Reference tables

REFERENCE

How to use this section

This is the part of the book you will still be using in ten years. It is not meant to be read through, but opened at the page you need, checked, and closed.

The order follows the way a cook needs it. Sections 1 to 4 are composition — what a food carries per 100 g, then the same foods ranked by energy density, by protein and by fibre. Section 5 is glycaemic index and load, including the values measured on actual Kerala preparations. Section 6 reprints the culinary-dose table from Chapter 9. Then sodium and potassium (7), the fats (8), conversions and scaling (9, 10), substitutions (11), the calendar (12), and storage and the freezer (13, 14).

All nutrient values in this section are estimates drawn from published food-composition tables — chiefly the ICMR-NIN Indian Food Composition Tables 2017 for Indian foods, and USDA FoodData Central for what those tables do not cover. Real food varies: by cultivar, season, soil and ripeness, by the age of a coconut and the fat cycle of a fish, and by how long it was cooked in how much water. Use these figures to compare one food with another and to decide what goes on your plate, not for clinical calculation. If you count carbohydrate to dose insulin, or work to a prescribed sodium or potassium restriction, use the numbers your clinical team gives you.

Two conventions. A dash (—) means the source does not report that value and the food is not a meaningful source of it. not established means no verified value exists in this book's sources. Those cells are deliberate, and better than a guess.

1. Nutrient composition, per 100 g

Reading these tables. Values are per 100 g of raw edible portion unless the row says otherwise. Energy is given in kcal; ICMR-NIN prints energy in kJ, and kcal here is that figure divided by 4.184. Carbohydrate is available carbohydrate, determined analytically, in the ICMR-NIN rows — this is not the same quantity as USDA's "carbohydrate by difference", which counts fibre and other components within the total. The two bases are not interchangeable, so the source column states which one every row comes from, and rows from different bases should not be compared digit for digit. ICMR-NIN codes and USDA FDC IDs are given so that every value can be traced back to its table.

Where the number of sampling regions behind an ICMR-NIN value is one (n=1), the figure rests on a single sample. Those rows are marked. They are the weakest values in the section.

1.1 Cereals and grains

¹ ICMR-NIN labels A017 Paspalum scrobiculatum (kodo millet) in one table and Setaria italica (foxtail millet) in another. One of the two is an error in the published volume. The macronutrient values above are as printed; the species name should not be relied on, and no foxtail millet values should be taken from this entry.

Rice, cooked. Cooked matta rice weighs about 2.6 times its dry weight by the absorption method and 2.5 times by the drained method (R015). Dividing the parboiled-milled figures by 2.6 gives roughly 135 kcal, 3.0 g protein and 29.7 g available carbohydrate per 100 g of cooked rice — the arithmetic is shown so you can redo it with your own measured yield. Every cooked-weight rice figure in this book is derived this way, not measured.

1.2 Pulses and legumes

Dry weight. A pulse takes up two to two and a half times its weight in water on soaking and cooking, so the per-100-g figures below are roughly three times what the same food delivers on the plate. Section 9.6 gives the yields.

Insoluble and soluble fractions, where ICMR-NIN reports them, are in Section 4.

1.3 Tubers and starchy vegetables

² Carbohydrate by difference, not available carbohydrate, and on a different moisture basis from the ICMR-NIN tapioca row above. The two entries for the same plant differ by a factor of two; for Kerala kappa use the ICMR-NIN row.

The one fact worth memorising from this table: all five Kerala starchy staples land between 17.5 and 17.9 g of available carbohydrate and 80 and 89 kcal per 100 g. They are interchangeable by weight. Choosing chena over kappa is a flavour decision, not a metabolic one — except that chena and chembu carry 2.5 to 3.3 g of protein against kappa's 1.0 g.

1.4 Vegetables

1.5 Greens

Moringa leaves are the highest-protein green in this section by a wide margin. That is a compositional fact and carries no therapeutic claim; see Section 6.

1.6 Fruit and plantain

³ The USDA entry labelled "raw" is clearly the ripe fruit, on its carbohydrate content. Tender green jackfruit and ripe jackfruit are two different foods in the kitchen and in the table: green carries roughly a quarter of the available carbohydrate and twice the fibre of ripe. Do not let a recipe or a label blur them.

