Idli batter and bread dough: same factory, different workers

Both turn sugar into carbon dioxide, trap it in a wet starch matrix and set the foam with heat. Bacteria run the idli line, yeast runs the bread line, and the maths of gas, bubbles and yield stress shows exactly where the two factories differ.

A plate of steamed idlis with a bowl of sambar
Photo: Sutapa Pal, CC BY-SA 4.0, via Wikimedia Commons

Idli batter and bread dough are the same factory: in both, microbes eat sugar and breathe out carbon dioxide, a wet starch matrix traps the gas as bubbles, and heat sets the foam into a sponge. The workers are different. Lactic acid bacteria such as Leuconostoc and Weissella leaven idli, making one CO2 per sugar molecule plus acid, while baker's yeast makes two CO2 per sugar. The scaffolds differ too: idli batter holds bubbles with a yield stress of 13 to 43 Pa, bread dough with a stretchy gluten network.

Key takeaways
  • 0.4 litres of gas is what a kilogram of 3:1 idli batter gains overnight, by the density data.
  • 1 CO2 vs 2 CO2 per glucose: bacteria pay half the gas per sugar that yeast does, and take acid as wages.
  • 13–43 Pa yield stress lets idli batter hold any bubble smaller than about 1 cm radius without gluten.
  • Strain hardening above 1 is the gluten trick that lets bread bubbles inflate without bursting.
  • Verdict: the link holds. Same chemistry and gas physics, different workers and different scaffolding.

Put a bowl of risen idli batter next to a bowl of proved bread dough and they look like cousins. Both have doubled in size, both smell faintly sour and alcoholic, both are full of bubbles. That is not a coincidence of looks. It is the same production line, built from the same three machines: a gas generator, a gas trap and a setting oven.

I'll prove the link the way an engineer would: count the gas, name the workers, test how each matrix holds a bubble, then say where the comparison breaks.

Who are the workers in idli batter and bread dough?

Bread is leavened by one hired worker, baker's yeast (Saccharomyces cerevisiae), added on purpose. Idli is leavened by a crew of wild lactic acid bacteria that come in on the urad dal and rice, with no yeast needed.

The story

In 1856, in Lille, a manufacturer named M. Bigot, whose son studied under Louis Pasteur, asked the chemist for help. His beetroot alcohol was going wrong, and some batches were turning sour. Pasteur started by repeating Theodor Schwann's work showing that yeast is alive, and then looked at what was in the spoiled vats.

His answer split the job between two workers. Yeast turned sugar into alcohol: fermentation was the work of living cells, not mere decomposition, as Liebig and Berzelius had held. When another microbe got in, it made lactic acid instead, and the product soured. In a memoir published on 30 November 1857 he described "a particular ferment, a lactic yeast", always present when sugar becomes lactic acid. Bread dough runs on the first worker. Idli batter runs on the second.

Wikipedia: Louis Pasteur (fermentation and germ theory, citing René Vallery-Radot)

In 1965 a Cornell team showed that idli batter rises on one lactic acid bacterium, Leuconostoc mesenteroides, with no yeast at all; it is told in full in the science of idli batter fermentation.

Modern sequencing has refined the cast but not the plot. A 2019 study from the National Centre for Cell Science in Pune followed batter for 24 hours and found lactic acid bacteria above 85% of the bacterial population throughout. Weissella, a close relative of Leuconostoc, jumped from 9% to 71% of reads between hour 6 and hour 9, just as the batter starts to rise.

Diversity and Succession of Microbiota during Fermentation of the Traditional Indian Food IdliMandhania MH, Paul D, Suryavanshi MV et al. · Applied and Environmental Microbiology 85(13):e00368-19 · 2019

The team fingerprinted 10 vendor idlis and a lab batter sampled at nine points over 24 hours, using 16S rRNA amplicon sequencing (3.5 million reads), genus-specific qPCR and 354 cultured isolates. pH fell from 6.3 to 4.5. Weissella counts rose from about 4.3 to 7.9 log10 gene copies per gram in 15 hours, and radiolabelled glucose confirmed CO2 came from heterofermentative bacteria. The limit: the primers targeted bacteria only, so the study says nothing about whether yeasts help.

Read the paper (free)
Leuconostoc mesenteroides bacteria seen in a wet mount under a light microscope at 1600 times magnification
Leuconostoc mesenteroides at 1600x. These lens-shaped cocci were the first bacteria shown to leaven a food on their own.Photo: Ajay Kumar Chaurasiya, CC0, via Wikimedia Commons

How much gas does each factory have to make?

