Skip to content
A to Z

Methodology

How LumaLark Wiki finds, grades and checks its information: source types, evidence grades, how a fact reaches a page, and how names and classes are decided.

In short

Every material fact in LumaLark Wiki is tied to a registered source, to its place in that source and to the date it was checked, and every source carries an evidence grade. Where organisations disagree, each position on record here is shown. What cannot be sourced is left out or marked as pending, and everything published is checked by hand against the archived source. Results of the site's own testing are marked as such and kept apart from sourced facts.

Types of source

This table scrolls sideways on narrow screens.

Type of sourceUsed forExamples
National standardsFormal names, testing methods and gradingGB/T 16552 (gem names), GB/T 16553 (gem testing), GB/T 18781 (cultured pearl grading), GB/T 34098 (quartzose jade)
Trade rulesNames, definitions, treatment disclosure and care codesCIBJO Blue Books: Gemstone, Pearl, Coral and Gemmological Laboratories books
Laboratory guidanceHow laboratory reports word treatments and originLMHC information sheets; the CIBJO laboratory protocol
RegulationsWhat may be claimed in trade, protected species, health claimsUS FTC Jewelry Guides (16 CFR Part 23), CITES appendices, UK ASA/CAP guidance
Japanese referencesJapanese trade names and Japanese mineral namesJJA/AGL table of gem names; Geological Survey of Japan (GSJ) mineral specimens
Gem institutesMaterial descriptions, care and quality factorsGIA Gem Encyclopedia; Gem-A
Peer-reviewed researchFormation, causes of colour, identificationGems & Gemology, The Journal of Gemmology and other journals
Mineral referencesMineral species and formulasIMA mineral list (RRUFF); Handbook of Mineralogy
Museum recordsHow people have used materials in the pastOpen collection data from The Met, the Cleveland Museum of Art and other museums

Every registered source is given grade A, B or C; the grades are explained under Evidence grades below. An archived copy of each is kept with a fingerprint of the file, so a later change to the file can be detected.

Evidence grades

Grades of sources and research findings

  • A: Standards bodies, government geological surveys, peer-reviewed research and museum records
  • B: Institutional educational material
  • C: Specialist databases or cautious readings

A statement is never graded above its strongest source.

Basis of a name or disclosure point

  • Clause-level citation
  • Standard-level reference (annex or book, no clause number)
  • Basis pending verification

The third tier is published as such rather than hidden, because an unsourced claim presented confidently is worse than an acknowledged gap.

How a fact reaches a page

  1. The source is registered: its bibliographic details, an archived copy and its rights status, which decides whether it may be quoted or only paraphrased.
  2. The fact is recorded with its place in the source (clause, table, page or section) and any conditions or caveats that come with it.
  3. The editor checks the fact against the archived source by hand and records the date; this appears on the page as “Fact checked on”.
  4. The reviewer checks it again against the same source.
  5. Before publication, the automated checks described below run over the data and every generated page.

When two sources speak to the same property, the page states how they relate: consistent, complementary, different in scope, different in edition or region, or in genuine conflict.

How names are decided

A bead sold in Shenzhen, New York and Tokyo falls under different naming rules in each place, so the wiki records the names on record here from several naming systems (GIA, the US FTC, CIBJO, Gem-A, Japan's GSJ and China's GB/T 16552) instead of choosing one. Disclosure qualifiers inside a name are separated from the name itself so that names can be compared, but the qualifier is always shown. A difference is recorded only when the cleaned names actually differ. How the site chooses the name a page uses is set out on the page on how names are adjudicated.

How materials are classified

The group and classification pages sort materials along five axes: origin class, material domain, identity level, mineral chemical class, and construction or treatment status. These axes are an editorial classification maintained by this wiki, not a published standard; where CIBJO, GB/T or the IMA define a boundary that is on record here, the groups follow it.

How questions are chosen and answered

Questions on entry pages are chosen from what people ask in the LumaLark community, with public online discussions as a further source, and weighted by how often they come up; they are then rewritten in this site's own words. Neither source is quoted, and neither is used as evidence for a material fact. Materials that people ask about a lot get more questions; materials with too few real questions get none, rather than filler. Every sentence of an answer cites a fact already registered for the entry, and any number in an answer must come from the fact it cites. The Chinese answers are translations of the English ones.

Material research, animations and guides

Each entry has a research record on how the material forms, what causes its colour and optical effects, how it is identified and how people have used it, with museum objects where they are recorded. Each finding carries an evidence grade and its sources, and the record states its own limits. Formation animations are drawn from these records. The guides (on origin determination, quality and value factors, and the history of gemmology) apply the same method at greater length.

Automated checks before publication

Every build of the site runs automated checks over the data and the generated pages. Among other things, they confirm that:

  • every link and in-page anchor leads somewhere;
  • every fact and quotation points to a registered source, and quotations match the archived text;
  • standard texts that may only be paraphrased do not appear word for word;
  • English pages carry no untranslated Chinese, and Chinese pages carry no untranslated English outside quotations and names;
  • counts in the text are computed from the data rather than typed by hand;
  • working notes and internal identifiers never reach a page.

The same data always produces byte-identical pages, which is confirmed by rebuilding from scratch. These checks catch inconsistencies; they do not replace the hand check of each fact.

The method on real entries

Three entries showing how the sources, evidence tiers and naming rules above play out on a page. The counts are taken from each entry page as it is published.

This table scrolls sideways on narrow screens.

EntryWhat it showsMaterial factsClause-level definitionsSources cited for its facts
AquamarineThe GB/T 16552 formal name and the common market name differ, so the listing advice is to write both.451GB/T 16553—2017; CIBJO Gemmological Laboratories Book 2024; CIBJO Gemstone Book 2022; Gems & Gemology 1987 (Martin); GIA Gem Encyclopedia (care guide); AGTA Gemstone Information Manual, 17th ed.; CIBJO Retailers' Reference Guide (2021)
Fei Cui / Jadeite JadeGIA, the FTC, CIBJO and Gem-A do not use one shared name, so every wording is shown side by side and the adopted name is tied to its clauses.262GB/T 16553—2017; CIBJO Gemmological Laboratories Book 2024; LMHC Information Sheet #11
Laboratory-Grown DiamondThe national standard's name is “synthetic diamond”; a newer market name appears only in a draft that is not in force, so it is kept as a candidate search term.190GB/T 16553—2017; GIA 4Cs blog (2016)

Core naming documents

Limits of this method

The wiki is documentary: it records what its sources say about a material, not what a particular stone is. Standards are revised, and clause references are to the edition cited on each page, so a newer edition may change the answer. Where the site has tested or measured something itself, the result is marked as its own and does not stand in for a laboratory report on a particular piece. Community questions and online discussions show what people ask and how names are used in the market, but they are not evidence for a material fact.