Facial thirds and fifths: the classic canons and real faces

Facial thirds and fifths are the two oldest rules for describing a face’s proportions. The thirds divide the face from top to bottom: hairline to brows, brows to the base of the nose, and nose to chin. The fifths divide it across, at eye level, into five eye-widths. Classical artists drew them as equal. Real faces almost never are, and that turns out to be fine. This guide covers where the rules come from, what measurements of real faces found, how to measure your own and what your proportions do and do not tell you.

The thirds

• Upper third: from the hairline (trichion) to the point between the brows (glabella). The forehead.
• Middle third: from the brows to the base of the nose (subnasale). The eyes, cheekbones and nose.
• Lower third: from the base of the nose to the bottom of the chin (menton). The lips, teeth, jaw and chin.

Clinicians sometimes split the lower third again: the upper lip takes about a third of it and the lower lip and chin about two thirds. Orthodontists and surgeons use the lower third most, because it is the part that bite, teeth and jaw growth change.

The fifths

The fifths are measured across the face at the level of the eyes: from the face edge to the outer corner of one eye, across that eye, across the gap between the eyes, across the other eye, and out to the other face edge. The rule says each of the five is one eye-width. Two related rules come from the same tradition: the gap between the eyes equals one eye’s width, and the nose is as wide as the gap between the eyes.

Where the canons come from

The thirds are at least two thousand years old. The Roman architect Vitruvius, writing in the first century BC, described the face as divided into three equal parts: from the chin to the underside of the nostrils, from there to the line between the eyebrows, and from there to the roots of the hair. He used it as an example of the proportion he wanted in temples.

Renaissance artists took the rules up again. Leonardo da Vinci’s notes and Albrecht Dürer’s studies of human proportion set out how to divide the head and face so that drawn figures looked right. These were drawing aids for artists. Later they were repeated in textbooks as if they described what a normal or ideal face looks like, which is where the trouble starts.

What measuring real faces found

In 1985 the anthropometrist Leslie Farkas and colleagues tested nine of these neoclassical canons on 153 young white North American adults, measured directly on the face. The canons fitted poorly. The vertical profile rules did worst: no more than two of the sections were ever equal in the same face. The best-fitting rules were horizontal, and even they held for a minority: the gap between the eyes equalled nose width in 40% of people, and nose width equalled a quarter of face width in 37%. The authors concluded that the canons “do not represent the average facial proportions”.

A 2002 study extended this to young Chinese, Vietnamese and Thai adults and compared them with North American Caucasians. In no group were the three sections of the profile equal. The other canons held in 17–37% of the Caucasian group and 2–27% of the Asian groups, mostly because the Asian faces were wider between the eyes, at the nose and across the cheekbones relative to the canon. A later international survey by the same research group measured 1,470 young adults across Europe, the Middle East, Asia and African populations and found large differences between groups, especially around the eyes and nose.

The pattern is consistent: the canons describe an artistic convention, not the average human face, and certainly not the average face of every population.

Our own numbers say the same. MediaPipe’s canonical face, the average face our scan is built on, measures 32% / 33% / 35% for its thirds, close to even. Its fifths are 21 / 17 / 24 / 17 / 21: the gap between the eyes is about 1.4 eye-widths, not one, and the face is about six eye-widths wide at eye level, not five.

Which third is usually longest?

There is no single answer, because it depends on where the hairline sits and on the population. In adults, the lower third is often the longest of the three when the face is relaxed with the lips closed, and the upper third varies the most because hairlines vary so much. On the average face our scan is built on, the lower third is the longest at 35%, the middle 33% and the (estimated) upper third 32%. A difference of two or three percentage points between thirds is well within ordinary variation; the 1985 study found large spreads in every facial measurement it took.

How proportions change with age

The thirds are not fixed for life. Farkas’s study also measured 6-, 12- and 18-year-olds and found age-related differences in how well the canons fitted. In young children the forehead and upper face take up more of the face and the lower third less; the jaw, nose and chin grow most in adolescence, which lengthens the middle and lower thirds. Through adulthood, tooth wear and tooth loss can shorten the lower third, which is one reason dentists restore the height of worn teeth, and a receding hairline lengthens the upper third on paper without any change in the face. If you compare your proportions over time, compare like with like: same age range, same hairline, same expression.

How to measure your own

1. Stand at least 1.5 m from the camera and zoom in, or prop the phone at that distance. A close phone enlarges the nose and middle of the face.
2. Hold the camera at eye level, chin level, looking straight ahead.
3. Relax your face with your lips closed. An open mouth lengthens the lower third.
4. Pull your hair back so the hairline is visible.
5. Mark the hairline, the point between the brows, the base of the nose and the bottom of the chin. Measure the three distances and divide each by the total.
6. For the fifths, mark the face edges and the inner and outer corners of each eye on a line at eye level, and divide each of the five widths by the total.

The pitfalls

• The hairline. It is the least reliable point: it moves with age, recedes, and is hard to define under a fringe. Many clinicians measure the middle and lower thirds and treat the upper third as a rough guide.
• Head tilt. Tipping your chin down shortens the lower third in the photo; tipping it up lengthens it.
• The face edge. Where the face ends at eye level depends on hair and ears, so the outer fifths vary most between methods.

How our tool measures them

The facial ratios calculator places the points for you from a short camera scan, after rotating your face to a frontal pose, and draws the thirds and fifths on your own face. It measures the thirds along the midline of your face and the fifths across it at eye level. It also reports the gap between your eyes relative to one eye’s width, and your nose width relative to your mouth width.

One limit to know: the face model has no hairline point. Its top point sits roughly halfway between the brows and the hairline, so we estimate the hairline at twice that distance above the brows. The upper third is therefore an estimate, and the result says so. The middle and lower thirds are measured directly.

What your proportions tell you

Read your thirds and fifths as a description of your face, not a grade:

• A longer lower third can come from a long chin, a vertical jaw growth pattern or lips that rest apart. A shorter one from a short chin or, later in life, tooth wear and tooth loss.
• A longer middle third usually means a longer nose.
• A longer or shorter upper third mostly reflects your hairline and forehead shape, and changes with hairstyle.
• Wider-set eyes (a gap well over one eye-width) and closer-set eyes are both common variations, and the typical gap differs between populations.

Do proportions affect attractiveness?

Somewhat, and in a different way than the canons suggest. The strongest finding in this research is about averageness: faces close to the population’s average shape tend to be rated more attractive than most individual faces, as Langlois and Roggman showed with composite faces in 1990. When researchers searched for the eye-to-mouth distance and eye spacing that made faces most attractive, they found it matched the average face, not a classical rule. And when a study tested 27 “ideal” frontal ratios and 26 profile angles from the literature against ratings of 64 adolescents, only two ratios and three angles related to the ratings at all.

Our attractiveness score uses proportions gently. How even your thirds and fifths are, and how close your eye spacing is to the average, together make up the proportions component, which carries 15% of the score. The tolerance is wide, because even the average face is not perfectly even. Symmetry and averageness each carry more.

The thirds and fifths are also where most “golden ratio” claims about faces start. The golden ratio guide looks at what that evidence actually shows, and the golden ratio face test measures your face against both. For left–right balance rather than proportion, see the facial symmetry guide.

The free result gives your thirds, fifths and key width ratios. The full report ($14.99) draws them on your own scan alongside your symmetry by region and where proportions sit in your score.