On this page
- The three things that decide a size set
- What grading is — and what it is not
- Where grade rules actually come from
- What the increments actually do
- Grade breaks: rules do not stay constant across a range
- Petite, regular, tall, plus and cup-size ranges
- Reference points: get one wrong and every size drifts
- Measurement chart and tolerances
- Pattern truing after grading
- Releasing a size set from one approved sample
- Which sizes should actually be sampled
- A worked example: one rule, three sizes
- How to specify grading
- Where grading meets the size chart you already sell
- FAQ
- The grading release pack
Lay a graded pattern on a table and you can see the whole range at once: four, six or eight outlines of the same shape, nested one inside the next, sharing a centre line and diverging as they move away from it. The base size is the one that was approved on a body. Everything else in that stack is an inference — a set of rules applied to a garment that has never been worn.
That is the part buyers underestimate. A brand approves a size M sample, the fit is right, the chart is confirmed, and the size set is released. When bulk arrives, the M is correct, the L is acceptable — and the XL has a problem nobody can name precisely: the neckline gapes, the armhole pulls, the waist sits in the wrong place relative to the body. Every measurement on the chart is within tolerance.
The garment is not incorrectly graded in the arithmetic sense. The increments are the ones that were agreed. What happened is that grading was applied to a set of measurements while the garment's behaviour was left to look after itself — and behaviour does not scale linearly. A size set is not a photograph of the sample enlarged and reduced. It is a set of rules applied to a body whose proportions change as it changes size.
This article sets out where grading rules actually come from, why they change across a range, how a graded pattern is trued, and which sizes have to be physically made before a size set is released.
The three things that decide a size set
| What decides it | What goes wrong if it is not agreed |
|---|---|
| The base size and the reference points | Grading runs from the wrong point; every size drifts in the same direction |
| The increments per point, including any break in them | Sizes grow at rates the body does not follow; extremes distort first |
| The tolerances per point | Every size is judged against a number with no stated acceptable range |
The size that fails first is almost always an extreme — not because the arithmetic is harder there, but because errors in the increments accumulate as the size moves away from the base.
What grading is — and what it is not
Grading takes an approved pattern and produces the other sizes by moving specific points by defined amounts. The pattern keeps its logic; the distances between its points change according to a rule set.
Not proportional scaling
The most common misunderstanding is that grading enlarges and reduces the whole pattern evenly. It does not, and it should not. Bodies do not scale evenly: a larger size does not have proportionally longer arms, a proportionally wider neck, or a proportionally deeper armhole. Grading rules exist precisely because growth is unequal between parts.
If a pattern is scaled proportionally, the result tends to be acceptable in the middle of the range and wrong at the extremes — which is the XL problem in the opening example. Proportional scaling is also the fastest way to produce a size set that passes measurement and fails fit.
Not identical to a size chart
A size chart lists finished measurements per size. Grading rules produce them. The chart is the output; the rules are the mechanism. Two factories can supply the same chart having arrived at it by different rules, which is why a chart alone does not make a size set reproducible.
Where grade rules actually come from
This is the correction that matters most, because there is a persistent habit of treating a published table of increments as if it were a standard.
There is no universal grade rule set. Increments are a design decision expressed numerically, and they come from six inputs.
| Input | Why it changes the increments |
|---|---|
| The brand's body standard | The measurement specification for the intended customer. A teen market and a mature market do not grow the same way even within one size label |
| The market and size system | US, EU, UK and Asian sizing differ in range width, where the range's centre sits, and how a label maps to a body |
| The fit block | The ease and proportion built into the base pattern. Two blocks with the same base measurement can grade differently and legitimately |
| The category | Fitted wovens, knits, outerwear and loose shapes grow differently. A jersey dress and a tailored dress do not share increments |
| The width of the range | A three-size range can be graded with a single rule set; a ten-size range usually cannot — see grade breaks below |
| The vertical policy | Whether the garment lengthens as it widens is a fit decision, not an arithmetic consequence |
The practical consequence: any table of "typical increments" — including the one below — is an illustration of the principle, not a rule set to adopt. What the table demonstrates is that different parts of the garment move by different amounts, and that the small-increment points are where errors become visible.
