Reading Notes: What Grows Where the State Has Been — Chinese Protestantism and the Register | ygpgsgl

These notes are compiled from a three-day reading programme (29–31 July 2026) that produced over 550 usable findings, 49 method disciplines, and over 40 retractions. The primary sources are: Nanlai Cao, Constructing China's Jerusalem (Stanford, 2011; ethnography of Wenzhou, quoted from the English original); Han Heng's three survey papers (2012, 2014, 2016); Wu Fei, 麦芒上的圣言 (ethnography of a Catholic village in Hebei); Duan Qi's Shantou–Nanyang comparison (2014) and her 宗教蓝皮书 chapters; Li Huawei's Nanyang fieldwork; and twelve current regulatory texts. I have not done fieldwork. Treat this as a map, not a territory.

Why This Material

The question that started this was: why does Henan, the cradle of Chinese civilisation, have more Protestants than any other province? That question turns out to be wrong in a productive way. The answer is not about Protestantism. It is about what kind of institution can grow in a place the state has been administering for three thousand years — and what happens to institutions the state does not recognise as existing.

That makes it a case study in the same problem The Prepared Soil and the 天灾 essay address from other angles: the relationship between registration, legibility, and institutional survival. The material below is organised by what it contributes to that line of work, not by source or by region.

A word about the title these notes no longer have. They began as notes on Henan and are not that: Zhejiang is named more often here than Henan, and roughly half the text is about somewhere else entirely. That is what the question did when it was followed. But a title claiming Henan describes the province the question started in rather than the material that came back, so it has been changed.

1. A Variable Worth Testing

Core, periphery, buffer

The literature offers many explanations for why Chinese Christianity looks different in different places — missionary history, terrain, the destruction of popular religion, regulatory intensity. The variable that survived the reading better than the others — though it does not explain every case, as the Wenzhou exception below makes clear — is: how the state treated the region over the long run.

Three positions. Core: the state is built on it, extracts from it, invests in it. Periphery: inside the map, outside the attention — neither extracted from nor invested in. Buffer: inside the attention, but as an instrument for protecting the core. The Central Plain is core. Wenzhou is periphery. Hebei is buffer.

The span is long — tenth-century hydraulics to congregations meeting last year — and the intervening steps are not all shown. A candidate variable that survived several attempts to kill it, then, not a demonstrated thousand-year chain.

The sentence that matters: it is not that the state cannot reach, but that it can reach and does not go. The Wuyue kingdom (907–978) — thirteen prefectures, roughly one province — built all its hydraulic works around the capital at Hangzhou. None went to southern Zhejiang. A state the size of a province allocates exactly as an empire does: toward the capital and the grain land. Core and periphery are a function of attention, not of distance.

An obvious hypothesis — south equals lineage waterworks, north equals state waterworks — collapsed on first contact. Wuyue organised the Lake Tai polder system through state power, which is why it was the more durable one. Southern waterworks were also state-organised where the south was core. The variable is not latitude but position. "The south" is not a unit.

What being core cost

Four costs, stacked.

The intermediate layer is thin. The plain is flat, so administration reached everywhere. Population swelled without matching resources, so people migrated constantly and durable single-surname lineages could not form. Dry-field agriculture requires little coordination, so a household can complete its cycle unorganised. And the lijia and baojia systems took over directly the functions lineages perform elsewhere. Both were registration systems. Where they operated thoroughly, the state was not merely present; it was the entity that recorded who existed.

The centre departed. In the thirteenth century the political axis moved east with the rerouting of the Grand Canal, taking the capital and the accumulated wealth. What remained was the population and a floodplain.

The floodplain is an engineering artefact. A breach in the Yellow River dikes is the failure of a state project — one that, across two millennia of dike-raising, progressively increased the cost of its own failure. The two largest breaches, 1128 and 1938, were cases of deliberate exploitation of that accumulated cost. This is the same structural shape as the tianzai essay's finding: infrastructure that produces the catastrophe it was built to prevent.

