Contour for a Square Face: The Four Corners, and the One Corner Worth Keeping

Contouring a square face starts with a decision that six guides never make

Start from what this outline has. Among the seven face shapes this site classifies, it is the only one whose defining feature is a corner you can put a finger on — the rule set places the square jaw angle at 127.9 degrees, the sharpest of the seven, published on the square face guide and the rule-set page and read at build time from the same constants the detector runs against. Every other shape's advice treats its defining feature as something to reduce: length on the oblong, width on the round, the point on the heart. On this face the record splits, and the split is the first thing to settle, because the placement changes depending on which side of it you are on.

One published source names the fork. An independent artist's guide from April 2020 gives the jaw two treatments and says plainly that they are alternatives: to de-emphasize the jaw, trade its highlight for contour shade; to keep the square readable, leave a soft highlight on the back of the jawline instead. Both arms, one page, the reader's choice. The other five dated positions do not offer the choice. A 2025 luxury-house guide says the aim on this face is not to sculpt but to soften squared-off features. A 2026 analysis site says to round the corners and manufacture length. Another 2026 guide opens with a blurring primer whose stated job is to diffuse the angles. A 2019 brand guide wants substantial contour on the outer jaw edges to soften the angles and pull attention to the lips. Only a mass-brand guide from July 2026 comes close to naming the fork from the other side — soften the edges while keeping the strong structure — and still does not say that keeping is an option.

The first-attempt write-ups take the keeping arm, and their reasoning is economic rather than aesthetic. On a face whose only surplus is structure, erasing the corner leaves nothing to replace it, and the write-ups that describe contour making a face look dirty, sunken or older are overwhelmingly the softening arm carried out at full strength. Keeping has a technique of its own, and the technique is not about the corner's existence but about its edges: what makes a corner read harsh is fuzz, shadow that stops dead at the angle, and a jawline that ends where the bone ends instead of running back. One person demonstrated the difference on their own face, one side finished and one side not: the side whose jawline stroke was swept back behind the ear read as a clean edge, and the side that stopped at the angle read as a stripe.

What neither arm can do is move the corner itself. The 127.9 degrees is bone, and no quantity of pigment relocates a bone — a point the round-face version of this page makes at length and this one will not repeat. What the two arms actually change is where the eye guesses the corner sits, how sharp it reads, and whether it is the loudest thing on the face. That is a real effect and a bounded one, and the rest of this page is about producing it deliberately instead of by accident.

The four corners, and the centre column between them

The placement itself is the most settled thing in seven years of material. Two corners sit high: the points where the hairline turns at the temples, at the outer flanks of the forehead. Two sit low: the angles of the jaw. The 2026 analysis site simply calls them the four corners; the 2025 house guide runs its wand diagonally across all four, naming the temples and the squarest points of the jaw; the July 2026 brand guide lists the sides of the forehead and the corners of the jawline; the 2019 brand guide shades the outer flanks of the forehead from the hairline down and the outer edges of the jaw; the 2020 artist shades the outer edges only and bans the centre of the forehead and the chin from receiving any shadow at all. Six sources, four points, no contact between them. If everything else on this page were lost, the four corners would survive.

The high corners are a gradient, not a band. The 2019 guide's stated reason for shading the outer forehead is that square foreheads are broad, and the shading both narrows the read and fades downward so the hairline stops looking like an edge. Its companion move is the centre column: a slim highlight down the middle of the forehead and the nose, kept deliberately narrow because a wide one widens the very feature the corners are working on. The 2020 guide reaches the same column from the other end — a triangle-ish highlight at the centre of the forehead, which lifts the eye upward and off the jaw — and adds the chin point for length, while keeping shadow away from both. The 2026 analysis site's version is a matte highlight down the mid forehead, nose bridge and chin, described as a vertical line of light that elongates and pulls attention off the angular sides.

