A cup of milk does not weigh 227 grams
16 September 2026
# A cup of milk does not weigh 227 grams
16 September 2026
King Arthur's ingredient weight chart is one of the most cited sources in home baking. It gives a cup of milk as 227 grams. A cup of yogurt as 227 grams. Heavy cream, light cream and half-and-half all as 227 grams. Vanilla extract as 227.
Any two of those figures being the same is a coincidence worth noting. All of them being 227 is not a coincidence at all. 227 is what you get when you take eight US fluid ounces of a substance and multiply by 28.35 grams per ounce, which is the conversion factor for weight ounces. Somewhere upstream of the chart, someone read the eight-fluid-ounce cup as if fluid ounces were weight ounces, and the number followed into every dairy row.
Why this is wrong for milk
Whole milk is roughly 88% water plus 3-4% fat, 3-4% protein and 5% lactose. Dissolved and suspended solids are denser than water, so the mixture is denser than water. USDA FoodData Central publishes the density directly: 1 cup of whole milk weighs 244 grams, which implies 1.031 g per millilitre. That figure comes from bench measurement on real product and has been verified by dairy laboratories for decades.
227 grams per cup implies a density of 0.960 g per millilitre. That is less dense than water. It is physically impossible for whole milk. The number is wrong because the method that produced it did not touch a scale — it touched a conversion table for the wrong unit.
The pattern across dairy
The 227-gram figure repeats across products that cannot all have the same weight. Whole milk has different fat and solid content from yogurt, which is different again from heavy cream at 36% fat. Half-and-half sits between milk and cream at 10-18% fat. Real densities range from about 1.00 for heavy cream, buoyed up by fat globules that are lighter than water, to about 1.04 for Greek yogurt, concentrated by straining. None of them are 0.960.
When a source assigns identical weights to products with different composition, the source is publishing a convention, not a measurement. This particular convention is the eight-fluid-ounce cup, converted to grams as if fluid ounces were weight ounces — a distinction fluid ounces exist specifically to make.
What to use instead
For dairy, use the USDA figures. They come from bench measurements on real product and are freely available at FoodData Central. Every ExactCups ingredient page links to the specific FDC entry.
- Whole milk: 244 g per cup
- Plain yogurt: 245 g per cup
- Heavy cream (36% fat): 238 g per cup
- Greek yogurt: 245 g per cup
These sit within a few grams of each other because they are all mostly water. That similarity is the physical truth the 227 figure was reaching for and missing.
Where 227 grams happens to be close enough
Some liquids really do weigh close to 227 grams per cup because they really are close to the density of water. Coconut milk lands between 226 and 241 depending on brand and fat content. Mashed banana at 225. Water itself, in a US customary cup, at 236.5882365. For these products, 227 grams is fortunate rather than corroborating — it happens to fall inside a physically plausible range, even though the method that produced it was blind to physics.
The way to tell the two cases apart is to ask what density the figure implies, and compare that to what the substance actually is. Anything denser than water that comes out lighter than water when weighed by cup has gone through the fluid-ounce filter.
The rule
If a chart gives every liquid in your recipe the same weight, the chart is publishing volume-as-weight, not measurement. Cross-reference against a source that names its density.
The label side of the same problem
Trusting a single source without checking is where recipe weights go wrong. Trusting the front of a food packet without checking the actual ingredient list is where diets and allergens quietly go wrong. That is why I also built SwapGenie — it reads the barcode, checks the ingredients against what you personally avoid, and suggests something better when the packet claims one thing and lists another.
Two ends of the same problem: knowing what is actually there, rather than what a convenient label implies.
