# Imperial examinations of China: the making of the exam hall

**What were the imperial examinations of China, and what do modern exams still take from them?** The imperial examinations of China chose officials by written tests from the seventh century until 1905. Candidates were searched, shut in rows of cells and asked to write on set texts. From the Song dynasty their names were hidden and clerks recopied each script before several examiners read it. Modern exams keep the anonymity, the second reader and the sealed room.

Published 2026-10-10 by EdProcto. https://edprocto.com/guides/the-imperial-examinations-of-china-and-the-invention-of-the-exam-hall

For some thirteen centuries, by the usual count, the imperial examinations of China chose the men who governed an empire by what they wrote in a guarded room. Much of a modern exam hall was already there: the search at the gate, the numbered place, the script with no name on it, the second reader.

This guide tells that history from museum and library pages, two encyclopaedias and a few studies, each read on 10 October 2026. "Abstract read" marks a study whose abstract or index record was all we could open, and a later account is named as one.

[EdProcto](https://edprocto.com/), which publishes this guide, makes quizzes sat under a clock and marked on a server. The wider story is in [the history of proof in education](https://edprocto.com/guides/the-history-of-proof-in-education-from-the-viva-to-the-server-marked-quiz).

## When did the imperial examinations of China run?

From the seventh century to 1905, and the sources we read disagree on the first date.

- **Columbia University's Asia for Educators (n.d.-b):** a man could become an official by written examination from the Sui dynasty (581–617), and only under the Song did that become the normal way up.
- **Hong Kong's Leisure and Cultural Services Department (2011):** AD 605, under Emperor Yang of the Sui.
- **Theobald (2011),** in the encyclopaedia ChinaKnowledge.de: the examination system (keju zhi 科舉制) was created under the Tang (618–907) and became the prevalent route under the Song (960–1279). The Sui examinations he calls its germs. The Museum of the Institute of History and Philology (n.d.) in Taipei also begins it in the Tang.
- **Chen (2019),** of Princeton's Cotsen Children's Library, citing Benjamin Elman: 650 to 1905.

### Why the starting dates differ

They count different beginnings: an examination that existed, a system that ran regularly, or the point at which it became the main way into office.

The end is less disputed. The curriculum was reformed in 1901 and the Qing court ended the system in 1905 (Theobald, 2011); Helliwell (2014) dates the end of the provincial examinations to 1903 and of the metropolitan ones to 1904.

The run had gaps: Theobald reports a break in the tenth century, and ten years without examinations from 1372 under the first Ming emperor.

## What did the examinations test, and how did that change?

They tested command of the Confucian classics and the skill of writing about them, in forms that narrowed over time. This section follows Theobald (2011) unless another source is named.

- **Tang.** A classicist was shown phrases from a classic and recited the passage, passing with five right out of ten. The harder examination for presented scholars (jinshi 進士) covered the classics, history and official documents, and from the 740s poems.
- **Song.** In 1071 the classicist and specialist examinations were abolished, and everyone who wanted office sat the jinshi. The reformer Wang Anshi argued that poetry was no use for administration, and it was given up after the Southern Song. Interpretation of the classics, read through Zhu Xi's commentaries, replaced it.
- **Ming and Qing.** Questions came from the Four Books and the Five Classics, and answers took the form of the eight-legged essay (bagu wen 八股文): eight set parts in paired sentences, 550 to 700 characters long. Helliwell (2014) says the form became compulsory in the fifteenth century.
- **After 1901.** The first session asked about political history, the second about the governments of foreign countries, and the classics came third.

Under the Qing the ladder had four rungs (Museum of the Institute of History and Philology, n.d.; the characters are Theobald's): local examinations for the status of government student (shengyuan 生員), a provincial examination every three years (juren 舉人), a metropolitan one in Beijing (gongshi 貢士) and a palace examination before the emperor (jinshi).

## What was the examination compound like, and how long did a candidate stay?

It was a compound (gongyuan 貢院) of single cells (haofang 號房) in rows, and a candidate stayed for days.

The largest was the Jiangnan compound at Nanjing, used for provincial examinations. The University of Bristol (n.d.), whose archive holds photographs of it dated 1913, notes that up to 20,446 cells were in use in the late Qing.

As Cartwright (2019) reports, a cell had an open front, so that invigilators could see in, and a few plain boards that made a desk and a seat or a bed.

