Best Neckline for Round Face: The Two Numbers You Already Have

Neckline Shapes for Face Shapes: The Two Numbers Behind the Words

Everything on this page starts from one measurement: cheekbone width, taken corner to corner across the widest part of the face with the tape held level. That is the same landmark the measurement page describes for the two ratios below, so if you have done it once for the ratio quiz you have done it for this.

For a round face this site's rule set puts the forehead at 0.88 of that width and the jaw at 0.85. Those are centres, not your measurements — they are the values that generate the ratio bands printed across the site, and the table below is those two numbers multiplied out against the widths this site works with. Each cell is one multiplication and one division, so any row can be redone on the spot with your own width in place of the printed ones.

Centres are 0.88088 (forehead ÷ cheekbone) and 0.84923 (jaw ÷ cheekbone) for a round face, on the soft-tape scale; every millimetre cell is that centre multiplied by the row's cheekbone width and rounded to one decimal, and each inch cell is that same unrounded product divided by 25.4 and rounded separately — so an inch can land one hundredth away from dividing the millimetres printed beside it. Those two centres are the same constants that generate the seven-shape ratio bands published elsewhere on this site — this page reproduces all twenty-one of those bands cell for cell before using any of them. The last column is one standard deviation of the jaw ratio inside the measured round-face group in this site's own sample (0.120 on this scale, ten faces), so it is a spread inside one shape, not a tolerance on your tape. 128.2 and 148.8 mm are the ends of the adult cheekbone band this site works from and 136 mm is its working line; the dataset behind that band is named where this site publishes that band and is not reprinted here.

What the Guides Already Own

Start with the part that is not a gap, because it would be easy to overstate one. Choosing an opening by face shape is written about heavily and at length. Every one of the eleven articles read for this page uses the word for an opening a dozen times or more, running from eighteen to fifty-six uses in a single article. The shape names get addressed individually: one styling site works through six of the seven outlines this site counts, round and oval and square and heart and oblong and diamond in order, and gives thirty-three uses of the phrase for the face itself. A 2026 retail guide carries a whole section titled by shape, and another publishes a do-not-do list for a round face specifically. The mainstream instruction is one sentence long and appears almost verbatim across the set: choose an opening that disagrees with your outline, so a round face wants a V.

Two of those pages go further than the rest, and both are worth naming because they are close to this page's argument. A dress retailer's 2026 guide states that the second mistake people make is ignoring the width of the opening — it puts that dimension in words, and then never gives it a number, and never asks you for one. An outfit blog from 2014 argues that an opening relates to the shape of your jaw rather than the shape of your body, and explains the familiar advice that a round body wants a round opening as an illusion caused by round bodies tending to have round jaws. That is the direction the jaw ratio in the table above points in. Its own ruler, though, is a mirror and a comparison of shapes; across three thousand words it never prints a millimetre.

Here is the part that is a gap. In those eleven readable articles, taken together this month, the count of size units attached to anything is zero. No millimetres, no centimetres, no inches — not for an opening, not for a drop, not for a width across the collar. Two of the eleven carry a table at all: one is a guide that lists opening styles against face shapes and prints no measurement in any cell, the other's single table is an address block from a page template. One article defines each style by its corner geometry — a straight horizontal run, a right angle — and gets all the precision it wants out of angle words without ever stating a width.

So the gap is not that nobody connects an opening to a face. It is that the connection is made entirely in adjectives, and nothing in that set lets you bring your own measurement to it. The one instruction closest to a measuring rule that this page found anywhere, in any year, is not in a styling article at all: it is a tailoring manual's method for cutting the neck of a waistcoat, and it derives the opening from your neck.

A Neckline Chart by Face Shape, All Seven at One Width

The taper column in the first table, cheekbone width minus jaw width, is what the shape words are circling when they call a face round instead of angular. Run the same multiplication through all seven outlines and it separates them in millimetres, at the 136 mm working line.

Read it as a scale, not as a prescription. The taper for a round face at that line is 20.5 mm, which is 15.1% of the width it came from; heart and diamond reach 38.1 mm, oval 26.8 mm, oblong 10.5 mm, square 6.7 mm, and triangle comes out negative because its jaw is wider than its cheekbones. The reason this site can print those differences and still refuse to publish a per-shape instruction is the size of the spread. At the same cheekbone width the jaw gap between round and square is 13.8 mm, which is 0.85 of one standard deviation inside a single shape; the forehead gap between them is 9.5 mm, or 0.86 of the spread on that column. Both differences are smaller than the variation you would find between two round faces. A number keyed to a shape label would be reading this sample as a rule, and it is not one.

Each row is the published centre for that outline multiplied by 136 mm, then subtracted from 136 mm; the last numerical column is that jaw width tripled, which is the step this page adds in the next section and is labelled there. Oval, oblong, round and square rest on measured faces (18, 12, 10 and 6). Heart, diamond and triangle do not: their centres are interpolated along a published ordering because the sample holds one, three and zero faces, so treat those three rows as positions on a scale, not as results. A round face's own published jaw band, 0.73 to 0.97, spans 99.3 to 131.9 mm at this width — 32.6 mm, or 1.29 in, of real variation inside the single label this table calls one row.