What had been there was cleared — and cleared incompletely. The process is usually compressed into one word, "the suppression of Buddhism," which conflates two operations across three periods. 1928: the campaign converting temples into schools — an attack on property. 1949–53: the suppression of the redemptive societies — an attack on organisation. These are different things. And the premise that popular religion was swept clean has counter-evidence from the same author's same fieldwork: Nanyang has 147,817 Daoists (1.4 per cent of population) against Kaifeng's 4,700 (0.09 per cent) — a factor of roughly fifteen in population share, and over thirty in absolute number. Whether Henan's popular religion was "swept clean" depends entirely on the comparison chosen.

The decisive counter-case is Shantou, where the institutional religions were hit harder, not more gently: the campaign against the redemptive societies there executed more than a hundred people — the source names the 老母会 specifically — and Daoism is described as never having recovered. Shantou's Protestant share is nonetheless 0.8 per cent against a national average of 1.8. The reason given is that roughly seventy per cent of the popular-deity temples came through: what was destroyed was the organised societies, and what survived was the layer beneath them. What withstood the blow had a corporate form. The useful move is to shift from "who struck" to "who could take the hit."

Wenzhou is the exception, and the exception must be stated

Wenzhou occupies the peripheral position on every structural measure. But that position does not explain its Christianity. Within Wenzhou, believers are more concentrated in the more economically developed areas — the opposite of what an abandonment-and-substitution mechanism would predict. Zhejiang's other two Protestant concentrations, Xiaoshan and Cixi, are neither remote nor mountainous but industrial zones. And Lanxi, the maximum of state penetration within Zhejiang, is not among the province's Protestant concentrations — which also disposes of the broader claim that Zhejiang's terrain explains weak state penetration.

The axis holds at the level of position and not at the level of explanation. Wenzhou's Christianity operates on a different axis — class — and these notes treat it separately in section 5.

2. Three Forms with Evidence, and Several Without

The same doctrine, arriving in different social conditions, produced congregations that share nothing but the book. No doctrinal dispute separates them; the variation is in everything except doctrine — composition, property, authority, gender arrangement, the route by which people arrive, and what the congregation can do if the building is taken away.

The reading programme turned up several distinct forms. Three of them have enough material behind them — fieldwork, surveys, mechanism literature — to describe with confidence. The rest are attested but thinly documented: a growth curve here, a single interview there. This section treats the three that earned their description. The others are noted at the end for what they are: leads, not findings. The number of forms is whatever the material happens to contain; it is not a typology drawn from a sampling frame.

The substitution form — rural Henan and Anhui

A farmhouse, a Bible, a person who can read. The formal clergy structure is almost absent: in Han Heng's 2016 survey (a convenience sample — 32 seminary students reporting on their home congregations, not a random draw), the share holding the office of pastor was 0.6 per cent in villages and zero in townships, with over ninety per cent being ordinary preachers. These figures indicate direction, not magnitude. There is no property and no address that matters.

The entry route is illness. In the Anhui survey, over eighty per cent of believers came to faith because a family member fell sick. The Anhui curve makes the shape visible: 350,000 believers in 1949, 13,000 by 1966, 141,000 by 1979–82. Whatever this is, it can be reduced by ninety-six per cent and return elevenfold.

Who is in the room follows from who is in the village. Men leave for migrant work at higher rates than women. A congregational leader in a migrant church outside Beijing (Liu Qi and Huang Jianbo, fieldwork 2004–05) put the mechanism without prompting: the men collecting scrap "are basically out all day, so very few keep the Sabbath… Sisters are different. Sisters stay home all day, nothing much to do" (laughing). The gender composition requires no assumption about women's religiosity. It requires a labour market and a bus timetable.

From the same 2016 survey: only 9.8 per cent of 174 congregations ran an enquirers' class — the institution whose whole purpose is receiving newcomers. (This is the one item in that chapter whose denominator is unambiguously N=174.) Fewer than one in ten of these congregations has a front door.

And when these people move to the coast, the form travels with them but mutates. Churches near Wenzhou's industrial zones turn on every exterior light and set off fireworks before meetings; the noise and heat remind arrivals of a temple fair at home. Many cannot understand why there is no admission fee. The church is where job information circulates. Preachers reach migrants through "scientific witness." Migrants call the preacher "teacher" and call attending church "going to class." To track a membership that moves constantly, congregations began issuing baptismal certificates with a photograph. A sealed, numbered document, issued to the people with the fewest documents.

