Every instruction in this family names a shape or a place (the layer begins around the cheekbone, the fringe frames the face) and none of them names a width. A width is what the cut actually consumes: the section divided off along the hairline, measured from one edge of it to the other, before anything has been shortened.
That width is published, though not by hair pages. The forehead-to-cheekbone ratio carries five centres on this site's rule set, the table of ratios every face-shape page here is built from, one centre for each face outline (the shape names this site sorts readers into), from 0.80 to 1.05. The distance between the two ends is a quarter of your own cheekbone width, 34.0 mm (1.34 in) at the 136 mm of cheekbone the table below works on. Multiply the figure that fits you, halve it, and the words curtain bangs and face-framing layers turn into two distances you can hold a ruler against.
One Strip of Hair, Measured Across
Take the hair that will fall against your face and mark how wide it is, edge to edge along the hairline. That is the number the whole family talks around. A curtain is this section parted down the middle and opened to both sides; a face-framing layer is the same section cut short in the middle and left long at the two edges. Both start from one width, and the difference between them is a distance.
The width has a published size. Forehead width divided by cheekbone width, the two lines the calculator on this site asks you to take against your skin, runs from 0.80 to 1.05 across the five centres. A ratio is not a length until you multiply it by something, so the table works on a single cheekbone width: 136 mm, which is 5.35 in. Substituting your own width is one multiplication, and section six says what the substitution does not give you.
Read the fourth column as this page's own choice rather than as a figure on the record: between two-thirds and three-quarters of one edge is the band the cut is made inside of. The fifth column is where that band's outer edge sits against the widest point of your own cheekbones, and it changes sign on the top row. At the widest published centre the edge of the section lies 3.4 mm (0.13 in) beyond the cheekbone's own widest point, while at every other centre it lies inside it. Nothing in the record draws that line, because nothing in the record measures the section. For a reader with a tape, what the table asks for is one line from column four, and column five is where you check whether that line sits inside your own face or past it.
The section both cuts are made from, at the five published forehead-to-cheekbone centres, worked on a cheekbone width of 136 mm (5.35 in). Column two is the ratio multiplied by that width; column three is half of column two; column four is the band this page cuts inside; column five is the difference between one edge of the section and half of 136 mm.
| Forehead-to-cheekbone centre, and the outlines that carry it | Section, edge to edge | Part line to one edge | This page's band, one edge | That edge against the cheekbone's widest point |
|---|---|---|---|---|
| 1.05 — heart | 142.8 mm (5.62 in) | 71.4 mm (2.81 in) | 47.6 to 53.6 mm (1.87 to 2.11 in) | 3.4 mm (0.13 in) beyond it |
| 0.95 — square, oblong | 129.3 mm (5.09 in) | 64.7 mm (2.55 in) | 43.1 to 48.5 mm (1.70 to 1.91 in) | 3.4 mm (0.13 in) inside it |
| 0.90 — oval | 122.3 mm (4.81 in) | 61.2 mm (2.41 in) | 40.8 to 45.9 mm (1.60 to 1.81 in) | 6.9 mm (0.27 in) inside it |
| 0.88 — round | 119.8 mm (4.72 in) | 59.9 mm (2.36 in) | 39.9 to 44.9 mm (1.57 to 1.77 in) | 8.1 mm (0.32 in) inside it |
| 0.80 — diamond, triangle | 108.8 mm (4.28 in) | 54.4 mm (2.14 in) | 36.3 to 40.8 mm (1.43 to 1.61 in) | 13.6 mm (0.54 in) inside it |
· 136 mm is a working line, not a prediction about you: substitute your own cheekbone measurement and every millimetre column moves with it. · Each inch figure is the millimetre product divided by 25.4 and rounded on its own, so the two columns print one number twice; 40.8 mm pairs with 1.60 in and 45.9 mm with 1.81 in. · Three cells land exactly on a half of a tenth of a millimetre (53.55, 64.65, 61.15) and are rounded up. · Column four is this page's rule, not a published band: two-thirds to three-quarters of one edge. Cut at one-half to two-thirds instead and the top row reads 35.7 to 47.6 mm (1.41 to 1.87 in), with the lower edge 11.9 mm (0.47 in) below the one printed. · One published standard deviation of the ratio (0.081, how far measured foreheads stray from the centre) moves the edge-to-edge width by 11.0 mm (0.43 in) at 136 mm, which is 5.5 mm on one edge. · Two rows carry two outlines each because those outlines share a published centre.
