{4}(0.215)^2 = 0.0363 \) m², so
\[ q'' = \frac{Q_{lat}+Q_{sen}}{A\, t_b} = \frac{72{,}400}{0.0363 \times 120} \approx 16.6\ \text{kW/m}^2 \]or about 19.5 kW/m² with the higher specific heat.
- Cross-check. The paper's text gives about 532 W latent and 192 W sensible heat per dosa; over 120 s that is about 87 kJ, consistent with my 72 to 85 kJ.
Scaling up to the record dosa
For the record dosa I have to assume a width, because none was published. At 0.3 m wide (my assumption), 37.5 m is 11.25 m². That needs:
- about 22 MJ of heat into the dosa;
- delivered at roughly 190 kW for two minutes;
- about 31 kg of batter at the paper's 2.75 kg per square metre.
A large commercial induction line can supply that; a row of home stoves cannot.

When does a dosa snap? The beam-bending limit
A crisp dosa snaps when its own weight bends it harder than its strength allows, and for a 3 mm rice-starch sheet that happens at an unsupported overhang of roughly 8 to 25 cm by my estimate. This is Euler–Bernoulli beam theory, the same model engineers use for a diving board.
How strong is a dosa?
No one has published a tensile test of dosa, so I borrow from the nearest material that has been tested: cooked rice noodles, also a rice-starch gel. Kraithong and Rawdkuen measured breaking forces of 14 to 125 grams-force for eight commercial Thai rice noodles.
Taking a strand cross-section of about 3 mm² (my assumption, since the paper does not give dimensions), that is a stress of roughly 47 to 410 kPa. A dry, crisp dosa is not a wet noodle, so treat this range as an order-of-magnitude estimate.
The cantilever calculation
- Load. A strip of width \(b\) weighs \( w = \rho g t b \) per metre of length.
- Bending moment at the support of a cantilever of length \(L\): \( M = w L^2/2 \).
- Section modulus of a thin rectangle: \( Z = b t^2/6 \).
- Bending stress equals strength at failure:
\[ \sigma = \frac{M}{Z} = \frac{3 \rho g L^2}{t} \quad\Rightarrow\quad L_{max} = \sqrt{\frac{\sigma t}{3 \rho g}} \]Width cancels out, so a narrow strip and a wide sheet snap at the same overhang.
- Numbers. With \(\sigma\) = 47 kPa, \(t\) = 3 mm, \(\rho\) = 680 kg/m³:
\[ L_{max} = \sqrt{\frac{47{,}000 \times 0.003}{3 \times 680 \times 9.81}} \approx 0.083\ \text{m} \]With 410 kPa it is about 0.25 m.
Sensitivity check: \(L_{max}\) grows only with the square root of strength and thickness. Doubling the thickness to 6 mm lifts the overhang by 41% but, as shown above, ruins the cooking.
A crisp dosa is a brittle beam with a short reach: support the middle, or lose the tip.
How did 75 chefs lift a 37.5 m dosa?
They lifted it with many supports, and the beam maths says they needed one about every 17 to 50 cm; 37.5 m divided by 75 chefs is exactly 0.5 m each.
A sheet resting on evenly spaced supports behaves like a chain of simply supported spans, whose maximum moment is \( w s^2/8 \) for spacing \(s\). Setting that equal to strength:
So the crowd was not ceremony; it was structural. If the dosa is still warm and pliable instead of crisp, the rules change. A flexible sheet hung from one edge fails in tension when its own weight exceeds its strength, a quantity engineers call breaking length:
That is why a soft dosa can be rolled or draped over a long plate, as in the four-foot dosas restaurants serve, while a crisp one must be carried flat.
The authors tested eight commercial Thai rice noodles, five white and three coloured, for composition, cooking loss and texture. Using a TA.XT Plus texture analyser with spaghetti tensile grips at 3 mm/s, they found tensile strengths from about 14 to 125 grams-force and extensibility from about 14 to 31 mm after cooking to optimum time, with large differences between brands. The limit for this post: strength is reported as force, not stress, and noodle cross-sections are not given, so converting to kPa needs an assumed area, and a crisp dosa is drier and more porous than a cooked noodle.
Read the paper (free)Design your giant dosa
No law of physics caps a dosa's length: heat has to cross only 3 mm, so a 37.5 m dosa cooks like a 23 cm one, square metre by square metre (Fourier number about 4.5 in a 120 s bake). Thickness is capped near 5 mm by the square-law diffusion time, and handling is capped by beam bending: a crisp dosa snaps beyond roughly 8 to 25 cm of overhang, so it needs a support every 17 to 50 cm. Give it a long enough griddle and one lifter per half metre, and the length is up to you.