The replication form — Wenzhou

This form does not fill a hole but reproduces the surrounding class structure: male, elite, propertied, seating separated by sex, authority flowing from commercial success. Nothing here resembles a farmhouse. A downtown congregation occupies fifteen hundred square metres inside an office tower and leases eight hundred of them to a garment company, taking 249,000 yuan in rent in 2005; the members regard the property as an appreciating collective asset. It requires a pre-existing merchant economy and local elite, and the material behind it — Cao Nanlai's nineteen-month ethnography — is treated in section 5.

The inheritance form — Catholic Hebei

People are Catholic because their village has been Catholic for four hundred years. This form answers a different question from every other: not "under what conditions do people change belief" but "how does a belief persist for four centuries without persuading anyone." Its answer is that it never presents itself for decision.

The material is Wu Fei's ethnography of a village he anonymised as "Duanzhuang" — fieldwork in 1996–99, published as 麦芒上的圣言. Its contribution is not about Catholicism as such but about two different structures of religious authority — one vertical, where salvation requires clergy and credentials can be issued and revoked, and one horizontal, which section 3 takes up because it is what the licensing apparatus fails to catch.

One detail from Wu's fieldwork: the local measure of religious standing is not "piety" (虔诚) but "热心" — a word that is behavioural, observable, and collective, with no component referring to inner belief. The criteria: goes to church often, attends mass often, takes an interest in church affairs, observes the commandments, does not marry outside the faith. Wu caught himself contaminating the vocabulary: "I always liked to use the word 'pious,' and they could understand it, and would follow my usage. But in other contexts they rarely used the word." The respondents adopted the researcher's term in conversation and dropped it when he was not there.

Forms observed but not deeply researched

The following are attested in the material but lack the fieldwork, survey data, or mechanism literature to describe with confidence. They are listed here as a record of what exists, not as findings.

The Jesus Family (Shandong, 1921–1952). Communal residence, common property, no pastors, a "family head." The growth curve — fourteen locations in 1937, sixty-four under Japanese occupation, 127 by 1948, dissolved in 1952 — tracks the inverse of state capacity. But why Shandong produced communal negation rather than the kind of elite replication seen in Wenzhou, no one has studied. The material is one chapter of a book not fully read.

Shimenkan and the Nujiang (Miao, Lisu). Samuel Pollard's and James O. Fraser's independent, isomorphic sequences — failure among the Han, creation of a script for a subordinated minority, mass conversion — are striking and well-attested in outline. But the material read was secondary summaries, not primary ethnography. What is documented is the act (giving writing to a people who had none) and the scale (sixty thousand believers by 1915 among the Miao; over 100,000 among the Lisu by 1995). What is missing is the internal structure of these communities and what happened to them after 1949.

Early Rain (Chengdu). The confrontation form — refused the grey market, attended worship in formal dress with name badges, lasted thirteen years, ended with a nine-year sentence. Documented from news sources and legal records, not from ethnography. What the congregation looked like from the inside is not in this material.

The Little Flock (Fujian). The only indigenous Chinese denomination that became worldwide. Watchman Nee refused release in 1967 and died in a labour camp. The domestic branch was destroyed; the overseas branches survived. Documented from biographical and church-historical sources.

Korean-Chinese congregations (Yanbian). Faith travelling with an ethnic group, worshipping in Korean, with 102 schools built. One paragraph in the material; no mechanism literature.

These are not nothing. Several of them — Shimenkan in particular — could be the subject of their own research programme. But they are not the subject of this one, and describing them as if they were would be claiming evidence that does not yet exist.

3. What Registration Does

The constitutive claim

Consider an unregistered building. It has no title deed and therefore does not exist in law. One may live in it, but one may not sell it, mortgage it, or assert anything about it as its owner when it is demolished. An unregistered congregation is in a comparable condition: it cannot assert rights in its collective religious identity on the same terms as a registered legal subject.

On the common-sense account, registration records something that already exists. On the account that emerged from this reading, registration is the procedure by which the object acquires legal reality. The Three-Self church is therefore not a substitute for the "real" church. It is the only form through which a Protestant congregation can acquire recognised institutional standing inside the religious-affairs system — which is not the same as being a single legal person, since the system distinguishes religious bodies, activity sites, and sites that have obtained legal-person status.

