Lessons · Lesson 2 of 6
What a scan captures, and what it cannot
Read a body scan as a measurement of a surface in one posture, and hold the four things it does not measure clearly enough to design around them.
Lesson 2 of 6 · 19 min
12 May to 20 June: eleven hundred people
Ostley asks Sonderby Body Lab to scan its own customers. Four cities, six weeks, 1,100 volunteers taken from the loyalty file and paid in shop credit. Start to finish, including the recruiting and the reporting, it costs USD 38,500 — about USD 35.00 a person.
What comes back is not a size chart. It is 1,100 surfaces.
A body scanner produces a three-dimensional surface of the body it saw. Some scanners project a pattern of light and read how it bends; some combine many photographs; some use radio-frequency imaging. Software then pulls measurements off that surface. It finds landmarks and runs set paths between them: a girth is the shortest closed path around the surface at a given height, a length is the distance between two landmarks.
Three things follow, and they are the honest case for scanning.
- It repeats. No two people with tape measures agree. The same software on the same surface agrees with itself exactly.
- It is fast enough to be a population study. A hundred and thirty dimensions in the time it takes a person to stand still.
- The surface is kept. This one is underrated. Six months later Ostley wanted a thigh girth taken at a different height. It got it out of the stored surfaces without calling a single person back. A tape study cannot do that at any price.
Now the other half, which is longer.
Four things a scan does not measure
A scan does not press. A tape presses into soft tissue; a beam of light does not touch it. So a scan girth and a tape girth of the same body are different numbers, and the difference is not one number. Ostley had 60 of the volunteers measured both ways on the same afternoon, by a trained measurer following a written procedure.
| Point of measure | Mean difference | What is under the tape there |
|---|---|---|
| Waist | 1.9 | The softest tissue on the body |
| Hip | 0.8 | Tissue over bone at the widest point |
| Thigh | 1.1 | Muscle and fat, no bone to sit on |
| Inseam | -0.6 | No pressing at all; the crotch point is placed differently |
Read the last row. The inseam goes the other way, because that difference is not pressing at all. It is software placing a crotch landmark on a surface, where a measurer places a tape against a body. Four points of measure, three sizes of gap and two directions. Anybody who applies one correction factor to turn a scan chart into a tape chart has invented one of these numbers.
A scan captures one posture. The volunteer is told where to stand, how far apart to put their feet, and to hold their arms clear of their body. That pose exists so the surfaces can be compared with each other, which is a real virtue and the reason the scan repeats. It is not how the person stands in a shop. It is certainly not how they stand at the end of a day.
A scan captures a body that is not moving. Ostley scanned 40 of the volunteers standing and again seated. Waist girth seated was 4.4 cm larger on average than standing. No static chart carries that number anywhere, and a trouser is worn sitting down for most of its life. The free-text return comments in lesson 1 said cannot sit down in them. The chart the trouser was made from had no seated number in it at all.
A scan does not know how the garment is worn. Ostley asked 300 customers to put a marker on themselves where they actually wear a trouser waistband. The marked heights spread over 7.5 cm, and the median sat 2.8 cm below the anatomical waist landmark the scanning software finds.
That last one deserves a sentence on its own. The scan measures the narrowest girth of the torso. The trouser sits somewhere else, on a girth that is larger and that the chart does not contain. The chart is measuring a place the garment never touches.
The mistake nobody made
On 4 August Tess issues a new body chart built from the scan medians — the middle value at each size, the number half the bodies fall below. On 19 August Yatawara's pattern room, run by Chamila, grades OS-2214 to it and applies Ostley's standing ease policy — the room the garment is meant to leave around the body: 2.0 cm at the waist, unchanged, because nobody had questioned the ease and nothing about it was wrong.
The first proto arrives on 8 September. Measured flat, the waist is exactly on specification. On the fit model it reads visibly loose, and the fit session's comment is waistband slack, needs taking in about a centimetre and a half.
Everybody did their job. Here is why the garment was loose anyway.
| Old chart | New chart | |
|---|---|---|
| Instrument behind the number | Tape, unknown date | Scan, 2026 |
| Chart waist | 76.0 | 75.9 |
| Ease applied | 2.0 | 2.0 |
| Finished garment waist | 78.0 | 77.9 |
| The same body's tape waist | 76.0 | 74.0 |
| Ease the person actually gets | 2.0 | 3.9 |
The two charts differ by 0.1 cm and look interchangeable. They are not, because the numbers in them came from different instruments measuring different objects: one a pressed body, the other an untouched surface. The ease was designed against the first and applied to the second. So the person wearing it got 3.9 cm instead of 2.0 cm — nearly double, from a chart that moved by a millimetre.
It cost two extra proto rounds, USD 1,940 and 17 days of the development calendar. It was found only because a fit model put the garment on. A brand that had gone fully virtual on this style would have carried it into bulk, because the pattern matched the chart and the simulation matched the pattern.
Check yourselfYour scan study says the median size-10 waist is 75.9 cm. Your supplier's tape reading on the fit model says 74.0 cm. Which is correct?Show the answer
Both are, and asking which is correct is the error. They measure two different things: the girth of an untouched surface, and the girth of tissue under the pull of a tape held by a person following a procedure. Neither is a truer body than the other. What matters is that a garment specification is a chain — body number, plus ease, equals garment number — and every link has to come from the same instrument, or the chain has an unrecorded 1.9 cm in it. If you must mix them, measure the gap yourself, on your own people, at every point of measure, and record it as a written conversion rather than a rule of thumb. And do not expect the gap to hold across the range: it is largest where there is most soft tissue, so it grows with size.
What to take to your own scan programme
- Ask for the surfaces, not only the table pulled out of them. The table answers today's question; the surfaces answer next year's.
- Pay for a subset measured both ways — a few dozen people, both methods, the same day. It is the cheapest line in the whole study, and the only one that connects the new chart to everything you already own.
- Get a seated scan for anything worn sitting down, and a marked-waistband exercise for anything with a waistband. Both are an afternoon.
- Write the instrument, the posture, the date and the sample size into the chart itself, as fields. A chart that cannot say how it was made will be inherited by somebody in five years, and lesson 1 is about what happens then.