On the length of the stay the sources count differently. Theobald (2011) gives three sessions for the Qing provincial examination, on the 9th, 12th and 15th days of the eighth lunar month, and the Taipei museum says each session lasted one day. Helliwell (2014) says the examination lasted three days, all spent in the cell, and quotes Ichisada Miyazaki's book on the subject, which describes the cell as "like a prison without bars". We could not confirm the size of a cell or the nights spent in one.

The compound was closed to the world: at the Song metropolitan examination, Theobald writes, candidates could neither leave nor meet anyone, and the compound was sealed (suoyuan 鎖院).

## How did the examiners guard against cheating and favouritism?

By a chain of procedures, most in place by the Song, which kept the candidate away from help and the examiner away from the candidate's name. The steps follow Theobald (2011) unless another source is named.

1. **A search at the gate.** Helliwell (2014) calls the body searches humiliating. Cartwright (2019) quotes the writer Pu Songling (died 1715), for whom the candidate being searched is "just like a prisoner".
2. **A numbered cell and official paper.** Song candidates sat by number and wrote on paper stamped with an official seal.
3. **The name removed.** Tang chief examiners knew who the candidates were; names were masked from 702. Under the Song a sealing office (fengmiyuan 封彌院) took the candidate's details off each script and gave it a registered number.
4. **The script recopied.** A copy office (tengluyuan 謄錄院) wrote out each script in full, so that no examiner could know a hand. Qing clerks copied in vermilion ink (Museum of the Institute of History and Philology, n.d.).
5. **More than one reader.** A first examining official read the copy, then a second, then a chief or vice examiner.
6. **Examiners kept apart.** From 1037 an examiner could not serve in his home prefecture, a rule of avoidance (huibi 回避).
7. **Quotas by region.** Of a Ming metropolitan quota fixed at 300 in 1475, 165 places went to a southern group of provinces, 105 to a northern and 30 to a central one, so that, as Cartwright (2019) explains, the richer regions did not take every post.

How two readers reach one mark today is in our guide to [moderation](https://edprocto.com/guides/moderation-two-markers-one-standard).

## What cheating went on anyway?

Substitutes, bribes and smuggled texts. Helliwell (2014) lists hiring someone to sit the examination, bribing officials, who risked caning or execution if found out, and carrying cribs into the cell on paper, silk or clothing.

### The objects that survive

- **Bodleian Library, Oxford.** Helliwell (2014), recording a display there, lists a silk crib sheet in the form of a napkin (Sinica 6096) holding about two dozen model essays, and a tiny printed Four Books (Sinica 2103), the smallest block-printed Chinese book in the library's collections.
- **Princeton University Library.** Chen (2019) describes a strip of silk satin, 6 by 29 inches and folded in three, covered in what Chinese calls "fly-head script" (蝇头小字): eleven essays, and on the back one character, yi (義), rightness. The East Asian Library there holds a silk robe, dated between about 1840 and 1905, inscribed inside and out with 722 essays.
- **Shanghai Jiading Museum.** A replica of a linen vest from its collection, pictured for a Hong Kong exhibition in 2011, hides 62 eight-legged essays of more than 40,000 characters. Its caption adds that "cheats were caught frequently" (Leisure and Cultural Services Department, 2011).

### An object is not proof of use

As Chen (2019) reports, Andrew Plaks studied the Princeton robe for a quarter of a century and found that much of its text came from widely published collections of model essays, which makes copying from it into a script doubtful. Helliwell (2014) notes that it may have been worn as a talisman, and reports the view that a candidate able to use the silk sheet hardly needed it.

A candidate caught, Cartwright (2019) writes, was thrown out, barred from the next examination and sometimes stripped of the first degree.

## How many sat, and how few passed?

Many sat and very few passed, in every period for which we read a figure.

- **Tang.** Between 800 and 1,200 men sat the jinshi each year, and a dozen or a few dozen passed (Theobald, 2011).
- **Song.** The numbers from Asia for Educators (n.d.-b), which reach about 400,000 competing, are in our history of proof. The same page adds that men were on average a little over 30 when they succeeded.
- **Qing.** The metropolitan examination passed 406 men in 1730, its highest number, and 96 in 1789, its lowest (Theobald, 2011).

Most boys never entered at all, says Asia for Educators (n.d.-a): many left school after five or six years.