The immediate objection — that unregistered house churches obviously exist — is correct, and forces the load-bearing correction:

Legal non-recognition purchases not certain destruction but widened discretion.

This needs stating carefully, because the strong version is not true. Unregistered congregations are not outside administrative law: the 2017 Regulations on Religious Affairs provide for unauthorised premises to be closed and unlawful income confiscated, and for activity at unapproved locations to be halted and fined. Members, premises, and the person in charge all remain reachable, and enforcement can leave decisions, penalties, and a route of appeal behind it. What non-registration does is narrower and more durable than lawlessness: it widens administrative discretion and weakens the congregation's ability to compel a procedure in its own name. A registered counterpart has standing to be dealt with as itself; an unregistered one is dealt with through its members, its landlord, and its building. This is why the unregistered condition is stable rather than terminal — and why it is frequently preferred by both parties.

Grey as an equilibrium

The most unexpected finding. A Wenzhou congregational leader had sought to register, and the religious affairs bureau refused. He then obtained informal recognition through commercial relationships. He is not evading registration. He was denied it and found a substitute.

On the other side, many older believers reject state involvement in church affairs on principle. For them the absence of a certificate is not a vulnerability but the point.

Neither party wanted the certificate, though for opposite reasons. One because he could obtain the substance without it. The other because the certificate is what they fear.

Set against this, Early Rain demanded exactly what the Wenzhou congregation was denied and the older believers refused. The result was nine years. One party actively sought legality and was eliminated; one actively declined it and persisted. That pair is difficult to accommodate in an account where regulation simply presses and religion simply resists.

The apparatus was installed on the wrong chain

The licensing regime — qualification requirements, seminary coursework, filing and certification, and where it operates, periodic assessment — acts on clergy. On the chain that carries actual weight in an unregistered congregation, authority is conferred by the congregation, evidenced by efficacy, and accumulated from below. The criteria are observable: the length and vocabulary of prayer, emotional intensity, "the atmosphere." Legitimacy is settled by whether prayers "work" — 灵验. The ultimate credential: "if Brother C goes out to drive away demons, it will certainly be effective."

There is no certificate to revoke, because none was ever issued. This explains why Catholicism could be constrained through its clergy — its clergy were the transmission mechanism — while the same clerical apparatus failed to reach the house church’s actual chain of authority. The licensing gate does not catch the load-bearing chain, not because enforcement is weak, but because the credential the gate administers does not exist in this form.

A Catholic village makes the contrast vivid. The huizhang (lay leader) is a position created by registration requirements: "In the government's files there must be a person in charge for each religious activity site, each parish must have a political consultative conference representative, so there must be a fixed leadership core. And so each parish must elect a huizhang." The election "makes no difference" — the same people lead before and after — but it gives the group a hook the state can grab. In the Protestant villages, there is no hook. The state can only count heads.

What congregations brought to the counter

If registration is the currency of legal existence, some congregations could purchase procedural obstruction and others could not. Wenzhou's downtown congregations had property registered in private names (requiring a civil action to remove), leadership seated on political bodies (requiring a political calculation), and enterprises that were significant taxpayers (requiring an economic one). The Henan meeting points possessed none of the three. There was nothing to convert into procedure.

The difference in outcome between the 2013–16 cross-demolition campaign in Zhejiang and the same measures from 2018 in Henan does not require a difference in enforcement intensity. What differed was whether the object of enforcement could compel a procedure.

The currency was replaced

But that ability had a shelf life. In 2014 the stated ground for demolition was "illegal construction" — a building category: it has dimensions, admits counsel, and can be argued against. By 2026 the stated ground was "refusal to fly the national flag" — a political declaration: it has no dimensions, admits no counsel, and cannot be argued. When "compliance" moved from building specifications to political allegiance, the assets that had bought procedural obstruction ceased to buy it.

The timeline: 1,700 crosses demolished by July 2015. In 2020, two churches in Yongjia that believers had defended in 2014 were demolished. In 2021, two churches in Longwan whose crosses had been rebuilt after the 2014 demolition were demolished again by crane at night. In January 2026, Yayang church was demolished, reportedly for refusing to fly the flag.