Green versus ripe, from the numbers: green plantain 17.58 g available carbohydrate against ripe nendran's 24.95, with about 60% more fibre. In the USDA pair, total carbohydrate barely falls on ripening (36.7 → 31.9 g) because carbohydrate by difference counts starch and sugar alike. What changes on ripening is the form of the carbohydrate, not the total.

1.7 Nuts, seeds and coconut

⁴ ICMR-NIN 2017 has no coconut milk entry. Every coconut-milk figure in this book is USDA 170172 scaled by extraction as set out in M08. Fresh milk pressed at home from a given weight of kernel will differ.

The comparison the whole coconut argument rests on: fresh grated coconut carries about 10 g of fibre and 3.8 g of protein per 100 g. Coconut oil carries none of either (Section 1.8).

1.8 Fats and oils

Coconut oil is the only oil for which this book has a verified per-100-g composition. For the other Kerala oils, ICMR-NIN publishes fatty-acid profiles as a percentage of total fatty acids but no per-100-g energy in the tables used here; those percentages, and the saturated–monounsaturated–polyunsaturated comparison, are in Section 8. Ghee is a working assumption, not a measured value: see Section 1.13.

1.9 Fish and seafood

Carbohydrate and fibre are not reported for animal foods in the ICMR-NIN table used here; fish, meat and egg are not sources of either.

⁵ The ICMR-NIN sardine entry is a single sample and is internally inconsistent: its total fat (8.99 g) does not agree with its own fatty-acid table, which sums to roughly 2.65 g. Treat the sardine row as indicative only. ⁶ Atlantic mackerel is a fattier species than Indian mackerel (Rastrelliger kanagurta). Do not apply this row to ayala. It is here because it is the methodologically stronger figure for the fish sold as mackerel outside India. ⁷ Also carries 382 mg calcium and 307 mg sodium per 100 g — the calcium because the bones are eaten, the sodium because it is a canned product.

EPA and DHA, mg per 100 g (ICMR-NIN 2017 Table 7, read directly; several rows are single samples):

1.10 Meat, poultry and egg

⁸ Chicken leg energy: 1,605 kJ as printed implies far more energy than 19.44 g protein and 12.64 g fat can account for (about 798 kJ). The protein and fat figures are usable; the energy figure is not. ⁹ These five rows are the USDA SR Legacy generics used to compute the panels in the meat and egg chapter, because ICMR-NIN's tables as extracted for this book contain no red meat, no poultry other than chicken and no organ meat. Energy marked "calc." is computed at 4 kcal per gram of protein and of carbohydrate and 9 kcal per gram of fat, which is the method stated for every nutrition panel in this book.

1.11 Dairy

¹⁰ ICMR-NIN 2017 has no curd and no buttermilk entry — its milk group covers buffalo milk, cow milk, paneer and khoa only. USDA is therefore the source for both, and Western cultured buttermilk is not identical to Kerala moru, which is curd let down with water. Note the sodium: cultured buttermilk carries about 148 mg per 100 g before a cook adds any salt at all, against 46 mg for plain yoghurt. That matters in Section 7.

1.12 Sugars

¹¹ᵃ Salt, stated once for the whole book: plain salt is costed at 400 mg of sodium per gram (1 level teaspoon ≈ 6 g ≈ 2,400 mg), and the M07 70:30 blend at 275 mg of sodium per gram. Every nutrition panel names which of the two it was computed on, and where a recipe offers both, the panel figure is the M07 one; with plain salt, multiply the salt-derived portion by about 1.45. The L icon and the [LOW SODIUM] tag are read off the panel as printed.

¹¹ The working assumption used in this book, stated plainly because it is an assumption and not a measurement: jaggery is costed as sucrose at 98 g of carbohydrate and about 390 kcal per 100 g. Real jaggery carries some moisture and a small non-sucrose fraction, so its energy per gram sits at or slightly below that of white sugar rather than above it. Costing it as sugar means the panels in this book never understate what is being eaten, which is the direction of error a health cookbook should choose. Nothing in this book's evidence base supports a claim that jaggery is metabolically preferable to sugar.