A kilogram of 3:1 idli batter needs about 0.03 mol of CO2 overnight, and a kilogram of bread dough about 0.06 mol during proving. The two factories run at the same order of output.

Assumptions and variables

SymbolMeaningValueSource
\(\rho_0, \rho_1\)Idli batter density, start and after 22 h (3:1)0.93, 0.682 g/cm³Ghosh and Chattopadhyay
\(T\)Fermentation temperature303 K (30°C)Durgadevi and Shetty used 30°C
\(s\)CO2 solubility in water near room temperatureabout 0.03 mol/LFrom 1.45 g/L at 25°C; slightly lower at 30°C (estimate)
\(w\)Water in the batterabout 0.5 L per kgEstimate for a typical home batter
\(k\)CO2 per glucose1 (bacteria), 2 (yeast)Stoichiometry of each pathway
  1. Gas volume from density. Mass stays the same while volume grows, so
    \[ \Delta V = m\left(\frac{1}{\rho_1} - \frac{1}{\rho_0}\right) = 1000\left(\frac{1}{0.682} - \frac{1}{0.93}\right) \approx 391\ \text{cm}^3 \]
  2. Moles of gas, ideal gas law. At 1 atm and 303 K:
    \[ n_{gas} = \frac{PV}{RT} = \frac{101{,}325 \times 3.91\times10^{-4}}{8.314 \times 303} \approx 0.0157\ \text{mol} \]
  3. Add the gas that dissolves first. Bubbles only grow once the water is saturated (Henry's law), so the factory must first fill about 0.5 L of water:
    \[ n_{diss} \approx s\,w = 0.03 \times 0.5 = 0.015\ \text{mol} \quad\Rightarrow\quad n_{total} \approx 0.031\ \text{mol} \]
  4. Sugar burned. Heterofermentative bacteria split glucose into lactic acid, ethanol and one CO2, so \(k = 1\):
    \[ m_{glucose} = \frac{n_{total}}{k} \times 180\ \tfrac{\text{g}}{\text{mol}} = 0.031 \times 180 \approx 5.6\ \text{g per kg of batter} \]
  5. Same sum for bread. Take 841 g of lean dough (500 g flour, 325 g water, salt and yeast) at an assumed 1.2 g/cm³, rising to 2.5 times its volume. Then \(\Delta V \approx 1.05\) L, \(n_{gas} \approx 0.042\) mol, plus about 0.01 mol dissolved, so 0.052 mol. Yeast makes two CO2 per glucose:
    \[ C_6H_{12}O_6 \rightarrow 2\,C_2H_5OH + 2\,CO_2 \qquad m_{glucose} = \frac{0.052}{2}\times 180 \approx 4.7\ \text{g} \]

So a bread factory burns under 1% of the flour weight as sugar to rise, and an idli factory a similar share of its solids. The output is comparable. The efficiency is not: per gram of sugar, yeast makes twice as much gas as the idli bacteria.

The cross-check with acid

The idli pathway also makes one lactic acid per glucose. At 0.031 mol that is about 2.8 g of lactic acid per kg, or 0.28% of the batter. Ghosh and Chattopadhyay measured titratable acidity rising from 0.44% to 0.91%, a gain of 0.47 points.

The measured acid is higher than the gas alone predicts. That fits the 1965 and 2019 findings: later in the night, homofermentative bacteria such as Enterococcus and Pediococcus take over and make acid with no gas at all.

Idli lineBread line1 glucose1 glucoseLeuconostoc, Weissellawild, from the dalSaccharomyces cerevisiaehired, added on purpose1 CO2lactic acid+ ethanol2 CO22 ethanolpH 6.3 to about 4.5, sourabout 10 to 14 h at 30°Ctwice the gas per sugarabout 1 to 3 h at 30°C
The two production lines. Same raw material, different workers and different by-products. The idli line pays itself in acid; the bread line in extra gas.

Where does each factory get its sugar?

Neither raw material arrives with much free sugar, so both factories run on starch broken down by enzymes and on the little sugar already in the grain.

  • Bread: baker's yeast eats glucose, sucrose, galactose and fructose within about an hour, then lives on maltose cut from damaged starch by flour amylases. Bakers often add fungal amylase to raise maltose.
  • Idli: soaking and grinding rice and urad dal releases sugars from the grains, and the bacteria's own enzymes keep the supply going. Lactic acid bacteria also break down some black gram protein; free amino nitrogen rises through the night.
  • Shared by-products: ethanol and organic acids in both, which is why both smell faintly of beer and yoghurt.
Non-Conventional Yeasts as Alternatives in Modern Baking for Improved Performance and Aroma EnhancementZhou N, Semumu T, Gamero A · Fermentation 7(3):102 · 2021

A review of how yeasts ferment flour sugars and survive baking stresses. It sets out the order in which S. cerevisiae uses dough sugars (glucose, sucrose, galactose, fructose, then maltose), notes that the first four go within about an hour, and that sweet doughs with 25–40% sucrose put the yeast under osmotic stress. Standard baker's yeast cannot break down starch itself. The limit: it is a narrative review, so it reports others' measurements rather than new ones.

