Techniques

Chapter 14 — Ten Techniques That Decide Everything

There are perhaps forty things a Kerala cook does with her hands, and ten of them decide almost everything. Learn the ten and the recipes in this book become instructions you can argue with rather than orders you must obey.

Each technique is given the same way: what it is, why it happens, how it fails, and what it has to do with the reason you picked up this book. That last part matters more than it looks. Almost every genuine fat-reduction lever in Kerala cooking is not a substitution but a technique — a change in when the fat goes in, how the body of a gravy is built, how much of a small quantity of oil you can persuade the tongue to notice. Nothing here asks you to cook a diminished version of Kerala food.

Two of these carry physics you cannot argue with: the coconut milk rule and the curd rule. Learn those first. They are the difference between a curry and a split, grainy disappointment.

1. Thaalikkuka — the tempering, and why its position changes everything

The thaalippu is the smallest complete act in Kerala cooking: fat, heated; mustard seed, popped; dried chilli, darkened; curry leaf, thrown in and made to crackle; often shallot, sometimes urad dal or fenugreek. Thirty to ninety seconds, start to finish. Everything about the technique is decided by what happens inside those seconds.

When a mustard seed hits fat at roughly 170–180°C (340–355°F), the moisture inside flashes to steam and the seed coat ruptures. That pop is not theatre. It is the moment the seed's interior becomes accessible, and the moment its pungent isothiocyanates — formed when the crushed, wetted seed's own myrosinase broke down its glucosinolates — become available to the oil. Popped mustard tastes nutty and rounded; unpopped mustard tastes raw and faintly bitter, and will taste that way for the life of the dish. Curry leaf is different: it is a leaf full of volatile terpenes held in oil glands, and hot fat both bursts those glands and dissolves what comes out. The leaf crackles because its water is leaving. Its aroma reaches you because the fat has taken up the terpenes and is now carrying them.

That single fact — the aromas are volatile, fat-soluble and short-lived — explains the whole grammar of tempering.

Tempering last, poured over a finished dish. Here the point is aroma and texture, both fragile. A tempering laid over thoran, pachadi, kichadi, pulissery, moru or kappa puzhukku arrives at the table still audible: the shallot crisp, the leaf intact, the volatiles undiminished. This is a perfume, not a foundation.

Tempering first, with the dish built on top of it. Here the point is foundation. The aromatics infuse the fat; the fat then coats every ingredient that follows, so flavour is distributed rather than layered. This is how you start sambar's base, a mezhukkupuratti, an ularthiyathu, most meat roasts.

Many good dishes have both. If a method makes you ask whether you have already done this, the answer is that the first tempering flavours the food and the second flavours the air above it.

Now the metabolic point, which is an advantage rather than a compromise. A tempering finished raw over a dish delivers far more aroma per gram of fat than the same fat cooked through, because none of the volatiles have been driven off and none of the oil has been absorbed into the food. Two teaspoons of coconut oil, properly hot, mustard actually popping, poured over a thoran as it leaves the pan, will read as more coconut oil than four teaspoons stirred in at the start. You are putting the oil where the tongue and the nose can find it.

How to do it properly

1. Heat the pan dry, then add the oil. Oil heated from cold spends longer in the range where it oxidises and less time where it works.
2. Wait for the shimmer, not the smoke. Coconut oil is ready when the surface moves in fine ripples and a single mustard seed pops within two seconds.
3. Add the mustard alone and let it finish — vigorous, then subsiding, 20–30 seconds. Add nothing while it is loud.
4. Then dried chilli, then urad dal or fenugreek, then curry leaf last. The leaf is wet, drops the oil temperature and burns fastest: 5–10 seconds, until the crackling turns fine and the leaf goes glossy and slightly translucent.
5. Get it out of the pan. Residual heat keeps cooking a tempering.

