[ { "id": "all-purpose-flour", "schema_version": 3, "name": "All-purpose flour", "synonyms": [ "plain flour", "AP flour", "white flour" ], "category": "flour", "default_state": "unspecified", "confidence": "low", "states": [ { "id": "scooped", "label": "Dipped and swept", "density_g_per_ml": null, "withheld": true, "withheld_reason": "One source. Rule 2 requires three.", "sources": [ { "name": "America's Test Kitchen", "ref": "How to Measure Flour, cooksillustrated.com/how_tos/6573, 1 cup = 5 oz (141.75 g)", "method": "dip-and-sweep", "density_g_per_ml": 0.5906 } ], "verified": "2026-09-14", "last_reviewed": "2026-09-14" }, { "id": "spooned", "label": "Spooned and swept", "density_g_per_ml": null, "withheld": true, "withheld_reason": "One source. Rule 2 requires three.", "sources": [ { "name": "King Arthur Baking", "ref": "Recipe Success Guide, measurement standards, 1 cup Unbleached All-Purpose Flour = 4 1/4 oz (120 g)", "method": "Spoon dry ingredients gently into the cup. Sweep off any excess.", "density_g_per_ml": 0.5 } ], "verified": "2026-09-14", "last_reviewed": "2026-09-14" }, { "id": "unspecified", "label": "Method not specified", "density_g_per_ml": null, "withheld": true, "withheld_reason": "One source, and it names no method. Rule 2 requires three.", "sources": [ { "name": "USDA FoodData Central", "ref": "FDC 168894, SR Legacy, 1 undetermined cup = 125 g", "density_g_per_ml": 0.5208 } ], "verified": "2026-09-14", "last_reviewed": "2026-09-14" } ], "yields": [], "notes": "Flour publishes no number, and that is the entry's finding rather than a gap in it. Three sources give 141.75 g, 125 g and 120 g per cup. They are not three disagreements about one quantity; they are three quantities, because each describes a different way of filling the cup. America's Test Kitchen name dip-and-sweep and give 5 oz, arrived at by having dozens of volunteers each measure a cup and averaging the results, and they state that technique alone can produce up to a 20% difference. King Arthur's Recipe Success Guide gives 120 g and instructs the reader to spoon the flour in gently and sweep off the excess. USDA records an 'undetermined cup' at 125 g and names no method at all, so its state stays 'unspecified' under decision 21. The spread from 0.5000 to 0.5906 g/ml is 18%, which stops the entry under rule 5 and splits it under rule 6, because two of the three sources name their own state rather than leaving it to be inferred from the figure. Splitting leaves each state with exactly one source, so all three are withheld under decisions 23 and 27 and the entry drops to low confidence. The earlier version of this entry paired USDA's 125 g with King Arthur's 120 g, called the 4% gap a difference of method and held it at medium confidence. That was wrong in a way worth recording: the two figures sat inside the 5% limit only because the third method was missing from the file, so a method difference was passing as a measurement agreement. Adding the source that names dip-and-sweep is what exposed it. King Arthur's own measuring guide separately states that a fluffed, scooped and levelled cup holds around 120 g while a more condensed cup can hold up to 160 g. That upper bound brackets the ATK figure rather than contradicting it, but it is the same publisher and confirms nothing independently. Note also that King Arthur's 120 g is stated for their own Unbleached All-Purpose Flour, not for all-purpose flour generally, and that they name the same technique three ways across their site: 'fluff, sprinkle and scrape' in their video, 'fluff, scoop and level' in their 2023 measuring guide, and the spoon-and-sweep wording quoted above. The method field records the wording of the page cited, not a tidied label. To publish, each state needs two further sources that name its method and measure it. Unverified lead: Mel's Kitchen Cafe reports testing dip-and-shake at 6 oz, dip-and-sweep at 5.25 oz and spoon-and-sweep at 4 oz. A blog, but it states method and describes testing, so it must be verified at