What would it take to make a 100-metre dosa?
A 100 m dosa, 0.3 m wide, needs a 100 m griddle delivering about 500 kW for two minutes, about 83 kg of batter and, by the beam estimate, 200 to 600 lifters, or a conveyor that never lifts it at all.
- Area: 100 × 0.3 = 30 m².
- Heat: 30 m² × about 2.0 MJ/m² = about 60 MJ into the dosa. At the paper's 51.5% thermal efficiency that is about 116 MJ, or 32 kWh of input energy.
- Energy cost: at an assumed ₹9 per kWh commercial tariff, about ₹300. The electricity is the cheap part.
- Batter: 2.75 kg/m² × 30 m² = about 83 kg.
- Spreading: the whole length must be spread well inside the 120 s bake. At half a metre per cook, as MTR did, that is 200 cooks spreading at once.
- The smarter design: a long moving steel belt over induction heaters, with batter sprayed on at one end and a continuous ribbon rolled up at the other. It is a dosa paper machine; length is limited only by how long you run it.

Where does the model break down?
The model is honest about three weak points: dosa strength is borrowed from noodles, properties change as the batter dries, and real failure starts at bubbles and cracks.
- Strength. The 47–410 kPa range comes from noodle forces divided by an assumed area. A texture-analyser test on dosa strips would replace the weakest number in this post.
- Changing properties. Conductivity, density and specific heat shift as moisture falls from 65% to 35%, so diffusivity is not constant.
- Flaws. The fermentation bubbles that make a dosa light also make it fail at weak spots, so real overhangs will be shorter than a uniform-beam model predicts. The science of idli and dosa batter fermentation explains where those bubbles come from.
And the best dosa is not the biggest. The onion ghee pudi dosa at Prasad Hotel in Mulbagal, on my ride there from Bengaluru, was a complete banger at a normal size.
What's new
- Guinness confirms the 37.5 m longest dosa by MTR Foods and Lorman Kitchen Equipments Guinness World Records, Mar 2024
- 75 chefs and a custom induction cooktop: how the 123-foot dosa was made Newsvoir, Mar 2024
What if next?
- Could you cook a chicken by slapping it? The dosa needs 2 MJ per square metre; a chicken needs about a third of a megajoule, delivered by hand.
- What if a samosa were the size of a house? Unlike a dosa, a samosa scales in three dimensions, and the square-cube law wins.
- Could you boil chai with body heat? From 190 kW griddles down to a human's 100 watts.
The drink that goes with it: what if you drank 10 cups of filter coffee a day? More from the Taste Lab on the Rasa world page.
Questions people ask
What is the world's longest dosa?
37.5 metres (123 ft), made by MTR Foods and Lorman Kitchen Equipments in Bengaluru on 15 March 2024 and certified by Guinness World Records. The previous record was 16.68 m.
Why does my dosa stay raw in the middle?
Usually the batter is too thick or the tawa too cool. Heat-crossing time rises with the square of thickness, so a 6 mm layer takes about four times as long as a 3 mm one.
What temperature should a dosa tawa be?
The heat-transfer study preheated its commercial plate to 170–180°C and baked each dosa for about 120 seconds. Home tawas vary, but that range is a good target.
Why does a crisp dosa break when you lift it?
A crisp dosa is a brittle beam. Its own weight creates a bending stress that rises with the square of the overhang, so beyond roughly 8 to 25 cm the unsupported part cracks off.
Sources
- Venkateshmurthy K, Raghavarao KS: Analysis of modes of heat transfer in baking Indian rice pan cake (Dosa), Journal of Food Science and Technology (2011) (thickness, diameter, batter mass, moisture, k = 0.42 W/m·K, cp, plate temperature, bake time, 98% conduction, 51.5% efficiency)
- Kraithong S, Rawdkuen S: Quality attributes and cooking properties of commercial Thai rice noodles, PeerJ (2021) (14–125 gf breaking forces, basis of the strength estimate)
- Guinness World Records: Longest dosa (2024) (37.5 m, date, record holders)
- Newsvoir: MTR celebrates 100 years with a Guinness World Records title for the longest dosa (2024) (75 chefs, induction cooktop, 16.68 m previous record)
- Wikipedia: Euler–Bernoulli beam theory (bending stress and section modulus)
- Wikipedia: Fourier number (dimensionless diffusion time)
- Wikipedia: Square–cube law