What the resistance of 2013–16 saved was taken back item by item once the scoring changed.

4. The Apparatus

Twelve texts, one model

Twelve regulatory texts govern the field. Ten were issued or revised after 2018 — a temporal fact, not a causal claim. They share a shape: the texts function less like a single decision procedure than like a stockpile from which several grounds of action are available for the same situation. What can be observed is the availability, and that the ground chosen varies while the outcome does not; the order in which decision and citation actually occur is not something these documents show.

Five gates, each conditioned on the one before it: entry to seminary, the 30 per cent public-affairs curriculum there, clerical rank, teacher qualification, clergy certification. Each level couples to the next; revoke any one and everything downstream fails. A sixth — assessment twice a year, with interviews, sealed ballots, cross-department file access and a grade entered in a personal record — is documented here in a Shenyang municipal measure whose direct force stops at that city, and is set out below rather than counted among the gates. Whether it is general practice is not something this material can establish; what it shows is the shape such a regime takes where one exists.

Two tracks: the conduit track (registered churches, shaped and audited) and the prohibition track (everyone else, punished). The Henan form is not inside the conduit.

Each form is attacked at its constitutive feature

This pattern deserves its own line because of what it implies about how the apparatus reads its objects:

Wenzhou = churches and crosses (the scoreboard of capital converted into religious authority) → demolished
Henan house churches = cross-village networks → one person per site, cross-county filing required
Hereditary transmission — which is not a Catholic peculiarity; Wenzhou's Protestant congregations run on it too, and Cao's own account of why the faith is now indigenous there is that believers inherited it from parents and grandparents → minors barred from participation, with non-religious pledges signed by parents. Documented for Zhejiang from around 2017; how far it is enforced elsewhere, this material cannot say.
Catholic episcopal appointment, a legitimacy source independent of the state → moved into a 2018 provisional Sino-Vatican agreement, renewed every two years since. The text has never been published. What can be said is that the question was moved into a bilateral instrument — not what the instrument concedes, and by whom.
Yanbian = cross-border personnel movement → deportation and entry bans
Shimenkan = education → the site is now a museum
Chengdu = public theological and constitutional discourse → inciting subversion

The minor-participation ban is the sharpest of these, because it targets reproduction rather than presence: there is no need to demolish a church if the congregation cannot replenish itself. The state lost the fight over the crosses and won the one over inheritance. Note also what the second and third entries do to the typology in section 2 — hereditary transmission turns out to cut across the forms rather than to distinguish one of them, which is part of why the nine-form scheme collapsed.

What does not need a decision about religion

Some outcomes require no religious decision at all.

A village clinic that would have been the Catholic parish's primary point of external contact was not closed by religious authorities. It was denied a permit by the county health bureau on the ground that the village already had enough doctors. The decision was internally rational and had no religious content whatever.

Fire codes, building codes, and urban-village renovation campaigns sweep through religious and non-religious buildings equally. The enforcement tools are general-purpose: the five laws cited in Henan's joint-enforcement provisions are a fire-safety law, a building code, a land-administration law, a planning law, and a public-security law. None is about religion.

Economic charges are a universal handle: four cases span Christianity and Buddhism. And the tool generates its own revenue — confiscation of illegal income has no ceiling. In 2024 January–September, non-tax revenue rose 13.5 per cent while tax revenue fell 5.3 per cent.

The sharpest formulation comes from a religious-affairs official himself, in a conversation over dinner recorded by Duan Qi: "The approval requirements for Buddhist and Daoist temples are high, and there have been multiple rounds of cleaning up small temples that supposedly don't meet the standard. But for Christian meeting points the requirements are low — there's really not much difference between a church and a meeting point, but as a meeting point it doesn't have to go through the provincial level for approval, which gives Christianity a lot of room to develop."

5. Wenzhou: What It Looks Like When There Is Something Underneath

The material here is Cao Nanlai's ethnography — nineteen months of fieldwork in 2004–06, about seventy life histories. His subject inverts the usual one. Where the field normally studies the elderly, rural, and marginal, his subject is the economically powerful, politically connected Christian entrepreneur: the "boss Christian."