Failure modes. Oil not hot enough: the mustard swells and sputters weakly, and the dish carries a raw seed bitterness. Oil too hot: the mustard blackens within seconds and there is no rescue — start again, it costs you a teaspoon. Curry leaf still wet: the oil spits and the leaf steams rather than crackles. Everything added at once: the mustard never gets its window.

2. Thenga aracha versus thenga chertha — the grammar of coconut

This is the most important technical distinction in Kerala cooking, and most books in English miss it. It is not a distinction between kinds of coconut but between two ways of building a gravy, and it organises the majority of Kerala's curries.

Thenga aracha means "with ground coconut." Grated kernel is ground to a paste — usually with shallot, cumin, green chilli and turmeric, sometimes fennel or garlic — and that paste is cooked into the dish. What goes in is not a liquid but a solid suspension: cell walls, fibre, protein, starch and fat still bound together in fragments. Cooked, it thickens a gravy the way a purée thickens a soup, by adding particles that hold water. The result is thick, opaque, faintly sweet, and it clings. This is the grammar of avial, erissery, kalan, the paste-based thorans, meen peera, and most sadya vegetable curries.

Thenga chertha means "with coconut added," and what is added is milk — fat and protein pressed out of the kernel into water, the fibre left behind in the spent gratings. What goes in is an emulsion, and an emulsion does not thicken a dish; it enriches it. Coconut milk makes a gravy paler, silkier, thinner in body and considerably more fragile. This is the grammar of olan, ishtu, mappas, moilee.

The consequences run all the way through. A ground-coconut curry holds together on standing and reheats without incident, because its body comes from insoluble particles that do not care about temperature. A coconut-milk curry separates if you push it, because its body comes from an emulsion that can break. A ground-coconut curry tastes of coconut and of whatever was ground with it, which is why the shallot and cumin go into the paste. A coconut-milk curry tastes of coconut alone, which is why olan needs almost no spice.

For this book the distinction is a lever. Fresh grated coconut carries about 10 g of fibre and 3.8 g of protein per 100 g alongside its fat (ICMR-NIN IFCT 2017). Coconut milk carries the fat and little of the fibre, because the fibre stays in the thengaappu you throw away. So a curry bodied with ground coconut delivers fibre and protein with its richness; a curry bodied with coconut milk delivers richness alone. Where a dish will accept either — and many will — the paste is the better choice on both flavour and composition. Where it will not, do not force it. An olan thickened with coconut paste is not a lighter olan; it is a bad erissery.

3. Varutharacha — the third register

Roast the grated coconut dark before you grind it and you are no longer making the same dish. Varutharacha is not a variant of thenga aracha but a third grammar with its own repertoire: varutharacha meen curry, varutharacha chicken curry, the roasted-coconut crown of an erissery, several Kottayam and central Travancore vegetable curries that are unrecognisable without it.

Roasting does three things at once. It drives off water, concentrating everything left behind. It runs Maillard reactions between the kernel's own sugars and amino acids, generating the brown pigments and the roasted, nutty, faintly meaty aroma compounds no spice will imitate. And it breaks cell walls so that some of the coconut's fat renders out — which is why properly roasted coconut goes glossy and looks oily just before it is done, and why the paste comes off the grinder smoother and darker than a raw one.

How to do it

1. Heavy dry pan, no oil, medium-low flame. Coconut roasts unevenly on high heat: the outside blackens before the inside dries.
2. Stir continuously with a flat turner — 12–18 minutes for 200 g, and it does not tolerate being left.
3. Read the three stages. White to ivory, smelling faintly of hot milk, 4–6 minutes. Then pale gold, smelling of biscuit, 4–5 minutes more. Then the colour of dark toast, visibly releasing oil, smelling nutty with a distinct edge of smoke, a further 4–6 minutes. That third stage is where you stop.
4. Add the whole spices only for the last two minutes — coriander seed, dried chilli, fennel, a curry leaf sprig. They will burn while the coconut is still pale.
5. Tip it out the moment it is right. A hot pan takes it past the point in ninety seconds.
6. Grind warm, with as little water as the machine accepts. Warm coconut grinds smooth; cold roasted coconut grinds gritty.