The low corners are worked in arcs. The oldest published row shades along the outer jaw edges; the 2025 guide blends in soft circles; the July 2026 guide traces the jaw corners and blends toward the earlobe; and the oldest technique in the practice record — written for this face in 2012 — has the brush sweep an arc across the angular parts on purpose, to fake the curve of an oval outline. Four wordings, one motion: the brush never travels along the straight run of the jaw the way a line is drawn, because a straight dark line beside a straight bone doubles the thing it was meant to soften. Section six takes the jaw stroke in full, including where it ends and what happens to the neck.

One warning that belongs at the placement stage rather than the blending stage. A 2017 account in the practice record singles out the standard jaw-angle method — shading hard at the corner — for a specific side effect: done with depth, it makes the cheekbones read more prominent, because dark immediately below a bone pushes that bone forward. The account treats this as a look that suits a different aesthetic rather than a mistake, and offers the alternative used throughout this page: small continuous strokes sweeping from the outer edge inward, darker nearer the jawline, travelling in one direction only and never back and forth. The cheekbone consequence is section five's subject; the one-direction rule is worth stating here, at the moment the brush first touches the face.

Seven-plus years of square-face placement, side by side

Six published guides for this face shape, spanning April 2019 to July 2026, are worth setting against each other for the same reason the round-face six were: they agree where you would not predict and disagree where you would. The span runs from seven years and three months to one day short of seven years and four, and the reason it cannot be narrowed is that the oldest row prints only a month — which day of April 2019 the page went up, the page does not say.

Dates: five of the six were read off the rendered page; the July 2026 brand row carries its date only in machine-readable form (published 30 July 2026, modified 1 September 2026), and the April 2019 row prints no day at all — its month and year appear once, in the footer tag list, which makes the oldest end of the span a footer tag rather than a dateline. Everything between them prints a full date in the body. Checked and excluded rather than padded into rows: a July 2024 artist guide that defines the square face and gives a tape-measure method for finding it but contains no square-face placement; a mineral-cosmetics guide with placement and no date anywhere on the page; a mass-brand magazine page taking the opposite stance — enhance the chiselled features rather than soften them — also with no date; an education platform whose only face-shape contour article covers the round face and not this one; and a recipe-style platform whose page could not be retrieved at all. No links are given to any source named in the table.

What did not move, and what did

Two things held across all six rows and the seven-plus years. The four corners are the first: every guide shades the temple-hairline turns and the jaw angles, and not one adds a fifth point or drops one. The centre column is the second, held by four rows and partly kept by a fifth — the July 2026 brand row retains only its nose-bridge segment — with the 2025 house guide the single exception, all perimeter and no light line. Where the column appears in full, its job is always the same: a vertical strip of light down the middle that gives the eye somewhere to go besides the corners.

The cheek stroke is where the movement is, and it moved in a direction. Into the hollow of the cheek in alignment with the ear, says the oldest row, April 2019. Angled toward the lip line, stopping at the iris, says April 2020. A C-shape around the top of the bone, says May 2025. Under the bone and down toward the jaw, blended inward, says April 2026. Under the bone, stopping halfway from ear to mouth, says July 2026. Under the bone, blended upward, says the July 2026 brand. Read in order, the hollow is losing: the oldest guide is now the only one putting product into it, and the three most recent all sit the stroke under or around the bone with the tail of the stroke pointing up. The practice record reached the same verdict years earlier and for a reason none of the six states — on this outline the face is already angular, so shading the hollow does not create structure, it subtracts volume from a face that has none to spare and reads back as sunken and older. The published drift and the first-attempt record are moving the same way without contact, which is as close to a settled answer as this subject produces.

Shade depth almost held, in the same pattern the round-face table showed. Four of the six commit to nothing. Of the two that do, one says one to one-and-a-half shades deeper and the other says one to two, cool-toned — and one-to-one-and-a-half sits inside one-to-two, so the two figures are closer than they look. The 2019 row also carries the only dose rule in the table: build the colour by adding product to the brush rather than to the face, which is the published version of the practice record's small amounts, many passes.

The word that moved underneath all of them is the aim. Soften appears in four rows; keep the structure appears in two; and only the April 2020 row holds both at once, as the fork in section one. There is also a warning worth keeping from the April 2026 analysis site, because it is the one sentence in the six that speaks to the other page on this subject: what sculpts a round face beautifully can make a square face look muddy. The placements are not interchangeable between the two outlines, and the round-face comparison lives on its own page.