1.13 Foods this book uses for which no verified value exists

Every row below is a gap, not an omission. Where a documented analogue exists it is named, and where it does not, the honest instruction is to look the food up in the primary tables rather than accept a figure from this book.

2. Energy density ranking

Why this is the most useful number in the book. People eat a roughly consistent weight of food. When the energy density of what is available falls, energy intake falls with it — an effect that holds across weight status, sex and behavioural type, and that appears in children as young as three (Rolls, Physiol Behav 2009). Trials that lowered energy density by adding water-rich foods produced substantial weight loss in people who were never told to eat less. This is the mechanism by which a recipe does the work that willpower is usually asked to do, and it is why half of the Kerala Plate is vegetables.

Read the ranking as a guide to volume. Everything under 50 kcal per 100 g can be eaten by the bowlful; the 50–150 band is where the protein lives and where a satisfying plate is built; above 250 kcal per 100 g, weigh it.

One warning about the bottom of the table. Dry grains and pulses appear high on the list because they are dry. Parboiled milled rice at 352 kcal per 100 g becomes about 135 kcal per 100 g cooked, because it takes up about 2.6 times its weight in water by the absorption method — the single conversion used for every cooked-rice figure in this book, set out in Section 1.1; a pulse behaves the same way. Compare cooked with cooked, or dry with dry, and never one against the other.

3. Protein per 100 g, ranked

For readers working to a protein target — 1.2 to 1.6 g per kg of body weight per day, 25 to 30 g per meal. The right-hand column is the weight of that food, as bought, that delivers 25 g of protein, rounded to the nearest gram. It is arithmetic on the column beside it, nothing more.

Two cautions before you use it. Grains and pulses are listed at dry weight, so the figure in the last column is dry weight too: 105 g of dry moong dal is a large quantity of cooked dal, not a spoonful. And no single plant food in this table is a practical way to reach 25 g — 105 g of dal will get you there on paper, but the plate that actually works pairs a pulse with fish, egg or curd.

The last three rows are the point of the table rather than a curiosity: kappa alone is a near-protein-free meal. The traditional pairing with a fish curry or a pulse is not decoration; it is what makes the meal coherent.

4. Fibre per 100 g, ranked

Total dietary fibre, determined analytically. Where ICMR-NIN separates the insoluble and soluble fractions, both are given, because they do different work: insoluble fibre for bulk and transit, soluble and viscous fibre for the slower glucose and lipid effects. Grains and pulses are dry weight.

Two readings worth keeping. Kadala carries more fibre than any other food in this book, and roughly nine times what raw milled rice carries. And fresh coconut carries 10.42 g of fibre per 100 g while coconut oil carries none — the whole of the book's position on coconut sits in the gap between those two rows.

5. Glycaemic index and glycaemic load of Kerala and Indian foods

5.1 What the two numbers are, and the formula

Glycaemic index (GI) ranks a food by the blood-glucose response to a portion containing 50 g of available carbohydrate, against glucose at 100. It says nothing about how much carbohydrate is on your plate. Glycaemic load (GL) puts the portion back in:

GL = GI × available carbohydrate per portion (g) ÷ 100

A food can have a high GI and a trivial GL because a normal portion contains almost no carbohydrate, and a moderate GI food can carry a very high GL because the portion is large. In a rice cuisine, GL is the number that decides the meal. The conventional bands are GL 10 or less low, 11 to 19 medium, 20 or more high, per portion.

5.2 The caveat that must not be buried

GI values vary enormously between determinations of the same food, and this is not a minor technical footnote — it is the first thing to know about the whole system. In the international tables, rice as a category has a mean GI of 67 with a standard deviation of 17 and a range of 19 to 116 (Atkinson et al., Am J Clin Nutr 2021). That range is far wider than the gap between "white" and "brown" rice, whose category means are 73 and 65. Variety, amylose content, degree of milling, parboiling intensity, cooking time, cooling, what else is in the meal and the individual eating it all move the number, in some cases by more than the whole white-to-brown difference.

