How long should you cook idlis?

Idlis are very popular in South India, mainly for three reasons – It’s an easily scalable recipe, so you can feed a crowd without a hassle. Idlis are easily digested and nutritious. But the most convenient reason is for the cook –  they cook in a jiffy! What more can you ask for? 🙂 

But exactly how long should you cook idlis? This is an important question because undercooking leaves the insides sticky. And overcooking makes the idlis dry and less fluffy. Let’s see if we can find out the optimal time for cooking idlis, for whichever mould you use to steam them. 

Stages in steaming

So, what exactly happens when you cook idli batter? You would ‘preheat’ your idli vessel by adding a cup or so of water into it and let it come to a boil. Then, you would pour a little of the batter into each cavity of your idli plate/mould. You would then place the idli plate/mould inside your vessel at such a height that the boiling water does not reach the plate/mould directly. You would close the lid of the vessel, which has a vent to let the steam escape (ie., you don’t pressure cook idlis). After a while, you open the lid and take out your idli plate, having no clue whether the idlis would have completely cooked or not. 

To find out how much time it takes to cook and also, to understand the stages that happen in the cooking process, I poured equal amounts of batter into 7 identical tiny bowls (katoris) and placed them on my idli plate. I placed the plate into my idli vessel which had a cup of water at boiling stage (this is important – that the water is boiling hot before you place the batter in the vessel). I closed the lid and after every passing minute, I took out one bowl. 

I de-moulded the idlis and compared them to understand how the batter cooks with each passing minute: (The idlis are place bottom-side up)

Here is their cross-sectional view (Not including the 1min and 2 min samples, since they did not hold their shape at all):

So, my understanding of the cooking stages is:

Step 1 – Hot steam circulates the vessel and heats up the top surface of the batter first. So the batter starts forming a cooked ‘skin’ on the top surface. 

Step 2 – The steam also heats up the metal plate which touches the bottom surface of the batter. The hot plate then transfers heat to this bottom surface and the sides, which start cooking.

Step 3 – The heat penetrates the batter through two sides – top (by convection of steam, which is faster) and bottom (by conduction from the hot plate, which is slower than the direct steam). So the top and bottom layers get cooked, the former more than the latter.

Step 4 – The middle layers cook the last. 

Step 5 – The idli cooks through all its layers. Also, by this time, the bubbles in the batter would have expanded fully. The idli reaches its maximum height.

Step 6 – On cooking further, the surface of the idli starts turning dry and the idli starts shrinking slightly in volume, losing some of its fluffiness. 

Conventional methods of checking ‘doneness’:

Some cooks try to check if the idlis are  thoroughly cooked by inserting a knife (some others use their fingers too!) into one of the idlis. If the knife comes out clean, the idlis are cooked. If it is coated with sticky batter, the plate is returned to the vessel to steam some more. But doing this in-and-out business leaves the idlis harder than they should be. Also, this trial and error method is no guarantee against overcooking, which also makes the idlis dry and hard. 

Is there a way to pre-calculate the optimal time for cooking idlis? It should be theoretically possible, right? Here’s my attempt at calculating and testing it:

Factors that matter:

Harold McGee is credited with being a pioneer in exploring the intersection between science and cooking. He is the famed author of ‘On Food and Cooking’, the ‘bible’ of food-science lovers. In his other book called ‘The Curious Cook’, Harold McGee talks about factors that determine the ideal cooking time. Though he explores the case of a steak, the equations he has used are applicable to other foods too, under certain conditions. 

According to his research, the time taken to cook a food (let’s call this ‘t’) is directly proportional to the square of the thickness of the food (let’s call the thickness ‘d’). 

So, t = Cd² , where ‘C’ stands for a constant that varies for each food, depending on its density, the specific heat and the thermal conductivity of the material of which the food is made and of the medium in which it cooks. 

Don’t worry about all of this for now – we will make it simple. You can measure ‘d’, the thickness of the batter, from your mould (use the shortest dimension). We want to find out the time needed to cook it, which is ‘t’. So, if we know what is ‘C’ in the case of idli steaming, we can calculate ‘t’ easily. 

In the experiment I did above, my batter was 2cm thick and it took 6 minutes to cook thoroughly. (5 min was fluffiest, but the bottom surface was a little bit sticky to the touch. So I am taking 6 minutes to be on the safer side). Using these two values, ‘C’ turns out to be 250. 

To test if this calculation holds good, I poured out 4cm thick idli batter into 4 identical tall ‘tumblers’. The tumblers are made of the same material as my idli plate (stainless steel). But they are slightly thicker than the plate –  I hope this doesn’t mess with the values too much. For 4cm thick batter, the time taken to cook should be 250 x (0.04)² , which is 0.4 hours. That is, 24 minutes.

I placed the tumblers into my idli vessel which was already ‘pre-heated with 2 cups of boiling water in it. I closed the lid and let the batter cook. I took out one tumbler each after 10min, 14 min, 18 min, 22 min and 26 min. 

I demoulded them at intervals of 4 min between each. Here is how these ‘idlis’ turned out:

When I placed them side-by-side for comparison, there wasn’t much visible difference to their exterior:

But when I cut them to view their cross-section:

The 10min sample had wet, uncooked batter at the centre. The 14 min sample had a relatively firmer but still-sticky centre. The 18 minute batter was cooked through well and was spongy. The 24 min sample had cooked through but had turned harder. The 26 min sample was also cooked through but hard. So, it turns out that the optimal cooking time was around 18 min and not 24 min as my calculations had shown. 

I thought about reasons why the idlis might have cooked faster than expected. And it dawned on me that there is a fundamental difference between the way these tumbler-idlis were cooked, compared to the katori-idlis I had used as the reference. The katoris were placed on a regular idli plate during steaming. So they did not touch the base of the vessel which was in direct contact with the fire. The heating and cooking of the batter happened mostly due to the circulating steam only. But in the case of the tumbler idlis, I had placed them directly in the idli vessel (since placing them on the idli plate would mean I couldn’t close the vessel with the lid, given how tall the tumblers were). So, the base of the tumblers were in direct contact with the base of the vessel which was directly exposed to fire. This may have caused the cooking of the batter to happen faster, due to conduction of the metal too. 

To check for this hypothesis and to make the cooking method as similar as possible, I am going to try this the next time – use the same tumblers and fill with 4cm batter. But instead of the idli vessel, I am going to place them on the idli plate in a taller, lidded-vessel which does have a similar-sized steam vent. Need to go through my stock of vessels to see which one will be best suited for this! Watch out for my update.