What the increments actually do
Grade rules define how much each point moves per size step. The relationships below are typical in womenswear ranges; the specific values are agreed per project.
| Measurement point | Typical relationship to the others | Why it behaves that way |
|---|---|---|
| Bust / chest | Largest horizontal increment | The torso widens most at this level across a size run |
| Waist | Smaller than the bust in most ranges | The waist widens more slowly than the bust |
| Hip | Broadly similar to the bust | But positioned differently in the pattern, so it drives a different part of the shape |
| Shoulder width | Very small increment | Shoulder width grows little between adjacent sizes |
| Neck width | Very small increment | Small change; over-grading here produces a gaping neckline |
| Across back | Moderate | Governs the armhole and sleeve relationship |
| Sleeve length | Small increment | Arm length changes far less than width across a range |
| Garment length | Depends on intent | Some ranges hold length and change width; others scale both |
Two rules that follow
Vertical and horizontal growth are separate decisions. A dress that lengthens proportionally as it widens often reads too long at the top of the range; a garment that holds its length while widening can read too short at the bottom. Neither is wrong — but the choice has to be made rather than inherited.
Small increments deserve as much attention as large ones. Neck width and shoulder width move very little, which means an error of a few millimetres in the rule produces a visible defect at the extremes, while the same absolute error at the bust is invisible. The worked example later in this article shows exactly how that accumulates.
Grade breaks: rules do not stay constant across a range
This is the piece most often missing from a grading specification, and it is the one that explains a large share of "the bigger sizes are wrong" complaints.
Grade rules do not have to be constant across the whole size range. A common and deliberate structure is one rule set across the core sizes and a different one from a defined point upward — because the body's proportions change, and because the market's fit expectation changes with it. The point at which the rule changes is a grade break.
Where breaks are normally considered:
- At the transition between a core range and an extended range. The increments that produce a good fit from XS to L are frequently not the increments that produce a good fit from XL upward.
- Where a garment would stop being wearable at the top of the range. Armhole depth, sleeve width and shoulder width are the usual places: a rule that is harmless over four sizes can make the largest size uncomfortable over eight.
- Where a design feature appears or disappears. If a style adds a panel, a yoke or a closure only above a certain size, the rules either side of that point are doing different work.
- Where the fit block itself changes. Plus ranges and cup-size ranges are usually graded from a different base — see the next section.
Two practical consequences.
A single linear increment column across a wide range is a warning sign. Either the range is genuinely narrow enough that one rule set holds, or the break was never considered. Asking the question costs nothing and often changes the answer.
Sample either side of every break. The sizes immediately below and above a break are where a rule change is most likely to be visible on a body, and they are the sizes that get skipped when a brand samples only the two ends of the range.
Petite, regular, tall, plus and cup-size ranges
These are habitually presented as extra sizes appended to a single run. They are not. Each is a separate fit block with its own base and its own rules, and treating them as an extension of the regular range is how a brand ends up with a range that fits its core customer and nobody else.
| Range | What actually changes | What it is not |
|---|---|---|
| Petite | Vertical references move — lengths, rise, waist position — with comparatively small width changes | Not a regular range shortened by a fixed amount at the hem |
| Tall | The same vertical references move in the other direction; arm and leg length, torso length and waist position shift | Not a regular range graded up |
| Plus | Different proportion changes, usually with grade breaks, and often a different block; horizontal and vertical growth diverge more than in a core range | Not a core range with increments added at the top |
| Cup-size ranges | The bust reference changes relative to the underband, so each cup is effectively a different base pattern | Not a core range with a bigger bust increment |
The commercial point: selling into a plus or cup-size market means approving a plus or cup-size block, not extrapolating. That is a development cost, and it is the honest reason those ranges cost more to open — not a pricing decision, a fit one.
