Idli batter fermentation: the microbes that make it rise

No yeast packet, no baking soda: idli batter rises because bacteria living on the grains turn sugar into acid and carbon dioxide, and urad dal holds the gas.

Steamed white idlis served with medu vada, chutney and sambar
Photo: OdetomyLife, CC BY-SA 4.0, via Wikimedia Commons

Idli batter fermentation is a wild lactic acid fermentation: bacteria that arrive on the rice and urad dal, mainly Weissella, Leuconostoc, Lactococcus and Enterococcus, eat the batter's sugars and make lactic acid plus carbon dioxide. In a 2019 study at about 30°C the batter's pH fell from 6.3 to about 4.5 and its volume doubled by 15 hours, while urad dal's proteins and gums trapped the gas that makes idli spongy.

Key takeaways
  • 1965: Cornell scientists showed idli is leavened by a lactic acid bacterium, Leuconostoc mesenteroides, not yeast.
  • 86% of the fermenting batter's bacteria are lactic acid bacteria; Weissella jumps from 9% to 71% between hours 6 and 9.
  • pH 6.3 to 4.6 by 12 hours, and volume doubles by about 15 hours at 30°C.
  • Urad dal holds the gas: its globulin protein and arabinogalactan gum coat the bubbles.
  • 2:1 to 3:1 rice to urad, 2% salt and 10 to 15 hours at 26 to 30°C is where the studies cluster.

Most idli advice is folklore:

  • "add a fistful of poha";
  • "keep it near the stove";
  • "never use a steel spoon".

The microbiology is better documented than most people realise. Food scientists have studied idli since the 1960s, and DNA sequencing in 2019 and 2022 mapped exactly which bacteria take over the batter hour by hour.

This post walks through that evidence, with the pH and volume curves, and turns it into rules you can use on a cold night or a hot Bengaluru afternoon.

86%share of lactic acid bacteria in fermenting idli batter (Mandhania et al., 2019)
6.3 to 4.5fall in batter pH over the first 15 hours
100%volume rise by 15 hours at about 30°C
0fermentation when the grains' surfaces were sterilised

What makes idli batter rise without yeast?

Idli batter rises because heterofermentative lactic acid bacteria split each glucose molecule into lactic acid, ethanol or acetate, and one molecule of carbon dioxide, and that gas is the leavening.

The story

In March 1965, five scientists at Cornell University's New York State Agricultural Experiment Station in Geneva, New York, published a short paper about a South Indian breakfast. S.K. Mukherjee, M.N. Albury, C.S. Pederson, A.G. van Veen and K.H. Steinkraus had followed the bacteria in fermenting idli batter stage by stage.

One organism did the work: Leuconostoc mesenteroides made both the sourness and the gas. Streptococcus faecalis grew in the later stages, and Pediococcus cerevisiae later still. Their conclusion, in Applied Microbiology, was that as far as they knew this was the first record of leavening produced entirely by a lactic acid bacterium, with no yeast doing the lifting. Fifty-four years later, DNA sequencing in Pune told the same story with a close relative, Weissella, in the lead.

Mukherjee et al.: Role of Leuconostoc mesenteroides in leavening the batter of idli, Applied Microbiology 13(2), 1965

Why some bacteria make gas and others don't

The pathway matters. The two kinds of lactic bacteria behave very differently:

TypeRouteProductsKitchen result
Homofermentative (most curd cultures)GlycolysisAlmost only lactic acidCurd sours but never rises
Heterofermentative (Weissella, Leuconostoc)PhosphoketolaseLactic acid, ethanol or acetate, and CO2Batter sours and rises

In 2019 Mandhania and colleagues proved this route in idli by adding glucose labelled with radioactive carbon-14 at the C-1 position: the label came out as carbon dioxide, exactly as the heterofermentative pathway predicts.

Yeasts can be present, especially in warm kitchens and in batters with poha or cooked rice, but the studies that sterilised or selectively cultured the batter show bacteria doing the main work.

Why the batter rises Bacteria on the grains Weissella, Leuconostoc, Lactococcus, Enterococcus Glucose phosphoketolase route (heterofermentative) Lactic acid: pH 6.3 to 4.5 CO2 gas: volume x2 Urad dal globulin and arabinogalactan gum coat the bubbles so gas stays in Steaming sets the starch around the bubbles: the honeycomb crumb of idli
Bacteria make the gas; urad dal holds it. Simplified from Mukherjee et al. (1965), Mandhania et al. (2019) and the TNAU Agritech note on black gram.

Which bacteria take over idli batter, and when?