How far, and how to know you overshot. The target is dark brown with a red cast, glossy, aromatic. The failure is black flecks and a smell that has stopped being nutty and started being burnt — a hard, ashy bitterness that reads as harshness rather than depth. Coconut has a real bitterness threshold, and once it is crossed every mouthful of the finished curry turns slightly astringent. There is no correction: no jaggery, acid or coconut milk will bury it. Throw it out and roast another batch.

The reason this belongs in a metabolic-health book is simple. Roasted coconut paste is the most flavour-dense thickener in the Kerala pantry. A curry built on 40 g of varutharacha tastes deeper than the same curry built on 150 ml of thick coconut milk, and carries less fat and more fibre doing it. Depth is not the same thing as richness, and roasting is how you buy depth cheaply.

4. The three extractions — onnaam, randaam and moonnaam paal — and the physics of why thick milk must never boil

Kerala kitchens do not have "coconut milk." They have first, second and third milk, and they use them for different jobs.

The procedure and the yields. From 400 g of freshly grated coconut — one large coconut, about four cups:

1. Grind the gratings with 120 ml warm water, 30–45 seconds in short bursts, stopping before the jar warms. Squeeze the mash hard through your fist or through muslin: about 200 ml of first milk (onnaam paal), thick, opaque, the colour of full cream, roughly a quarter fat by weight.
2. Return the mash with 300 ml warm water, grind briefly, press again: about 320 ml of second milk (randaam paal), thinner, distinctly coconut-flavoured, perhaps 8% fat.
3. Repeat with 400 ml water: about 420 ml of third milk (moonnaam paal), barely more than flavoured water at 3% fat or so.

The spent gratings, thengaappu, still hold most of the coconut's fibre and are worth stirring into a thoran rather than discarding.

Why the first milk must never boil. Coconut milk is an oil-in-water emulsion: droplets of coconut fat suspended in water, held apart by a thin interfacial film of coconut proteins, mainly globulins and albumins, assisted by phospholipids. That film is the only thing between you and a broken curry. Heat denatures protein — the molecules unfold, lose the shape that let them sit at the oil–water boundary, and aggregate with each other instead. Sustained boiling therefore destroys the stabiliser and supplies the agitation that lets naked fat droplets coalesce. What you see is clear oil on the surface and grainy white curds beneath. It is irreversible. Second and third milk survive boiling far better, simply because there is less fat to separate and more water to hold it.

The rule is absolute. Cook the vegetables, fish or meat in the second and third milk, which may simmer freely. Add the first milk at the end, off the heat or at the barest tremble where a bubble breaks the surface every few seconds and no more. Then swirl the pan; do not stir it. Do not reheat a first-milk curry at a rolling boil the next day either — warm it to just steaming.

Now the consequence this book cares about. A curry finished properly with thick milk needs measurably less of it than one boiled into submission, because an intact emulsion coats the tongue and a broken one does not. Fat that has separated and floated is fat you can see and cannot taste; fat still in suspension reads as body across the whole mouth. In practice a mappas or an ishtu finished off the heat with 120 ml of first milk tastes richer than the same dish boiled with 200 ml. The technique is the fat reduction.

Same physics, different protein. Curd and buttermilk dishes — pulissery, moru curry, kalan, kichadi — are stabilised by milk proteins that behave the same way and split at even lower temperatures, helped along by their own acidity. Warm them through, never boil them, temper afterwards. If you learn one sentence from this chapter, learn that thick coconut milk and curd both go in at the end.

5. Choosing the acid

The souring agent is the most reliable community marker in Kerala cooking, and choosing it accurately is a craft skill on the level of seasoning. Six acids do almost all the work, and they are not interchangeable.