## What did the system do for, and to, Chinese society and learning?

It opened office to talent in principle, spread one curriculum across an empire and narrowed what was studied. Historians still argue over how far the ladder reached.

### Social mobility: a ladder with conditions

Asia for Educators (n.d.-a) calls the system "an important vehicle of social mobility", and adds that a youth from the poorest family could rise in theory. Only men sat. In Song times most scholars came from the landholding class (Asia for Educators, n.d.-b), and Theobald (2011) lists those barred, tradesmen and craftsmen among them.

The modern argument is known by the title of Ping-ti Ho's 1962 book The Ladder of Success in Imperial China, which we did not read. Jiang and Kung (2021) tested the ladder on records from 1796 to 1905. Ability and an upper gentry family predicted success in the jinshi; direct measures of wealth did not. In their words the system "unwittingly facilitated mobility" while passing on hidden advantages (abstract read).

### The narrowing of study

Because the examination set the curriculum, the curriculum shrank to fit it. Theobald (2011) writes that Ming schools became "just an annex to the examination system", and cites the scholar Gu Yanwu's verdict that the eight-legged essay was worse than the First Emperor's burning of the books.

The same fact had another face: a farmer's son who never sat an examination still learned to read from the texts the highest graduates had studied (Asia for Educators, n.d.-a).

## Did other countries take up the Chinese examinations?

In East Asia they did; for Britain the evidence is put more carefully than the claim usually is.

Liu (2007) writes that Japan imitated the examinations from the eighth to the tenth century, that Korea's were the longest and most comprehensive among the other East Asian countries, and that Vietnam was the last to abolish them (abstract read).

For Britain, Kim and Massey (2025) write that the United Kingdom "may have referenced" China's system when the Northcote–Trevelyan Report of 1854 moved it from patronage to competitive examination. Missionaries, merchants and diplomats had carried word of the Chinese system to Europe and books described it, which the authors call circumstantial evidence; they then searched parliamentary debates from 1853 to 1860 and report finding relevant evidence of Chinese influence (abstract read).

Stronger wording is common. Opening the 2011 Hong Kong exhibition, an official called the examination a forerunner of Western civil service recruitment; the page offers no evidence (Leisure and Cultural Services Department, 2011).

## What can a teacher recognise, and what should not be copied?

A teacher can recognise most of the fairness of a modern examination, and should copy none of the hardship.

| In the compound | The doubt it answered | In a school today |
| --- | --- | --- |
| A number in place of the name | Favour to a known candidate | Anonymous marking by candidate number |
| The script recopied by a clerk | A recognised hand | Typed answers |
| A second and a third reader | One examiner's taste or slip | A second marker and moderation |
| One paper for all, on set days | Scores that cannot be compared | A common paper sat together |
| A sealed gate, a search, a watched cell | Outside help and smuggled texts | An invigilated room; on a screen, an [EdProcto](https://edprocto.com/) quiz that arrives only when the attempt begins |

What should not be copied:

- **The confinement and the search.** They bought security at the candidates' cost. Supervision should fit the stakes and be announced beforehand.
- **One form for every answer.** The eight-legged essay made scripts easy to compare and teaching narrow.
- **Treating an object as a verdict.** The Princeton robe was long taken for a cheating tool and may never have entered a hall. A suspicious record is a reason to look, not a finding.

No safeguard ended the cheating: every one was in force while the crib sheets were being written.

## How do the same ideas appear in an EdProcto quiz?

EdProcto keeps the parts of the compound that protect fairness and none that confine the candidate.

- **The sealed paper.** Before an attempt begins, a student's device is sent the title, the rules and the count of questions. The questions arrive with the attempt, without an answer key: the marks are made on the server.
- **An order of one's own.** "Shuffle the questions for each student" and "Shuffle the options" are on from the start; see [what shuffling prevents](https://edprocto.com/guides/shuffling-questions-and-answers-what-it-prevents-and-what-it-breaks).
- **The watched cell, as a record.** Measures can be switched on one by one, among them "Full screen", "Copy and paste off", "Right-click off", "Keyboard shortcuts off", "Printing off" and "Leaving the tab". An event is a record for the instructor, never proof.
- **Scores held.** With "Show the score" at "When released", nothing of the score reaches a student until the instructor presses "Release the scores".
- **Names.** The grading page is headed with the student's name, so marking there is by name.

