A ratio can guide a pastry, but it cannot replace a tested recipe

Ratios make scaling intelligible; a recipe records the equipment, sequence, yield, and endpoint that let the result arrive. The difference is why a clever percentage can still leave a pie crust sulking in its tin.

Published 19 August 2026

There is a particular confidence that comes from a percentage. It has a decimal point. It has survived a calculator.

It appears, briefly, to have removed the human condition from pastry.

So you work out that a 9 x 9-inch pan has 81 square inches of area, halve the batter by weight, find a pan near 40.5 square inches, and feel unusually well prepared.

Then the filling is too deep. The shell bakes before the center behaves. The spare portion sits on the counter looking like a small accounting error.

The arithmetic was sound.

A ratio preserves proportion; a recipe preserves the conditions that make the proportion work.

That isn’t an argument against ratios. They are among the clearest tools you have.

It is an argument against asking them to do the job of a tested recipe, which is a slightly unfair promotion for a column of numbers.

A percentage tells the ingredients how to travel

Baker’s percentage starts with flour at 100%. Everything else is expressed against the flour weight. A plain French bread might run 65% water, 2% salt, 1.25% yeast.

Those relationships stay legible whether you are making one loaf or forty.

That is the great usefulness of the system.

You can resize the batch without reinventing the bargain between water, salt, yeast and flour every time.

It is proportion, not prophecy.

The percentage does not know whether your flour is strong or soft, and flour strength changes the whole negotiation between tenderness and handling.

Water changes it too. A formula at 65% may want a different final adjustment when the flour absorbs differently.

In short dough the liquid activates gluten, so you add only enough to bind. A crumbly mass can still come together with pressing and chilling.

A smooth wet one has probably already gone too far. Milk, buttermilk, egg, vinegar and vodka are not interchangeable cells in a spreadsheet.

They bring different amounts of water, and also protein, acid, alcohol and fat.

The ratio has preserved your plan.

The recipe still has to say what the plan is asking the materials to do.

Pan geometry changes what the arithmetic means

The most useful small-batch example is not glamorous, which is why you can trust it.

A 9 x 9-inch pan gives you 81 square inches. So a half batch wants roughly 40.5.

That is far better than picking any pan that looks smaller and hoping geometry feels generous.

It also does not finish the calculation.

Area preserves depth, and depth is the thing you were really protecting.

But fill level moves depth too, and a shallow batter doesn’t bake like a deep one at the same setting.

Then the material joins in.

Dark metal, glass, ceramic and insulated pans all brown differently and bake at different paces.

Your oven setting has not become a law of nature because the ingredients were weighed correctly.

Which is why “halve the recipe” can be a perfectly good formula instruction and an incomplete batch plan.

The missing questions are practical. What pan. What depth. What rack. Preheated or not.

Baked empty, filled, covered or exposed.

And what change tells you it is done.

None of that is fussiness.

It is the recipe admitting that heat has a job too.

A count can hide a scaling problem

Eggs are where a respectable ratio develops a small limp. A large egg is about 50 g out of the shell.

A half batch calling for half an egg becomes executable the moment you beat one and weigh what you need.

That is not a demand that you turn breakfast into laboratory work.

It is a recognition that a counted fraction of an egg is ambiguous, and that a weighed portion tells you what “half” was supposed to mean.

Choux makes the principle obvious.

Egg addition controls fluidity, steam potential, protein setting and whether the thing can be piped at all.

Egg size varies. Panade dryness varies.

So the condition of the paste matters more than forcing in a fixed count. Before you commit a whole tray, pipe one or two test mounds.

The paste is ready when it reaches the ribbon or the peak the formula describes, and not when the recipe runs out of eggs.

The same ingredients in the same ratio can want a different stopping point when their earlier cooking has changed the paste.

That isn’t the ratio failing. It is the recipe doing the harder work of naming an endpoint.

Pastry has material states

Cold fat in a short crust stays in separate pieces, so its water makes small steam gaps as it bakes.

Fat melted fully into the flour has lost that structure. Fat too hard to roll tears the dough or breaks the layers.

For lamination, the butter and the dough need compatible flexibility. Resting and chilling let the gluten relax and the butter firm up before the next turn.

No percentage has enough information to tell you whether your butter is currently bendable, breaking, or quietly escaping.

Which is why temperature belongs in a recipe as an observation about a material, not as a decorative number.

The same thing continues in the oven.

A shell needs enough early heat to set before the fat runs out.

Puff needs a hot start while its layers are still distinct. Custard wants gentler, more even heat.

One bake profile can’t be handed between them because the batch was multiplied correctly.

Pastry is not being difficult about this.

It is being extremely literal.

A tested recipe reports what a formula cannot

A formula tells you what went in.

A tested recipe also tells you what came out.

Yield is the first report, and it is less obvious than it looks.

Raw dough weight, baked weight, finished volume, portion count, trim loss. Those are not the same number.

A batch may give you twelve hot muffins at the oven door and ten intact ones after the drive.

The useful yield is the one whose state somebody named.

So a scaled batch needs its yield stated again, rather than inherited from the original with a hopeful shrug.

Then comes the endpoint you can see. Golden-brown edges. A crisp exterior. A glossy finish.

A specified degree of thickening.

“Delicious” is nice news. It does not tell you whether to stop baking.