Kudampuli, the sun-dried rind of Garcinia gummi-gutta, is smoky, rounded and deep rather than sharp, and its acidity integrates into a coconut gravy instead of sitting on top of it. It firms fish protein, which is why a meen vevichathu holds its steaks while a tamarind curry tends to shred them. And it is the one acid in the Kerala pantry that makes a curry improve overnight: its acids keep diffusing into the flesh while the gravy's soluble compounds migrate in and the fish's migrate out, so the two converge. Tamarind flattens instead. Its brightness is a top note, and top notes dissipate; a tamarind curry on day two is rounder but duller. This is the single most useful thing to know about Kerala's two main acids. Rinse kudampuli and soak it 15–20 minutes; two or three pieces will sour a whole pot. Commercial smoke-curing can mask poor fruit, which is why one bag tastes of clean fruit acid and the next of a bonfire.

Tamarind is flat, bright, frankly acidic and fruity, with real body from its pulp, so it thickens slightly as well as souring. It belongs in sambar, rasam, chutneys, inji puli and some meat dishes — the sadya and Tamil-inflected acid, and wrong in a Kottayam fish curry.

Raw mango is fresh, sharp and seasonal, and brings fruit flesh as well as acid, contributing body and a little sweetness. In the pre-monsoon glut it stands in for kudampuli in coastal households, and it defines kichadi, pachadi and manga curry.

Kokum is mild and fruity, and marks the boundary of a different culinary language inside Kerala — the Konkani and Gowda Saraswat kitchens of Kochi, sol kadhi and ambat.

Lime is pure top note, gone within minutes of heating. A finishing and pickling acid, never a cooking acid.

Curd and buttermilk bring lactic acid, soft and cooling rather than sharp, plus milk protein and a faint sweetness. They are the Brahmin and sadya acids and they carry the boiling constraint above. Note what acid does to protein generally: it firms fish, it loosens collagen in tough meat over long cooking, and it sets curd gravies if you are careless with heat.

The metabolic use of all this is the most valuable thing in the chapter after the oil table. Acid is the most powerful single replacement for salt available to a home cook. Acidity amplifies perceived saltiness and lifts the whole flavour profile, and a properly soured dish will take a genuine 20–30% less salt without anyone at the table noticing. Kerala cooking is unusually well equipped to exploit this, because its souring agents are already the best salt-reduction tools in any Indian pantry. Learning to sour a dish accurately is learning to salt it less. When a curry tastes flat and your hand goes to the salt, reach for the lime or a fourth piece of kudampuli first, and taste again.

One boundary must be stated, and Chapter 9 sets it out in full. Kudampuli in a fish curry is a fruit rind used in gram quantities as a flavouring, with centuries of culinary use and no association with liver injury. Kudampuli sold as "Garcinia cambogia" capsules is a concentrated hydroxycitric acid extract linked to rare but severe liver injury (Onakpoya et al., 2011; NIH LiverTox). Cook with the fruit. Do not take the capsules.

6. Clay-pot cooking and residual heat

An unglazed earthenware meen chatti changes a fish curry, for three reasons that are all straightforwardly physical.

It is chemically inert to acid. Fired clay does not react with kudampuli or tamarind; iron and aluminium do, and the metallic edge they lend a strongly soured curry is real and unpleasant. That alone justifies keeping an unglazed pot or a kalchatti for sour dishes.

It has high thermal mass and low conductivity. Clay heats slowly, cools slowly, and spreads heat gently across its wall rather than concentrating it at the base. The result is a simmer soft enough that fish holds its shape — no localised boiling to knock the steaks apart, no scorched patch under the flame.

It is microporous. A little of every curry soaks into the wall and returns to the next one. A well-used meen chatti is a flavour archive, which is why families refuse to throw one out and why they cook nothing but fish in it. Never scour it with detergent; you would be washing away the reason you own it.