The cheek stroke: above the hollow, not in it

The failures in the practice record almost all start the same way: a placement copied from someone with different bones. The write-ups that describe what went wrong converge on the same repair, and the repair is a landmark rather than a rule. Put the index finger on the highest point of the cheekbone — the bone itself, found by touch — and sweep inward from there with the contour brush, stopping below the middle of the eyeball. One write-up reached that landmark after a period it describes as going from breakdown to enlightenment, having first made the two standard errors: product laid on heavy and never blended, and product laid where a different face's bone put it.

Why the finger and not the mirror. Sucking in the cheeks finds a hollow that moves with the expression and is not where the bone is; the finger finds the bone, which does not move. And why above the hollow on this outline specifically: the square face's problem is never a lack of angles, so a stroke into the hollow adds a shadow the face was not missing and takes volume the face was not spare of. A first-person account from 2020 — a person who names this outline and harsh features as their own — puts the whole subject in one line: contour does not need depth, but the position has to be right. Their highlight goes slightly above the apple of the cheek, which pushes the mid-face forward and does for the front of the face what the corner strokes do for its edges.

The sweep itself has three rules, and each one appears in both records independently. Direction: from the outer edge inward, darker nearer the outer contour, the front of the face barely touched — and one direction only, never back and forth, because a reversing brush redeposits what it just laid. Stop: below the middle of the eyeball, says the practice record; the outside edge of the iris, says April 2020; halfway from ear to mouth, says the July 2026 guide. The three stops are not three points on one path — a stroke angled downward and a stroke running under the bone end in different places — but they agree on the two things that matter: none of them reaches the mouth, and all of them keep the stroke's end in the back half of the cheek. Dose: small amounts, many passes, built from the outside inward — which is the practice record's wording for the 2019 guide's add-product-to-the-brush rule. The failure the dose rule prevents has a vivid description attached to it in one write-up: product piled unblended into the pores reads as stubble rather than as makeup.

One exception, and it is the square-face exception. The part of the cheekbone that faces the camera straight on cannot take contour powder at all — on this face the powder there reads as dirt rather than as shadow, which is the complaint behind the search phrase this page was written for. The workaround in the record is a change of product rather than of position: an orange-toned eyeshadow over the front-facing high point, because orange reads as warmth where grey-brown reads as grime. The same account extends the logic outward: shading the cheekbone along its run toward the outer and lower rim of the eye socket makes the eye area read larger, which is the frame-and-features trade in section ten, arriving early.

The brush discipline that sits underneath all of it costs ten seconds and prevents most of the dose failures. Check the brush is clean. Wipe off excess against a tissue. Pick the product up. Even it out on the back of the hand before it goes anywhere near the face. The last step is the one that gets skipped and the one that matters: a brush loaded in one spot lays that spot down in one pass, and the pass that cannot be blended out is the pass that was too heavy before it touched the skin. The same account's companion rule for the nose — shading it in three zones as a surface, never as one vertical line — is the general form: contour is an area with a gradient in it, not a stroke with a position.

The jaw stroke: one surface, not a line

The named failure of the jaw stroke is that it reads as makeup from anyone else's viewing angle. A few heavy strokes along the jawline, on the line and nowhere else, look like a line drawn on the face — the write-up that describes this had seen the result on other people and then produced it themselves. The repair is to stop treating the jawline as a line. The shadow extends off the bone and onto the neck as one continuous surface, so what the eye reads from across a table is not a stripe but the face's own underside continuing into shadow. The same account adds the maintenance that comes with owning the surface: the neck wears makeup now, and the neck gets cleaned now.