Repeatability is not mechanical sameness. It means separate attempts, under described conditions, land close enough for the stated purpose.

And the described conditions are the entire point. Ingredient state, pan, equipment, sequence, oven profile, yield, endpoint.

A revision date belongs there too, because a changed pan or package size or brand is a changed experiment.

Scale it, then ask what else moved

Scale the ingredients by weight. That part is easy and it is the part everyone does.

Then ask whether the pan area, the mixing time and the stated yield moved with them. A two-hour pastry may want fifteen minutes of your hands.

The rest is still part of the batch plan, because the chilling and the cooling are what make the result possible.

If those things did not scale, what you have is a formula. Not yet a batch plan.

That doesn’t make the original instruction worthless.

It just tells the truth about which part has been established and which part you are about to test.

Ratios travel well within a method.

Laminated dough, sugar work and small egg-based batches travel worst, because the material state can change before the arithmetic has finished congratulating itself.

Weigh the flour, by all means.

Then look at the butter, which has an opinion the spreadsheet was never going to hear.

Questions this raises

What is the difference between a formula and a recipe?

A formula records ingredient relationships, often by mass or percentage. A recipe also records the sequence, equipment, batch size, yield, quality checks, and endpoints that make those relationships executable. A formula can preserve proportion when scaled, while a tested recipe establishes what pan, handling, bake profile, and finished result make the scaled batch work.

Why does baker's percentage make scaling easier?

Baker's percentage sets flour at 100 percent and expresses every other ingredient relative to flour weight. King Arthur's sample French-bread formula lists 65 percent water, 2 percent salt, and 1.25 percent yeast. Those relationships can be recalculated at another batch size without changing the formula's proportion, although the production method still needs its own conditions and checks.

Can a pastry recipe be doubled just by doubling every ingredient?

Doubling every ingredient preserves the ingredient ratio, but it does not automatically preserve pan depth, mixing behaviour, oven exposure, or finished yield. A deeper batter receives heat differently from a shallow one, and pan material changes browning and bake pace. The doubled batch becomes a working recipe only after its equipment, sequence, endpoint, and yield are tested and recorded.

How should a smaller baking pan be chosen for a half batch?

Pan area provides a practical first check. A 9 by 9-inch pan has 81 square inches, so a half batch seeks roughly 40.5 square inches of area. That comparison helps preserve batter depth, but it does not prove identical baking because pan shape, material, fill level, and oven behaviour still change the surface and centre of the pastry.

Why does pan depth matter after the ingredients have been scaled?

Pan dimensions and fill level determine batter depth and heat exposure. A shallow batter does not require the same bake time as a deeper batter at the same oven setting. Dark metal, glass, ceramic, and insulated pans also change surface browning and bake pace. A scalable plan therefore records pan material, shape, fill depth, preheat state, and bake endpoint.

How can half an egg be measured in a scaled pastry batch?

A standard large egg weighs about 50 g out of shell, including approximately 14 g yolk and 35 g white. Weighing beaten egg makes a partial batch operational where a counted half egg is ambiguous. The technique preserves the intended ingredient proportion without pretending that a divided whole egg has a naturally uniform composition.

Why can the same flour ratio make a pie crust handle differently?

Flour protein, water, fat state, and handling determine how a short dough behaves. King Arthur lists particular products at 11.7 percent protein for all-purpose flour, 10.3 percent for a pastry-flour blend, and 8.0 percent for pastry flour. Those values illustrate a tenderness-and-handling trade-off rather than universal flour categories, so a ratio alone cannot predict the dough's exact response.

Why is cold fat part of the method and not just an ingredient choice?

Cold fat in short crust remains in separate pieces, allowing its water to create small steam gaps during baking. Fat that melts fully into flour loses that discrete structure, while fat that is too hard can tear dough or break laminated layers. The recipe must therefore state the required fat condition and the chilling steps that keep it workable.

What should a tested pastry recipe say about yield?

A tested recipe identifies the measurement boundary for its yield: raw dough weight, baked weight, finished volume, portion count, unit size, trim loss, or service count. “Serves four” remains an editorial judgment unless portion mass and meal context are stated. Clear yield lets a cook plan ingredients, vessel capacity, servings, and leftovers without guessing which state was counted.

What counts as a useful pastry endpoint?

A useful endpoint describes an observable result rather than a preference such as “perfect.” It can name golden-brown edges, a crisp exterior, a glossy emulsion, a specified degree of thickening, or another stated change in texture, appearance, aroma, or internal temperature. Such language tells a cook whether to continue, stop, cool, or adjust the process.

Does a recipe have to produce exactly the same pastry every time?

No. Repeatability means that independent attempts under described conditions reach an outcome similar enough for the stated purpose. A pastry recipe can report its operating range and observed result without claiming that every flour, oven, pan, and kitchen behaves identically. Mass-based formulas reduce some measurement ambiguity, but they do not eliminate material and equipment variation.

What needs to be recorded when a pastry recipe is revised?

A publication record identifies material changes rather than silently replacing the earlier instruction. Ingredient brands and package sizes can change, oven assumptions can age, and reader feedback can reveal ambiguity. For a scaled pastry, the useful record includes the formula, pan, equipment, sequence, stated yield, sensory endpoint, and revision date so later testing remains interpretable.