source and judged against the independence test in decision 25. Cook's Illustrated is the same organisation as America's Test Kitchen and is not independent of it. The eatyourbooks summary reports other people's figures and is not a source." }, { "id": "granulated-sugar", "schema_version": 1, "name": "Granulated sugar", "synonyms": [ "white sugar", "table sugar" ], "category": "sugar", "default_state": "unspecified", "confidence": "medium", "states": [ { "id": "unspecified", "label": "Method not specified", "density_g_per_ml": 0.8333, "sources": [ { "name": "USDA FoodData Central", "ref": "FDC 169655, SR Legacy, 1 undetermined cup = 200 g", "density_g_per_ml": 0.8333 }, { "name": "King Arthur Baking", "ref": "Ingredient Weight Chart, granulated white sugar, 1 cup = 198 g", "density_g_per_ml": 0.825 } ], "verified": "2026-09-14", "last_reviewed": "2026-09-14" } ], "yields": [], "notes": "Sources agree within 1%, in sharp contrast to flour. Sugar crystals are hard and free-flowing, so they do not compact meaningfully whether the cup is spooned or scooped. This is why granulated sugar needs only one figure while flour needs several. USDA's own teaspoon entry (4.2 g) implies 0.8521 g/ml, a 2.2% disagreement with their cup figure; this is rounding on a small measure, not a real difference, so the cup figure is used. Two sources only, so this entry does not meet rule 2 and is held at medium confidence until an independent third is found." }, { "id": "brown-sugar", "schema_version": 1, "name": "Brown sugar", "synonyms": [ "light brown sugar", "dark brown sugar", "soft brown sugar" ], "category": "sugar", "default_state": "packed", "confidence": "medium", "states": [ { "id": "packed", "label": "Firmly packed", "density_g_per_ml": 0.9167, "sources": [ { "name": "USDA FoodData Central", "ref": "FDC 168833, SR Legacy, 1 undetermined cup packed = 220 g", "density_g_per_ml": 0.9167 }, { "name": "King Arthur Baking", "ref": "Ingredient Weight Chart, brown sugar (dark or light, packed), 1 cup = 213 g", "density_g_per_ml": 0.8875 } ], "verified": "2026-09-14", "last_reviewed": "2026-09-14" } ], "yields": [], "notes": "The first ingredient where USDA names the measuring method rather than recording an 'undetermined cup'. Packed is the default because recipes almost always specify firmly packed brown sugar; the molasses content makes the sugar sticky, so it compresses in a way granulated sugar does not. USDA also publishes an unpacked figure of 145 g per cup, a 52% difference from packed, but that state is not included here: King Arthur's chart lists packed only, and no second published source gives an absolute unpacked density. America's Test Kitchen documents testing the difference with ten cooks and reports the packed cup averaging about 2 ounces heavier, but publishes a difference rather than an absolute figure, so it cannot serve as a source under rule 3. Unpacked is withheld under rule 8. Published converter sites give unpacked values of 130, 145, 160, 170 and 180 g per cup with no sourcing, which is precisely the disagreement this dataset exists to avoid reproducing. Two sources only, so this entry does not meet rule 2 and is held at medium confidence until an independent third is found." }, { "id": "butter", "schema_version": 1, "name": "Butter", "synonyms": [ "salted butter", "unsalted butter", "sweet cream butter" ], "category": "fat", "default_state": "solid", "confidence": "medium", "states": [ { "id": "solid", "label": "Solid, cut from the block", "density_g_per_ml": 0.9458, "sources": [ { "name": "USDA FoodData Central", "ref": "FDC 173410, SR Legacy, 1 undetermined cup = 227 g", "density_g_per_ml": 0.9458 }, { "name": "King Arthur Baking", "ref": "Ingredient Weight Chart, butter, 8 tablespoons (1/2 cup) = 113 g", "density_g_per_ml": 0.9417 } ], "verified": "2026-09-14", "last_reviewed": "2026-09-14" } ], "yields": [ { "id": "stick", "from": "1 stick", "grams": 113, "sources": [ { "name": "USDA FoodData Central", "ref": "FDC 173410, SR Legacy, 1 undetermined stick = 113 