No account · No ad pixels · No photo upload
Detection runs on-device with MediaPipe. Nothing is uploaded — no account, no email, no storage.
The Forehead Width Curtain Bangs Are Cut Against
The pages that answer this question do use width words. One tells you to consider the width of your forehead, your cheekbones and your jawline, along with the length of your face. Another, written for a different question, tells you to compare the widest point of your forehead with your cheekbones and jaw. Those are the two operations the table above performs, left as sentences: no figure, no unit, nothing to hold a ruler against.
The older books are no more numeric, and they do not pretend. A barbering manual from 1905 parts the front hair from the forehead about halfway back towards the crown, then makes the amount divided off a proportion whose variable is the thickness of the hair: in thick hair, a small proportion. A hairdressing text from 1923 gives the section a width directly (about the breadth of a hand, measured from the forehead to the place where the parting turns) and adds that a broad forehead is handled by moving the middle of that parting back towards the crown so the waves fall over the sides. A hand is a body part, and it differs between people: two people with the same forehead can bring different hands.
Where a curtain stops being a curtain is the question the record never answers, so this page draws the line and says that it is drawing it. Take one edge of the section as a share of your own half-width, which is what column four does, and then hold a fixed opening against it. An opening of 45 mm (1.77 in) on an edge covers 63.0 per cent of the widest published line and 82.7 per cent of the narrowest. At 55 mm (2.17 in) the same opening covers 77.0 per cent of the top row and 101.1 per cent of the bottom one, which is past that row's own edge: the hair being pulled forward is no longer a curtain but the whole front section and a little of what was meant to stay behind it. That is the difference between the two cuts, on the same ruler as everything else here.
One caution about that crossing. The arithmetic is about a section of hair; whether a face suits a cut is a different question, and this page has no figure for it. The record does make claims about what an appearance does: one page says that a sweep to one side reduces forehead width while adding movement, another that a shorter layer or bang placed at eyebrow level offsets the width of the forehead. Those are those pages' claims, and nothing here can repeat or check them. What a ruler does settle is narrower: how much of the strip the cut takes, and whether that leaves an opening in the front or pulls the whole section down with it.
What One Millimetre Costs, Edge by Edge
A search for curtain bangs face shape is asked in the other direction: an outline first, then the cut. The arithmetic below is why one fixed number cannot serve both directions. A single millimetre of extra opening is 1.40 per cent of one edge at the widest published centre and 1.84 per cent at the narrowest. The same millimetre costs about a third more of the strip at the narrow end, because the strip it comes off is narrower.
The band moves the same way. It is one-twelfth of an edge at either end, so 6.0 mm of room to cut in at the top row and 4.5 mm at the bottom one, and the widest published centre has a third more slack than the narrowest. Two readers can each say they took the middle of their band and still land 12.0 mm (0.47 in) apart on an edge, one at the widest centre and one at the narrowest. Same words, one multiplication apart, and the band itself is the only style opinion on this page.
What the percentages do say is that a rule written as a fixed number is a rule written for one outline and quietly applied to seven. A page that told you to take 52 mm (2.05 in) of the front would be spending 72.8 per cent of one edge at the top row and 95.6 per cent at the bottom one. Neither figure is a style opinion; both are the division this page does in public.
