Sizing and fit for a womenswear range

01Separate body measurements from finished garment measurements

A size chart is unusable until it says which of two things it publishes. A body measurement describes the person a size is intended to fit. A finished garment measurement describes the made garment, taken on the cloth. They are different numbers for the same size, and publishing one while implying the other is the most common sizing mistake a first range makes. It produces returns that look like a fit fault when they are a labelling fault.

Read the difference through one value. A bust of 92 on a body chart says the size is drawn for a 92 bust, and the garment will measure more than that, or less where the cloth stretches. A bust of 92 on a garment chart says the garment measures 92; the intended body measurement depends on its ease and stretch. A customer holding a tape around herself reaches the right size in one case and the wrong size in the other, having done exactly what the chart asked.

Run both charts and label every column. The garment chart is what a factory cuts to and checks against, it moves whenever the pattern moves, and it exists per style. The body chart is the brand’s statement about who each size is for, it changes rarely, and it is the same document across the range. Collapsing them into one table is how a range ends up with a size 12 that means one body in the specification and another on the site.

02Fix the point of measure before anything is measured

A measurement is not a number until the method is written down. The natural waist is a judgement, the fullest part of the hip sits at a different height on different people, and a garment can be measured flat or around. Two competent people measuring one garment without an agreed method produce two results, and the argument that follows cannot be settled because neither of them is wrong.

Bust

Taken at the fullest point with the tape level and parallel to the floor, over the underlayer the garment is intended to be worn with. State that underlayer.

Waist

Either the natural waist or a stated distance below a fixed reference. The second is less elegant and far more repeatable between two people taking it.

Hip

The fullest point, or a stated distance below the waist. Where a style is fitted through the seat, publish the distance so the value can be checked.

Lengths

Name the start and end of every length: centre back neck to hem, shoulder point to cuff, waist seam to hem. A length without a named origin is not measurable.

Flat or around

Say whether a garment value is the full circumference or a half measurement taken flat, since a specification written flat reads as half the number a customer expects.

Relaxed or extended

For ribs, elastics and knits, give the relaxed measurement and the extended one. Both are useful and neither can be inferred from the other.

Use one method at every stage. A measurement taken one way at a fitting and another way on the production floor manufactures a fault that does not exist in the garment. Put the method in the specification as a flat sketch with the points of measure marked on it, and reproduce that drawing on the product page, so the factory and the customer are given the same instruction.

03Ease changes with the silhouette and with the stretch of the cloth

Ease is the designed difference between the body a size is intended to fit and the garment made for it, decided per style and per measurement point and never once for a range. Wearing ease is the amount that lets a person breathe, sit, reach and move. Design ease is everything added on top to make the silhouette look the way it was drawn.

Two garments cut from one body chart can therefore measure very differently and both be correct. A fitted shift dress may carry a few centimetres at the bust. A shirt intended to be worn oversized may carry twenty or more. Neither figure can be judged without knowing the silhouette it serves, which is why a range-wide ease allowance is not an instruction a patternmaker can use.

Fabric stretch changes the arithmetic completely. A jersey top with reliable recovery can be cut with negative ease, the garment measuring less than the body, and fit correctly because the cloth extends and returns. Extension and recovery are separate properties and both have to be tested. Cloth that extends without returning grows through a day of wear and reads as a size too large by the evening, which arrives as a complaint about sizing and is a fabric decision.

A fitted woven dress and a jersey top drawn from the same body chart are two patterns, not variations of one. Ease is not uniform around a single garment either: a dress can be close at the bust, released at the waist and full at the hip, and that distribution is the design. Record the intended ease against each controlled measurement, beside the body chart it is measured from, so a later fit comment reads as a grading fault or a change of intent.

04Decide the size range as a question of depth

How many sizes to offer at launch is a commercial decision about depth. Every size multiplies against every style and every colourway. Six styles in three colourways across eight sizes is one hundred and forty-four lines to buy, cut, hold and sell. The same range across five sizes is ninety.

The failure shows up as a broken size run. If the most requested sizes sell out while slower sizes remain, customers may be unable to buy the style they want. Which sizes sell fastest depends on the target customer and the initial allocation. Use that evidence when planning depth; unsold stock and markdowns can reduce the cash available for the next range. The same logic governs style count, and launching with fewer styles buys the same protection on the other axis.

Choose the run as a continuous span centred where the intended customer sits, and decide it from evidence, not from a wish to be available to everyone in the first season. Order quantity is the constraint underneath it, so what a given quantity makes possible should be settled first. Ask whether the maker's minimum applies per style, colour, size or total order, and record the agreed size allocation.

