This has attracted my interest of late:

http://www.perceptionweb.com/ecvp97/98tue.html

One eye is usually centred horizontally (and near the golden section vertically) in portraits over the past 500 years. C W Tyler (Smith-Kettlewell Eye Research Institute, San Francisco, CA 94115, USA; fax: +1 415 561 1610; e-mail: cwt@skivsski.org; WWW: http://www.ski.org/cwt)

Although the eyes are a key feature of facial portraits, compositional rules for the placement of the eyes relative to the frame are obscure. Two hypotheses of eye position in the portrait frame were compared: that the pair of eyes were symmetrically placed or that one eye was centred in the frame. Portraits were defined as paintings of a single person from the waist up without other dominant objects or animals in the scene. From all artists represented in seven source-books on portraiture over the past five centuries (eg from van Eyck to Picasso), the first portrait in which both eyes were visible was analysed (170 portraits). Horizontal and vertical eye and mouth positions were measured as a proportion of frame width and height.

The eyes in portraits tend to cluster horizontally around the centre vertical, with one eye centred in a normal distribution with a sigma of only 5% of the frame width. The binocular mean had a bimodal distribution implying that one or other eye was usually centred. Conversely, the eye height distribution was not centred vertically but peaked close to the classic golden ratio of 0.618 (where the smaller portion has the same ratio to the larger as does the larger to the whole), with virtually no eyes below the vertical centre. The mouth distribution was much broader than that of the centred eye. The eye centring with an accuracy of ~1 eye width seems not to be mentioned in art criticism, which suggests that unconscious functions operate in our aesthetic judgements.

[Supported by NIMH 49044.]
The paper is here:

http://www.ski.org/cwt/CWTyler/Art%2...enterSPIE.html