Idli batter goes through a succession: soil and grain bacteria dominate the first six hours, Weissella takes over between hours 6 and 12, and Enterococcus and Streptococcus rise later as the batter turns sour. That is the picture from Mandhania and colleagues, who sampled lab batter every three hours for 24 hours and sequenced 16S rRNA genes, then cross-checked against batters from vendors in Bengaluru and Pune.

Their numbers are striking:

  • 48% of all sequence reads were Weissella, and it jumped from 9% to 71% of the community between hour 6 and hour 9, the window when leavening is fastest.
  • 104.3 to 107.9 gene copies per gram: Weissella's rise by quantitative PCR between hour 0 and hour 15, a several-thousand-fold increase.
  • 0.06 million to 77.2 million colony-forming units per gram: lactic acid bacteria grown on MRS agar over 24 hours.
Diversity and Succession of Microbiota during Fermentation of the Traditional Indian Food IdliMandhania et al. · Applied and Environmental Microbiology · 2019

The team made batter from parboiled rice and black gram at 2:1, soaked each for 4 hours, fermented it overnight at about 30°C with no starter, and sampled every 3 hours for 24 hours. Using amplicon sequencing (about 3.5 million usable reads), qPCR, culturing of 354 isolates and a radiolabelled glucose test, they found lactic acid bacteria made up 86% of the community and Weissella was the key gas producer. pH fell from 6.3 to 4.6 at 12 hours and 4.2 at 18 hours; volume doubled by 15 hours. Batter made from surface-sterilised grains did not ferment at all. Limits: one lab recipe for the time series and a modest number of vendor samples.

Read the paper (free)
Idli batter pH during fermentation at about 30°C
See the data as a table
Batter pH (pH)
0 h6.3
6 h6.0
12 h4.6
18 h4.2
24 h4.5

Source: Mandhania et al., Applied and Environmental Microbiology 2019, Table 2 · Parboiled rice and black gram 2:1, no starter culture

The 2022 Illumina study by Kavitake and colleagues, also using Indian batter, found the same shape from a different angle. At hour 0 the batter was 93% Proteobacteria (environmental genera such as Enterobacter, Serratia and Klebsiella) and only 7% Firmicutes, the phylum that holds the lactic bacteria.

By hour 15 Firmicutes made up 68%. The populations settled between hours 12 and 15, which is the practical end of a good ferment.

Who is in the batter: bacterial phyla at the start and end of fermentation
See the data as a table
Firmicutes (lactic acid bacteria) (%)Proteobacteria (%)
0 h793
15 h6831

Source: Kavitake et al., Journal of Food Science and Technology 2022 · Other phyla make up the remainder at 15 h

StageHours (about 30°C)Dominant microbesWhat you see
Lag0 to 6Mixed grain and soil bacteria (Proteobacteria, Bacillus)Batter flat, pH near 6.0 to 6.3
Leavening6 to 12Weissella and other Leuconostocaceae (71 to 85% at 9 to 12 h)Bubbles, rise, mild sour smell, pH drops toward 4.6
Souring12 to 18Enterococcus, Streptococcus, Lactococcus risingPeak volume, tangy smell, pH about 4.2
Over-fermented18 to 24+Acid-tolerant lactic bacteriaBatter may collapse, sharp sourness, flat idlis
A 24-hour idli ferment at about 30°CStages and pH, after Mandhania et al. 2019LagLeaveningSouringOver0 h6 h12 h18 h24 hpH 6.36.04.64.24.5Lag: grain and soil bacteria, batter flatLeavening: Weissella 9% to 71% of bacteria by 9 hSouring: Enterococcus and Lactococcus rise, peak volumeOver: foam weakens, batter may collapse
Bar height is a rough guide to how much gas is being made in each stage; the pH values are measured ones from the 2019 Pune study.

Why does urad dal matter so much?

Urad dal (dehulled black gram) matters because it supplies the gas-trapping structure: a surface-active globulin protein and a mucilaginous polysaccharide called arabinogalactan that form films around bubbles.

The TNAU Agritech note on rice fermentation lists both components and says the mucilage is believed to help retain carbon dioxide in the thick batter. Rice starch alone cannot do this; a rice-only batter ferments but stays dense.

That explains two kitchen rules:

  1. Grind urad dal fine and long, until it is light and fluffy, because that whips air into the protein foam before fermentation starts.
  2. Grind rice coarser, as TNAU recommends for idli, so the starch sets around the bubbles during steaming rather than turning gluey.

Where the bacteria come from

Mandhania's sterilisation test settles where the microbes come from. Batter ground from surface-sterilised grains did not ferment at all, so the inoculum is on the rice and dal themselves, not in the air, the grinder or your hand.