g", "grams": 113 }, { "name": "King Arthur Baking", "ref": "Ingredient Weight Chart, butter, 8 tablespoons (1/2 cup) = 113 g", "grams": 113 } ], "verified": "2026-09-14", "last_reviewed": "2026-09-14" } ], "notes": "Sources agree to within 0.4%, the closest agreement in the dataset so far. Salted and unsalted are identical by weight: USDA records 227 g per cup for both FDC 173410 (salted) and FDC 173430 (unsalted), so one entry covers both. Butter is the one ingredient where cup measurement is genuinely reliable, and the reason is visible in the data itself: USDA gives a stick as 113 g and a cup as 227 g, which is twice the stick. The cup figure reflects the packaging convention rather than a measurement of how butter fills a cup. A US butter wrapper is printed with cup and tablespoon markings, so the cook cuts a pre-portioned block instead of filling a vessel, and the packing variability that affects flour and brown sugar does not arise. Whipped butter is a separate product, not a state of this one: USDA records it at 151 g per cup (FDC 173411), roughly a third lighter because air is beaten into it. USDA also publishes a pat of butter at 5 g, defined as a 1-inch square a third of an inch high. It is not included as a yield here because King Arthur's chart does not list it and no second source was found, so rule 8 applies. The pat is a real US measure, used for restaurant portions and in older cookbooks, and is worth adding if a second source turns up. Two sources only, so this entry does not meet rule 2 and is held at medium confidence until an independent third is found." }, { "id": "olive-oil", "schema_version": 2, "name": "Olive oil", "synonyms": [ "extra virgin olive oil", "EVOO", "virgin olive oil" ], "category": "fat", "default_state": "liquid", "confidence": "high", "states": [ { "id": "liquid", "label": "Liquid, at room temperature", "density_g_per_ml": 0.913, "sources": [ { "name": "USDA FoodData Central", "ref": "FDC 171413, SR Legacy, 1 undetermined tablespoon = 13.5 g", "density_g_per_ml": 0.913 }, { "name": "FAO/WHO Codex Alimentarius", "ref": "CXS 33-1981, Appendix section 3, relative density 20C/water 20C", "range_min": 0.91, "range_max": 0.916, "note": "Standard specifies a range, not a figure, determined per IUPAC 2.101. Serves to confirm or reject under rule 4; it does not supply the published value." }, { "name": "King Arthur Baking", "ref": "Ingredient Weight Chart, olive oil, 1/4 cup = 50 g", "density_g_per_ml": 0.8333, "rejected": true, "rejection_reason": "0.8333 g/ml falls 8.4% below the Codex Alimentarius range of 0.910-0.916 at 20C. Olive oil is a liquid with no packing state, so no measuring method can account for the gap. The figure cannot be correct for olive oil at any normal temperature." } ], "verified": "2026-09-14", "last_reviewed": "2026-09-14" } ], "yields": [], "notes": "The first entry where an independent third source rejected one of the other two rather than breaking a tie. USDA publishes three measures for FDC 171413: a cup at 216 g (0.9000 g/ml), a tablespoon at 13.5 g and a teaspoon at 4.5 g (both 0.9130 g/ml). The tablespoon figure is published here rather than the cup, because Codex Alimentarius CXS 33-1981 specifies olive oil relative density as 0.910-0.916 at 20C and 0.9130 falls inside that range while 0.9000 falls below it. This reverses the usual preference for the larger measure: 216 g is a rounded figure and the rounding is visible once an independent standard is available to check it against. King Arthur Baking's chart gives 1/4 cup as 50 g, implying 0.8333 g/ml. That is 8.4% below the Codex range. A liquid has no sifted-or-scooped variation, so unlike flour there is no measuring method that reconciles the two figures; the chart value is simply wrong for olive oil. It is recorded as a rejected source rather than deleted, under rule 7, because a published figure that is wrong is worth knowing about. Density falls as temperature rises, so oil measured warm weighs slightly less per cup than the same oil measured cold." } ]