Facial proportions, and which of them you can actually measure

Every article about facial proportions quotes the same three rules — thirds, fifths, and the golden ratio — as if they were one coherent system handed down from antiquity. They are not. They come from different centuries, they make different claims, and only some of them survive contact with a measuring tool. This is what each one actually says, where it came from, and which parts of it you can check on your own face in about a minute.

What facial proportions actually are

A facial proportion is a ratio between two distances on the same face. That last part is the whole point: a ratio needs no scale. It does not matter how far you stood from the camera or what lens you used, because both distances shrink or grow together. This is why almost everything on this site is reported as a ratio, and why pupillary distance — a genuine millimetre figure — is the odd one out, only measurable because the iris happens to be a near-constant 11.7 mm across adults.

It also explains why proportion rules are stated the way they are. Nobody says an attractive nose is 47 mm long. They say the lower third should equal the middle third, because that comparison holds whatever size the head is.

The rule of thirds: the oldest facial proportion

Divide the face horizontally at three landmarks — the hairline, the brow line, and the base of the nose — and you get three bands running down to the chin. The classical canon, formalised in Renaissance drawing manuals and still taught in life-drawing classes today, says the three should be roughly equal.

In practice the upper third is the unreliable one, because it is bounded by a hairline rather than by bone. A receding hairline lengthens it, a fringe hides it, and pulling hair back changes the number you get. Surgeons who use thirds in planning often skip it for exactly this reason and work from the brow down.

The lower third has an internal division of its own that is more useful than the rule it sits inside: from the base of the nose to the lip line should be about a third of it, and lip line to chin the other two thirds. That upper segment is the philtrum, and it is the one part of face height that reliably lengthens with age.

The rule of fifths: face width in eye widths

The vertical counterpart divides face width into five columns, each one eye wide. Columns two and four are the eyes themselves. Column three is the gap between them — the rule says the intercanthal distance equals one eye width — and the outer columns run from the outer corner of each eye to the edge of the face.

This one is more measurable than the thirds, because every boundary sits on a corner of an eye rather than on hair. It is also where two common labels come from: when the central column is clearly wider than one eye, the eyes read as wide-set; clearly narrower and they read as close-set. Both are the same measurement, reported in opposite directions, and neither is a defect.

Where the golden ratio fits into facial proportions

The golden ratio, phi, is approximately 1.618. Unlike the thirds and the fifths it describes an unequal division, and it arrived in this subject far later — its systematic application to faces dates to the 1990s, most visibly through Stephen Marquardt's phi mask, which overlays a decagonal grid built from phi onto a photograph.

The mask is genuinely striking when it lines up, and that is most of its persuasive power. It also has a documented problem: Holland's 2008 analysis in Aesthetic Plastic Surgery found the mask was derived from a narrow set of fashion models and fits some populations far better than others, which makes it a description of one aesthetic rather than a universal constant. Quoting phi as though it were a law of biology skips that finding entirely.

Facial proportions and attractiveness: what the research supports

Three findings recur across the literature and are worth separating from the folklore around them.

• Averageness. Langlois and Roggman (1990) found that composite faces — many faces mathematically averaged together — are rated more attractive than most of the individual faces that went into them. The effect is robust and repeatedly replicated. It also implies something people find counterintuitive: unusual proportions cost more than they gain.
• Symmetry. Grammer and Thornhill (1994) reported a correlation between facial symmetry and rated attractiveness. The correlation is real and modest. Perfect symmetry is not the goal state — nobody has it, and faces made artificially symmetrical tend to be rated as slightly uncanny.
• Skin, not geometry. Fink, Grammer and Matts (2006) found that skin colour distribution alone shifted ratings of age, health and attractiveness, with no change to the underlying shape. In several studies of this kind, skin outperforms proportion.

Rhodes' 2006 review in the Annual Review of Psychology is the standard summary if you want one source: averageness, symmetry and sexual dimorphism all contribute, all with modest effect sizes, and none of them behaves like the single lever the popular accounts describe.

Proportion explains part of the variance in how faces are rated. It has never explained most of it, and no published result claims that it does.

How to measure your own facial proportions

You need one photo taken straight on, at eye level, with even light on both sides. Head rotation is the dominant error source — a turn of ten degrees changes every horizontal ratio on this page while leaving the vertical ones alone, which is exactly the pattern that makes two runs disagree in a way that looks mysterious.

1. 1Hold the camera at eye level, not above or below. A raised camera shortens the lower third; a lowered one lengthens it.
2. 2Face the lens squarely. If one ear shows more than the other, you are turned.
3. 3Relax your expression and close your mouth. A smile moves the lip line and shortens the lower third.
4. 4Pull hair off the forehead if you want the upper third to mean anything.
5. 5Compare the three vertical bands, then the five columns. Differences under about five per cent are not visible to anyone, including you.

Facial proportions a front photo cannot measure

This is the part the drawing guides leave out, because a drawing can simply choose a profile. A single front photo carries no depth information at all, and several of the proportions people most want are depth measurements.

Four of the seven named nose types are defined by exactly these profile traits, which is why that page tells you which of them it can and cannot decide from your photo instead of picking one anyway.

Proportions by feature

Each feature has its own conventions, its own measurable ratios and its own blind spots. These pages carry the arithmetic for each one:

What to do with the numbers

Almost nothing, most of the time. A face that misses equal thirds by four per cent is a face nobody will ever describe as disproportionate, because four per cent is below the threshold at which anyone sees it. The measurements are worth having for two narrower reasons: they tell you which of the labels you have been reading about actually apply to you, and if you are considering a procedure they let you describe what you want in terms a clinician can act on.

If that is where you are headed, the simulations render one change on your own photo so you can see the proportion shift rather than imagine it. It is a visualisation and nothing more — no image on this site is a surgical plan, and no measurement here is a diagnosis.