Seasoning a new pot. Soak it in water for 12–24 hours. Then fill it with water, add a handful of rice or a spoon of rice flour, and simmer 40–50 minutes. Discard, dry thoroughly, rub the inside with a little coconut oil, and use it for three or four ordinary curries before trusting it with anything you care about.

What cracks one. Thermal shock, almost always: a cold pot onto a high flame, a hot pot onto a wet counter, cold water into a hot pot, or a pot left dry over heat. Bring it up slowly, always with liquid or food in it, and cool it on a wooden board. A hairline crack in a well-seasoned pot often seals itself with use; a crack you can see light through is finished.

Cooking with residual heat. A clay pot at a simmer holds enough stored heat in its wall to keep cooking its contents for 8–12 minutes after the flame is off. So this book's fish curries tell you to cook until nearly done and then turn the flame off: the fish sets, the gravy settles, the tempering aroma stops being driven off, and the emulsion is never pushed past its limit. Add thick coconut milk at this point and it warms through without approaching a boil. Cook Kerala food in clay and you will find yourself turning the heat off five minutes earlier than you would in steel.

7. Fermenting the batters

The fermented Kerala breakfast is not slow food. It is food that works while you sleep, and it is retreating from Kerala kitchens because people have started treating it as a project rather than a routine.

The craft. For idli or dosa, the ratio that matters is rice to black gram dal, and 3:1 by weight is the workhorse — 300 g parboiled rice to 100 g urad dal, plus a teaspoon of fenugreek soaked with the dal. Move towards 2.5:1 for softer idlis and a faster ferment. Soak rice and dal separately, 4–6 hours. Grind the dal first, with cold water, until it is white, aerated and holds a peak; this is where the lift comes from and it is worth two minutes of patience. Grind the rice coarser. Combine by hand — your palm mixes air in and warmth from your hand starts the fermentation. Water is the commonest error: an idli batter should fall off the ladle in a thick ribbon, not pour, and a dosa batter is thinned after fermentation, never before, because thin batter cannot trap gas.

Reading a risen batter. Three signals, and you need all three. Volume: risen by half to double, surface domed rather than flat. Bubbles: tilt the bowl and look at the side of the vessel for a fine even honeycomb through the body of the batter, not a few large cavities, which mean over-fermentation and collapsing structure. Smell: clean, faintly sour, faintly yeasty, like the air in a bakery. Sharp, acetic or cheesy means it has gone too far — it will still make dosa, which forgives sourness, but it will make sullen idlis.

Temperature and timing. At 28–32°C (82–90°F), 8–10 hours is right; leave it 14 and you will have a sour flat batter by morning. At 18–20°C (64–68°F) — most of the diaspora most of the year — the same batter needs 14–20 hours and may still not rise. The fixes, in order of usefulness: an oven with only the light on, which holds about 30°C (86°F); grinding with water at 40°C (104°F) rather than cold; a longer, more aerated dal grind, so you build physical lift you are not relying on biology to supply; a higher dal fraction; and a larger vessel, so the batter has a deeper mass to hold its own warmth. If nothing has moved in 20 hours, a quarter-teaspoon of instant yeast finishes the job. That is not cheating; appam batter in many households was leavened with toddy, and later with yeast, anyway.

Where the safety guidance stands, which is thinner than you might expect. No national food-safety authority — not the UK Food Standards Agency, not the FDA, not FSSAI — publishes guidance on fermenting an idli or dosa batter at home. The one public-health document that addresses this batter by name is the BC Centre for Disease Control's Safety of Fermented Foods, which describes the batter as "left to ferment naturally overnight (typically at least 12 hours) at around 30°C [86°F]", says "fermentation should take place within 12-14 hours (at ±30°C [86°F])", records that after twelve hours "the pH of the batter ranged from around 4.3-4.6", and names E. coli, Staphylococcus aureus and Bacillus cereus as the organisms that matter. Three practical consequences follow, and this book adopts all three.

