Mian Xiang

Early references and the Han period

Xing Wang's 2017 Oxford thesis on Ming physiognomy says the Zuozhuan records a court historian predicting the fortunes of two sons of an aristocrat, citing among other features one son's plump chin. Early records show physiognomy used in political advice and in predicting disease and longevity among aristocrats. Yang Gang and Christoph Anderl (2026) say references before 221 BCE were scattered across texts, that dedicated treatises such as the 24-fascicle Xiangren listed in the Hanshu appeared from the Qin and Han periods, and that all such pre-Tang treatises are lost.

Xu Fu, a female physiognomist of the Han period named in the Shiji, gave her name to many later manuals, including several found at Dunhuang. The Eastern Han thinker Wang Chong (27 to about 97 CE) held that each person receives a fixed allotment of qi at birth that sets body and lifespan, which Yang and Anderl describe as the classical grounding of Chinese physiognomy. Wang's thesis notes that Xunzi criticized the practice, arguing that judging the heart matters more than reading outward form.

Tang, Song and the Ming compendia

Yang and Anderl identify the Dunhuang manuscripts, copied in the late Tang and Five Dynasties period from older sources, as the earliest surviving works devoted to physiognomy. Wang concludes that Chinese physiognomy became more systematic during the Song and Ming. He finds that six of the seven schemes for dividing the face used in Ming manuals, including the Twelve Palaces, are not attested in earlier manuals, and that none appears in the Dunhuang manuscripts. He suggests they are Ming inventions or at least Ming systematizations.

Wang describes Chen Tuan (872 to 989 CE) as historically famous for physiognomy and as the ascribed author of several later manuals. A Wellcome Collection record for an 1807 Japanese edition of the Shenxiang quanbian attributes the compendium to a Chen, spelled Chen Zhuan in the record, names Yuan Zhongche as editor and possible compiler, and says the first Japanese edition appeared in 1651. Such attributions are traditional claims, and the sources reviewed do not settle authorship.

Practitioners and attributions

Wang sorts recorded Song and Ming practitioners into full-time physiognomists, Buddhist monks, Daoist priests, diviners with several skills, and literati or officials, and notes that the roles overlapped. Chen Tuan's own identity is disputed: Livia Kohn, as cited by Wang, treats him as a Daoist master, while a study by Liao Hsien-huei, also cited, stresses a more complex background. The Ming compendium also carries texts under the name Guigu, which Wang says names a Warring States strategy text and later a Daoist deity.

Names and scope

Mian Xiang, xiangmian and mien shiang are spellings of the Chinese terms for face reading found in English-language writing. The scholarship reviewed uses xiangshu for the wider technique, which Wang defines as telling fortune by inspecting the body and the material world, including hands, animals, houses, tombs and written characters. A chapter titled Mianxiang zonglun appears in a Ming manual, the Water Reflection (Shuijing shenxiang), that Wang cites. This page uses Mian Xiang for the facial branch only.

Core vocabulary

Qi is the vital energy whose state the manuals say shows in color and voice. The Three Sections (santing) divide the face into upper, middle and lower parts. The Twelve Palaces (shier gong) name facial points for areas of life, and the Five Mountains and Four Rivers (wuyue sidu) map landscape onto the face. The Five Phases Shape (wuxing xingxiang) sorts bodies into Wood, Fire, Earth, Metal and Water types. Twelve Palaces also names the houses of the Ziwei doushu astrology chart, whose layout Wang says the facial scheme shares, so a source should be checked for which system it means.

Sections, ages and mountains

In Wang's account of Ming manuals, the Three Sections run from the hairline to the eyebrows, the eyebrows to the nose tip and the nose tip to the chin, standing for youth, middle age and old age. The forehead is Heaven, the nose Humanity and the chin Earth. A thirteen-point scheme of flowing years ties positions on the face to ages from 16 to 71. The Five Mountains scheme assigns the forehead to a southern mountain, the chin to a northern one, the left cheekbone to Mount Tai in the east, the right cheekbone to Mount Hua in the west and the nose to Mount Song at the center, which should be high.

The Twelve Palaces

In the Water Reflection manual, as Wang describes it, the glabella is the Palace of Life and the nose the Palace of Wealth. The eyebrows are the Palace of Fraternity, the upper eyelids the Palace of Estate, the lower eyelids the Palace of Descendants, the chin the Palace of Servants and the eye corners the Palace of Wives. The nose root is the Palace of Diseases and Mishaps. Lists differ between manuals: the Water Reflection manual has a Palace of Parents where the Shenxiang quanbian has a Palace of Appearance.

Body types, color and qi

Another scheme sorts bodies into five types. Wang notes that the Lingshu chapter of the Inner Canon on twenty-five types of people already describes shapes in Five Phases terms, for example earth-type people with yellow, round faces, and that Ming manuals quote it as a guide. The manuals summarize Wood as thin and straight and Metal as square with tight flesh and thick bones.

Color is read alongside form. Wang says Ming manuals define color as light radiating from bodily qi, recommend examining it by candlelight in the morning and tie facial colors to seasons and directions. He counts more than two hundred color correlations in the Shenxiang quanbian. The Ming manuals he cites also place a woman's fortune on the right side of the body, unlike a man's, so right-side features weigh more in her reading.

Fortune, career and marriage

The manuals Wang describes read the face for wealth, siblings, children, a spouse's conduct, health and future mishaps. He says Ming literati consulted diviners about examinations, careers, marriage, funerals and health, and that rulers and officials turned to physiognomy partly because examiners were thought unable to see through a candidate. Ming manuals treat women as a separate case, with fortune defined mainly through influence on husband and children.