See [integrity](https://edprocto.com/integrity) and [pricing](https://edprocto.com/pricing).

## Quick questions

### When did the imperial examinations begin and end?

In the seventh century, under the Sui or the Tang depending on the historian, and they ended in 1905.

### Were the examination scripts marked anonymously?

From the Song, yes. A sealing office replaced the name with a number, a copy office rewrote the script, and more than one examiner read it.

### Did candidates really cheat with clothing covered in writing?

Garments and silk sheets covered in essays survive in libraries and museums. Whether the best known, at Princeton, was ever used in a hall is doubted.

### Does EdProcto mark answers without the student's name?

[EdProcto](https://edprocto.com/) heads its grading page with the student's name, so marking is by name. What it keeps back is the answer key, which is never sent to a student's device.

## References

- Asia for Educators. (n.d.-a). *The Confucian classics & the civil service examinations*. Columbia University. [https://afe.easia.columbia.edu/cosmos/irc/classics.htm](https://afe.easia.columbia.edu/cosmos/irc/classics.htm)
- Asia for Educators. (n.d.-b). *The Song Confucian revival*. Columbia University. [https://afe.easia.columbia.edu/songdynasty-module/confucian-scholar.html](https://afe.easia.columbia.edu/songdynasty-module/confucian-scholar.html)
- Cartwright, M. (2019, February 8). *The civil service examinations of imperial China*. World History Encyclopedia. [https://www.worldhistory.org/article/1335/the-civil-service-examinations-of-imperial-china/](https://www.worldhistory.org/article/1335/the-civil-service-examinations-of-imperial-china/)
- Chen, M. (2019, March 29). *Cheating in examinations for cheapskates? A centuries-old tip from the Chinese collection of the Cotsen Children's Library*. Cotsen Children's Library, Princeton University Library. [https://cotsen.blogs.princeton.edu/2019/03/cheat-sheet/](https://cotsen.blogs.princeton.edu/2019/03/cheat-sheet/)
- Helliwell, D. (2014, July 11). *Proscholium exhibition*. Serica: Some notes on old Chinese books. [https://serica.blog/2014/07/11/proscholium-exhibition/](https://serica.blog/2014/07/11/proscholium-exhibition/)
- Jiang, Q., & Kung, J. K.-S. (2021). Social mobility in late imperial China: Reconsidering the "ladder of success" hypothesis. *Modern China, 47*(5), 628–661. [https://doi.org/10.1177/0097700420914529](https://doi.org/10.1177/0097700420914529)
- Kim, P. S., & Massey, A. (2025). New development: Re-investigating the influence of China on the British civil service examination system. *Public Money & Management, 45*(2), 139–143. [https://doi.org/10.1080/09540962.2024.2334524](https://doi.org/10.1080/09540962.2024.2334524)
- Leisure and Cultural Services Department. (2011, November 8). *Exhibition to highlight Chinese imperial examination system (with photos)*. Government of the Hong Kong Special Administrative Region. [https://www.info.gov.hk/gia/general/201111/08/P201111080259.htm](https://www.info.gov.hk/gia/general/201111/08/P201111080259.htm) Photograph caption: [https://www.info.gov.hk/gia/general/201111/08/P201111080259_photo_1032950.htm](https://www.info.gov.hk/gia/general/201111/08/P201111080259_photo_1032950.htm)
- Liu, H. (2007). Influence of China's imperial examinations on Japan, Korea and Vietnam. *Frontiers of History in China, 2*(4), 493–512. [https://doi.org/10.1007/s11462-007-0025-5](https://doi.org/10.1007/s11462-007-0025-5)
- Museum of the Institute of History and Philology, Academia Sinica. (n.d.). *Archives of the Grand Secretariat*. [https://museum.sinica.edu.tw/en/exhibitions/32/](https://museum.sinica.edu.tw/en/exhibitions/32/)
- Theobald, U. (2011, November 26). *The Chinese imperial examination system*. ChinaKnowledge.de: An encyclopaedia on Chinese history, literature and art. [http://www.chinaknowledge.de/History/Terms/examination.html](http://www.chinaknowledge.de/History/Terms/examination.html)
- University of Bristol. (n.d.). *Examination cells at the Jiangnan imperial examination centre, Nanjing* (OH03-016). Historical Photographs of China. [https://hpcbristol.net/visual/OH03-016](https://hpcbristol.net/visual/OH03-016)