State the intention publicly as well. A brand that launches with five sizes and says the run is being extended is understood. One that says nothing has made a statement about who the clothes are for whether or not it meant to, and that statement is read from the size chart first.

05Extend the run as development work, not as a reprint

Adding a size at either end of the run is a development job with its own fit sessions. It needs a fit body at or near the new size, because a size derived from a rule and never worn is a guess with a barcode on it. Past a certain distance from the base it needs its own block, since proportion changes across bodies and not circumference alone.

What changes is specific and adding girth does not solve it. The bust point drops and moves outward, the waist sits differently in relation to it, back width and shoulder slope change, and the armhole has to be redrawn instead of enlarged. That work produces a second base and a second grade running above or below a break size: a normal way to build a wide range, and an expensive surprise after the sizes have been sold.

Time the extension against evidence the first season gives away cheaply: the sizes that sell out first, the exchange reasons, the waitlist requests and the enquiries asking for a size. Extend a proven style first. Adding sizes to a garment whose fit is already approved isolates one variable, so a fit problem in the new size can only have come from the new work.

06Choose a base size where the fit is most stable

The base size is the one size drafted and fitted first, and every other size is derived from it. Choose it near the middle of the intended run. Grading error accumulates with distance from the base, so a base sitting at one end leaves the far end as the least tested garment in the range and the one most likely to reach a customer wrong.

Choose it also where a consistent fit body is available: someone who matches the body chart at the controlling points and can be booked for every session the style needs. Where the intended customer is concentrated in one part of the run, the base belongs there even if that is not the arithmetic centre, because the most tested size should be the one most people receive.

Consistency of that body matters more than any single measurement being ideal, because comments taken on two different people are not comparable. A stand set to the chart holds measurements but cannot show movement, so it settles proportion and leaves fit unanswered until a person puts the garment on.

07Approve the base pattern before anything is graded

Approve three things together: the pattern version, the fit on the body and the measured record of the finished garment. Approval is a recorded state rather than a good fitting. It names the pattern file, the sample it produced, the cloth, the underlayers and footwear it was judged with, and the measurement sheet taken afterwards. Anything unwritten at that moment is re-argued later from memory.

Fit in the intended cloth wherever the calendar allows it, or in a substitute matched for weight, drape and stretch. A fit settled in a stiffer toile moves when the real cloth arrives, and it moves most where the design is most fitted. Reading the sample properly at this stage is a distinct skill, and how a sample is read decides how much of the range is built on a sound base. The first sample is a question rather than a verdict, which is why it is never the product.

Hold the sequence firmly, because a change to the base after grading invalidates everything derived from it. The grade has to be reapplied, every size-set sample cut from the old rule is superseded, and any measurement already published against the old base is wrong on a live page. Refusing to grade an unapproved pattern is the cheapest discipline in the process. The wider sequence this sits inside is set out in From pattern to approved sample.

08Grade each measurement at its own rate

A grade rule states how far each pattern point and each controlled measurement moves from one size to the next. It is not a uniform enlargement, and a pattern scaled uniformly is visible on a body within seconds, because the parts of a person that vary most between sizes and the parts that vary least have been treated as though they vary equally.

Bust, waist and hip take different increments, and the relationship between those increments is what makes one silhouette read correctly on a range of people and not only on the fit model. A dress that is close at the waist and released at the hip stops being that dress if the two move at the same rate. The proportions of the design live in the differences between the increments.

Some measurements barely move at all. Shoulder width, armhole depth, neck and sleeve cap change very little between adjacent sizes, because those parts of the body vary far less than girth does. Graded at the rate of the bust, a shoulder runs off the arm in the larger sizes, the neckline gapes, and the sleeve head stops matching the armhole it has to be set into, which turns a fit problem into a sewing problem on the production floor.

Lengths follow their own logic, and whether a length grades at all depends on the design. A hem meant to sit at a fixed point on the leg moves with the body; a hem meant to be a fixed length from the waist does not.

Details divide into placement and dimension. A patch pocket usually keeps its own size and changes position. Button spacing follows the length grade while the buttons stay the same, and the button at the bust has to stay at the bust in every size. Zips, elastics and trims step in the lengths they are stocked in, so the rule names the step instead of interpolating one. Write the grade point by point on the pattern and measurement by measurement in the specification, then confirm the two agree before anything is cut.

09Fit the size set on bodies that match the chart

A size set is a group of physical samples in selected sizes, commonly the two ends of the run and one or two between, made in the intended cloth and fitted on people who measure to the chart for those sizes. It is the only test of the grade that exists, because a grade rule is a set of intentions until a garment made from it is worn.

