Chinese Physiognomy: Primary Sources and Further Reading

Read this one first

If you read only one text on this subject, read the chapter "Against Physiognomy" (非相) in the Xunzi.

It is short, its reasoning is transparent, and its method is to produce counter-examples: sages of antiquity who were ugly or deformed, tyrants who were tall and impressive. Its conclusion has been quoted for over two thousand years — judging the form is inferior to judging the heart, and judging the heart is inferior to judging the methods a person chooses.

Reading it first changes how everything else lands. You will approach the manuals knowing that the objections to them were written down a thousand years before they were, by a canonical author, in China, on empirical rather than moral grounds. That knowledge does more to orient a reader than any modern critique could.

Primary material in the histories and catalogues

Before the manuals there are scattered records, and these are more informative about the early practice than any later compilation.

• Zuozhuan, first year of Duke Wen — the court historian Shufu reads the faces of Gongsun Ao's two sons; one of the earliest such records.
• Shiji, biography of Fan Ju and Cai Ze — Tang Ju's reading of Cai Ze, a complete narrative with named participants.
• Shiji, annals of Emperor Gaozu — Lord Lü reads the face of Liu Bang. Read against the annals of Xiang Yu, this shows clearly how physiognomic anecdote was made to serve arguments about the mandate of heaven.
• Hanshu, bibliographic treatise, "form methods" category — a lost Xiangren in twenty-four scrolls is listed alongside works on siting dwellings, appraising swords and appraising livestock. That single classification tells you how the Han understood the art.
• Wang Chong, Lunheng, chapter on bone physiognomy — read together with his chapters attacking calendrical taboos, to see why the same sceptic rejected one and accepted the other.
• Liu Shao, Renwu Zhi — the nine indicators, belonging to the empirical assessment of candidates for office. Reading it is the quickest way to separate personnel appraisal from fortune-telling.

These sources share a limitation worth stating. What the histories preserve are the readings that came true; failures were not recorded and would not have been copied if they had been. However many such anecdotes accumulate, they cannot measure the art's accuracy — they measure only how seriously it was taken.

The manuals themselves

• Taiqing Shenjian, Yuguan Zhaoshen Ju, Yuebo Dongzhong Ji — early compilations from the Five Dynasties and Song, all pseudonymously attributed. Their value is in showing what the categories looked like when they first became fixed.
• Shenxiang Quanbian (神相全编) — credited to Chen Tuan, compiled in the Ming with the involvement of Yuan Zhongche and others. This is the most widely circulated compendium of face reading, and most manuals now on sale descend from it.
• Yuan Zhongche, Gujin Shijian (古今识鉴) — the author's record of his own cases, and the most direct material available on physiognomy at the Ming court.
• Chen Zhao, Xiangli Hengzhen (相理衡真) — Qianlong period, reorganising the scattered mnemonic verses into ordered categories; the clearest of the Qing manuals.
• The Siku Quanshu abstracts for fate and physiognomy books — reading the imperial cataloguers' reservations shows what Qing scholarship actually thought of this material.
• Chen Gongdu, Gongdu Xiangfa (公笃相法) — Republican era, the representative attempt to place the art on a scientific footing. The manner of its failure is itself instructive.

A word on how to approach the manuals as documents. They are best read not as instruction but as evidence: each one records what its compiler thought a life consisted of, which qualities deserved a name, and what a respectable person was supposed to look like. Read that way they are excellent social history, and the contradictions between them stop being a problem and become the most interesting thing about them. Read as instruction they are a closed system with no way of correcting itself.

Chinese-language scholarship

Rigorous studies devoted specifically to physiognomy are not numerous, and the relevant material is mostly distributed through histories of the divinatory arts, social history and the history of the body.

Li Ling's Zhongguo Fangshu Kao and its sequel deal with the origins and transmission of the divinatory arts as a whole, including excavated material, and are the foundation for understanding the knowledge system to which physiognomy belongs. Jiang Xiaoyuan's history-of-science work on astrology and the divinatory arts supplies a methodological model for assessing this kind of material.

