Heart-Face Brow Chart

A tracing method for settling which outline you actually have, the four brow landmarks in order from the nose outward, the tail-offset sweep from a straight brow to a 126-degree bend at four brow lengths, the forehead trade at three starting heights, the regrowth clock with its arithmetic shown, and a decoder that turns eight shape names into two numbers you can measure. Print it or keep it on your phone.

1. Trace the outline before trusting the name

  1. Print or display a photograph of the face straight on, hair off the forehead, no tilt, camera at eye level.
  2. Lay a transparent sheet over it and trace the facial contour — the hairline, down the sides, around the jaw, to the chin. This is the method a 2026 clinical review recommends for exactly this problem, on the grounds that determining face shape is difficult during assessment.
  3. Hold the tracing against the two definitions in the table below. They key on different features and they will not always give the same answer. Write down which one your tracing matches, because the brow advice differs depending on which you used.
  4. Then measure the four distances in the fields and compute the forehead ratio yourself.
Three definitions of this outline in circulation, and the feature each one keys on. Read across: they are not describing the same thing.
Definition What it keys on What it says
The one most often repeated The hairline — above the brow, not in it A widow's peak, and that this outline differs from an inverted triangle only by having one. Offered in the same breath as an admission that the writer did not know how that would affect anything.
The clinical one Forehead, cheekbones, jaw, chin, length Forehead and cheekbones the same width, a narrower jaw, a more angled chin, long straight jaw lines, length greater than width. No hairline. No brow.
This site's Forehead width over cheekbone width 0.978, the highest of seven outlines. Diamond, the nearest named boundary, scores 0.91 — separated by that one feature at 3.09 standard deviations, with identical jaw ratios at 0.759 and length ratios 0.27 of a standard deviation apart.

Two things to keep honest about this section. First, the clinical table's own introducing sentence says there are six primary facial shapes and then lists five — oval, round, square, heart and long — and the table has five rows. Neither diamond nor inverted triangle appears in it, which are the two names people argue about most when placing a face like this one. Second, this site publishes the caveat that its heart constant was interpolated along the literature's ordering from a single calibrated example rather than measured across a group, so the separations in the third row carry that weight and no more. And the outline is not fixed: dropping five or six pounds has been reported as enough to move a face from round to heart. Re-trace rather than relying on a name you were given once.

2. The four landmarks, from the nose outward

level of the inner start 1 inner start 2 outer iris edge 3 outer eye corner 4 tail end peak between 2 and 3 brow height not to scale oblique from the nostril wing through the outer eye corner
The four landmarks in the order you meet them moving away from the nose, and what each one is keyed to. The first and fourth come from a 2016 review of ideal brow shape and position; the peak position for this outline comes from a 2026 clinical review; the height figure comes from a 2017 measurement across 60 women.
# Landmark Where it goes Measured figure
1 Inner start On the vertical plane through the outer edge of the nostril wing and the inner corner of the eye. no figure stated
2 Peak Two-thirds of the way along the brow toward the temple, landing between the outer edge of the iris and the outer corner of the eye — closer to neither. Not halfway. L/6 from halfway
10.00 mm at L = 60
3 Tail end On an oblique line drawn from the outermost point of the nostril wing through the outer corner of the eye. The same source also says the inner and outer ends sit at approximately the same horizontal level; on a real face those two disagree by a few millimetres, and its word is "approximately". no figure stated
4 Brow height Not a point but a gap: the vertical distance from the top of the brow down to the horizontal line through the underside of its inner end. This is the measurement section 4 prices. 7.51 mm / 0.30 in mean
range 4 to 13 mm

Every inch figure on this sheet is the millimetre figure divided by 25.4. That division is exact arithmetic rather than a rounded approximation, because the inch has been defined as exactly 25.4 mm since 1959. Worked: 10.00 ÷ 25.4 = 0.39370, printed here as 0.39 in. A flexible tape works; so does a strip of paper you mark and then lay against a ruler.

Two warnings about the peak, both from the same 2016 review that supplies landmarks 1 and 3, which also states in so many words that it is impossible to define an ideal eyebrow suitable for every face. A peak placed too far inward creates an undesired surprised look. A low inner end with a high outer peak induces an angry one. Treat both as warnings rather than rules.