That is why washing and soaking water matter less than people think, and why old, very clean or irradiated grain can be a slow starter.

A heap of white dehulled urad beans, the black gram used in idli and dosa batter
Dehulled urad (black gram). Its globulin and gum hold the gas; its surface carries part of the bacteria that make it.Photo: BaluBFA12, CC0, via Wikimedia Commons

What ratio, soaking time and temperature work best?

Published studies cluster around 2:1 to 3:1 rice to urad by weight, about 4 hours of soaking at 30°C, 2% salt and 8 to 14 hours of fermentation at 26 to 30°C. Below that temperature the lactic bacteria slow sharply; above it, souring outruns rising.

Preparation of idli batter, its properties and nutritional improvement during fermentationGhosh and Chattopadhyay · Journal of Food Science and Technology · 2011

The authors soaked polished parboiled rice and decorticated black gram for 4 hours at 30°C, ground them, added 2% salt by weight and fermented blends of 2:1, 3:1 and 4:1. Over fermentation, batter density fell from 0.93 to 0.59 g/cm³, pH from 5.9 to 4.21, and titratable acidity rose from 0.44% to 0.91%. Shear stress peaked at 7 hours. The 3:1 blend reached 0.76 mg riboflavin and 0.73 mg thiamine per 100 g, and folic acid peaked at 0.75 mg per 100 g after 10 hours. Limits: one lab setup, and the vitamin figures are unusually high compared with food composition tables, so treat them as relative changes.

Read the paper (free)

A sensory study from Tamil Nadu by Durgadevi and Shetty used response-surface methods to find the best-scoring idli: IR20 rice and ADT3 black gram at a ratio of 3:1.475 (roughly 2:1), fermented for 10.2 hours. That lines up with the microbiology, since 10 hours sits in the middle of the Weissella leavening window.

VariableValue used in studiesWhat happens if you push it
Rice : urad (weight)2:1 to 3:1More rice gives denser idli; more urad gives softer but flatter, stickier idli
Soaking4 h at 30°C (Ghosh 2011; Mandhania 2019)Longer soaking in a hot kitchen starts souring before grinding
Salt2% of batter weight (Ghosh 2011)Much more salt slows lactic bacteria; none makes flavour flat
TemperatureAbout 26 to 30°CCold rooms stall the lag phase; 35°C+ sours fast
Time10 to 15 hPast 18 h pH bottoms out near 4.2 and the foam weakens

The 2% salt figure is easy to get wrong when batch sizes change, so here is a quick calculator.

Salt for your idli batter at the studied 2%

Why does salt and temperature change the ferment?

Salt and temperature change how fast the bacteria work. The studies here all use modest salt, around 2% or less, because salt draws water away from microbes and high doses slow the gas producers along with everything else.

Temperature sets the speed: Mandhania's batter at about 30°C took 6 hours just to get going, and the same team ran a 4°C control, the temperature of a fridge, which is how households pause a batter.

Winter and summer fixes

For winter nights in North India or a cold air-conditioned flat, the fix is heat, not chemicals:

  • an oven with only its light on;
  • a warm corner near the fridge's compressor;
  • a casserole wrapped in a towel.

The winter fermentation post goes deeper into those tricks. Summer has the opposite risk; a batter left 24 hours at 35°C can go past pH 4.2 and start to break down.

Studies on a cousin batter show how tightly the variables interact. In a 2018 response-surface study of dhokla batter (Bengal gram and rice), Sharma and colleagues found the best fermentation at 8 g/kg of added salt, 32°C and 18 hours, which also cut phytic acid by 94% in the final steamed product.

Salt, warmth and time are three dials on one microbial process.
A crisp golden dosa served with coconut chutney and sambar
Dosa batter is the same ferment run thinner and often with more rice; its sourness and browning come from the lactic acid and sugars the bacteria leave behind.Photo: Roland, CC BY 2.0, via Wikimedia Commons

Does fermentation make idli more nutritious?

Fermentation does change the batter's nutrition: lactic bacteria produce phytase that breaks down phytic acid (which otherwise binds iron and zinc) and they synthesise some B vitamins. Mandhania's team detected phytase activity throughout fermentation and found vitamin B12 production by bioassay, plus amylase activity in the early phases that frees sugars for the bacteria.

The size of the benefit varies by study. Ghosh and Chattopadhyay reported riboflavin and thiamine climbing during fermentation, with folate peaking at 10 hours. A 2016 study by Chelliah and colleagues added 5% dried curry leaf powder to idli batter, which raised calcium about tenfold over plain idli and stretched batter shelf life at 30°C from 2 days to 5, a reminder that additions change both nutrition and microbiology. For blood sugar effects of idli and dosa, see the glycaemic index of Indian staples.