Three consequences, stated plainly:

1. A published GI belongs to the specific food, prepared the specific way, measured in the specific group of people the study used. It is not a property of an ingredient name.
2. Many determinations are underpowered. FAO/WHO methodology calls for 8 to 12 subjects; a review of 65 millet studies found most in-vivo work fell short of that (Anitha et al., 2021).
3. This is why the book does not print a GI on any recipe. A GI for a mixed dish is not scientifically defensible, and printing one would be a false precision. Use the tables below for staples, and use portion size — which you control exactly — for everything else.

5.3 Measured on actual Kerala preparations — the values to use first

These are the only values in this section measured on Kerala recipes, cooked as Kerala cooks them, in a 24-week randomised intervention in Kerala (Pavithran et al., Int J Environ Res Public Health 2020; doi 10.3390/ijerph17176232). That is what makes them worth more to this book's reader than any category average.

In that trial, the arm eating South Indian food built from these low-GI staples for 24 weeks showed HbA1c −0.93%, fasting glucose −21 mg/dL, postprandial glucose −37.87 mg/dL, HOMA-IR −6.51 and weight −1.87 kg, with reductions in triglycerides, hs-CRP and apolipoprotein B. That is a study finding in one trial, not a promise about your own numbers.

Two honest qualifications. First, these are single determinations in a single trial; Section 5.2 applies to them as much as to any other GI value. Second, no other peer-reviewed GI determination for matta rice was found anywhere in this book's sources — the "matta is 70% healthier" line that appears on packets and websites traces to retail marketing, not to a measurement, and this book does not repeat it.

5.4 Polishing: the contrast that matters more than the grain's name

Indian parboiled Indica rice (variety BPT-5204), one variety, three degrees of milling (Shobana et al., Asia Pac J Clin Nutr 2017):

The authors' conclusion is the sentence to remember: "any degree of polishing leads to higher glycaemic responses." Even minimal polishing — 2.3%, which a shopper cannot see and a packet does not declare — wiped out most of the advantage. This is why the sourcing instruction throughout this book is unpolished or single-polished parboiled rice, and why "silky" and "extra polished" go back on the shelf.

Related, and separate: pressure parboiling reduced GI by nearly 30% in some studies through resistant-starch formation, while traditional mild parboiling had minimal effect (Boers et al., Br J Nutr 2015). Kerala matta is traditionally parboiled, and whether a given commercial sack was pressure-parboiled is not knowable from the packet. Do not assume the benefit.

5.5 Category means, for comparison only

Use these to compare food groups, never to describe a dish. A category mean applied to a specific preparation is exactly the error Section 5.2 warns against.

5.6 What processing does — the counter-evidence to the word "millet"

Eight healthy adults, 50 g available-carbohydrate portions served as thick porridge (Shobana et al., Int J Food Sci Nutr 2007):

The glycaemic advantage belongs to the grain in an intact or coarsely milled state, not to the word on the packet. Decorticate a millet, mill it fine, pop it, or cook it into a thin porridge and the advantage disappears. Ragi puttu made from very fine flour and eaten hot is not automatically low-GI.

5.7 Glycaemic load, worked

The arithmetic below is transparent and reproducible from the composition tables in Section 1 and the GI values above. It is arithmetic done here, not a published finding, and no source states it as such.

Portion is a bigger lever than colour. The same grain at 50 g dry and at 150 g dry moves the load of the meal by a factor of three, which no grain swap on earth achieves. Both levers are worth pulling; only one of them is under your complete control.

5.8 Values that are not established

6. Culinary dose versus trial dose

How to read this table. "Trial dose" is the dose at which the cited meta-analyses and trials found an effect. "Realistic amount in a serving" is an estimate from ordinary Kerala recipe practice — typical quantity per dish divided by four servings. Those per-serving figures are estimates for comparing orders of magnitude, not measured literature values, and should be read as such. Curcuminoid content of turmeric powder is taken at about 3%, within the commonly reported 2–5% range. The verdict column uses three categories only: meaningful at culinary doses, only at supplement doses, insufficient evidence.