What it tests is what the numbers cannot show: whether the silhouette survives away from the base. Where the waist actually sits, whether the arm can come forward, how the skirt moves when the wearer walks, and whether a pocket that looked considered on the base has climbed towards the bust at the top of the run.

Fit the set in one session where the calendar allows it, with the underlayers and footwear used for the base. Differences between sizes are visible when the garments are seen together and become a matter of memory when they are not.

10Correct the rule when a derived size fails

When a size in the set fails, establish which of three things happened before touching anything. A point moving at the wrong rate is a grade correction and the cheapest of the three. A design that stops working past a certain size needs a split rule above or below a break size, which is development work and has to be scheduled. A fault that was always in the base sends the whole range back a step, and it is worth finding now.

Correct the rule, not the sample. Altering one garment and letting the others through means the fault ships in every size that was not fitted, and it ships with a measurement sheet saying it is correct. Re-cut the corrected size and fit it again before bulk, because a corrected paper pattern is a proposal and not evidence.

Record the failure and its cause beside the grade rule so the next style inherits the answer. The same block produces the same armhole problem in the next style that uses it. Keeping that record is part of what makes the second order easier than the first.

11The number on the label is a brand decision

Womenswear sizing is not standardised between brands, so the number printed on the label is a positioning choice and not a technical output. A size 12 in one brand and a size 12 in another are drawn from different body charts, with different ease, for different customers, and both are internally correct. Customers already know this: they carry two or three plausible sizes for themselves and check a chart before ordering, which is why the chart earns more attention than the number.

Vanity sizing is the practice of attaching a smaller number to a given set of body measurements than a brand previously did, or than its neighbours do. It flatters the customer who is pleased to take a smaller number and it is expensive on the customer who orders the number they have always worn and receives something too large. Both of those customers are in the same range at the same time, so the choice cannot be settled by deciding which of them to please.

Choose the naming against the reference set the intended customer already uses. If they shop across a group of brands whose 12 sits at a particular measurement, sitting a full size away from that group produces a steady stream of exchanges however well the garment is made. Being deliberately different is defensible and can belong to a position, provided the page says so plainly and the measurements back it up. Being accidentally different is a running cost with nothing to show for it.

Letter sizing moves the problem without removing it. S, M and L widen the span of bodies each size is asked to cover, which makes the published measurements more load-bearing, not less. Numeric, lettered or named, the naming still has to be declared against a body chart the customer can reach.

Internal consistency comes before any of it. A 12 that means one body in a dress and a different body in a trouser is the fault a customer notices fastest, because it stops them learning their size in the brand at all. Agreement with the rest of the market is worth less than agreement with the rest of the range.

12Publish the measurements a customer needs to buy

A product page has to let someone choose a size without handling the garment. That takes published numbers, a stated method for taking them, and a plain sentence about how the style is meant to sit. Every size decision the page leaves unsupported is made by guessing, and the response to guessing is to order two sizes and send one back.

The size chart

Labelled as body or garment measurements, with the point of measure defined for each value and flat or full circumference stated.

Per-style measurements

Finished measurements for the controlling points of that garment, in the size shown, since two styles in one range are meant to fit differently.

How to measure

Short instructions for each body measurement, with the drawing from the specification, so the customer measures the way the factory does.

Fit intent

A sentence in ordinary words: close through the waist, cut to be worn loose, shorter than it looks. It does work no table can do.

The model and their size

The model’s height and key measurements with the size they wear, wherever a photograph on a body is carrying the selling.

Fabric behaviour

Whether the cloth stretches, recovers, softens or relaxes with wear, because that decides whether sizing between two sizes is sensible.

The model note is worth insisting on. A photograph of a garment on a person is a fit claim, and two customers reading one image reach opposite conclusions about length and volume. The model’s height, measurements and worn size give both a useful reference.

Use fit returns to check the chart, the pattern, product consistency and customer expectations. A missing measurement is one possible cause, not the only one. It has taken the outbound delivery, the return, the handling and the time the garment spent out of stock, and it has spent the goodwill of someone who wanted the clothes enough to order them once. The rest of what a product page has to answer behaves the same way: questions left unanswered are answered by the customer guessing.

Keep the published figures tied to the approved size set and reissue the page whenever a pattern is revised. Where measurements are published finely, say that production varies within an agreed tolerance, so a garment inside that band is not received as a fault. The page and the garment drift apart quietly, and the first sign of it is a return rate nobody can account for.