Role of Leuconostoc mesenteroides in leavening the batter of idli, a fermented food of IndiaMukherjee, Albury, Pederson, van Veen and Steinkraus · Applied Microbiology · 1965

The first microbiological study of idli. The authors isolated and identified the bacteria present at each stage and found a sequence: Leuconostoc mesenteroides dominated and produced both the acid and the gas, followed later by Streptococcus faecalis (now Enterococcus) and then Pediococcus cerevisiae. They concluded it was the first known leavening done solely by a lactic acid bacterium. Limits: culture-based methods from 1965 miss organisms that are hard to grow, which is likely why Weissella, a close relative of Leuconostoc, only showed up clearly with DNA sequencing decades later.

Read the paper (free)

Should you add baking soda, yeast or poha?

You do not need any of them if the grains are fresh and the batter is warm, because the native bacteria supply both the gas and the flavour.

AdditiveWhat it doesCost
Baking sodaInstant CO2 by neutralising acid; lifts a weak batterSoapy taste; destroys some of the B vitamins the ferment built
YeastSpeeds risingBread-like smell
Poha or cooked riceEasily fermented starch and softnessAdds no new microbes to speak of

The better fix for a batter that will not rise is the one the 2019 study implies: give it a starter. Their inoculation tests with 10% of already-fermented batter and with a pure Weissella culture were designed for exactly this, and backslopping (saving a spoon of good batter for the next batch) is how many households and vendors keep a consistent ferment. The step-by-step method is in how to make idli and dosa batter. More on Indian food science is on the Rasa world page.

Small button idlis on a plate, showing their soft spongy texture
Button idlis. The fine holes are CO2 bubbles from bacterial fermentation, set in place by steaming.Photo: Randhirreddy at English Wikipedia, CC BY-SA 3.0, via Wikimedia Commons

Questions people ask

Why is my idli batter not rising?

Usually the room is too cold, the dal was under-ground (not enough foam), or the grains carry few bacteria. Warm it to about 30°C, grind urad until fluffy, and add a spoon of previously fermented batter as a starter.

Should salt go in before or after fermentation?

Most studies, including Ghosh and Chattopadhyay (2011), add about 2% salt before fermenting. In hot weather some cooks add it after to avoid slowing the ferment in cold weather; the effect is modest at normal salt levels.

Can idli batter ferment in the fridge?

Barely. At refrigerator temperatures lactic bacteria are nearly dormant, which is why the fridge is used to stop fermentation once the batter has risen.

Is sour idli batter safe to eat?

Sourness is lactic acid and is normal. A pH near 4.2 to 4.5 is typical of well-fermented batter. Putrid or rotten smells, slime or coloured spots are signs of spoilage, and that batter should be discarded.

Why do dosa batters use more rice than idli?

Dosa needs a thin batter that spreads and crisps rather than a foam that holds gas, so more rice starch and less urad gum suits it. TNAU's note on rice fermented foods describes dosa batter as finely ground and very thin.

Sources

  1. Mandhania M.H. et al., Applied and Environmental Microbiology: Diversity and Succession of Microbiota during Fermentation of the Traditional Indian Food Idli (2019; pH, volume, Weissella, sterilised grains)
  2. Kavitake D. et al., Journal of Food Science and Technology: Bacterial diversity of traditional fermented food, Idli by high throughput sequencing (2022; phyla shift 0 to 15 h)
  3. Mukherjee S.K. et al., Applied Microbiology: Role of Leuconostoc mesenteroides in leavening the batter of idli (1965)
  4. Ghosh D. and Chattopadhyay P., Journal of Food Science and Technology: Preparation of idli batter, its properties and nutritional improvement during fermentation (2011; density, pH, acidity, salt 2%)
  5. Durgadevi M. and Shetty P.H., Journal of Food Science and Technology: Effect of ingredients on sensory profile of idli (2014; 3:1.475 ratio, 10.2 h)
  6. Chelliah R. et al., Journal of Food Science and Technology: Bio-fortification and shelf-life extension of idli batter using curry leaves (2016)
  7. Sharma A. et al., Journal of Food Science and Technology: Preparation of antinutrients-reduced dhokla using response surface process optimisation (2018)
  8. TNAU Agritech Portal: Fermented products based on rice (black gram globulin and arabinogalactan)
  9. Deccan Herald: 97 employees fall ill after suspected food poisoning at Bengaluru office canteen (2026)
  10. Metro India: MTR launches Minute Fresh Batter range in Hyd (2026)
idlidosafermentationlactic acid bacteriaWeissellaLeuconostoc mesenteroidesurad dalbatter pH
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