Reference points: get one wrong and every size drifts
Grading does not operate on the whole pattern. It operates on named reference points — the ones corresponding to the measurements on the chart — and everything between them follows.
| Reference point | Why it is foundational |
|---|---|
| Centre front / centre back | Sets the vertical axis; errors here shift the whole garment |
| Shoulder point | Governs neckline width, armhole position and the sleeve relationship |
| Bust point | The horizontal and vertical datum for the whole upper body |
| Waist line | The narrowest reference, and the one fit sessions judge most strictly |
| Hip line | Sets the relationship between torso and skirt across the range |
| Side seam | Where vertical and horizontal growth meet |
Why an error here is expensive. Because every other point is positioned relative to these. An error at the bust point does not produce one wrong measurement; it produces a garment whose neckline, armhole, waist and hip are all slightly wrong in a consistent direction — by an amount that grows with the distance from the base size. That is why a grading error is often invisible at the base size, marginal one size away, and obvious at the extreme.
It is also why the base size matters. Where the base is mid-range, errors distribute in both directions and are easier to spot. Where the base sits at one end — a small base for a wide range — every other size accumulates the error, and the far end fails first.
Measurement chart and tolerances
A size set is released with a measurement chart: finished measurements per size, at named points. The chart is necessary and it is only half of the specification, because it states what each measurement should be and not how much variation is acceptable.
Why tolerances belong with the chart. Two garments can both be "correct" against a chart and measure differently, because manufacturing has natural variation. Without a stated tolerance per point, every measurement is either perfect or wrong, and inspection becomes an argument.
Tolerances are usually set per point rather than applied uniformly, because points differ: a bust or waist measurement on a woven garment is easier to hold closely than a measurement across a curved seam, and a knit behaves differently again. As with grade rules, the values are agreed per project — what matters is that they are written next to the chart rather than assumed.
| Quantity | What it is | Where it lives |
|---|---|---|
| Measurement | What the size should be at a named point | The chart |
| Tolerance | How much variation is acceptable at that point | Next to the chart |
| Fit intent | How the garment should relate to the body at that size | The fit comments and the approved sample |
A size set that states only the first is a drawing. One that states the first two can be inspected. One that states all three can be repeated.
Pattern truing after grading
Grading moves points. It does not, by itself, guarantee that the pieces still make a garment. Truing is the check that happens between grading and cutting, and it is the step that catches the failures which show up as twisting, pulling and dipping rather than as wrong measurements.
| What is checked | The failure it prevents |
|---|---|
| Seam lengths match across pieces | Side seams, shoulder seams and waist seams that are unequal, so one side has to be eased or stretched during sewing |
| Notches align | Pieces that go together but land out of position, which shows as a shifted seam or a visible mismatch at a join |
| Armhole and sleeve-cap relationship | A sleeve cap whose length no longer matches the armhole, which is what produces a pulling or gaping sleeve at the extremes |
| Side-seam balance, front to back | A garment that rotates on the body, so the side seam sits forward or back of where it should |
| Curve quality | A neckline or armhole that was a smooth curve becoming a corner because a point moved and the intermediate curve did not follow |
| Grain lines | Grain that drifts across the range, which changes how the garment hangs at some sizes and not others |
| Dart, pleat and seam positions | Shaping that stops sitting where the design intended because its position scaled differently from the panel around it |
Two of these explain symptoms that look mysterious in production. A sleeve that pulls at the top of the range is usually a truing finding, not a rule finding — the cap and the armhole are graded against each other, and that relationship is easy to lose when the armhole depth uses a different increment from the cap height. And a garment that twists is usually a balance finding: front and back side seams of unequal length, or a grain that drifted, makes the garment rotate, which is invisible on a form and obvious in wear.
Which is why truing is a gate before a size set is cut rather than a formality. On structured styles it is more involved, because internal layers have to be trued alongside the shell — how structured bodices are made in womenswear covers why the build and the grading have to be agreed together.
Releasing a size set from one approved sample
The practical sequence is short, and each step depends on the one before it.
1. Confirm the base size and its chart. The approved sample is measured against the chart, and any differences between the sample as made and the chart as written are resolved — because grading from an unconfirmed base reproduces the discrepancy across the range.
2. Confirm the reference points. The points that will be graded are identified on the pattern and checked against the chart's named points. A mismatch here is the most common cause of a size set that measures correctly and fits inconsistently.