The summary line for the whole table: one spice has an effective dose that overlaps ordinary cooking, one has a reachable food dose if you go looking for it, one is borderline, and the rest are flavour. That is not a disappointing finding. It means the levers that work — the grain, the portion, the pulse, the oil, the salt — are the ones a cook actually controls.

7. Sodium and potassium quick reference

7.1 The arithmetic, which is worth memorising

To go the other way: sodium in mg ÷ 400 = salt in g. UK and EU labels give salt; US and Indian nutrition panels and every clinical target give sodium. Mixing them up is a common and consequential error.

7.2 The targets

On that last row, honestly: no Kerala 24-hour urine data were identified in this book's sources. Andhra Pradesh is the nearest southern data point. Given Kerala's consumption of dried fish, pickle and restaurant food there is no reason to expect Kerala to sit below the national picture, but this book will not print a Kerala figure it does not have.

7.3 Sodium in the salt-dense items of a Kerala kitchen

What to do instead of guessing. For anything cooked at home, the honest and easy method is arithmetic: salt added to the pan (g) × 400 ÷ number of servings = mg sodium per serving, plus whatever the ingredients bring. Every recipe panel in this book is computed that way. For dried fish, the instruction is procedural rather than numeric: rinse and soak it before cooking, and add no further salt to the dish. For pickle and pappadam the instruction is portion-based: a teaspoon of achar as a condiment, not a spoonful as a side dish, and pappadam as an occasional item rather than a daily fixture.

7.4 Potassium-rich Kerala foods

The cuisine is already built on them, which is the useful part. From this book's sources, the potassium-dense Kerala foods are: tapioca (kappa); plantain and raw banana; plantain stem and plantain flower; yam and elephant foot yam (chena); colocasia (chembu); drumstick and drumstick leaves; the thoran greens — cheera, ponnanganni, thakara; coconut water; jackfruit; papaya; tomato; and every pulse — parippu, cherupayar, kadala, vanpayar, muthira. Vanpayar and kadala are the most useful of all, because they deliver potassium, fibre and protein in the same spoonful.

The framing is swap, not supplement: raise potassium by eating more of these foods, not by taking potassium pills — and read 7.6 before doing either.

7.5 What the evidence supports

Two teachable points from the sodium rows. The pattern works without cutting salt or losing weight, and cutting salt works on top of the pattern. And the sodium effect is non-linear: most of the gain in DASH-Sodium came from the step below about 2.3 g of sodium a day, which is exactly the step most households never take.

7.6 Potassium-enriched salt: the carve-out, in full

The Salt Substitute and Stroke Study replaced ordinary salt with a 70% sodium chloride / 30% potassium chloride blend in 600 villages — 20,995 people with prior stroke or aged 60 and over with poorly controlled hypertension — for a mean of 4.74 years. Stroke fell 14%, major adverse cardiovascular events 13%, and all-cause mortality 12%. Hyperkalaemia events did not differ significantly between groups (3.35 versus 3.30 per 1,000 person-years). It is the single strongest one-ingredient swap in this book, and it requires no behaviour change at all.

The carve-out, which appears every time this swap does

Potassium-enriched salt is not for everyone. Do not use it — and do not raise dietary potassium deliberately — if you have chronic kidney disease, or if you take an ACE inhibitor (ramipril, enalapril, lisinopril and the rest), an angiotensin-receptor blocker (losartan, telmisartan and the rest), a potassium-sparing diuretic (spironolactone, eplerenone, amiloride), or a potassium supplement — unless your doctor has approved it. Those conditions and medicines raise the risk of potassium building up in the blood, which is the reason the trial screened for renal function. This book does not tell anyone to start or stop a medicine. It tells you which question to take to the person who prescribed it.

Two practical notes for the cook. The trial was run in a setting where nearly all salt was added at home during cooking, which is where this swap does its work. And potassium chloride tastes faintly metallic and bitter as its share rises, which is why 30% is the practical ceiling — and why M07 is blended at 70:30 and finished with plain salt where a dish needs a surface crystal.