Weber, used as evangelistic material

Local preachers invoke Weber's thesis on Protestantism and capitalism from the pulpit as a reason to convert. "Elective affinity," which Weber hedged, is taken as intrinsic: Western capitalist modernity is a consequence of Christian faith, therefore to be modern is to be Christian. If one wants the extreme form of borrowing prestige, this is it: the borrowed language is not management jargon but Weber himself, recycled into a sermon proof that joining makes you modern.

Not cover, but convergence

Seating is separated by sex and strictly enforced. A young preacher's reasoning: women are emotional and sinned first; brothers are rational, because man was created first; woman is the hands and feet, man is the head. Cao's reading: there is a "striking parallel between traditional Chinese patriarchal ideology and conservative evangelical ideals — both stress female submission, subordination, chastity, piety and domesticity." The modern discourse does not conceal the traditional core. It ratifies it. An imported religion and a local patriarchy arrived independently at the same gender ideology, and they fit without a gap.

Martha became a job

Women do the cleaning, cooking, and hospitality, and are called Marthas. The role is institutionalised in nearly every Wenzhou church. Then the detail: the Marthas are busy throughout the service and so often cannot concentrate on the sermon; attending an entire service from beginning to end is, for them, a luxury. The people who make the service possible cannot attend it.

The hinge between forms

Two million migrants from Henan, Anhui, Hubei, Jiangxi, and Sichuan work in Wenzhou. They arrive from the substitution form. Before Wenzhou they took Christianity to be superstition for illiterate old women. In Wenzhou they discovered that the richer and better educated a man is, the more likely he is to believe. Male migrants who were ashamed of the female-led churches at home accept the Wenzhou version, because it reads as elite, rational, and male.

One young woman from rural Anhui had heard her grandmother praying at midnight before an exam and thought the old woman was talking to the air; after months of Bible study in Wenzhou she came to admire the same commitment. A man from Henan, ten years in the city: "last year I didn't even know Jehovah was God."

They rejected one form and converted to the other, and they converted to the second because it is not the first.

The mission arm and its economics

Wenzhou money goes inland: entrepreneurs organise mission work with strategic planning, using the existing commercial network to survey conditions first — "no different from ordinary market research." A cross-church mission body founded in 1989 counted among its aims building a mission network of Wenzhou merchants nationwide.

Both sides of the arrangement

The arrangement requires maintaining the boundary between urban civiliser and migrant to be civilised. Local leaders hold that evangelising locals has little effect — locals do not believe because "their hearts are hard." God brought migrants to Wenzhou as part of a "kingdom plan," so that Wenzhou Christians no longer need to travel inland. Migrants are described as simple, humble, with teachable hearts. A single male migrant preacher from Guizhou whose mother told him not to forget who he was: the church's training course fulfilled his university dream — more satisfying, he said, than getting into Peking or Tsinghua. But local believers repeatedly questioned whether he had truly received a vision. Those doubts "deeply hurt" him.

Cao's judgment: most migrants' hope of thoroughly remaking themselves may never be realised. The same institution does two opposite things at once. It gives dignity, education, a sealed document and a public space, and positions them as objects of instruction and limits their possibilities.

6. How Numbers Lie

This section exists because every quantitative claim in this literature has at least one thing wrong with it, and the wrongness is systematic.

Three definitions of "believer"

"How many Christians are there in Henan?" has no single answer because "believer" has at least three incompatible operational definitions: registration (the religious affairs bureau counts only registered baptised members), baptism, and attendance (in rural areas there is no registration procedure, so attendance is operative). Among women attending regularly in one county town, the baptised share was 44.4 per cent in one district and 28 per cent in another. The available precision about Wenzhou and the irreducible vagueness about Henan are not an artefact of research effort. They are a consequence of the argument. What entered the books can be described; what did not, cannot.