Read the paper (free)
Oval baker's yeast cells of Saccharomyces cerevisiae seen under differential interference contrast microscopy
Baker's yeast, the bread factory's hired worker: a single-celled fungus, where the idli crew are bacteria.Photo: Masur, Public domain, via Wikimedia Commons

How does idli batter trap gas without gluten?

Idli batter is a yield-stress fluid: below a threshold stress it behaves like a soft solid, so a small bubble cannot push its way to the surface. Bread dough instead uses an elastic gluten film that stretches around each bubble.

The theory: the critical yield number

In a plain liquid, every bubble rises; Stokes' law only sets the speed. In a material with a yield stress \(\tau_y\), a bubble stays put if the yield stress beats its buoyancy. Fluid dynamicists measure this with the yield number \(Y\), the ratio of yield stress to buoyancy stress, where \(\ell\) is the bubble's equivalent radius:

\[ Y = \frac{\tau_y}{\Delta\rho\, g\, \ell} \qquad \text{static if } Y > Y_c, \quad Y_c = 0.132 \text{ for a sphere} \]

Rearranged, the biggest bubble a batter can hold is

\[ \ell_{max} = \frac{\tau_y}{Y_c\,\rho\,g} = \frac{13}{0.132 \times 1000 \times 9.81} \approx 10\ \text{mm} \quad\text{to}\quad \frac{43}{1295} \approx 33\ \text{mm} \]

Nagaraju and Manohar measured idli batter yield stress at 13 to 43 Pa. Real batter bubbles are fractions of a millimetre, so the matrix holds them all with a wide margin. That is how a batter with no gluten at all can still double.

Flow onset for a single bubble in yield-stress fluidsPourzahedi A, Chaparian E, Roustaei A, Frigaard IA · arXiv 2107.07580 (Journal of Fluid Mechanics draft) · 2021

Using computation, the authors found the minimum yield stress needed to hold a buoyant bubble still, for many shapes. A sphere is held when the yield number exceeds 0.132, and surface tension does not change that limit for a sphere; long, prolate bubbles need more yield stress than flat ones. As practical guidance they note that about 1.5 Pa holds a 1 mm bubble in a water-based gel. The limit: their test fluids are ideal viscoplastic gels, and real batter also thins with shear and changes by the hour.

Read the paper (free)

Why over-fermented batter goes flat

The same measurements explain a kitchen failure. Yield stress peaked at 7 hours and the flow index fell significantly by 23 hours, as acid and enzymes break down the matrix. Push fermentation too long and \(\tau_y\) drops, \(\ell_{max}\) shrinks, bubbles merge and escape, and the batter slumps.

A steel vessel of fermented idli batter, thick and bubbly, photographed in Bengaluru in 2026
Fermented idli batter in Bengaluru. It pours, yet its yield stress keeps the bubbles from rising out.Photo: Gpkp, CC BY-SA 4.0, via Wikimedia Commons

How does bread dough trap gas?

Bread dough holds bubbles with a gluten–starch film that gets stiffer as it stretches, a property called strain hardening. A strain hardening index of 1 or more lets a bubble wall thin without tearing.

Every bubble fights its own internal pressure. The Young–Laplace equation gives the extra pressure inside a bubble of radius \(r\) with surface tension \(\gamma\):

\[ \Delta P = \frac{2\gamma}{r} \qquad \gamma \approx 0.04\ \tfrac{\text{N}}{\text{m}}: \quad r = 50\ \mu\text{m} \Rightarrow 1.6\ \text{kPa}, \quad r = 1\ \text{mm} \Rightarrow 80\ \text{Pa} \]

The surface tension here is my estimate for a protein-rich liquid. The point holds anyway: small bubbles sit at higher pressure, so CO2 diffuses from small bubbles into large ones. This is disproportionation, the foam version of Ostwald ripening, and it coarsens any foam left alone.