Position and feathering come from the same record. The shading sits on the bone rather than above it, and it feathers both up and down, which closes the colour step between face and neck instead of adding a second step beside the first. The July 2026 guide gives the same ending from the other direction — trace the jaw corners and blend toward the earlobe — five years after the practice record's behind-the-ear demonstration and fourteen years after its arc. Where the stroke finishes is the tell of which arm of section one's fork you are on. Stopped at the angle, it is a softening stroke with the corner as its target. Swept back behind the ear, it is a keeping stroke — the corner stays and the edge gets clean, which is the half-face demonstration's whole finding: same face, same product, one side swept behind the ear and one side stopped at the angle, and the stopped side is the one that reads heavy.

There is a depth ceiling on this stroke and the record states it twice. The 2017 account's warning — shading hard at the jaw angle makes the cheekbone above it read more prominent — is the structural reason; the product-quality write-up's observation that cameras eat makeup while daylight does not is the practical one. A jaw stroke that photographs as definition can read at arm's length, in a window's light, as dirt. The check is therefore not a mirror check and not a photo check: it is a glance from the distance other people stand at, taken before leaving the house, which is two seconds of the routine and catches both the cheek and the jaw failures at once.

What the jaw stroke buys, stated once by the published record and never again by anyone: the 2019 guide claims the substantial contour along the outer jaw edges draws attention to the lips. It is the only row that offers a payoff beyond narrowing, and nothing in the practice record confirms or contradicts it. Treat it as the one open claim in the table — everything else in this section is corroborated from two directions.

The base underneath decides whether any of this reads as bone or as dirt

Two finishes were tested against each other on one face in the practice record, and the result was decisive enough to be worth stating before any placement: a luminous, dewy base makes square bone more prominent, not less, and turns every furrow — the tear trough, the fold beside the nose — into a shaded groove, because a glossy surface announces its own topography. A fully matte, powdered base does the reverse: it flattens the furrows and lowers the bone's presence. The published record lands in the same place without ever running the comparison: both 2026 guides keep the light at the centre and nothing glossy where the corners are — one specifies a matte highlighter for its centre column, the other opens with a blurring primer around jaw and forehead. Matte at the edges, controlled light in the middle — on this outline the finish is doing structural work before a single contour stroke exists.

Colour comes before finish, and the rule is counterintuitive enough to have its own failure write-up: the lighter the face against the neck, the larger it reads. A base chosen a shade brighter than the neck separates the face from the body as its own surface, and a separate surface is a bigger surface — the contour that then goes on top is narrowing a face the base just widened. The repair in the record is to match the neck rather than the wrist, mixing two foundations when no single one lands: in the worked example, one formula too light and too dry and another too deep and too moist, mixed until the result disappears into the neck. Only on top of that match does one-to-two shades deeper mean anything, because the contour depth is measured from the base, not from the bare skin.

Order is the oldest settled thing on this page. The 2012 technique gives it as two rules and nothing in the fourteen years since has contradicted either: cream or liquid highlight goes on after foundation and before setting powder, where it can still be pressed in and melded; shading powder goes on after the setting powder, on a dry, set surface, where it moves until it is blended and then stops. The same account's tools are still the cheap ones — a damp sponge to press and blend, a damp brush that behaves like a cream product and gives control back to a heavy hand.

The product failures in the record are more specific than the technique failures, and they cluster on one format. The dual-ended stick bought on impulse: its highlight end was sparkly and near-white, showing every pore it passed, and its shade end laid down too heavy to blend — discarded after one use, with the conclusion that when the powder is bad, no amount of technique rescues it. The same write-up's verdict on the format in general is that a stick highlight reads like white paint in any light short of a camera's, which is why the luminous finishes the published record recommends keep getting discarded in the practice record: the two are describing different lights, and one of them is describing a photographed face. For a beginner the record's recommendation is a liquid contour instead of a stick or a loose powder — it adheres to the base, shows its colour without needing depth, and fails soft.

Every number on this page, in millimetres and inches

This page's measurable content is small, and the smallness is honest: placement on this face is mostly landmarks — the finger on the cheekbone's highest point, the iris line, the vertical dropped from the outer corner of the eye — and landmarks do not convert. What does convert is the three-court division and its two stopping rules, which come from the oldest technique in the record and are the only distance figures on the page.

