Size charts that lie: measuring the garment you already own

A size chart is a claim. It is the seller statement about an object that was measured somewhere by somebody under conditions nobody wrote down, and it may describe the object, the body the object is cut for, or something in between. The only way to check a claim is against an object you can measure yourself, and the object you already own is the only one available to you before you buy anything.

Choose the garment that fits the way you want the new one to fit, which is not necessarily the one you wear most often. A shirt you like but that is slightly tight across the chest is the wrong reference, because aiming at it will deliver a second shirt that is slightly tight across the chest and you will have paid for the privilege. Pick the garment whose fit you would accept again without thinking about it.

Two properties make a reference better or worse, and both are worth checking before you trust a number taken from it. The first is how much the garment has been washed, because shrinkage is a one-way process and a shrunken reference biases every measurement you take from it. The second is how structurally similar it is to what you are buying: a woven shirt is a poor reference for a knit, and a cotton tee is a poor reference for a fleece carrying the same nominal size.

Lay the garment on a flat surface and let it rest without pulling at it. Pulling is the most common measurement error and it always biases in the same direction, because a stretched measurement makes a reference look larger than it is and pushes your eventual decision toward a smaller size. If a seam will not sit flat, smooth it with your hand rather than tensioning the fabric, and accept the number you get.

Measure the chest across the front, seam to seam at the armholes, then double it to get a circumference. That doubling step is where errors enter most often, because a seller chart listing a chest circumference and a buyer measuring across the front are quoting two numbers that differ by a factor of two. Measure shoulder width from seam to seam across the back, and measure length from the point where the collar meets the body down to the hem.

For knitwear, add a fourth measurement: the relaxed hem width, taken across the bottom edge without stretching the ribbing. For anything with sleeves that matter, take the sleeve length from the shoulder seam to the cuff, and note whether the cuff is ribbed, because a ribbed cuff measures shorter relaxed than it wears. Write everything down in centimetres and note the labelled size of the reference garment, because that label is the field that lets you compare your numbers against the seller chart later.

The first question is not how the numbers compare but what kind of chart you are looking at. A body chart describes the person the garment is cut for, and its numbers are meant to be matched against a body. A garment chart describes the object itself, and its numbers are meant to be matched against other garments. Reading a body chart as a garment chart produces errors of about one size, and sellers routinely publish one while the buyer assumes the other.

The second question is the unit. A chart in inches compared against measurements taken in centimetres is not a close call; it is a different order of magnitude, and it is the error most likely to survive a careful reading because each number looks plausible in isolation. Confirm the unit printed on the chart before comparing anything, and if the chart states no unit, ask rather than assuming centimetres because the numbers look large.

Then compare term by term and mark the largest disagreement rather than the average. A chart that matches on length and misses on chest by five centimetres describes a different cut rather than a different size, and the correct response is to expect a different fit rather than to size up and hope. If the chart does not say which kind it is and the difference is material to your decision, ask the seller which chart the listing uses. That question is answered faster than a return is processed, and it is answered before you have spent anything.

The garment that has to fit is not the one that arrives. It is the one that exists after a season of washing, and for a natural fibre that garment is measurably smaller than the one in the bag. Cotton knits typically lose a visible amount of length and a smaller amount of width on a first hot wash. Synthetics move very little. Wool is the extreme case: a wool garment that meets a hot cycle can lose several sizes, and the change is usually not reversible by any means available at home.

Because your reference garment has probably already shrunk, your baseline is post-shrink while the new garment is pre-shrink. That is the right direction for a knit you intend to wash normally, because you are matching washed to washed. It is the wrong direction for a garment you intend to wash cold and hang dry, because in that case your reference shrank and the new one will not, and the gap between them is the amount your reference lost.

The adjustment is a decision rather than a formula, and there are three branches. If the fibre is cotton and you will wash it normally, accept a reference number taken from a washed garment and buy to match it. If the fibre is wool or a wool blend, treat any pre-shrink number as unreliable, buy on the generous side, or accept that the garment will need cold hand washing for its whole life. If the fibre is synthetic, skip the adjustment entirely and buy to the measurement.

You now have three inputs: your measured reference, the chart, and a shrinkage direction. Where the chart and the reference agree within a centimetre or two, the decision is made and there is nothing left to weigh. Where they disagree by more than that but in a consistent direction across every field, you are almost certainly reading a different chart type, and you should resolve that before choosing anything.

Where they disagree by a lot and you cannot establish why, the answer is not a compromise size between the two. A compromise size produces a garment that is wrong in a way you cannot describe, which is worse to live with than one that is clearly too large, because it cannot be returned as a fit failure and cannot be worn comfortably. Choose the branch you can verify, which in practice means buying from a seller who publishes a garment chart, or buying a size you can exchange.

A useful rehearsal before paying: predict the number rather than the letter. Say out loud what you expect the chest of the arriving garment to measure. If you can state a figure in centimetres, your reasoning is complete and the purchase is a decision. If you can only state a letter, you have not finished the measurement step, and the size you are about to select is a guess wearing a chart-shaped justification.

Each step above has one test, and the tests are deliberately cheap because the expensive failure is the return. Step one passes if you can name the fibre of your reference and describe its wash history. Step two passes if you can state whether the chart is body or garment and in which unit, without going back to look at it a second time.

Step three passes if you can say which direction shrinkage moves your decision, rather than merely acknowledging that shrinkage exists. Step four passes if you can state the expected measurement in centimetres instead of a label, and if you can name what you would do if the seller published no chart at all, which is a situation that occurs often enough to deserve a prepared answer.

The pattern behind the tests is that a sizing decision is finished when it becomes falsifiable. If a garment arrives and you cannot say what you expected it to measure, then nothing that arrives can surprise you in a useful way, and the measurement work you did cannot teach you anything that improves the next order. The tests exist to make sure the work is reusable rather than merely done.