The Wide vs Narrow Neckline Face Shape Question

Width is the dimension that gets named and not numbered, and it is the harder one for a simple reason: an opening is cut on a neck. Nothing about your face determines how wide a garment's neck hole is, any more than it determines a sleeve. What your face does supply is a set of widths to compare an opening against once you know how the opening is built — and that comparison has been made in millimetres for a very long time, just not against a face.

The oldest instruction in this set is a trade manual for waistcoat cutters published in London in 1902, and it is the only document here that prints a size rule for an opening at all. Its method is procedural: pass the tape round the neck, bring the ends together, and work from half that measure. Cut each front run of the opening to one sixth of the neck measure, then run the tape back over the finished line and check it against the halved neck. It finishes the top with a little spring at the shoulder seam — about an inch, which is 25.4 mm — because a flat front run would ride up at the throat.

That gives three quantities you can put numbers on, and one identity worth noticing. Written as arithmetic: half the neck is N ÷ 2, each front run is N ÷ 6, so the two front runs together come to N ÷ 3 — and the shortfall between that pair and the halved measure is exactly one front run, since a half minus a third is a sixth. That identity holds in every row of the table below, which is how you can check the table without taking it on trust. The doubling to two runs and the subtraction are this page's own steps, not the manual's.

The rule is quoted as a procedure: the tape goes round the neck and the measure is halved, each front run is set at one sixth of the neck and checked against that halved measure, and a spring of about an inch at the shoulder seam is added on. The sixth, the halving and the inch are the manual's own numbers; the fourth column (two runs together, a third of the neck) and the fifth (their shortfall against the halved measure) are this page's arithmetic, and the fifth column is simply the sixth again, which is the check. The inch figures are the millimetre cells divided by 25.4. The neck measures in the first column are a plain 20 mm ladder chosen to straddle a range of adult necks; they are not typical values and nothing here depends on which row is yours — the manual works from whatever the tape says, and so does this table. The 25.4 mm spring appears once, in that one document, and is reported as a single source rather than as a convention.

Where Your Jaw Width Enters the Opening

Now bring the two halves together, which is the step nobody in this set takes. Take your jaw width from the first table — 115.5 mm at the 136 mm working line — and triple it. That gives 346.5 mm, and 346.5 mm is the neck measure at which the two front runs of an opening, built by the 1902 rule, come out exactly as wide as your jaw. Below that neck measure the pair of points lands inside your jaw corners; above it, it runs past them. At 340 mm of neck the two runs total 113.3 mm, which is 2.2 mm inside 115.5 mm, close enough that the crossover is a real thing to check and not a curiosity.

The ratio is the useful part because it does not depend on which row you are reading. Three times the jaw centre is 2.55 times the cheekbone width, at every width and for every outline: 326.6 mm of neck at the narrow end of the cheekbone band, 346.5 mm at the working line, 379.1 mm at the wide end. Put your own two numbers in and the comparison is one multiplication and one division.

How much to trust it is the other half of the same arithmetic, and it deserves millimetres instead of adjectives. One standard deviation of the jaw ratio inside this site's own round-face sample is 16.3 mm at the working line, so the crossover moves by 49.0 mm for that much spread — the crossover is a band roughly 297 to 395 mm of neck measure, not a line. Reading the cheekbone tape badly matters less than the spread does. A 2 mm error on that width shifts the implied jaw by 1.7 mm (0.07 in), 5 mm shifts it by 4.2 mm (0.17 in) and the crossover by 12.7 mm, and 10 mm shifts it by 8.5 mm (0.33 in), which is 0.52 of one spread unit. The band is wide because the sample is small and because a soft tape on a face is not a caliper, and both of those are said here, not smoothed over.

One scale trap needs its own paragraph because it produces the same 8.5 mm as a bad reading and looks like proof of a difference that is not there. The ratios quoted on this page are on the soft-tape scale. The detector's own raw quotient for a round face's jaw is 0.787, not 0.849; use it as though it were the tape ratio and at 136 mm you get 107.0 mm instead of 115.5 mm. That is an 8.5 mm error made by a calculator, not by a hand, and the two are easy to confuse.

What this comparison does not claim: it is a statement about where two cut points fall relative to two corners of your face, and it is checkable. It is not a claim about how your face reads, and no row here should be quoted as though a wider opening would make a face look longer or narrower. That question belongs to the next section, which is where this page runs out of numbers.

High vs Low Neckline Face Length: Why This Page Stops Short

Depth is where the advice is densest and the measurements are thinnest, and this site cannot derive it from your face. The five readings published here are face length, forehead width, jaw width, jaw angle and chin width. Every one of them is taken on a face. Not one is taken on a neck, and a neck's length is not a function of the widths above it — it is a body measurement, and it behaves like one.