  • Primary defence: the gluten–starch matrix, which must strain harden so a thin spot stiffens instead of tearing.
  • Secondary defence: thin liquid films lined with surface-active proteins and lipids, which take over late in proving when the matrix starts to crack.
  • Idli's equivalent: the yield stress, plus urad dal's surface-active globulins and sticky arabinogalactan, which help stabilise the foam.
Mechanism of gas cell stabilization in bread making. I. The primary gluten–starch matrixSroan BS, Bean SR, MacRitchie F · Journal of Cereal Science 49:32–40 · 2009

Using two wheat flours, defatting and lipid add-back, protein fractions and biaxial extension tests, the Kansas State team supported a dual mechanism: the gluten–starch matrix is the main bubble wall, and thin liquid films take over when gaps open during late proving and early baking. Loaf volume and strain hardening rose with protein molecular weight up to an optimum, then fell. The limit: two flours and test-bake loaves, so the optimum is specific to those wheats.

Read the paper (free)
Idli: a yield-stress cageBread: a gluten balloonCO2buoyancyyield stress holdsstays if τy > 0.132 ρ g r13–43 Pa holds r up to about 1 cmCO2 pushes outgluten–starch wall (solid line)strain hardening index ≥ 1inner liquid film: backup seal
Two ways to keep a bubble. Idli batter is a soft solid that a small bubble cannot push through; bread dough is a rubbery wall that stiffens where it thins.

Does the recipe ratio change the factory's output?

Yes. More urad dal gives a lighter, more aerated batter, which points to the dal as the source of both the workers and the foam-stabilising scaffold.

Idli batter density after 22 hours of fermentation, by rice to black gram ratio
See the data as a table
Mean density at 22 h (g/cm³)
2:10.592
3:10.682
4:10.831

Source: Ghosh and Chattopadhyay, Journal of Food Science and Technology (2011) · Lower density means more gas trapped. Fermentation at an uncontrolled 20–26°C.

A separate response-surface study at a steady 30°C found the batter rose 65% at 3:1 but only 31.3% at 3:1.5, and picked 3:1.475 fermented for 10.2 hours as the best sensory balance. The two studies disagree on whether more dal always means more rise, a reminder that rise and sponginess are not the same thing.

Preparation of idli batter, its properties and nutritional improvement during fermentationGhosh D, Chattopadhyay P · Journal of Food Science and Technology 48(5):610–615 · 2011

Batters of parboiled rice and dehulled black gram at 2:1, 3:1 and 4:1 were fermented for up to 22 hours at 20–26°C and sampled every three hours or so. pH fell from about 5.9 to 4.2, acidity rose from 0.44% to 0.91%, and volume grew 1.6 to 3.2 times, mostly in the first 13 hours. Riboflavin, thiamine and folic acid roughly doubled at their peaks. The limit: the paper has internal inconsistencies in units and temperature labels, so I use only its density and acidity trends.

Read the paper (free)

Sensitivity: what if the kitchen is colder?

Temperature changes the speed of both factories far more than their output. A common biology rule of thumb, the Q10 coefficient, says reaction rates roughly double for every 10°C rise across the comfortable range; I use Q10 = 2 as an assumption.

\[ t_{T_2} = t_{T_1}\, Q_{10}^{\,(T_1 - T_2)/10} \qquad t_{20^\circ C} = 10.2\ \text{h} \times 2^{(30-20)/10} \approx 20\ \text{h} \]
  • 30°C to 20°C: the 10-hour Bengaluru-summer batter needs about 20 hours in a Delhi winter kitchen.
  • Water share from 0.4 to 0.6 L/kg: dissolved CO2 moves from 0.012 to 0.018 mol, so total sugar burned stays at 5 to 6 g/kg.
  • Yield stress from 43 Pa down to 13 Pa: the trap still holds bubbles up to 1 cm, so collapse needs far more breakdown than a few extra hours.
  • Swap the workers: run the idli sum with yeast (\(k = 2\)) and sugar burned halves to about 2.8 g/kg. Run bread with bacteria only and the dough would need about 9.4 g of sugar, and turn sour.

Use this to check what you just read with your own batter or dough.

Run your own gas factory

Verdict: the link holds, with two changes

Idli batter and bread dough really are the same factory. Both burn roughly 5 g of sugar per kilogram to make 0.03 to 0.06 mol of CO2, both trap it in a starch-water matrix, and both are set by heat. Two parts differ: the workers (bacteria making one CO2 plus acid per glucose, against yeast making two CO2) and the scaffold (a 13–43 Pa yield-stress batter, against a strain-hardening gluten film). The analogy breaks at time, acid and texture, not at the physics.

Where does the analogy break down?

The comparison is exact in chemistry and gas physics, and loose in three places.

