Clothing size labels

Why can two garments labelled the same size fit two completely different people?

You take three garments off three rails, all labelled the same size. One is snug at the shoulders, one gapes at the waist, one fits nobody you have ever met. Nothing is broken, nothing was mislabelled, no factory made a mistake.

Before blaming the manufacturers, it is worth asking what kind of thing a size label is. Is it a measurement, like the litres on a bottle? Or is it something else wearing a measurement's clothes?

Let us test the measurement idea. A measurement has to be a number of something. Litres are a number of litres. So: a size 12 is twelve what?

There is no answer. The number is an index, not a quantity — an arbitrary label on a rung of a ladder. That alone should make us suspicious, but it is not yet the deep problem.

Here is the deep problem. How many numbers does it take to specify a body well enough to cut a garment for it? Not one. A pattern cutter needs bust, waist, hip, height, back length, shoulder width, arm length, the ratios between them. Call it a dozen numbers that vary substantially and somewhat independently between people.

Now ask what the label carries. One number.

So we are compressing roughly a dozen dimensions into one. What must happen? Information is destroyed. And the specific way it is destroyed matters: many genuinely different bodies get mapped onto the same label, and there is no rule that any two of them resemble each other. Two people who are "a 12" may agree on one measurement and disagree on every other.

That is the whole answer in outline, but it leaves a question. If the compression is inevitable, why do the garments not at least fail consistently? Why is one brand's 12 not simply another brand's 12 plus a bit?

Because compressing many dimensions to one requires a choice, and each brand makes its own. Every label rests on a fit model — a real person or a mannequin whose measurements define the sample size — plus a grading rule for how the pattern scales up and down from there. Choose a different fit model, or a different grading rule, and you get a different garment under the same number. Nobody is doing anything wrong; they are simply making different choices about which information to keep.

And this is where the history matters, because there was once an attempt to standardise. The American commercial standard CS 215-58, published in 1958, rested on a large anthropometric survey of women conducted in 1939-41 — a sample now understood to have been badly unrepresentative, and in any case describing bodies of ninety years ago. That standard was withdrawn in 1983, leaving voluntary ASTM tables that no one is obliged to use.

Now add the last ingredient. Since the label is not tethered to anything, it is a free parameter — and a free parameter attached to something people feel about themselves does not stay put. If a smaller number is more pleasant to read, the mapping drifts downward over time. That is vanity sizing, and the drift is measurable: comparisons of retailers' own size charts across decades show the body measurements assigned to a given number growing steadily. Even nominally honest labels drift; men's trousers marked with an inch waist have repeatedly been found, on measurement, to run larger than the number printed on them.

The misconception to let go of is that somewhere out there is your true size, and the brands that do not fit you have made an error. There is no true size, because a single ordinal number is not the kind of object that could encode a body. The fitting-room experience is not a diagnostic of your body; it is a report on the distance between your proportions and a particular brand's fit model.

Notice too that the two problems are separable, and the fix for one is not a fix for the other. Vanity sizing is a drift problem — the mapping moves over time and between brands, and it could in principle be cured by an enforced standard. The compression is a dimensionality problem, and no standard cures it, because the trouble is that one number cannot represent twelve. This is why the genuine improvements in fit have all come from carrying more numbers, not from disciplining the one: separate cup or leg or back-length runs, garments sold by actual measurements, and made-to-measure.

• Whether 3D body scanning and per-brand fit prediction actually beat the label, or just move the
• Why menswear kept nominal measurements and womenswear did not, and what that difference cost.
• Return rates as the hidden price of the label: who ends up paying for the information the tag

Every term the collection defines is gathered in the glossary.