7.7 Cutting sodium without cutting flavour

Kerala cooking already owns the whole toolkit. In rough order of usefulness:

1. Acid — kudampuli, tamarind, raw mango, lime, tomato, curd and moru. Acidity amplifies perceived saltiness. The cuisine's souring agents are the best salt-reduction tool available to it.
2. Umami — dried fish used as a seasoning rather than a portion, finely powdered; tomato; roasted onion; coconut browned to nuttiness. Dried fish is itself high in sodium, so use it as a flavour concentrate with no additional salt.
3. Aroma — curry leaf, coconut oil added off the heat at the end, freshly ground pepper, cardamom, clove. Fragrant food tastes more seasoned than it is.
4. Heat — chilli's pungency raises flavour intensity at lower salt. This is capsaicin's real job.
5. Texture — a crisp element, a raw element, a coarse grind. Uniformly soft food reads as bland.
6. Salt placement — finish with a few crystals on the surface, where they meet the tongue directly, rather than dissolving every gram into the mass of the dish.
7. Gradual reduction — salt-taste preference adapts over roughly 8 to 12 weeks. Cutting 10 to 15% a month is more sustainable, and far more likely to go unnoticed, than one steep cut.
8. Take the shaker off the table, and treat pickle and pappadam as occasional condiments.

In trained-panel testing on two Asian dishes, a 40% sodium reduction with umami enhancement maintained the perception of saltiness against the full-salt control (Leong et al., 2015/16 — noting that one author was affiliated with an MSG manufacturer, so read the direction of the finding rather than its precise size).

8. Fat and oil comparison

8.1 Fatty-acid profile, percentage of total fatty acids

ICMR-NIN 2017 Table 12. These are percentages of total fatty acids, not grams per 100 g — every oil in this table is close to pure fat, so the percentages are what distinguish them.

The reading, and it is consistent with both WHO 2023 and the AHA 2017 advisory: replacing part of the coconut oil with gingelly, groundnut or rice-bran oil moves a Kerala kitchen from roughly 91% saturated fat to 16–24%. WHO's strong recommendation is replacement of saturated fat with polyunsaturated fat; plant monounsaturated fat is a conditional recommendation; and replacing saturated fat with refined carbohydrate did not lower cardiovascular disease. Trials that lowered saturated fat and replaced it with polyunsaturated vegetable oil reduced cardiovascular disease by about 30%.

8.2 Coconut oil in detail

USDA puts the same molecule in grams per 100 g: 892 kcal, 99.1 g fat, 82.5 g saturated, lauric 41.8 g, myristic 16.7 g, palmitic 8.64 g, C8:0 6.8 g, C10:0 5.39 g, monounsaturated 6.33 g, polyunsaturated 1.7 g (SR Legacy 171412). The two sources agree on the shape of the molecule.

Why the "MCT" label does not fit. Genuinely medium-chain C8 and C10 together are about 8% of coconut oil's fatty acids (2.76 + 5.18). Lauric acid, which is half the oil, sits at the boundary of the medium-chain definition and does not behave like C8 in the outcome that matters. That follows from the composition alone and needs no further argument. In 16 trials of two weeks or longer, coconut oil against non-tropical vegetable oils raised LDL cholesterol by 10.47 mg/dL and total cholesterol by 14.69 mg/dL, raised HDL by 4.00 mg/dL, and did nothing to body weight (−0.23 kg, not significant), fasting glucose or CRP (Neelakantan et al., Circulation 2020). One well-run four-week trial at 50 g/day found coconut oil comparable to extra-virgin olive oil for LDL and better than butter (Khaw et al., BMJ Open 2018); it belongs here for honesty. The net cardiovascular effect of coconut oil has never been tested in an outcome trial.

8.3 Smoke point: an ordering, not a table of numbers

No numeric smoke points are printed in this book, deliberately. Published values vary widely between sources, depend heavily on refining level and free fatty acid content, change as soon as an oil has been used once, and the most-cited comparative paper has an industry connection. Smoke point is also a poor proxy for oxidative stability: an oil high in polyunsaturates can oxidise well below the temperature at which it smokes. Any precise figure would be false precision.

