Take every chart on this site, line up each size against the one directly above it, and ask how far apart they are on each field. Do it 148 times, once for every adjacent pair of sizes we hold, and two fields behave completely differently from the rest.
On sleeve length, the typical gap between one size and the next is one centimetre. On shoulder width it is two. On the waist of a pair of trousers it is four, every single time, without one exception in 95 pairs.
That arithmetic is the whole of this page. It tells you what shoulder and sleeve are for, and it is not what most people use them for.
Where the numbers come from
We hold 57 size charts, read out of the images one seller published beside their items, all from MVT and published per item on the shelf. One store is a narrow base and we say so on every sizing page here.
Of those 57, 25 print a shoulder figure and 23 print a sleeve figure; the rest are lower-body cards carrying waist, hip and leg opening instead. That leaves 55 adjacent shoulder steps and 51 adjacent sleeve steps to work with.
Nearly all of those charts carry the seller's own footnote about a one-to-three-centimetre margin of error on manual measurement. Hold that number next to the step figures below, because the comparison is the point.
Sleeve cannot choose a size for you
Of the 51 adjacent sleeve steps in the set, the median is one centimetre and the largest is three. Not one sleeve step in the entire corpus exceeds the seller's own admitted measuring error.
Across a whole size run — smallest size to largest, on the same chart — sleeve moves a median of two centimetres, and the widest run only moves six. A chart that offers you XS through M offers you two centimetres of sleeve. If your reference garment sits two centimetres off, you have not learned which size to buy. You have learned that the chart does not know either.
So the sleeve row is not a selection field. It is a description of the garment's proportions, and it earns its keep in exactly one situation: when the number is nowhere near yours. Sleeve figures in our set run from 22.5 to 75 centimetres, because the same field name covers both short-sleeved and long-sleeved garments with nothing in the row to say which. A sleeve of 52 against your reference of 66 is not a sizing decision. It is a different garment than you pictured, and catching that before you pay is worth more than any size choice.
Shoulder is mostly a veto, occasionally the whole decision
Shoulder behaves the same way but not quite as strongly. Fifty of the 55 adjacent shoulder steps are three centimetres or less. Only five clear the stated tolerance and actually separate two sizes.
Compare that with chest, where 19 of 55 steps exceed three centimetres, or with waist, where all 95 do. On upper-body charts, chest is the field that usually resolves a decision and shoulder usually is not.
That does not make shoulder unimportant — it makes it a different tool. On structured outerwear the shoulder is the seam nobody can alter, so its job on a chart is to rule a size out, not to pick between two neighbours. Read it as a constraint: if the shoulder figure is far from your reference, stop, regardless of what the chest row says. If it is close, it has told you everything it can and the decision moves to another field.
When the shoulder number stops describing a shoulder
Shoulder figures in the set run from 43 to 81 centimetres. Part of that is genuine variety in cut. Part of it is that the field is not always measuring the same thing.
In 11 of the 80 rows that carry both a shoulder and a chest figure, the shoulder number equals or exceeds the chest number. On a flat measurement of a normal garment that is only possible on a very dropped shoulder, where the seam has been pushed halfway down the arm as a styling choice. It is also what you would see if those two rows were taken differently from each other. Nothing on the card says which, and not one of the 57 charts states its measuring convention in words.
One chart in the set makes the problem visible without any interpretation at all. Its shoulder row reads 64, then 55, then 56 as the sizes go up, while length climbs 65, 66, 67 and sleeve climbs 73, 74, 75. Every other field grades normally and the shoulder runs backwards. Either the smallest size is a different cut, or a number on that card is wrong. You cannot tell from the table, and if the shoulder had been the field you sized on, you would not have known to look.
What to do with both fields
- Measure the shoulder and sleeve of a garment you own that sits correctly, using the same landmarks each time — shoulder seam to shoulder seam across the back, and shoulder seam down to the cuff edge. The flat-measure walkthrough covers the landmarks and the raglan exception.
- Use shoulder to eliminate. If the chart's figure is more than a few centimetres from yours, that size is gone whatever the chest says.
- Use sleeve as a sanity check on what kind of garment you are looking at, not as a tiebreaker between two sizes. It is mathematically incapable of being one here.
- Then decide on the field that actually moves: chest for upper body, waist for anything with legs.
- Ask your agent to lay a tape across the shoulder seam in the QC photos before it ships. That request is the only measurement in the process taken from the object you are buying rather than from a card.
The limits of this
Fifty-seven charts from one store is not a market, and none of these garments have been bought and re-measured by anybody here. What we publish is the seller's printed numbers with a link back to the image they came from — the reasoning for keeping it that narrow is on the methodology page.
Another seller's shoulder rows could step differently. What would be surprising is a chart where sleeve stepped further per size than waist does, because a sleeve does not need to grow four centimetres to fit a slightly larger person.