The same author, three reliabilities

Han Heng published three studies from the same research grant. The 2012 paper (journal article, 14 villages, N=170) declares N separately for every table; all seven tables close exactly when denominators are back-calculated. The 2014 paper (journal article, 19 villages, N=262) likewise declares N per table and gives raw frequencies. The 2016 chapter (blue-book contribution, 174 churches) declares N once in eleven pages while its items rest on six different denominators; its introductory sentence — "of the 174 churches, rural churches account for 65.1 per cent" — has no integer solution at 174; and it states in its abstract, a section heading, and its conclusion that the proportion of church workers under thirty is higher in villages than in cities, with no figure anywhere in the text supporting the comparison.

Trust by venue, not by author. Journal articles from the same person, same project, are clean. Blue-book chapters are not.

What a comparison group does

"Rural Christians are mostly illiterate" appears at least four times in this literature without a comparison group even once. Han Heng's 2014 study gives an illiteracy rate of 28.6 per cent among 262 rural Christians. The 2010 census gives Henan's rural illiteracy rate at 7.01 per cent for ages 15 and over. That is 4.08 times, and it looks conclusive.

It is wrong. Of those 262 people, 29.4 per cent are over sixty and 83.2 per cent are women — and illiteracy depends heavily on both. Rural women aged 15–29 are 0.93 per cent illiterate; rural women over sixty, 33.43 per cent. Comparing an old, mostly female sample to the whole rural population charges age and sex to the account of belief.

Standardised cell by cell on age and sex: expected rate 12.5 per cent against observed 28.6. The claim holds — standardised, still 2.3 times — but the crude comparison overstates by roughly double. The citable figure is 2.3, never 4.08.

A table that does not exist

While retrieving census data, a search tool returned: rural illiterates, Inner Mongolia, 6,753,661 persons, rate 27.67 per cent, men 27.49, women 27.85. The numbers are on the page, but the table is "registered population by sex and region" and the figures are Inner Mongolia's population absent six months or more. The table does not exist. The numbers looked entirely plausible — right magnitude, male and female rates close, correct format. The only way to catch it was to fetch the original and decode it cell by cell. Every figure in these notes comes from an original table or text.

7. What Both Sides Actually Refused

One of the research programme's most unexpected findings: what both Protestants and Catholics refuse is not the state but the intermediary the state created.

Underground Catholic priests: "They are very willing to register directly with the government, rather than register with the Patriotic Association… The Patriotic Association is just a lay believers' organisation; under no circumstances can it be placed above the Church… It is a lay organisation placed above the Church. No Catholic canon law provides a basis for this."

Protestant house-church leaders (four or five, in Fangcheng): "They have never been anti-government, even when the government suppressed them. Their only objection is to the Three-Fixed policy, which they regard as incompatible with the Great Commission."

Both sides reject the intermediary. Both give doctrinal reasons. Both explicitly say they are willing to face the government directly. This inverts the model in which the state manages through religious associations while the managed party evades the state. What is evaded is the association.

And a Three-Self leader who was compelled to denounce the faith lost all authority — because in a system where legitimacy rests on congregational recognition, official appointment cannot compensate. The state's management tool is self-cancelling: using it destroys the authority it was meant to co-opt.

8. A Chinese God

There is a standing claim that all this is foreign infiltration. It fails from several directions. By 1949 the whole of Henan had about fifty thousand believers after seven decades of missionary presence — the missions failed. Growth followed rejection: the Boxers swept Henan in 1900 and missionaries fled, yet the independent churches grew in those same provinces within two decades. And the social-position evidence of section 2 disposes of the question from the demand side.

But the direction that settles it is not about Christianity at all.

Huang Daxian. A Daoist deity born in Zhejiang in the year 328. The Song court canonised him. The temple that housed his cult was called the finest Daoist temple south of the Yangtze. Today the Wong Tai Sin Temple in Hong Kong covers eighteen thousand square metres; at his birthplace the buildings standing now were reconstructed in 1996, and one source describes his influence in contemporary practice as "relatively small."

A Chinese god, born in a Zhejiang village in 328, received the same treatment. The question was never about foreignness.

And the obvious formulation — inside the country every religion is managed into a scenic area, outside it stays alive — is false. The 1996 reconstruction was funded by Hong Kong's Sik Sik Yuen and registered as a site for religious activity. It was never a scenic area. What is actually the case is sorting: the same god, within one county-level city, carries four simultaneous legal identities — site for religious activity, national intangible cultural heritage, seventy-year residential land use, and scenic area. Registration does not record what a thing is. It assigns what a thing is.