One, the soak. BCCDC: "if soaking takes place at room temperature, limit soaking time to a maximum of 4 hours." So soak four hours on the counter, and if you want six, soak in the refrigerator.

Two, the long cold ferment. A 14–20 hour ferment in an 18–20°C kitchen is longer than the BCCDC's window, which is written for a warm room where the acid arrives fast. In a cold kitchen the batter is spending more time at a pH that has not yet fallen. This book still gives the cold-kitchen timing, because it is what the food requires, but the honest framing is: warm the batter rather than extend the clock. The oven-light trick and the 40°C grinding water are safety measures as much as convenience ones.

Three, afterwards. Refrigerate as soon as the ferment is done. The BCCDC gives fermented batter "approximately 1 day at room temperature", extending to "approximately 5-7 days" refrigerated. And a batter that has not risen, has not soured and smells of nothing after twenty hours has not fermented; it has simply sat. Discard it rather than adding yeast to a batter that has been at room temperature for the best part of a day.

Now the evidence, honestly. Fermentation of a rice-and-dal batter is driven by Leuconostoc mesenteroides and related lactic acid bacteria, with wild yeasts contributing to appam. What it genuinely does: it raises some B vitamins measurably — riboflavin, thiamine and folate all increase during fermentation (Ghosh and Chattopadhyay, 2011); it partially hydrolyses protein and degrades legume trypsin inhibitors and tannins, raising protein digestibility; and the falling pH activates phytases, degrading phytate and improving the bioaccessibility of iron and zinc. The rice-and-black-gram pairing also complements limiting amino acids, rice being short on lysine and black gram on methionine. A long ferment and a generous dal fraction both help.

What is routinely overstated: fermented batter is not a meaningful source of vitamin B12. The compositional study usually cited for it reports riboflavin, thiamine and folate and does not report B12 at all; the dominant fermenting organisms are not efficient cobalamin producers; and where B12 has been detected in Indian batters the values are small, wildly variable between households, and much of what an immunoassay picks up in a bacterially fermented plant food is corrinoid analogue rather than bioactive cobalamin — analogues can be inert in humans. B12 is also heat-labile, and steaming or griddling takes more. If you are vegetarian, do not rely on idli for B12; Chapter 32 covers what to do instead.

On glycaemic index the honest answer is narrower than you will read elsewhere. The mechanism is sound: fermentation's lactic and acetic acids slow gastric emptying and blunt the postprandial glucose rise, which is the best explanation for sourdough outperforming yeast bread at matched composition, and a fermented batter sits at pH 4.2–5.9 with 0.44–0.91% total acidity. That is fair to state as mechanism. This book will not state a number, because a clean measurement of the same batter fermented against unfermented does not exist in the published literature. What is firmly established is that the grain matters far more than the fermentation: in Indian parboiled rice, glycaemic index rose from 57.6 unpolished to 73 at 2.3% polish and 79.6 at 9.7% polish, and the authors concluded that any degree of polishing raises glycaemic response (Shobana et al., 2017). So ferment fully, raise the dal, include the fenugreek — noting that fenugreek has real glycaemic effects at 10–25 g a day and can compound insulin and sulfonylureas, which the teaspoon in a batter does not approach but Chapter 9 sets out in full — then spend your real effort on buying less-polished rice and on never eating idli or appam without a pulse-and-vegetable accompaniment. Three appams are three servings of rice, whatever the batter did overnight.

8. Building body without fat

This section addresses the actual reason people pour oil and coconut milk into Kerala food: they are chasing body. Viscosity, not fat quantity, is what the tongue reads as richness. Six methods deliver it, each at a different cost.

1. Slow-reduced shallots and onions. Sliced shallots cooked on a low flame in a covered pan for 18–25 minutes, until they collapse, lose their water and turn the colour of weak tea, become a purée that thickens a gravy substantially and contributes a sweetness and depth that briefly fried onion cannot. Another 8–10 minutes uncovered takes them jammy and darker. Buys: real body, sweetness, and a savoury base that reduces how much salt the dish wants. Costs: half an hour of low-attention time. The most transferable technique on this list, which is why M15 exists.