Who Said What, In Order
One page in this family already owns the two words this table is built from. It puts landing points in a heading and asks, in the body, how wide the fringe section should be, then decides it in a sentence: a narrow triangular section makes a lighter fringe and leaves more hair at the sides, while a wider or deeper one fills out the curtain and takes more density off the front perimeter. That is the right question, asked first, and the missing number. Its own table lands at the brow, at the highest point of the cheekbone and at the jaw corner, and prints no figure against any of the three rows. What follows claims only the half that page left open: the width as a figure you can substitute into, with the tolerance named (how far either side of that figure the cut is still the cut you asked for) and the choice marked as a choice.
Fifteen addresses were sent on this page's own pass. Ten gave text on 20 October 2026, at between 485 and 5,231 words. Two more had been read earlier that same morning, in a pass about twenty minutes before this one, and did not resolve on the counting pass, so their figures come from the attempt that read them: a fringe guide at 2,246 words with one table, and a styling guide at 3,688 words with none. Three returned nothing, a 404, a refusal and a pause screen. Across the twelve pages whose text reached a reader, not one prints a length in a unit. The longest runs to 5,231 words, and no figure in it is followed by a length unit. Two of the twelve print a table, and neither of those tables holds a number. Four further positions named below come from the same morning's wider sweep and are kept out of those counts: one was readable there at 1,261 words, one returned only a picture index of 104 words, one only a discussion page's rules, and one is an address that sweep carries cut short.
What has been said about the width of the front section, sorted by when it arrived. The undated rows sit at the bottom, counted by the words on the page that was reached; they are the parts of the record that could not be dated, and they are not discarded. The five rows above this page's own are positions where no text reached a reader in either pass, kept so that the count above can be checked.
| Date, and what kind of source | What it says about the width of the section | What it leaves you holding |
|---|---|---|
| 1905 — a trade manual for barbers and hairdressers | Part the front hair from the forehead about halfway back towards the crown; how much is divided off is a proportion that shrinks in thick hair | A place and a proportion. No width, no unit |
| 1923 — a hairdressing and cosmetology textbook | The front section is about the breadth of a hand, measured from the forehead to the cross-parting; a broad forehead is handled by moving the middle of the parting back | The only direct width in the record, given as a body part. Two people, two hands |
| 28 November 2022 — a beauty magazine's fringe column, 1,487 words | How to get the centre-parted shape and keep it through a day; the section is where the styling starts | A routine, and a date printed only in abbreviated form: the passes that scanned for month names in full found none on it |
| 16 January 2023 — a salon's note on layering against face framing, 2,189 words | The difference told as movement: layers go through the shape, framing holds it | Two names and no measurement |
| 15 January 2025 — a magazine's gallery of face-framing layers, 2,410 words | The shortest layer placed around the cheekbone; a shorter piece at eyebrow level to offset the width of the forehead | Two positions and one width word, no number attached to either |
| 13 February 2025 — a fashion magazine's guide to the centre-parted fringe, 2,495 words | Answers its own heading asking what separates face-framing layers from curtain bangs, in blow-drying steps | A routine. The section is never divided off in words |
| 25 April 2025 — a salon's styling guide for every outline, 5,231 words | Consider the width of your forehead, your cheekbones and your jawline, as well as the length of your face. The same page also prints 1 February 2026 | The widest sentence in the family, in both senses, and no figure in 5,231 words |
| 14 July 2026 — a stylist's blog on the three front-hair options, 1,751 words | Curtain, blunt or face-framing, and what decides it before your outline does | Three options sorted by treatment; the width is not one of the variables |
| 22 September 2026 — a fringe guide, 2,246 words, one table | Landing points in a heading; how wide the fringe section should be decided in a sentence: a narrow triangle gives a lighter fringe, a wider or deeper section fills the curtain and removes density from the front perimeter | The right question and the missing number: its table lands at brow, cheekbone and jaw with no figure printed in any of its four columns |
