How these tools work

Every number on this site comes from one calculation engine with sourced data. This page explains where the data comes from, how the math works, and — just as important — what we refuse to guess at.

Ingredient data & sources

Volume ↔ weight conversion needs a density: how much a cup of an ingredient actually weighs. Our dataset (currently v1.0.0) covers 47 ingredient records. The primary source is USDA FoodData Central (SR Legacy household measures), a U.S. government publication in the public domain; where the USDA doesn't publish a usable pair, values are derived arithmetically from those same USDA records.

Every record carries:

  • The measuring convention — "spooned into the cup and leveled" weighs quite differently from "scooped" or "sifted". The convention is shown with every conversion result.
  • A confidence levelhigh where sources agree, medium where published values vary meaningfully (flour is the classic case: scooping can add 20% or more versus spooning). Medium-confidence records include a plain-language variability note.
  • Full provenance — the organization, document, and review date behind the number.

Ingredients whose published data we couldn't verify to a usable standard are left out entirely rather than guessed at. If you can't find an ingredient in the picker, that's deliberate — you can still convert it within volume units (cups → mL) or within weight units (oz → g), which need no density.

Regional cup standards

A "cup" isn't one size. The active standard is always shown next to results and only ever changes when you change it:

Standard1 cup equals
US customary (cup ≈ 236.59 mL)237 mL
US legal/nutrition (cup = 240 mL)240 mL
Metric (cup = 250 mL)250 mL
Imperial (cup ≈ 284.13 mL)284 mL

Densities were measured in US customary cups, so converting under another standard applies the exact volume ratio — and tells you so with a notice when the difference matters.

Fractions & rounding

Scaling math runs on exact rational arithmetic — no floating-point drift. For display, results snap to kitchen-usable fractions (halves, thirds, quarters, sixths, eighths; teaspoons additionally to sixteenths) only when the snapped value stays within a small tolerance of the exact value. Whenever display rounding moves a number, a notice says so, and the fraction/decimal toggle is formatting only — both views come from the same exact value.

Recipe scaling rules

  • Quantities multiply exactly; ranges ("2–3 cups") scale both ends.
  • Eggs and other indivisible items get a warning when the scaled count isn't whole — we never round your eggs silently.
  • Lines we can't fully parse are preserved unchanged and labeled, so nothing disappears from your recipe.
  • "To taste" and similar non-numeric quantities pass through untouched.

Pan size math

By default pans are compared by surface area (round: π·r²; square: side²; rectangular: width × length). When you enter depths for both pans, you can switch to volume comparison. The multiplier is the exact ratio; the percentage difference tells you how much more or less your pan holds. Large jumps come with a baking-time reminder — thicker or thinner batter bakes differently, and that part is genuinely an estimate.

Oven temperatures

°F ↔ °C uses the exact affine formula. The "set your oven to" value is rounded to the nearest 5° because that's what dials do — the exact value is shown beside it. Gas marks use the standard anchors:

Gas mark°F°C
1/4225105
1/2250120
1275135
2300150
3325165
4350175
5375190
6400205
7425220
8450230
9475245

Fan/convection guidance (−20 °C rule of thumb) is displayed in its own clearly-labeled block, separate from the measured conversion.

Baker's percentages

Every ingredient is expressed relative to total flour weight (flour = 100%). Hydration is the weight of rows you mark as water divided by total flour. Preferments and starters aren't decomposed into their flour and water parts — that's noted right in the tool. Scaling to a target dough weight applies one exact factor to every row.

What we deliberately don't do

  • No guessed densities — unsupported ingredients are absent, not approximated.
  • No silent unit assumptions — ambiguous input produces a question or a labeled pass-through, not a guess.
  • No nutrition, substitution, or "AI recipe" advice — these are measurement tools.