2. Ground coconut paste as thickener, rather than coconut milk as enrichment. Forty grams of coconut ground with shallot and cumin thickens a curry for four more effectively than 200 ml of coconut milk, and brings fibre with it. Buys: opacity, cling, coconut flavour, fibre. Costs: ten minutes and a clean grinder — and it changes the dish's identity. This is an aracha curry now, and should be named as one.

3. A spoonful of cooked dal, blended in. Two tablespoons of cooked toor or moong dal, blended smooth with a little gravy and stirred back, thickens like a starch and adds protein without announcing itself. Buys: silkiness, protein, and a lower glycaemic load for the whole meal — pulses have the best evidence base in this book. Costs: almost nothing, if you keep cooked dal in the fridge. Which is the point.

4. Vegetable purée. Ash gourd, pumpkin or bottle gourd, cooked soft and blended, gives thickness at 17–21 kcal per 100 g. Ash gourd is nearly flavourless and vanishes; pumpkin brings colour and sweetness and will show. Buys: volume and body at almost no energy cost, and it is how a curry for four stretches to six. Costs: 15–20 minutes, plus accounting for pumpkin's sweetness.

5. Roasted-coconut paste for depth. Where the others add thickness, varutharacha adds thickness and the roasted, faintly smoky depth that otherwise comes only from fat and long cooking. Buys: the most flavour per gram of anything here. Costs: 15–18 minutes of continuous stirring and a real risk of overshooting into bitterness. Make it in batches and freeze it, which is what M09 is for.

6. Reduction by patience. A gravy simmered uncovered for an extra 10–15 minutes loses water, concentrates everything dissolved in it, and thickens without gaining a calorie. Buys: intensity and body, free. Costs: time, attention, and a slightly smaller quantity of food than you started with.

The principle to carry away: when a dish tastes thin, the reflex is to add fat, and the fat is almost never the answer. Ask instead whether it needs reduction, a purée, a starch or a paste — then add the half-teaspoon of coconut oil off the heat at the end, where it will be noticed.

9. The measured-oil method

Every recipe in this book gives oil in grams. A tablespoon of coconut oil poured from a bottle is anywhere between 12 and 20 g depending on how you hold it, which is 110–180 kcal of variance in a dish for four, repeated three times a day. A digital scale is the highest-return purchase in a fat-loss kitchen, and oil is the reason.

The technique that makes small quantities work has five parts.

1. Heat the pan before the oil goes in, so the oil reaches working temperature immediately and spends no time pooling and soaking into food while it warms.
2. Use a wide, heavy pan for shallow work and a narrow one for tempering. Ten grams of oil covers the base of a 26 cm pan as a film; the same 10 g in a 20 cm pan is a puddle that some of the food sits in and the rest sits above. Match the vessel to the quantity, not the quantity to the vessel.
3. Swirl and tip. Add the oil, lift the pan, swirl so the film runs up the sides and back, and tip out anything that pools. You will be surprised how little stays and how completely it covers.
4. Brush rather than pour, for anything with a surface. A silicone brush puts 2–3 g onto a whole fish or a dosa griddle; pouring puts 10 g on and 7 g of it into the food. For appam in a seasoned appachatti, a brush wiped almost dry is enough, and often nothing is needed.
5. Distinguish the oil that cooks from the oil that flavours, and finish raw. Oil used to conduct heat and prevent sticking is a functional quantity and should be as small as the pan allows. Oil used for flavour goes in off the heat at the end, where its volatiles survive and the tongue meets it directly. Half a teaspoon of coconut oil stirred into an avial off the heat is perceived more strongly than two teaspoons cooked in from the start. This is the highest-leverage oil-reduction move in the book, and it is also how avial, olan and kappa puzhukku were always finished.