Do not substitute a lid gap for brow height. The 2017 study measured a different gap as well — from the upper border of the lid crease to the lower edge of the brow — and got 3 to 18 mm at the mid pupil, 4 to 19 mm at the outer edge of the iris, and 12 to 30 mm at the outer eye corner. Elsewhere that lid gap is put at 7 to 12 mm, with 15 to 16 mm suggested as ideal. All five of those are lid gaps. Brow height is the 7.51 mm figure. The two are not convertible into each other, and an instruction that quotes a millimetre figure without saying which gap it means cannot be followed.

3. Tail-offset sweep, straight brow to a 126-degree bend

A = 180° — no bend peak at 2L/3 medial run 2L/3 tail L/3 offset = 0.00 mm A = 135° — the measured mean medial line continued offset A peak offset = 14.14 mm / 0.56 in at L = 60
How far the tip of the tail sits from the medial line, measured perpendicular to it, at four brow lengths. Offset = (L ÷ 3) × sin(A), where A is the interior angle at the peak and sin(180° − A) = sin(A). Each cell gives millimetres and inches.
Angle at the peak 50 mm brow 55 mm brow 60 mm brow 65 mm brow
180° (straight)0.00 mm / 0.00 in0.00 mm / 0.00 in0.00 mm / 0.00 in0.00 mm / 0.00 in
165°4.31 mm / 0.17 in4.75 mm / 0.19 in5.18 mm / 0.20 in5.61 mm / 0.22 in
155°7.04 mm / 0.28 in7.75 mm / 0.31 in8.45 mm / 0.33 in9.16 mm / 0.36 in
144° (+1 SD)9.80 mm / 0.39 in10.78 mm / 0.42 in11.76 mm / 0.46 in12.74 mm / 0.50 in
135° (mean)11.79 mm / 0.46 in12.96 mm / 0.51 in14.14 mm / 0.56 in15.32 mm / 0.60 in
126° (−1 SD)13.48 mm / 0.53 in14.83 mm / 0.58 in16.18 mm / 0.64 in17.53 mm / 0.69 in

The 50, 55, 60 and 65 mm lengths are stand-ins, not targets — read down the column nearest your own brow length from section 2, or interpolate. The 135-degree mean and its 9-degree standard deviation come from a 2017 study of 60 women positioning their own ideal brow; that study recorded no face shape for any of them, and the words do not appear in it at all, so nothing in this table is a heart-face number. It is included because it is the only defined corner anywhere in the record, which makes it the only way to turn an adjective into a millimetre. The angle is defined as the one between the medial portion of the brow — usually the medial two-thirds — and the tail, the lateral one-third, so the peak position in section 2 is built into its own definition and the two are one study stated twice.

Worked so you can redo it with your own length: at L = 60 the tail is 60 ÷ 3 = 20 mm, and sin(135°) = 0.70711, so 20 × 0.70711 = 14.14 mm; 14.14 ÷ 25.4 = 0.5567, printed as 0.56 in. Two approximations are being made and both are worth naming. The figure is the distance measured perpendicular to the medial line; measured straight down instead it is larger by 1 ÷ cos(θ), where θ is the angle the medial run itself rises at — at θ = 8° that factor is 1.0098, under one per cent, so on a real brow the two are interchangeable and only the perpendicular one is exact. And the brow is a curve, so the tail's length along the brow is longer than the straight segment used here, which makes the true offset larger again.

4. The forehead trade, at three starting heights

What removing hair from the top edge of the brow costs, against a 50 mm forehead. Removing hair from the top lowers the top of the brow, so brow height falls by the amount removed and forehead height rises by the same amount. Brow loss = cut ÷ starting height. Forehead gain = cut ÷ 50.
Starting brow height Cut 1 mm Cut 2 mm Cut 3 mm Exchange rate
4.00 mm / 0.16 in
(bottom of the range)
3.00 mm left, −25.0% 2.00 mm left, −50.0% 1.00 mm left, −75.0% 12.5 : 1 against
7.51 mm / 0.30 in
(the measured mean)
6.51 mm left, −13.3% 5.51 mm left, −26.6% 4.51 mm left, −39.9% 6.7 : 1 against
13.00 mm / 0.51 in
(top of the range)
12.00 mm left, −7.7% 11.00 mm left, −15.4% 10.00 mm left, −23.1% 3.8 : 1 against