The method. Measure the face's length in three courts, hair pulled back: from the hairline — the real one, not the styled one — to where the brow starts; from there to the base of the nose; from there to the tip of the chin. An even split means each court is the total length divided by three. The long court decides which highlight gets shortened: if the middle court runs long, the nose highlight stops 5 mm short of the tip; if the lower court runs long, the chin highlight stops 5 mm short of the chin's point. The stop exists so the light line ends before the feature it would otherwise elongate further. In inches the stop is 5 ÷ 25.4 = 0.19685, so 0.20 in — every conversion here is exact division by 25.4, since the international inch has been defined as exactly 25.4 mm since 1959. Worked through at one length: a 120 mm face gives courts of 120 ÷ 3 = 40 mm, and 40 ÷ 25.4 = 1.5748, so each court is 1.57 in and the stop is 0.20 in of that.

The upper court's rule is a fraction rather than a distance, so it needs no conversion: when the upper court runs long, the forehead highlight occupies the lower two thirds of the forehead's height and leaves the top third empty, which shortens the court by declining to light the part of it that is already too long. The angle figures on this page — 127.9 degrees at the jaw corner, the measured band in section nine — are degrees in every unit system and convert to nothing.

Every figure is exact division by 25.4 and then rounded to two decimals — 105 ÷ 25.4 = 4.1339, so 4.13; 35 ÷ 25.4 = 1.3780, so 1.38 — so any cell can be re-derived with one division. The stop column repeats because it is a constant: 5 mm ÷ 25.4 = 0.19685, so 0.20 in, whatever the face length. A court is the classical three-part vertical division of the face, measured as described in the text; the even split is the reference case, and the point of measuring is to find which of your three courts departs from it, because that is the court whose highlight gets the shortened stop.

Put the whole placement on paper before the first attempt. The sheet below carries the keep-or-soften decision with both arms' steps, the four-corner map with every stopping rule, the cheek and jaw strokes with their landmarks and direction arrows, the base and layer order, this three-court table with the arithmetic beside it, the failure checklist, and a log to fill in on the day so the third attempt can be compared against the first: Download the square-face contour placement chart (printable) ↓

The jaw-angle claim that does not survive checking

There is a figure about this face that circulates in beauty teaching without ever being attributed to a measurement: that a woman's jaw angle runs between 120 and 140 degrees and a man's between 100 and 120 — a twenty-degree gap, with the square jaw framed as the male end of it and every square-jawed woman cast as an outlier from her own band. The claim matters to this page because it is a premise of the softening arm: if the corner is masculine bone on a feminine face, softening stops being a preference and becomes a correction. So it was checked against the measured literature, and it does not survive.

A peer-reviewed imaging study published in June 2025 measured the gonial angle — the interior angle at the jaw's corner, the same angle this site's rule set places at 127.9 degrees for this outline — in 141 adults, 55.3 per cent of them women. Right side: women 128.66 degrees with a standard deviation of 5.50, men 125.68 with 5.10. Left side: women 128.84 with 5.97, men 125.26 with 4.89. The difference is significant at p < 0.005, and its direction is the opposite of the circulating claim: on this measurement women's corners are the more obtuse ones, by about three degrees. The claim's male band fares worse than its female one. Men's angles in the study span, at one standard deviation either side of the mean, 120.6 to 130.8 degrees — so the claimed male range of 100 to 120 contains essentially none of the measured men, and its sharpest edge, 100 degrees, sits more than twenty degrees below anything the band describes.

The honest limits, because one study is one study. The literature disagrees about the direction itself: some measurements find men more obtuse, and a 2024 cohort of adults aged 61 to 70 reports women at 128.43 against men at 124.05 — the same direction as the 2025 study, in a different age band. Populations, imaging methods and age ranges all vary between studies, and the angle changes with age. What no measurement in the checked literature supports is a twenty-degree gap between the sexes, in either direction. The real figure is a difference of a few degrees around shared means.