One point about the oil itself. Coconut oil is about 91% saturated and raises LDL cholesterol relative to unsaturated oils (Neelakantan et al., 2020); sesame, groundnut and rice bran run 16–24% saturated. So keep coconut oil where it is the flavour — tempering, thoran, a finishing drizzle, the dishes that are unrecognisable without it — and use groundnut, sesame or rice bran for volume work and shallow frying. Do not reuse frying oil more than once, never top up old oil with fresh, and discard oil that has darkened, thickened, foams or smells acrid.

What each dish is usually cooked with, and what it actually needs. Quantities are per four servings.

The honesty in that table is deliberate. A book claiming every dish can be made with two teaspoons of oil is a book you should not trust on anything else. Two of Kerala's best things — a properly deep-fried mackerel and a plate of upperi — need oil, need a lot of it, and are worth eating occasionally as themselves. What this book does about them is tell you the truth and then tell you how often.

10. Doneness by sense, not by clock

Every recipe here pairs a sensory cue with a time range, in that order, and asks you to trust the cue. A time is a property of a particular pan on a particular flame with a particular quantity of food at a particular starting temperature, and you have none of those. A cue is a property of the food.

When the raw smell lifts off a masala. Ground spices in fat go through a distinct transition. At first the smell is dusty and slightly harsh — the raw note, largely unhydrated ground spice plus the pungency of raw chilli and coriander. Then it lifts, and what replaces it is rounder, warmer, more integrated. On the tongue the difference is clearer still: the raw paste tastes of separate spices, the cooked one tastes of a masala. Reckon 4–6 minutes on a medium flame for a wet paste, 2–3 for powders in oil.

When a paste darkens by a shade. Shallots, ginger-garlic, coconut paste: the change you want is one step down the colour scale, not two. Pale gold to gold; gold to light brown. The interval is short — 60–90 seconds at the end of a 15-minute cook — and it is where the flavour arrives.

When oil separates at the edge of a gravy. A thin rim of fat around the circumference of a simmering masala means the water has largely gone and the emulsion of spice, aromatic and fat has tightened: the base is cooked and ready for liquid. Note the exception — oil separating on the surface of a finished coconut-milk curry is the opposite signal. That is a broken emulsion, and it means the milk boiled.

When a fish is set. Raw flesh is translucent and slippery; cooked flesh is opaque and flakes along its lines. Press the thickest part with the back of a spoon: it should feel firm and give slightly, not squash. In a clay pot at a simmer, steaks take 8–12 minutes and keep cooking after the flame is off, so take them off before they look done rather than after.

When a thoran is dry enough. Tilt the pan; if water runs to the low side, keep going with the lid off. A finished thoran is dry to the point where the coconut looks like coconut again rather than a wet paste, and the vegetable is cooked but still has bite — usually 6–9 minutes covered plus 2–3 uncovered.

When a batter is spread thin enough. For dosa, the test is light: hold the griddle up to a window and a properly thin dosa shows a faint translucence before it browns. For appam, the lace is the test — swirl once, and the batter should climb the sides in a thin film that sets into holes within 20–30 seconds while the centre stays thick. An opaque edge means the batter is too thick or you swirled twice.

Two habits make all of this reliable. Taste at every stage rather than at the end — you cannot correct a dish you have met only once. And smell the pan deliberately, before and after every transition, so you build a memory of what a lifted raw note actually smells like. The time ranges in this book exist so that you know roughly when to start paying attention. The cue is what tells you the dish is ready.

A standing note. Techniques 4 and 7 are referenced by name throughout Part Three, and the mother preparations in Chapter 15 are this chapter made portionable: M08 is technique 4, M09 is technique 3, M10 to M12 are technique 7, M15 is technique 8, and M06 and M07 together are technique 5 turned into two jars you can keep on a shelf.