3. Apply the grade rules — including any break. The pattern is graded in the agreed steps, with the break (if there is one) applied at the agreed size.
4. True the graded pattern. Seam lengths, notches, armhole-to-sleeve, balance, curves and grain are checked across the range before anything is cut.
5. Check the sizes that matter, on a body. Not only the two ends: the base is already approved, so the informative sizes are the extremes and the sizes either side of a grade break.
6. Release the chart, the tolerances, the rules and the break points together. A size set released as a chart alone invites re-interpretation at the next order, because the rules that produced the chart travel nowhere.
Which sizes should actually be sampled
"Sample the smallest and the largest" is the most common instruction on a size set, and it is an incomplete one. The sizes worth making are the ones where the rules are doing the most work, or where the risk is highest.
| Decide by | What it implies for sampling |
|---|---|
| Range width | A narrow range can be validated at the ends; a wide range needs more than two points |
| Grade breaks | Sample immediately either side of every break — the sizes where the rule changed |
| Construction complexity | Structured, gathered, bias-cut and highly shaped garments need more validation points than a simple loose shape |
| Risk points | The areas the rules move most — bust, armhole, waist — deserve a size that exercises them |
| Design features that appear mid-range | If a panel, yoke or closure is added at a larger size, that size has to be made |
| The customer's range of bodies | Where the range is intended to cover a wide span of wearers, one mid-range body does not validate it |
The honest minimum for a range with no breaks and a simple construction is the base plus both ends. Add: one size either side of each grade break; any size where the construction changes; and at least one size at the top of the range for any style with a fitted armhole or a shaped bodice, because that is where accumulated error becomes visible first.
Two constraints on doing this well. Size-set samples have to be read on production-intent material, not substitutes, because a grade is validated on the cloth it was calculated for. And a size set is a development cost that has to be planned: the more points you sample, the longer the validation. How a first sample is developed covers why a size set is a different exercise from a fit sample.
A worked example: one rule, three sizes
Take a fitted dress graded from a mid-range base, with the neck-width rule set a few millimetres too generously per step — and nothing else wrong. The bust, waist and hip all follow their own rules correctly.
| Size | Neck width | On a body |
|---|---|---|
| Base | Correct | Correct |
| Two steps up | Visibly wide | Slight gaping at the shoulder |
| End of range | Clearly wide | Gaping neckline; the fit reads as "the pattern is wrong" |
One small increment, set wrongly, accumulating with distance from the base. That is why grading problems are so often described as "the larger sizes are wrong" rather than "one rule is wrong". The reverse happens too: where a rule is too tight, the extreme size reads as restricted — an armhole too small, a sleeve too narrow, a waist that fits but does not sit. Both directions are fixed at the rule rather than at the garment.
How to specify grading
A grading specification is short, and it should travel with the chart rather than replace it.
| # | What to state | Why |
|---|---|---|
| 1 | Base size | Grading runs from it; everything else is relative |
| 2 | Size range | Determines how far errors accumulate, and whether breaks are needed |
| 3 | Reference points | The points being graded, matched to the chart's named points |
| 4 | Increments per point | The rule set itself; without it the chart cannot be reproduced |
| 5 | Grade breaks | The sizes at which the increments change, and what they change to |
| 6 | Vertical policy | Whether length scales, holds, or changes at defined points |
| 7 | Tolerances per point | Turns the chart into something inspectable |
| 8 | Fit intent per size tier | How the garment should behave at the ends of the range and either side of a break |
The lines buyers forget
The increments. Charts are sent; rules usually are not. Without the rules, the next size set is a fresh interpretation.
The grade breaks. A specification that names a range but not where the rules change describes only part of the range.
The vertical policy. Whether a style grows in length as it grows in width is a design decision that is often left unstated and then inherited by accident.
Fit intent at the extremes. The base size is approved on a body. The extremes are where the design intent has to be restated, because that is where the rules do the most work.