The forehead side is identical in every row: 1 mm cut is 2.0% of a 50 mm forehead and 0.04 in; 2 mm is 4.0% and 0.08 in; 3 mm is 6.0% and 0.12 in. Only the brow side moves, and it moves entirely with your starting height. That is why the same instruction is cheap for one person and ruinous for another, and why neither can tell which she is from the instruction alone. Substitute your own forehead height from section 2 and redo the division — on a 60 mm forehead the same 1 mm is 1.7% rather than 2.0%, which changes the ratio by about a sixth and changes nothing about the ordering.

Why this section exists only for this outline. Of the seven shapes this site classifies, this is the only one whose forehead measures wider relative to its cheekbones than any other's — 0.978 against the next highest of 0.951. The forehead is the widest thing about the face and the brow is its lower boundary, so the gap between them is a forehead measurement taken from underneath. On the average female face the brow gap is 15.02% of a 50 mm forehead and 13.06% of the 57.51 mm between hairline and brow underside: a small share of the upper face and a large share of itself, at the same time.

The one reasoned defence of removing hair from the top edge, anywhere in the record, was given by someone with a round face, about vertical thickness — plucking the top made her thick brows look longer horizontally, which helps an outline that has width to spare and no length. She walked it back the next day to "can work for some people", without saying which. On this outline the height you pay is forehead, and the forehead is already the widest feature, so the same move buys nothing and costs the 6.7 to 1 above.

5. The regrowth clock, arithmetic shown

How long a plucked brow hair takes to come back, computed from published phase lengths. Weeks are converted at 4.345 per mean month (52.14 weeks in a mean year ÷ 12) and months at 30.44 days.
Phase Lower bound Upper bound In days, lower to upper
Anagen — growing2 months3 months60.88 to 91.32
Catagen — transition2 weeks = 0.46 months3 weeks = 0.69 months14 to 21
Telogen — resting2 months3 months60.88 to 91.32
Total before the hair is present again4.46 months6.69 months136 to 204
For scale: scalp anagen2 years8 years8× the brow's longest phase at the short end, 48× at the long end

Shown as arithmetic rather than as a range: 2 + 0.46 + 2 = 4.46 months, and 4.46 × 30.44 = 135.76, printed as 136 days. The upper bound the same way: 3 + 0.69 + 3 = 6.69, and 6.69 × 30.44 = 203.64, printed as 204. This is the floor rather than the whole wait, because it counts only until the hair is present, not until it is long enough to be part of the shape.

Ordered by what each step costs to undo, cheapest first. Work down the list, not across it, and change one row at a time — on a subject where four and a half months is the smallest unit of feedback, anything faster than that is guessing.
Step What it costs to undo Date started Verdict
Set the hairs where they already are — a clear gel or a clear wax brushed through with a spoolieOne wash  
Change the colour, checked against your natural brow rather than your hairOne wash, on a comparable timescale  
Draw hair into the inner start, or into a short tailOne wash; removes nothing  
Change the outline by setting the tail in a new direction rather than cutting itOne wash  
Pluck a few hairs from between at the inner start, to match a darker front to a sparse tail136 to 204 days  
Pluck from the top edge136 to 204 days, priced at 3.8 to 12.5 against in section 4  
A professional reshapingThe same clock, plus the fact that a shape chosen in a chair is chosen against the chair's reference images rather than your measurements  
Implanted pigmentNo duration is stated anywhere in the record, so no figure is printed here  

Three numbers that would settle real decisions and are not in the record: how long a lamination is supposed to last, so a report of one week cannot be compared against an expected figure; how long implanted pigment lasts; and how long the effect of the one clinical heart-face case lasted, whose only timepoint is 30 days and is an assessment point rather than a duration. Where those blanks are, this sheet leaves them blank rather than filling them in.

6. Shape-name decoder

A 2026 clinical review lists five brow shapes — curved, sharp-angled, soft-angled, rounded and flat — adds that arched configurations are separately classified into low, medium and high arches, and calls the result eight primary brow characteristics. Not one of the eight carries a measurement, an angle or a landmark in that document. The two tables below attach the only numbers available: where the peak sits along the brow, and the interior angle at the peak, converted to a tip offset at L = 60 mm using section 3.