| 28 September 2026 — a salon's comparison of the two cuts, 1,108 words | Which of them changes your routine, and how each behaves through a day | A maintenance comparison. No section, no width |
| 9 October 2026 — a hair group's comparison page, 485 words, one table | A table whose row asks where the shortest framing piece should sit | The question in a table cell, and no number in the cell next to it |
| No date — 3,688 words — a hair group's styling guide for the centre-parted fringe | A styling guide read at 3,688 words, with no table and no width figure in it | Read on this page's first pass; the pass that counted the fifteen could not resolve the address |
| No date — 2,444 words — a hairdresser's guide to face-framing cuts | The difference between curtain bangs and face-framing layers, then: section the hair | Three words of instruction about the section, and no width |
| No date — 2,289 words — a consumer blog on fringes and outlines | Compare the widest point of your forehead with your cheekbones and jaw, then pick the fringe | The division, told as a gesture. No result printed |
| No date — 1,600 words — an app's feed page on long layers with face framing | Read at 1,600 words by the same morning's wider sweep, with no figure in it | This page's own pass found only a 404, so the count of twelve does not carry it |
| No date — 1,261 words — a style news explainer on the two cuts | Read by the wider sweep at 1,261 words, with no figure in it | Kept out of the count of twelve: this page's own pass never sent that address |
| No date — 362 words — a discussion page on the same choice | Only the panel of discussion rules came back, in either attempt | The discussion itself was never read, so it is not evidence of anything |
| No date — 104 words — an image-idea shell page for the comparison | Nothing: a shell that collects pictures, with no article behind it | Excluded by the sweep's own rule for picture indexes, and kept out of both counts |
| No date — 8 words — a same-family tool site's curtain-bangs page | Deployment paused | No article exists to read; recorded so the count of fifteen closes |
| No date — 8 words — a magazine's face-framing layers page | Refused the reader | Same: a position that produced no text, kept in the count |
| No date — no count — one address the record itself carries cut short | Nothing read: the tail this page guessed for it came back as a 404 | A gap in this record, recorded as a gap |
| 20 October 2026 — this page | The section as five published centres multiplied by one working width, in millimetres and inches on each edge, with the cut's own band marked as chosen and what moving it costs printed beside it | A number you can substitute into, and the two places where the substitution has to stop |
Word counts and table counts come from the pass this page sent over its fifteen addresses on 20 October 2026, with scripts and styles removed. A pass a few minutes earlier measured two of those pages 2 and 3 words differently; the later pass is the one printed here. Two of the fifteen would not resolve on that later pass, and their figures are the earlier one's, row by row. The rows drawn from the wider sweep the same morning carry that sweep's counts and say so. Source kinds are given instead of names by this site's rule. The last row is this page's own date, and it moves if the page ships on a different day than the one it was written.
A Part That Is Not in the Middle
The same question gets asked as side parted fringe face shape, and the arithmetic there is friendlier than the answers suggest. Move the part line sideways and one edge of the section lengthens by exactly what the other gives back. On the 0.88 row, whose edge sits 59.9 mm (2.36 in) from a centre part, a part shifted 10 mm (0.39 in) off centre leaves 49.9 mm on one side and 69.9 mm on the other, and the two still add out to the same 119.8 mm across.
That is the useful half of the sentence, and the other half is the warning in it. An off-centre part cannot widen the section; it can only move the edge. At 69.9 mm the wide side has gone 25.0 mm past the band this page cuts inside of (69.9 against 44.9) and sits 1.5 mm short of the widest published row's own edge, while the narrow side is still 10 mm short of its own. A deep side part therefore spends the same amount of hair as a centre part and charges all of it to one edge.
Whether you want that is not in this table. What is in it is that the hair ending up framing one side came out of the back, and that the sum did not move when you decided it.
What This Table Cannot Give You
The five centres are published rules about outlines, not measurements of your forehead. If your own two widths divide to 0.93, use 0.93. The outline names in the first column are there to help a reader find the nearest line; the row you work from is settled by your own division.