  1. Acid. Idli batter ends near pH 4.5 by design, because its workers make acid; a plain yeast dough has no acid-making crew, so it stays far less sour. Sourdough is the bridge: wild yeast plus lactic acid bacteria, both crews on one line.
  2. Setting. Idli is steamed, so the surface never browns; bread is baked, so the crust gets Maillard browning.
  3. Hygiene. Idli's long, warm fermentation is a risk as well as a recipe. The acid helps, but a batter left warm for days can still be colonised by the wrong microbes.
Same machines, different crew: one makes twice the gas, the other makes the acid that keeps the batch safe.The verdict, in one line
Slices of sourdough bread showing an open, irregular crumb of gas bubbles
Sourdough crumb. This loaf used both crews at once: wild yeast for gas and lactic acid bacteria for sourness, the idli line and the bread line merged.Photo: Angel Ganev, CC BY 2.0, via Wikimedia Commons

The steps for a reliable batter are in how to make idli and dosa batter, the full microbiology is in the science of idli batter fermentation, and the cold-kitchen problem is solved in why idli batter won't ferment in winter. The rice side of the recipe is in rice science: amylose, basmati and sona masoori, and the far end of the timeline is what if idli batter fermented for a year. More of the Taste Lab lives on Rasa.

What if next?

Questions people ask

Is there yeast in idli batter?

Some studies have isolated yeasts from idli batter, but their role is not well defined. The classic 1965 Cornell work and the 2019 sequencing study both trace the rise to heterofermentative lactic acid bacteria, mainly Leuconostoc and Weissella.

Can I add baker's yeast to idli batter?

You can, and it will make gas faster, since yeast makes two CO2 per sugar. The batter will taste less sour, because the acid comes from the bacteria, and the texture changes because yeast does not add the dal's stabilising compounds.

Why does idli batter rise without gluten?

Because fermented batter has a yield stress of about 13 to 43 Pa. A bubble only escapes if its buoyancy beats that stress, and at that level anything smaller than about 1 cm radius stays trapped.

Why does over-fermented batter go flat?

Yield stress peaks around 7 hours and falls as acid and enzymes break down the matrix. Bubbles then merge and escape, and the batter collapses.

Sources

  1. Mukherjee SK, Albury MN, Pederson CS, Van Veen AG, Steinkraus KH: Role of Leuconostoc mesenteroides in leavening the batter of idli, Applied Microbiology (1965) (first record of leavening by a lactic acid bacterium alone)
  2. Wikipedia: Louis Pasteur, fermentation section (Bigot 1856; lactic ferment memoir 1857; source for the story)
  3. Mandhania MH et al.: Diversity and succession of microbiota during fermentation of idli, Applied and Environmental Microbiology (2019) (pH 6.3 to 4.5; Weissella 9% to 71%; LAB above 85%)
  4. Ghosh D, Chattopadhyay P: Preparation of idli batter, its properties and nutritional improvement during fermentation, J Food Sci Technol (2011) (densities 0.93 to 0.592–0.831 g/cm³; acidity 0.44% to 0.91%)
  5. Durgadevi M, Shetty PH: Effect of ingredients on sensory profile of idli, J Food Sci Technol (2014) (65% rise at 3:1; optimum 3:1.475 for 10.2 h at 30°C; arabinogalactan and globulins)
  6. Nagaraju VD, Manohar B: Rheology and particle size changes during idli fermentation, Journal of Food Engineering 43:167–171 (2000) (yield stress 13–43 Pa, peak at 7 h)
  7. Pourzahedi A, Chaparian E, Roustaei A, Frigaard IA: Flow onset for a single bubble in yield-stress fluids, arXiv (2021) (critical yield number 0.132 for a sphere)
  8. Sroan BS, Bean SR, MacRitchie F: Mechanism of gas cell stabilization in bread making I, Journal of Cereal Science (2009) (dual stabilisation; strain hardening index of 1)
  9. Zhou N, Semumu T, Gamero A: Non-conventional yeasts as alternatives in modern baking, Fermentation (2021) (sugar use order; maltose; yeast cannot break down starch)
  10. Wikipedia: Carbon dioxide (CO2 solubility 1.45 g/L at 25°C)
  11. Department of Science and Technology: Fermented food can personalise nutrition for India's diverse population (2025)
  12. Onmanorama: Kerala's new state microbe has been in our food all along (2026) (Bacillus subtilis and bread rope)
idli batterbread doughfermentationLeuconostoc mesenteroidesWeissellaSaccharomyces cerevisiaeyield stressglutenideal gas lawheterofermentation
Sapta

Sapta (Saptarshi Sarkar) is a mechanical engineer turned marketer in Bengaluru and a solo touring rider. He writes the seven Saptaverse blogs. Work with him or read more of Foodblog.

Keep reading in Foodblog

From the other blogs