Facial thirds: a centuries-old rule the researcher who modernized it found doesn't hold.

Facial thirds divide a front-facing face vertically into three bands, using landmarks that are easy to spot on a photo:

• Upper third — hairline (trichion) to the space between the brows (glabella).
• Middle third — glabella to the base of the nose (subnasale).
• Lower third — subnasale to the bottom of the chin (menton).

The classical version of the rule, tracing back to neoclassical canons of proportion applied to the face, holds that a balanced face has all three bands at close to equal height — a clean 1:1:1 split.

The equal-thirds idea predates modern measurement — it's a classical proportion canon, not a clinical finding. The researcher who put it to an actual anthropometric test was Leslie Farkas, widely credited as the founder of modern facial anthropometry. His landmark 1985 study measured over a hundred young Caucasian adults using standardized landmarks and found something the canon didn't predict:

• The classical neoclassical canons, including equal thirds, described only a minority of real faces — even within that single ethnic group his study sampled.
• Only about a third of his Caucasian subjects met the related orbital canon, and roughly 40% met the orbitonasal canon; conformity was lower still in the Afro-American subjects in the same study.
• Across later population studies, the mismatch runs in a consistent direction: the lower third measures longer than the middle third far more often than a strict 1:1:1 split predicts.

Farkas's own body of work is what modern facial plastic surgery actually relies on — not the classical canon it replaced, but the age-, sex-, and population-specific norms he spent decades compiling to correct it.

The pattern Farkas found in his original Caucasian and Afro-American samples has since shown up across other populations studied the same way:

• A study of young adults from the Arabian Peninsula found roughly nine in ten did not conform to the equal-thirds canon — the same direction of mismatch (longer lower third), and the subjects were independently rated as aesthetically pleasing anyway.
• Research on facial feminization planning has documented meaningful lower-face proportion differences across ethnic groups, reinforcing that population-specific norms fit better than one global target.
• Measurement itself is sensitive to head position: manual anthropometry needs a standardized, level head pose to compare against published norms at all, and clinical literature flags natural head position versus a strict reference plane as a real source of error even in controlled settings.

None of this says a longer or shorter lower third looks worse. It says the "equal thirds" target being repeated as a universal rule is the part of the story Farkas's own research spent decades correcting, not confirming.

Mogalytics does not compute or display a separate facial-thirds number. Its six-measurement composite reads live from a front-facing camera feed, and the closest measurement by axis is:

• Jaw ratio — jaw height relative to total face height, occupying the same lower-third region the thirds rule debates, though it's a two-point ratio, not a three-band split.

Jaw ratio and facial thirds both concern vertical face proportion, but they measure different things — one is a specific two-point ratio Mogalytics reads live, the other is a three-way classical split it does not compute. Mogalytics makes no claim of parity between the two, or with any other site's specific thirds assessment.

Facial thirds sits alongside a few of the six measurements Mogalytics tracks. If you're auditing the rest:

• Jaw ratio — the closest camera-read measurement by axis, sensitive to chin pitch rather than head position.
• Golden ratio face — another classical proportion canon with its own peer-reviewed critique literature.
• SNA/SNB angle — a different skeletal-projection standard with the same photo-versus-clinical-landmark limitation.
• Why face-rating scores swing between scans — the camera and pose sensitivities behind every one of these self-measurements.