What can be said, qualitatively, is the ordering — lowest to highest:

Unrefined sesame · unrefined virgin coconut ≈ unrefined mustard · refined coconut ≈ refined sesame ≈ refined mustard · groundnut · rice bran · ghee

Two working rules that matter more than the ordering. Composition beats smoke point when you are choosing an oil for health reasons: use the columns in 8.1. And do not reuse frying oil more than once — repeated heating drives oxidation and generates compounds associated in animal and human mechanistic studies with raised blood pressure, raised total cholesterol and vascular inflammation. Discard oil that has darkened, thickened, foams or smells acrid; never top up old oil with fresh; and fry a small quantity in a small pan so there is less to be tempted to save.

8.4 Where each oil belongs in a Kerala kitchen

9. Weights, volumes and conversions

The convention this book uses. Weights come first and volumes second, and where the two disagree the gram figure is the one to follow. A cup of rice that is 15% heavier than the one used to write a recipe moves the carbohydrate load of the meal by 15%, which over a week is the difference between a plan working and not working. The tables below are the working conversions used throughout the book, measured with a levelled 240 ml cup and levelled standard spoons. Scoop density varies by about ±10% with how you fill the cup, and grated coconut varies by nearly half its weight between loose and firm packing — which is why the recipes state grams.

9.1 Weight

Exact: 1 oz = 28.35 g · 1 lb = 454 g · 1 kg = 2.205 lb.

9.2 Volume

Exact: 1 US cup = 236.6 ml · 1 US fl oz = 29.57 ml · 1 UK pint = 568 ml · 1 US pint = 473 ml.

The Indian cup is not the American cup. The measuring cup sold in India and used in most Indian recipe writing holds 200 ml; this book uses the 240 ml cup. That is a 20% difference — on rice, the difference between 200 g and 165 g in a single scoop. If your cup came from an Indian kitchen shop, weigh instead.

9.3 Cup to gram, by ingredient

One levelled 240 ml cup, unless the row says packed. Source for the whole table: the paired gram-and-cup figures used across this book's recipes.

9.4 Spoon to gram

9.5 Temperature

Conversions are exact arithmetic (°F = °C × 9/5 + 32). Recipes in this book round the °F figure to the nearest number an oven dial is actually marked with, which is why a recipe may say 200°C (390°F) where the table says 392°F.

9.6 Cooked yields, dry weight to cooked weight

Above about 500 g of dry rice the absorption ratio drops slightly, because proportionally less water evaporates from a deeper pan: R015 scales 100 g of rice to 250 ml of water but 500 g to 1.15 litres, which is 2.3 parts rather than 2.5.

10. Scaling reference

10.1 What scales, and what does not

10.2 Worked example: a kudampuli fish curry from four servings to eight

This is the pattern behind every SCALE IT table in the recipe chapters, and it is why those tables print real figures rather than telling you to multiply.

10.3 Scaling down

Halving is harder than doubling, for three reasons worth knowing. A small quantity in a large pan evaporates faster, so cut the liquid by less than half — about 60% of the original. Tempering does not halve below a floor: you cannot fry mustard seed in half a teaspoon of oil in a 26 cm pan, so use a small pan rather than less oil. And whole spices do not halve into fractions of a clove — round down rather than up, and take one cardamom pod out of the pan early if it is dominating.

11. Substitution matrix

Every row states the consequence, because a substitution that changes nothing does not exist. Where this book's sources give no ratio, the ratio column says not established rather than a guess — in those cases start with equal weights, taste, and write your own figure in the margin.

11.1 The Kerala-specific ingredients

11.2 The health-directed swaps

12. Seasonal calendar

What this calendar can and cannot tell you. The rains, the Malayalam month anchors and the festival food seasons below are drawn from this book's sources. A full month-by-month list of when each Kerala crop is at its best is not established in those sources — the closest thing to one is the Namboothiri household calendar's division of the year by what the land is giving: plantain and yam for three months, jackfruit and mango for another three, leafy vegetables for one quarter. Treat the produce column as anchors rather than as a market schedule, and trust your own fishmonger and vegetable seller over any printed table. Kerala has two rains and no dormant season, which is why it never built a preserving culture on the North Indian or Northern European model: what it preserves, it preserves for a specific glut — chakka varatti, kondaattam, kadumanga — not for a winter it does not have.