The consequence for this page is narrower than it looks and worth stating exactly. This site's square centre of 127.9 degrees sits inside the women's band from that study — 123.16 to 134.16 degrees at one standard deviation — and 0.76 degrees from its mean. A corner at this outline's angle on a woman's face is not a masculine feature; it is within a degree of the measured female average. No placement above changes as a result: the corner is bone, it does not move, and section one's fork is a preference either way. What changes is the premise one arm of the fork is usually sold on. Softening the corner is a legitimate choice. Correcting a male bone that happens to be on a female face is not a thing the measurements describe.

Shrink the frame or enlarge the features — the trade nobody states

A face reads as a frame with contents, and there are two ways to make the frame read smaller: shrink the frame, which is what contour does, or enlarge the contents, which is what everything else on the face does. The practice record's most systematic account of this outline is organised entirely around the second move — its working principle, stated once and then applied to every feature in turn, is that when the features get bigger the face gets smaller — and the two moves compound, because a frame is judged against what it contains. None of the six published guides mentions the trade. Two of them, both from 2026, assign blush a supporting role beside the contour; none asks what the blush is doing to the face's apparent size on its own.

The largest single instrument is blush, because it is the biggest area of colour on the face. The record's rules for it: sweep toward the mid-face, never crossing the vertical line dropped from the outer corner of the eye — past that line the colour adds width, which is the one thing this page's contour is removing; and connect the blush to the eye colour with no gap between them, so the two read as one area rather than two small ones. The 2012 technique adds the transition: blush placed in front of the contour shading, worked in upward circles, which is also what stops the shading edge from reading as an edge. The July 2026 guide draws the same boundary in terms — never past the outer eye corner, because the width it adds is width the face cannot spare — and the April 2026 guide's avoid-too-wide is the same boundary stated loosely, which makes this the one blush rule both records keep.

The mid-face moves are highlighting moves and they belong to the contour routine rather than beside it. Matte light into the hollow at the centre of the mid-face — the dip under the eye and along the cheek pad's upper edge — makes the pad read smaller, which makes the frame read smaller; the fold beside the nose takes its highlight in short up-and-down zigzags rather than one stroke, because the fold runs in a hook and a straight line of light beside a curved line of shadow announces the curve. The nose's shading starts from the hollow of the eye socket, not from the head of the brow, and keeps a gap between its root shading and the brow itself. The lips: this outline carries a larger mouth better than most, widened vertically rather than horizontally, with the lip line deliberately softened — a crisp drawn line adds visible pale space around the mouth, and pale space is frame.

One line about the hair, because it undoes the whole routine from outside. Hair worn down beside the face contrasts against the skin and re-draws the outline the corners were shaded to quieten; hair tied back removes the second frame. The account that makes this observation treats the tied-back version as the standing recommendation and keeps the rest of the face work unchanged.

How to practise this without wasting a face

The placement cannot be learned from a diagram, including this one, and the two failure write-ups in the record agree on why: the diagram's stopping points were found on someone else's bone. What can be learned is the search procedure, and the record's version of it is slow on purpose. Work one side of the face at a time — finish one side completely, look at the other side raw, then match the finished side to it rather than to the diagram. Both of the record's best findings came out of exactly this asymmetry: the behind-the-ear jaw sweep was demonstrated one side against the other, and the matte-versus-dewy base result was the same kind of comparison.

The checks are ordered by how often they lie. The morning mirror lies most: it is the instrument with the least independent information, operated at the hour with the least light. Photographs lie next, in the specific direction the record names — cameras eat makeup, so a result that reads correct in a photo is often a result that is too heavy in daylight. The check that lies least is the arm's-length glance in window light, from the distance other people stand at, and it takes two seconds. If a stroke survives that glance it is not too heavy; if the whole face survives it, the base colour matches the neck.

Everything on this page washes off. Unlike this outline's brow page — where the record's hardest lesson was that removal is permanent and the reversible methods come first — there is no irreversible move here, which means the third attempt costs exactly what the first one did: an evening and the products. The record's summary of the whole subject, from the first-person account that named position over depth, is that only slow trial finds the placement that suits a given face. The practical form of that is a log: date, product, where each stroke stopped, and the arm's-length verdict — written down on the day, so that a week later the comparison is between two recorded attempts rather than between a memory and a mirror.