One caution applies when reading Chinese secondary literature in this field. A number of authors are not only describing the art but recommending its use. Such books can be read as sources — they accurately report what practitioners believe — but they cannot be read as scholarship, and the distinction is not always signalled on the cover.

Excavated and manuscript material deserves a mention even though it is thinner here than for date selection. Physiognomic content appears among Dunhuang manuscripts and in miscellanies alongside other divinatory texts, and where it survives it offers the same advantage as the Daybooks do elsewhere: a record of what was actually in use at a datable moment, rather than a later compilation claiming antiquity.

International scholarship

• Richard J. Smith, Fortune-tellers and Philosophers — divination across Chinese society, placing physiognomy within the whole picture rather than treating it in isolation.
• Charlotte Furth, A Flourishing Yin — the body and gender in Chinese medicine, useful for understanding the conception of the body that medical inspection and physiognomy share.
• Johann Caspar Lavater, Essays on Physiognomy — the representative work of eighteenth-century European physiognomy. Read alongside a Chinese manual, it shows the same habit of thought in two independent forms, which is more illuminating than either alone.
• General histories of phrenology and criminal anthropology — necessary for understanding where this line of thinking led in Europe, and why the modern versions deserve suspicion.

Psychology and empirical testing

The literature that bears most directly on whether any of this works is not about China at all.

Bertram Forer's 1949 experiment is the starting point: a single vague personality description was accepted as accurate by nearly an entire class, the effect now called the Barnum or Forer effect. Work on thin-slice judgements examines what can genuinely be inferred from brief observation, and should be read together with the recent replication literature, which has tempered several of its claims.

Alexander Todorov's research on facial impressions shows how rapidly and how consistently people form judgements of trustworthiness and competence from faces — and is routinely misreported as showing that those judgements are correct. The critical literature on machine learning claims to infer criminality or character from photographs completes the picture, and the analyses of dataset bias in those studies are worth reading in full, because they explain a failure mode that will keep recurring.

Cautions for using this material

A few practical warnings apply to reading in this field specifically.

• Attributions are conventions. Books here are routinely credited to Chen Tuan, the Yellow Emperor or other legendary figures. This signals membership of a tradition; it says nothing about authorship or date.
• Separate description from recommendation. A great deal of what is published looks like history and is actually instruction.
• Treat undated claims sceptically. Material from the Republican period and later is frequently presented as ancient.
• Be wary of anything that promises to teach you to assess people, particularly in a professional context. This is where the practice does its current harm.
• Do not use any of this on named individuals, and do not upload photographs to services offering analysis.

The purpose of reading here is to understand a system that shaped a civilisation's literature and social life for two thousand years. That is a good reason to study it and not a reason to apply it, and the distinction is the one these pages have tried throughout to keep visible.

One last suggestion about order. Begin with the Xunzi chapter, then the scattered records in the histories, then one manual read straight through — the Shenxiang Quanbian serves — so that you see the categories in their own voice before anyone interprets them for you. Follow with the historical scholarship, and finish with the psychology, which will explain the sense of plausibility that reading the manual will by then have produced in you.

Taken in reverse, beginning with a modern popular guide, the usual outcome is that a set of social conventions from the Ming is mistaken for a set of observations about human beings. The order of reading does much of the interpretive work by itself.

Key points

• These pages draw on scholarly studies, primary Chinese texts and general reference works.

• Richard J. Smith is the natural English-language starting point for Chinese divination and popular culture.

• The Shenxiang Quanbian is the most famous classical compendium of physiognomic lore.

• The standard histories and the great novels show physiognomy at work in literature.

• Psychology-of-belief literature explains why readings feel convincing.

• Together these let a reader see the tradition as history, as text, as literature and as a set of claims.

Sources drawn on for this article include: the works of Richard J. Smith on Chinese divination and popular culture, the Shenxiang Quanbian and other classical physiognomic manuals, the Records of the Grand Historian, general scholarship on Chinese fiction and drama, and the psychology-of-belief literature.