Where grading meets the size chart you already sell
Most brands are not grading into a vacuum. They already have a size chart — their own, or a market convention — and the graded set either matches it or does not.
| Situation | What it means | What to decide |
|---|---|---|
| Factory grades to the brand's chart | The brand's fit is the reference | Provide the chart with tolerances, and confirm the base size it is written against |
| Brand has no chart | The factory's convention becomes the fit | Approve a chart before grading, not after the size set arrives |
| Two charts disagree | The brand's chart and the sample's measurements differ | Resolve before grading; grading reproduces whichever is used |
| The brand sells a plus or cup range | The core block cannot be extrapolated | Approve a separate block and its own rules |
The third situation is the expensive one, because it stays invisible until a later order — the graded set follows one chart, the inspection follows the other, and the discrepancy returns as a "measurement issue" that looks like a factory problem rather than a specification gap. The practical rule: the chart and the base sample have to agree before the size set is graded.
FAQ
What is garment size grading? Grading is the process of producing the other sizes in a range from one approved pattern by moving defined reference points by defined increments. It applies rules to a pattern; it does not enlarge the sample.
Why can't I just scale the pattern proportionally? Because bodies do not scale proportionally. Shoulder width, neck width and sleeve length change far less between sizes than bust or waist. Proportional scaling produces garments that are acceptable in the middle of the range and wrong at the ends.
Are there standard grade rules I can just adopt? No. Increments come from the brand's body standard, the market and size system, the fit block, the category, the width of the range and the vertical policy. Published tables of typical increments illustrate the principle; they are not a rule set that transfers between projects.
Which size should be the base? Usually a mid-range size, so that errors accumulate in both directions and are visible earlier. A base at one end of a wide range concentrates error at the far end, where it is hardest to correct cheaply.
What is a grade break? A size at which the increments change. Rules do not have to be constant across a full range — a common structure is one rule set across the core sizes and another from a defined point upward, often around the transition into plus or extended sizing. The break and its position belong in the grading specification, and a range should be sampled either side of it.
Why does my largest size distort when the smaller sizes are fine? Because a rule is slightly wrong and its effect accumulates with distance from the base. Small increments — neck width, shoulder width — are the usual culprits, since a few millimetres per step becomes visible over several sizes. Where the smallest sizes are fine and the largest are not, it is also worth checking whether a grade break was needed and was never set.
Do I need tolerances if I have a measurement chart? Yes. The chart says what each measurement should be; tolerances say how much variation is acceptable. Without them, every measurement is either perfect or wrong, which is not how manufacturing works.
What is pattern truing? The check that a graded pattern still makes a garment: that seam lengths match, notches align, the sleeve cap agrees with the armhole, front and back side seams balance, curves remain smooth and grain stays correct. Truing catches the faults that appear as twisting, pulling and dipping rather than as wrong measurements.
The grading release pack
Everything above reduces to one set of documents. Released together, they make a size set reproducible; released as a chart alone, they make it a fresh interpretation at every reorder.
| # | Document | What it must contain |
|---|---|---|
| 1 | Base confirmation | The approved sample, its size, the chart it was measured against, and any resolved differences |
| 2 | Measurement chart | Finished measurements per size at named points |
| 3 | Tolerances | A stated acceptable range per point, not one global figure |
| 4 | Grade rules | Increments per point, plus any vertical policy |
| 5 | Grade break schedule | The sizes at which increments change, and what they change to |
| 6 | Truing record | Seam lengths, notches, armhole-to-sleeve, balance, curves and grain checked across the range |
| 7 | Size-set sample list | Which sizes were physically made, and why those — including the sizes either side of any break |
| 8 | Fit intent note | How the garment should behave at the base, at the extremes and either side of a break |
Eight rows, and the point of writing them down is not compliance. It is that a size range is the one part of a garment that no single person ever sees. The pattern is graded, the sample is cut, the chart is written, the sizes are sewn — and the first time the whole range exists as garments is in the customer's hands. The release pack is what stands in for having seen it.
If you want a size set released against agreed rules and a truing record rather than a chart alone, send the approved sample and your target range. OEM manufacturing covers how the sample, the chart and the size set are kept in agreement, quality control covers how measurement is inspected against tolerances per point, and pattern making covers the block the whole range is built on. Where a size range also carries structure or shaping that has to be graded with it, how dress construction relates to fit and cost explains why those decisions cannot be priced separately.