Editorial standards

Every substantive article carries a Sources & further reading drawer at its foot. It is collapsed by default — a reader should meet an article as prose, not as a bibliography — but it is always there, and it names the works the page rests on.

What the archive accepts as a source, in rough order of preference:

• Peer-reviewed scholarship — academic monographs and journal articles, preferred for any contested claim.
• Primary documents — statutes, inscriptions, manuscripts, contemporaneous accounts, named and dated.
• Reference works of established standing, used for uncontroversial framing rather than for the substance of a disputed point.
• Recorded oral testimony, labelled as such and never silently promoted to documentary fact.

What it does not accept as the basis for a factual claim: unsourced popular history, content generated by a language model, social-media assertion, and any source whose provenance cannot be established. Where the archive is summarising a tradition rather than asserting a fact, it says so with a label.

Two independent systems run across the archive, and they answer different questions.

Knowledge-type labels answer what kind of claim is this? — whether a passage rests on scholarly consensus, on competing readings of the evidence, on folklore, on mythology, on religious tradition, on oral history, or on a modern political narrative. The seven labels are set out in full, with what each one means, on the About page.

Confidence levels answer a different question: how sure are we? They sit on a three-point scale and are stated in plain words rather than implied by tone.

• Well-established — strong, corroborating evidence; little serious dispute.
• Probable — supported by evidence, but with gaps or some scholarly disagreement.
• Contested — thin, conflicting, or heavily debated; treat with caution.

The two are orthogonal. A folklore passage is not thereby low-confidence: the archive can be well-established about what a tradition holds while making no claim at all about whether the events described took place. Reading the label and the confidence together is the point.

Competing interpretation Where historians genuinely disagree, the archive presents more than one credible reading and names who holds which. It does not adjudicate scholarly disputes, and it does not present a majority view as though it were unanimous.

Punjabi on this site is written in two scripts — Gurmukhi and Shahmukhi — alongside a romanisation. A great deal of it is drafted before a native reader of the relevant script has checked it.

Rather than publish that silently or withhold it until some indefinite future, the archive marks it. Passages awaiting review carry a data-review="pending" flag in the page source. The material is readable, and it is honest about its status.

One rule follows from this and is enforced automatically: unreviewed script never becomes the name of a page. A pending romanisation or spelling may sit in a page body where it is marked; it is never used in a page title, a meta description, or structured data, because those are the strings search engines and other systems index and repeat, stripped of the marking that made them honest.

Where a Punjabi word matters, the archive aims to give: the word in ਗੁਰਮੁਖੀ, the word in شاہمکھی where appropriate, a romanisation, pronunciation guidance, a literal translation, the intended meaning, and the cultural context. Not every entry has all of these yet; the ones that do not are incomplete rather than finished.

The romanisation is a reading aid, not a standard. It is meant to let a reader who knows neither script say the word approximately right. Two conventions are worth stating because they are easy to misread:

• Nasal vowels are written with ṅ — trinjaṇ, ghoṛīāṅ — rather than with a tilde over the vowel. The tilde form is common elsewhere; it renders unreliably across fonts and breaks text search, so this archive does not use it.
• Long vowels carry a macron (ā, ī, ū) and retroflex consonants a subscript dot (ṛ, ṇ, ṭ, ḍ). These are not decorative.

Both scripts are marked in the page source with their language and, for Shahmukhi, their direction, so that screen readers and translation tools handle them correctly rather than guessing.

Punjab is divided by an international border and has been since 1947. The archive covers the whole region and the whole inheritance: East and West, Gurmukhi and Shahmukhi, Sikh and Muslim and Hindu and Christian, those who stayed and those who left.

In practice this means several standing commitments:

• Place names are given in both their historical and modern forms, with the modern state named, so that a reader can locate a place that has been renamed or repartitioned.
• Religious traditions are described from within as belief, and distinguished from external historical claims about the same events. Reverence is not agreement, and description is not endorsement.
• Partition is treated as a shared catastrophe rather than as one community's grievance against another.
• Coverage is checked for lopsidedness as the archive grows — whether one side of the border, one faith, or one century is being served better than the others. Where a gap is known, it is stated on the page rather than left for the reader to notice.

The archive is written in English, which is itself a limitation, and one it does not pretend away.

Where scripture appears in this archive it is held to a stricter standard than anything else on the site.

• Scriptural text is never typed by hand. It is extracted programmatically from an open, versioned corpus, and every line carries a cryptographic checksum recorded at extraction. A verification pass re-computes those checksums and fails loudly if a single character has changed.
• Scripture is not decoration. It appears where it belongs, in pages built to carry it, and is not placed on general surfaces to lend them weight.
• No measurement runs in front of it. No analytics beacon, tag or tracking script will be placed on a page displaying scripture. A reader in front of sacred text is not a measurement subject.
• Translation is marked as translation, and never presented as equivalent to the original.

Images are used only where their copyright status can actually be established: public domain, or a Creative Commons licence permitting reuse. The licence and the author are recorded for each one.

Anything whose provenance or copyright status cannot be verified is not used, however well it would have illustrated the page. Where a page is waiting for an image it does not yet have, it is better for it to be plainly missing than to be filled with something the archive cannot account for.

Illustrations drawn for the archive itself — maps, diagrams, decorative panels — are original work and are identifiable as drawings rather than as photographs of objects.

This archive will contain errors. Some will be typographical, some will be errors of fact, and some will be errors of framing — a claim stated more confidently than the evidence supports, or a tradition described in a way a member of it would not recognise.

All three are worth reporting, and the third is the most valuable.

Send a correction naming the page and, where you can, the source that supports the correction. Corrections that are accepted are made to the page itself. A correction that changes the substance of a claim is noted on the page rather than applied silently, because a reader who relied on the earlier version deserves to know it changed.

Disagreement about interpretation is not the same as an error, and is welcome on its own terms. Where a serious alternative reading exists and the page does not mention it, that is an error — of the third kind.

• No advertising, and no sponsored or paid placement of any kind.
• No engagement metrics feeding editorial decisions. What readers look for is useful information about what to build next. It is not an argument about what is true, and it does not decide what the archive says.
• No gamification — no streaks, no scores, no rewards for returning. Learning here is not a competition.
• No invented content. Nothing on this site is published as fact because it sounded plausible. Where the archive does not know, it says so.
• No tracking cookies, and no third-party advertising or profiling scripts.

These are permanent commitments rather than current circumstances. They are listed here so that a departure from them would be visible.