How to scale a recipe

Doubling a recipe is not multiplying every number by two. Most of the numbers, yes. But four of them will quietly ruin dinner, and they're the four nobody warns you about.

Every ingredient quantity scales linearly. 200 g flour doubled is 400 g flour. This is the easy 90% and it's why scaling feels like it should be trivial.

One wrinkle worth knowing: measure by weight if you can. Cup measures compound their own error as you multiply — a cup of flour scooped straight from the bag can be 20% heavier than one spooned in and levelled, and at 4× that gap is most of an extra cup. Weight has no such ambiguity, which is the real argument for grams over cups and has nothing to do with the metric system.

A 9×5 inch loaf tin doubled is not an 18×10 inch loaf tin. It is two 9×5 tins, or one larger tin and a longer bake.

The reason is that batter depth, not volume, sets the baking time. Pour double into the same tin and the outside is done long before the middle sets. Spread it across two tins and nothing about the bake changes — same depth, same time, same result. That's almost always the move.

If you do go up a pan size, work in area. A 9-inch round is 64 square inches; a 13×9 is 117 — roughly 1.8×, near enough to double for most batters. Two 8-inch rounds are about the same total area as one 11-inch. Depth stays constant, time stays roughly constant.

Cooking time tracks thickness and surface area, not quantity. Twice the soup in the same pot still boils at 100°C, it just takes longer to get there — and then simmers for exactly as long as the recipe says. Twice the chicken thighs on two trays instead of one roast in the same 35 minutes.

What genuinely does change is anything involving evaporation or browning. A sauce reducing by half takes much longer at double volume, because you've doubled the liquid without doubling the surface it evaporates from. Same for browning meat: a crowded pan steams instead of searing, which is why every recipe that knows what it's doing says “in batches”. Scaling up turns “in batches” from advice into a requirement.

Salt scales, but taste doesn't scale with it, so scale it and then taste — season to about 75% on the way up and correct at the end. Chilli and strong spices compound noticeably; start at 75% of the scaled amount for those too.

Baking soda and baking powder are the ones that will actually fail. They're chemistry, not flavour: too much and the crumb goes coarse, the rise collapses, and you get a distinct metallic taste. Scale them straight, never round up, and be suspicious of any batch where the leavening is doing a lot of work.

Yeast is the opposite — a double batch does not need double the yeast, because yeast multiplies as it works. Same quantity, watch the dough rather than the clock.

“1 4-pound roast” doubled is 2 4-pound roasts. It is not one 8-pound roast, and if you scale it that way the cooking time is wrong by an hour and a half.

Same with “2 (5-ounce) cans of tuna” (4 cans, still 5 oz each), “1 3-inch cinnamon stick”, “cut into 2 cm cubes”. Numbers in a recipe describe two completely different things — how much you need, and how big something is — and only the first kind scales. This is the single most common way automated scaling gets it wrong, and it's worth a second look whenever you double something by hand.

• Amounts: multiply. Weigh rather than measure by volume.
• Pans: more pans, not bigger ones. Keep the depth the same.
• Times: unchanged, except reductions and anything you brown.
• Salt and spice: 75%, then taste.
• Baking soda and powder: exact. Yeast: leave it alone.
• Sizes of things (a 4-pound roast, a 3-inch stick): don't touch.

Every table on RecipeTables has a ½× / 1× / 2× / 3× control that applies all of the above — it scales the amounts, leaves pan dimensions and item sizes alone, and keeps a measurement and its converted twin (“1 4-pound / 1.8 kg roast”) in step rather than letting the two halves drift apart.

It won't taste the sauce for you. Nothing will. Here's how the format works, or browse the library and try the control on something real.