The 1902 manual is the evidence for that, because it treats the vertical exactly as it treats the horizontal: as something you measure rather than infer. Alongside the halved neck for the opening, it takes a separate length round the neck and down to the bottom wanted, and another from the nape to the depth of the armhole. The vertical is on the body, in the body's units, and the book never once asks what shape the wearer's face is — a manual that uses the word for a neck seventy-five times and the word for an inch forty-three times, and has no occurrence of the word for a face at all.

What a face measurement can legitimately do is keep you honest about a proportion you already have. This site publishes a length-to-cheekbone ratio for every outline, with its own band and its own spread, and if you know your face length and your cheekbone width you can work out where a garment's lower edge falls relative to your own proportions. That is a comparison, not a prescription: nothing in this page converts a ratio into a depth, and the articles that recommend a depth for a round face are not doing arithmetic either.

If you want a depth number that is actually yours, measure the thing you are asking about. Sit down, put the tape at the nape, and let it run to the point on your chest where an opening would end — the same way the drafting manual instructs — and record the number where it stops feeling like nothing and starts feeling like a costume. That number is a measurement of you. The one in the table above is a measurement of a sample, and the two should never be confused for each other.

What Changed Between 1902 and August 2026

Ten dated entries, sorted by the day each one was published. Two rows carry a date that appears only inside the page's own machine-readable fields rather than printed on it, and that is said in the row. Two pages from the same list served a browser-verification wall to both attempts made this month and are absent for that reason — a limit on this page's reading, not evidence about them.

The point of running them in date order is what stays still. The face-shape vocabulary grew: the 1902 rule names no shape at all, a 2025 guide works through six of the seven outlines this site counts in a single run, and by 2026 a retail guide titles a whole section by shape. The enumeration got denser, and the advice got shorter — at least a hundred and twenty-three years apart, the instruction is still essentially pick the opening that disagrees with your outline. What did not happen in any of those ten rows is the one this page is about: not one printed a millimetre, centimetre or inch for an opening. The only size numbers anywhere in the span are 1902's sixth-of-a-neck and its inch of spring, and both are taken off a neck.

Read the 1902 imprint as late as the year allows and the span from it to the newest row is still 123 years and 7 months; the nine rows published on the web span 12 years and 7 months, and the two newest of those are 7.6 months apart. Dates were taken from each page's own statement — a printed line, its metadata or its structured data — and the three columns above describe what was read; they are not quotations. Eleven readable write-ups were gathered for this comparison and nine are dated here. The two that are not: a first-person piece on a discussion site, which one pass this month read in full but never located a day on, and one page that came back in a second pass as a shell with thirty-four words of visible text. Two further pages in the same list served a browser-verification wall to every attempt, so nothing about their contents is claimed here — including whether they print a number.

How Far to Trust Each Row

Three things in this page's arithmetic are this site's own, and they are named so you can discount them. The first is the step that doubles a front run into a pair and triples a jaw width into a crossover neck measure: the 1902 rule only ever states a sixth, and the comparison to a jaw width is this page's idea, not a convention anyone else uses. The second is the 20 mm ladder of neck measures, which is a convenient spacing and not a set of typical values. The third is the choice to compare openings against jaw width rather than cheekbone width, which follows the taper column but is a choice.

Two more sit behind those centres, and both push toward less confidence, not more. Sample size: the standard deviations here come from the four groups with five or more faces — eighteen oval, twelve oblong, ten round, six square — while the heart, diamond and triangle centres in the third table are interpolated along a published ordering, not measured. Then a column this page deliberately does not use as a dividing line. The jaw angle differs by 24 degrees between a round and square face in the ordering the rule set follows, but on this site's own fifty photographs those two groups average to the same angle, around 133 degrees, and cannot be separated by measuring them. An instruction keyed to an angle would therefore be keyed to something this site's data cannot resolve, which is why nothing on this page turns one.

And one caution about the tables themselves: they are built to be redone, not to be trusted. Every cell is a multiplication by a published centre or a division by 25.4. The moment you have your own cheekbone width and your own neck measure in millimetres, the printed rows are redundant — the crossover is 2.55 times your width, the pair of front runs is a third of your neck, and you can stop reading here.

• Measure cheekbone width, level tape, corner to corner. Read your jaw width off the first table, or better, measure the jaw the same way and use your own number instead of the 0.85 row.
• Measure round the base of the neck without pulling the tape tight, the way the drafting rule asks. Halve it; that is the number an opening is actually built against.
• Triple your jaw width. If your neck measure comes in below that, the two front runs of an opening fall inside your jaw corners; if above, they run past them.
• Want a depth? Measure it on your own body, from the nape down to where you want the edge, sitting as you normally sit. Your face has no number for it, and neither does this page.
• Carry the spread with the result. At the working line one spread unit of jaw width is 16.3 mm, so a crossover of 346.5 mm is honestly 297 to 395 mm — and if you are outside the 128.2 to 148.8 mm cheekbone band, use your own width, because all of this is arithmetic and none of it is a rule about you.