Scaling a recipe sounds trivial: multiply everything by a factor. The factor is indeed trivial. Everything that makes scaling fail is a property of the ingredients or the physics of the cooking, not of the arithmetic.
The arithmetic, and where it stops working
For a recipe yielding Y servings scaled to Y′, the multiplier is:
k = Y′ ÷ Y
Every ingredient becomes k × original. If you double a 4-serving recipe, every number doubles. That part is exact and easy.
Then there are four things that do not follow the rule:
- Leavening agents. Doubling baking powder in a doubled cake does not double the rise — it changes the pH and the crumb, usually making it bitter or soapy. Most baking sources recommend scaling baking powder by about 1.5x for a doubled recipe, and salt by less than 2x.
- Seasoning and salt. These scale sub-linearly for the same reason: flavour compounds saturate. A dish seasoned for 4 that is scaled to 8 usually wants about 1.6x the seasoning, not 2x.
- Cooking time. Time depends on thickness, not volume. A doubled stew takes somewhat longer, but a doubled cake takes barely longer, because the distance heat must travel is unchanged.
- Pan size and heat. Doubling volume in the same pan changes the depth from 2 cm to 4 cm, which doubles the time for the heat to reach the centre. A doubled bake in the same pan is undercooked in the middle even if the timer says the original time.
A cup is not a unit of volume
This is the part that quietly ruins conversions. The word "cup" means different volumes in different places:
- US legal cup — 236.6 ml (16 × 14.79 ml tablespoons)
- Australian tablespoon — 20 ml, so an Australian "cup" is 250 ml
- "Metric cup" — commonly 250 ml, matching the Australian tablespoon
- Traditional British cup — 284 ml
The spread from 236.6 to 284 ml is 20%. A recipe written in Australian cups and interpreted as US cups produces 20% less liquid than intended, which for a cake is the difference between a good one and a dense one. This is why almost every serious recipe gives weight in grams alongside volume in cups.
Weight is not automatically better either
Volume measures space; weight measures mass. For water-like ingredients they agree, and for sugar and water the difference is small. For flour they are not interchangeable at all:
A "cup of flour" is defined by scooping and levelling a packed measure. Depending on humidity and how firmly it was packed, the same "cup" of flour can be 120 g or 160 g — a third more mass. Recipes that list "1 cup flour, 120 g" are giving you the density they expect; measuring by volume can miss it by a third.
Density is the conversion factor that matters
Converting between volume and mass always needs a density, and each ingredient has its own:
mass = volume × density
Useful anchors at room temperature: water 1.00 g/ml, milk 1.03, honey 1.42, granulated sugar 0.85, all-purpose flour 0.53, butter 0.91, and salt 1.2. These are the numbers that let you substitute one measuring system for another, and the reason a "cup of honey" is so much heavier than a "cup of flour" despite being the same volume.
Low-and-slow cooking beats arithmetic
For anything where heat transfer is the limiting factor, a larger batch is not harder — it is the same problem with more mass. A braise that takes 3 hours at 1 kg takes about 3.5 hours at 2.5 kg, not 7.5. The relationship is roughly linear in thickness, and thickness grows as the cube root of volume, so a doubled recipe takes only about 1.26x as long per unit thickness — in practice, a little more than that because the pot holds heat differently.
The practical rule: scale the seasoning sub-linearly, scale the time by roughly the cube root of the volume increase, and check the thickness rather than the volume.
Common substitutions, and the ratio that actually matters
Substituting an ingredient is a density problem wearing a costume. The useful table is mass per unit volume, because that is what lets you swap one measure for another:
- Butter to oil: 1 cup butter (227 g) ≈ 3/4 cup oil, because oil is less dense and butter contains about 20% water.
- Flour to almond flour: use 85% of the weight, not 100%. Almond flour has less gluten structure and absorbs less liquid, so a straight swap gives a dry, crumbly result.
- Brown sugar for white: a straight 1:1 by weight works, but brown sugar packs 15-20% looser by volume, so measuring brown sugar by the cup gives noticeably less mass.
- Cocoa powder for flour: 1:1 by weight is fine, and it reduces the gluten content — which is exactly why brownies do not need wheat flour at all.
The pattern: flours and sugars can be swapped by weight, fats need a volume correction, and leavening never scales linearly.
Scaling a big batch for the freezer
Cooking in bulk is where scaling gets tested, because the errors concentrate. Three things matter:
Pan depth governs time, not volume. A stew at 4 cm deep simmers in 3 hours; the same stew at 8 cm needs about 6, because heat has twice as far to travel and evaporation keeps the top cooler. If you double a recipe, double the pan area rather than the depth and the time barely moves.
Seasoning in layers. A double batch seasoned all at once and frozen is measurably worse than the same batch seasoned in three stages during cooking. Salt and aromatics need time to distribute through a cold mass; a pot that was salted before freezing tastes under-seasoned after reheating no matter how much you add later.
Cool before you freeze. A hot pot goes into the freezer as a warm mass, which thaws slowly in the danger zone. Cooling in shallow containers cuts the cooling time dramatically, and it is the single biggest safety improvement in bulk cooking.
Why scales are not linear
It is tempting to say "double the recipe, double the yeast" and stop thinking. The mechanism is saturation: a dough with 2% yeast rises fine, and the same dough with 4% does not rise twice as fast, because the limiting factor becomes the available sugar and the dough's own structure. The same applies to salt in a brine and to spices in a curry.
The empirical rule used by most professional kitchens and baking sources: scale yeast and leavening by roughly 1.5x, salt by 1.6-1.8x, and spices by about 1.5x, when you double a recipe. These are not laws, but they are far better than 2x, and they are the difference between "good" and "too sour" or "inedibly salty".
The general principle behind all of it: a recipe is a set of ratios plus a process, and scaling changes the ratios of the things that act non-linearly while leaving the arithmetic of the rest intact.
Frequently asked questions
How do I scale a recipe?
Multiply every ingredient by target servings divided by original servings. Crucially, do not scale the baking time or the leavening agent by the same factor — a doubled cake still needs close to the same baking time, and doubling the baking powder changes the texture, not the size.
Why is a US cup not the same as a metric cup?
The US legal cup is 240 ml, the Australian tablespoon is 20 ml, and many 'metric cup' recipes use 250 ml. Meanwhile a British cup is traditionally 284 ml. Using the wrong one shifts a recipe by 4-18%, which is enough to change a cake from good to dense.
Should I scale by weight or by volume?
By weight, whenever the recipe gives weights — it is more accurate and scales cleanly. Volume measurement is only reliable for water-like ingredients, since flour packed loosely and packed firmly differ substantially.
What scales and what does not when doubling a recipe?
Ingredients scale linearly, except leavening and seasoning, which often scale sub-linearly because their effect is not proportional. Time scales with thickness, not quantity, so a doubled roast takes roughly 1.5x as long, not 2x, and a doubled cake barely takes longer at all.