Houses and landscape

The manuals describe the body as a replica of natural landscape, a system similar to fengshui. Wang reports that the Iron Halberd manual links the Five Mountains scheme to geomantic theory: a person whose face has one mountain notably stronger should live in a house facing the same direction, so that good fortune is magnified. These are claims in the manuals, and the sources reviewed do not test them.

Complexion and health

Physiognomy and medical diagnosis share vocabulary but have different aims. Lu Luming and colleagues (2009) describe complexion inspection as a characteristic method of Chinese medical diagnosis, discussed in the Neijing, that reads changes in facial color through yin-yang and Five Phases theory matched to the five zang organs. A 1817 Qing woodcut in the Wellcome Collection charts facial sectors that correspond to internal organs. Wang notes that Ming physiognomy manuals tie facial color to the Five Phases, and that most known masters of pulse physiognomy were local physicians, one of whom, Wu Kun, criticized the practice from a medical standpoint.

Moral cultivation and change

Wang argues that Ming physiognomy did not treat fortune as wholly fixed: manuals hold that moral cultivation can alter the body and so change fortune. Yang and Anderl contrast this with Wang Chong's view that bodily form cannot change, and suggest the idea that conduct alters appearance may reflect Indian Buddhist influence.

Reading character from faces

Psychologists have long documented that observers agree in attributing traits to faces. Todorov and colleagues (2015) review this research and argue that the diagnostic validity of such attributions has been greatly overstated. In a developmental study, 141 children aged 3 to 10 and 99 adults judged pairs of computer-generated faces. Attributions of trustworthiness, dominance and competence at ages 3 to 4 converged with adults, and those of 5- to 6-year-olds were at adult levels of consistency (Cogsdill and colleagues, 2014). Agreement shows that people share impressions, not that the impressions are correct.

Belief and what has not been tested

Jaeger and colleagues (2025) state that earlier studies found facial impressions of traits such as trustworthiness show little to no accuracy, and that physiognomy is widely regarded as pseudoscience in academic circles. Their four studies found that 43.5% to 55.2% of participants at least somewhat endorsed physiognomic beliefs, including 52.1% of a representative Dutch sample of 2,624. None of the sources reviewed tested Chinese face-reading systems, such as the Twelve Palaces or five-element face types, against later life events or character measures. PubMed searches on physiognomy and accuracy returned no such study, which is an absence of evidence in either direction.

Measuring complexion

Studies of complexion inspection in Chinese medicine measure color, not fortune. Zhao and colleagues (2014) used facial images of 122 people, labeled by three Chinese medicine physicians (a label required agreement of at least two), to train a classifier for six complexion classes. The best configuration reached 86.89% accuracy under leave-one-out testing. The authors note a small dataset, weak results for the normal class and lighting variation, and describe traditional facial diagnosis as experience-based and subjective.

Xu and colleagues (2012) measured facial color with a spectrophotometer at eight points in 183 healthy, 287 sub-healthy and 370 diseased people and reported significant differences among the groups at six points. Only the abstract was read, and it reports no diagnostic accuracy. In a cross-sectional study of 64 adults, Mun and colleagues (2017) found lower facial redness was associated with a higher self-reported cold pattern score (B = -1.048, P = 0.021). The measure was a questionnaire, and the study was small, single-center and lacked strict lighting control.

Facial shape and type 2 diabetes

Cheng and colleagues (2022) studied 2,574 people at one hospital in Lanzhou, China, 1,590 with type 2 diabetes and 984 without, a comparison group that included people with other conditions. They derived 18 shape measures from 81 facial landmarks and fitted a regression model. On the test set the area under the curve was 0.686 and 0.668 for two versions of the model, which indicates modest discrimination, and the authors conclude that facial features can only give an auxiliary reference. This concerns shape, not complexion, and the abstract and results report different values, so the figures need checking against the full text.

The introduction of Zhao and colleagues describes earlier computerized work distinguishing hepatitis from health by facial color, without giving accuracy figures. No free full-text hepatic study was verified for this page.

Fatalism, bias and decisions

Wang reports that Ming physiognomists often used euphemism and vague wording when predicting misfortune, partly from fear of the questioner's temper, and that skeptical literati called the practice psychological trickery. The Song scholar Kong Zhongping argued that people with similar features had very different fates. None of the sources reviewed measured how receiving a reading affects a person's decisions or wellbeing, so the size of any harm from fatalistic predictions is unknown.

Jaeger and colleagues (2025) report earlier studies linking stronger physiognomic belief to support for facial profiling algorithms in hiring and policing and to rating applicant photos as more useful for hiring, and they describe historical uses of face-based judgments in criminology and eugenics. Wang documents historical Chinese use of face reading in choosing officials and in marriage consultations. No study of present-day hiring or marriage practice involving Mian Xiang was found.

Health and delayed care

A face reading is not a medical assessment. Zhao and colleagues describe traditional facial diagnosis as experience-based and subjective, and Cheng and colleagues call facial features an auxiliary reference only. The group differences in Xu and colleagues' study were not presented as a screening test. The sources reviewed report no cases of delayed care caused by face reading, so the size of that risk is unknown. New or worsening symptoms warrant assessment by a qualified clinician, and a reading should not replace or delay necessary medical care.

Training and regulation

The sources reviewed describe lineages of masters and pupils and attributed manuals, not a professional register, protected title or training standard, and none was found for Mian Xiang in any jurisdiction. Practitioner backgrounds vary widely, from monks and officials to full-time diviners in the historical record.