13. Storage, refrigeration and freezer guide

Times are the ones used throughout this book's recipes and pantry chapter, and they are real limits rather than cautious ones. Two general rules sit behind all of them. Cool fast, in a shallow covered container, and refrigerate within one to two hours — a warm pot cooling slowly on the hob is the single most common food-safety error in a home kitchen. And fat is what goes rancid, which is why oily fish, copra, dried fish and roasted-dal podis have shorter lives than lean or dry things.

13.1 Raw ingredients and the pantry

13.2 The mother preparations

13.3 Cooked dishes, by category

13.4 Cooked rice: the food-safety rule

This is not a quality note. It is the one genuine hazard in a rice kitchen, and it applies to everyone. Every limit below is the UK Food Standards Agency's, from its Home food fact checker, with the reheating temperature from the FDA Food Code.

Cooked rice carries Bacillus cereus spores that survive cooking. The FSA states it plainly: "uncooked rice can contain spores of a bacterium called Bacillus cereus … the spores of Bacillus cereus can survive being cooked." Left to cool slowly at room temperature they germinate, and the emetic compound they produce is not reliably destroyed by reheating — the FDA's Bad Bug Book records that cereulide "is stable after heating at 121°C [250°F] for 30 minutes, cooling at 4°C [39°F] for 60 days, and at a pH range of 2 to 11".

• Chill "as quickly as possible, ideally within one hour" (FSA). Spread it shallow, in a covered container; the FSA notes that "dividing it into smaller portions can help with this".
• Chill to 4°C (39°F) or below.
• "Keep rice in the fridge for no more than one day until reheating" (FSA).
• Reheat once, until "it is steaming hot all the way through" (FSA) — 2–3 minutes in a pan with a tablespoon of water, or 90 seconds to 2 minutes in a microwave. The FDA Food Code puts a number on "steaming hot": reheated food should reach 74°C (165°F) in all parts.
• "You should never reheat rice more than once" (FSA), and never leave cooked rice out overnight to chill in the morning.

Two notes on the sources, because they do not agree in every particular. The FDA Food Code's own cooling schedule for commercial kitchens is more generous than the FSA's — from 57°C (135°F) to 21°C (70°F) within two hours, and to 5°C (41°F) or below within six hours in total. This book follows the stricter FSA figure, because it is the one written for a home kitchen. And the Food Code's seven-day limit for refrigerated ready-to-eat food is a commercial date-marking rule; it is not a licence to keep cooked rice a week.

The same rule governs the chilled-and-reheated rice method (R016). Chilling rice for 24 hours at 4°C raises its resistant starch from 0.64 to 1.65 g per 100 g and lowered the glycaemic response by about 18% in one small study — and the cooling has to be fast for the technique to be safe. Chilling at room temperature for 10 hours gets you roughly half the resistant-starch gain and all of the risk, which is why this book does not recommend it. Anyone taking insulin or a sulfonylurea should read the warning at the head of R016 before making this a routine: in 32 adults with type 1 diabetes the method produced 12 hypoglycaemic episodes against 3, because the dose had been calibrated for the un-chilled food.

13.5 The does-not-freeze list

Coconut-milk curries thickened with first extraction · curd- and moru-based dishes · fried fish and any battered or crisp-crusted item · cooked egg white · raw grated vegetable salads and pachadis with raw cucumber or onion · appam, idli and dosa batter (the cooked items freeze; the batter does not) · M07 the 70:30 salt · pickles · sambharam · anything already thawed once.

14. Batch and freezer planning

Three hours on a Sunday, done in this order — batter and pulses first because they need time, pastes and podis last because they need attention — sets up a week. The portion sizes below matter more than the freezer times: a batch frozen in one large block is a batch you will not use.

One planning rule worth more than the table. Cook the pulse in quantity, not the finished dish. Four hundred grams of dry vanpayar cooked on a Sunday, drained and frozen in 150 g portions, becomes four different dishes across the week with fresh coconut in each one — and it is the habit that Mohan's 22.9% finding turns into an actual plate.