The next limit is the ruler. This site's classifier works in a scale built from points a photograph can locate, where the same seven lines run 0.910 to 0.978 with a spread of 0.022; every figure printed above is on the scale a tape gives, where they run 0.80 to 1.05 with a spread of 0.081. The conversion goes end to end (widest line to widest line, narrowest to narrowest, nothing in between re-scaled), and it is why a millimetre from this page cannot be checked against a classifier's number without converting it back first. The millimetres here are on the tape scale because that is the scale you can put a ruler against.
Two more. The band in column four is this page's own: two-thirds to three-quarters, chosen because it is the range in which a curtain can open without taking hair a framing layer leaves at the edges, and it is published nowhere; the sensitivity (what each row becomes if the band is moved) is printed with it, and a reader who prefers one-half to two-thirds has 11.9 mm (0.47 in) of room at the top row and 9.1 mm (0.36 in) at the bottom one. And nothing here answers which cut suits which face. The record's perceptual claims are the record's, and a table of widths cannot check them. The length a fringe falls to is a different axis; the page on how long bangs should be measures that one and does not measure this.
Cutting to It
The two widths this table multiplies are two of the four the calculator takes against your skin, in the same order it asks for them. Open the face shape calculator ↓
Take the two widths the calculator asks for (forehead across its widest part, cheekbones across the tops) and divide the first by the second. That quotient is the line in column one you are working from; the outline name printed beside it is a label for the same row.
Then multiply it by your own cheekbone width and halve the result. That is the distance from your part line to one edge of the section, and two-thirds to three-quarters of it is how far this page says to cut inside of. If your part is not in the middle, move that one figure: the edge you lengthen is the edge you shorten, and the sum stays where the ratio put it.
If the decision is still between the two cuts, the crossing is the cheaper test. Fifty-five millimetres on an edge is 77.0 per cent of the widest published edge, past the top of this page's own band yet still inside the strip, and 101.1 per cent of the narrowest one, which is more hair than that strip holds. Marking 55 mm on each side of your own part, once, tells you which side of it you are on, and that is the only thing this page asks you to do with a ruler.
Curtain bangs and face-framing layers are the same piece of hair treated two ways, and the treatment is fixed before the scissors move, by the width of the section it is cut from.
That width is not nothing. It is a ratio with published centres, a published spread and two ends of the earth to map between, and multiplying it by a measurement you can take yourself gives millimetres, which is more than any of the twelve pages whose text reached anyone gave.
Everything above is one multiplication, one halving and one subtraction, done on lines that are on the record. Where the page had to choose, it said so and printed what the choice costs.
Questions this page answers
How wide should the section at the front of the head be?
Between 108.8 mm (4.28 in) and 142.8 mm (5.62 in) across, on the published centres, measured against a cheekbone width of 136 mm — your own width moves every figure with it. Divide your forehead width by your cheekbone width, multiply the quotient by your cheekbone measurement, and that product is the section. Curtain bangs and face-framing layers both come off it.
Does the number change when curtain bangs vs blunt bangs is the actual decision?
It changes by how much of the section is taken. At the widest published centre a blunt fringe uses the whole 142.8 mm (5.62 in) across, while the band on this page stops at 47.6 to 53.6 mm (1.87 to 2.11 in) on each edge. The hair a blunt fringe keeps off the sides is 17.9 mm (0.70 in) per edge at that centre, or 35.7 mm across.
Is a face-framing layer the same thing as a curtain?
Both start as the section divided off along the hairline. A curtain is that section parted in the middle and opened to both sides, shortest at the part. A framing layer is the same section cut short in the middle and left long at the edges, so the length sits on the two sides rather than between them. The width you divide off is the figure both cuts spend.
Why does no hair page give the section width in millimetres?
Because the instruction works as a style sentence. A page can say the layer starts at the cheekbone and be understood, and the same words cover 54.4 to 71.4 mm depending on whose forehead is being cut. The number is only needed by someone doing the arithmetic, and the twelve readable pages in the record above were not written for that person.