9. What Connects to What

To The Prepared Soil

That essay argued that China's failure to produce religious institutions independent of the state was produced rather than accidental. Henan's house-church networks are the strongest available counter-example — independent, numbering in the millions, possessing their own authority structure. But the counter-example may confirm the thesis. The networks grew after repeated state failure, and in a thoroughly de-institutionalised form. Yang's mechanism holds: independent institutional religion does not grow in China under normal conditions. It grows in the cracks left by state failure — and must give up the form of an institution in order to survive.

The Japanese comparison extends. The Prepared Soil treated Tokugawa's terauke seido — temple certification — as absorption: Buddhism was institutionally preserved and spiritually hollowed. The point that essay did not make is that the instrument was registration itself. Japan issued the certificate universally; China withholds it selectively. Same instrument, two settings. Japan's version preserves the institution and hollows its content. China's version leaves the content untouched and denies it recognised institutional standing. Japan's costs a clerical payroll. China's costs nothing and purchases discretion.

To the 天灾 essay

The closure of the interpretive loop — Heaven judged the ruler, and the officials who read Heaven's signs were the ruler's employees — reappears here at regional scale, since to register is to accept the state's description of what you are. What that essay lacked and this material supplies is a case where the loop failed to close: Cao argues that Wenzhou's dialect barrier weakened the central state's capacity to act as interpretive authority for local society at all.

To Lanwei

The Yellow River dike system is a lanwei structure: infrastructure that cannot be completed, whose incompletion generates the next obligation, and whose failure cost rises with every increment of construction. 花园口 is the extreme case — a deliberate exploitation of accumulated incompletion.

10. What Is Missing, and What Was Learned About Looking

No fieldwork. Everything here is secondary.

No distribution map. The county-level distribution of Henan's Christians is not in the public domain. If needed, it requires local gazetteers or field surveys.

No explanation for most of the forms observed. Only three have mechanism literature behind them; the rest are described in section 2 and explained nowhere.

The registration thesis is an insight, not a demonstrated argument — and it is nonetheless in the title, which is worth saying plainly rather than hoping no one notices. It is the through-line because it is what made the rest of the material legible, not because it has been proved: the direct evidence for it is one county-level relocation case, one leader's interview, and one blue-book sentence with no N. The parts that carry more weight are the ones where it predicts something checkable — that the licensing gate should miss unregistered congregations entirely, and that a Daoist deity should be sortable into four legal identities at once.

Wenzhou after 2018 is under-documented. The material covers Wenzhou through Cao's fieldwork (2004–06) in rich detail and through news reports thereafter in thin detail. The transition from building-code enforcement to loyalty-declaration enforcement is attested but not ethnographically described.

The hymn corpus. 2,246 songs, sung in unregistered congregations for thirty years, written by a woman with one year of secondary education who cannot read notation and holds no credential. The licensing apparatus has no purchase on her because there was never a certificate to revoke. This is the limit case of the argument, and these notes do not treat it — it is treated separately, in Reading Notes: The Canaan Hymns, because it is the one part of this material that required leaving the literature and counting something directly. Those notes are the only place in the three days where the finding rests on primary-source analysis rather than on reporting, and folding them in here would have put one section on a different evidentiary footing from everything around it.

The rules that came out of it

Fifty accumulated across the reading, but most are already at work above rather than stated — trust by venue is what section 6 does to Han Heng, composition control is the illiteracy standardisation, and compared to what? is the question that produces the Shantou counter-case. Listing those again would only describe the article to itself. These five appear nowhere in the body, and they belong here rather than in a section of their own, because each one is a rule about not claiming more than you have:

"Pointed out" requires a sentence. If you cannot point to the sentence in the original, delete "pointed out."
Capability is verifiable; intent is not. "The state intended to…" becomes "the state demonstrated the capability to…" or "the result was…"
Statutes are inventory, not reasons. A regulation found in the stockpile after the fact is not the reason for the action.
"Cannot find" is not "does not exist." Before writing "structural absence," list where you have looked.
Do not de-anonymise. If the author anonymised the site, following the clues back is an act of direct opposition to that protection.