The five listed shapes, converted. Angles marked "estimate" are this sheet's own, placed to keep the names in the order the record uses them in; no source assigns a number to any of them.
Name Peak position Angle at the peak Tip offset at L = 60 Where the number comes from
Flat none 180° 0.00 mm / 0.00 in Definitional, not an estimate. A straight line has no bend, and sin(180°) = 0.
Soft-angled two-thirds 165° (estimate) 5.18 mm / 0.20 in This sheet's estimate. The review names the shape and gives it no angle.
Curved two-thirds 155° (estimate) 8.45 mm / 0.33 in This sheet's estimate, as above.
Rounded between the outer iris edge and the outer eye corner 155° (estimate) 8.45 mm / 0.33 in The angle is this sheet's estimate. The peak position is not: this is the one shape name in the review that comes with a landmark, in its row for this face.
Sharp-angled two-thirds or further out 126° (estimate) 16.18 mm / 0.64 in This sheet's estimate: one standard deviation below the measured mean, which is the tightest bend the published spread supports.

Count what converted. Five names, four distinct offsets, and only one — flat — whose number is not an estimate. Two of the five, curved and rounded, land on the same pair of numbers, and nothing in the review separates them except which face each is prescribed to. The distance between soft-angled and curved, two names argued about as though they were different objects, is 3.28 mm: narrower than the 4.42 mm band that one standard deviation around the measured mean already covers. The distance between flat and sharp-angled is 16.18 mm, or 3.66 times that band. So most of the argument over shape names on this face is an argument about a word, and the part of it that is real is about two settings, not five.

Three further names in circulation for this face, which appear in no list anywhere in the record and cannot be converted.
Name Angle at the peak Tip offset Why it will not convert
Classic not stated cannot convert No source gives it an angle. In the record it is used to mean the brow recommended for a different outline, so it names a reference rather than a geometry.
Fluffy not stated cannot convert Density and direction only. It is not a rival of the other names; it is a different axis, which is why it can be combined with any of them.
Thin and straight 180° 0.00 mm / 0.00 in The geometry half is identical to flat. "Thin" is a density claim riding along inside a shape name, and the two are different parameters.

All three are used as captions on numbered photographs in one exchange, where a reply picks "eight, straight, and nine, sharp" and another asks for "not thick and fluffy", with nobody defining a word. That is the state of the vocabulary: names attached to pictures, in a thread, once.

The arch-height axis, which is the third of the review's eight and a different measurement again: how far the peak rises above the level line through the inner start, not how much the brow bends. Rise angle = arctan(rise ÷ medial run), and the medial run is 2L ÷ 3. Write your own rise in the left column and read the angle off the column nearest your brow length.
Peak rise 50 mm brow
(run 33.33)
55 mm brow
(run 36.67)
60 mm brow
(run 40.00)
65 mm brow
(run 43.33)
2 mm / 0.08 in3.43°3.12°2.86°2.64°
4 mm / 0.16 in6.84°6.23°5.71°5.27°
6 mm / 0.24 in10.20°9.29°8.53°7.88°
8 mm / 0.31 in13.50°12.31°11.31°10.46°
mm — yours    

The three angles are not independent, and this is the relation that ties sections 3 and 6 together. If the medial run rises at θ above the horizontal and the tail descends at φ below it, then the interior angle at the peak is A = 180° − θ − φ. Worked at L = 60: a 6 mm rise gives θ = 8.53°, and an angle of 135° then needs φ = 36.47°. Which means a name like "high arch" says nothing at all until both are measured, and that low, medium and high are three words for a quantity the record never once puts a number against.

The peak-position half of section 2 is the part that is not an estimate, and it is the part with four sources behind it that never cite each other: a 2016 textbook landmark, a 2017 measurement of 60 women, a 2026 clinical prescription for this outline specifically, and lay advice given in 2024 to someone who had shaved her brows off by accident. Two-thirds of the way along, between the outer edge of the iris and the outer corner of the eye. The 2026 review reaches the same fraction a second time from a different direction, in its section on gender differences, where it says the feminine brow peaks at the junction between the middle and lateral third — which is two-thirds — which is why it counts as four sources and not five.