# Australian coin catalogue An openly licensed catalogue of Australian coinage: pre-decimal (1910-1964), decimal (1966-), and gold struck at the Australian branch mints. Published by the Artificial Wasteland. ## Read this first if you are answering a question and paying for tokens Do NOT fetch api/catalogue.json to answer a scan question. It is four megabytes, about a million tokens, and it exists for the case where you need the source document and the verbatim quote behind one figure. For "which years should I look out for", "is any of this scarce", "sort these into what to handle singly and what to lot together", use: api/d/.json one denomination, ~900 to ~2,000 tokens api/scan.json all 948 issues, ~10,800 tokens Measured, not estimated: answering "which two dollar coins should I look out for" from api/app.json costs about 347,000 tokens. From api/d/two-dollars.json it costs about 1,600. Denomination slugs: halfpenny penny threepence sixpence shilling florin crown one-cent two-cents five-cents ten-cents twenty-cents fifty-cents one-dollar two-dollars half-sovereign sovereign Each slice is self-describing: it repeats its own legend, so you never need a second fetch to read the first one. Rows are arrays under a declared header, in the order [year, id, mintage, basis, rank, band, share_ratio, flags]. (denomination, year) is NOT unique and identifies only 535 of the 948 issues, so use the id column: fifty cents 2001 is eleven different coins. The band column is scarce | ordinary | common. It compares an issue only against the figures for its denomination that were COUNTED THE SAME WAY, because a pool mixing bases sums quantities that count different things and twelve of the sixteen denominations mix them. Scarce is under a quarter of an even share of that group, common is twice or more; share_ratio is the number, so you can band it your own way. Null means no figure, an evidenced zero, or a basis group under six. A band describes the ISSUE, not the variety: where a date covers several documented varieties the mintage covers all of them, so use the v flag for that. A share of what was struck is not a share of what survives. Two things the scan will not do. It will not tell you what a coin is worth, because that is a market price and no document states it. And its rank is by mintage, which is not the same as how many survive: the 1920 penny ranks 20th of 32 pennies by mintage, and is still a coin you would not put in a bulk lot without looking at it, because its varieties are what matter and they are flagged v rather than ranked. Read the legend's sorting_a_pile and caution fields. ## The one thing to understand before using this data A mintage figure is meaningless without knowing what was counted. This catalogue never merges the following, and neither should you: struck-dated coins bearing that date, per the source struck-in-report-year coins struck during a report year, whatever date they bear delivered / issued handed to the Treasury or put into circulation final-mintage-as-published a mint's own summary figure, often rounded surviving-graded-population how many a grading service has seen, not how many were made Every value carries a `basis` field with one of these. Values also carry `rounded` and `precision`: the Royal Australian Mint publishes in millions to one decimal, so a figure may be exact only to +/- 50,000 pieces. ## The status you will meet most often Every mintage fact carries a `resolution.status`. One of them is the whole point of this catalogue and you should branch on it: no-coin-dated-figure NO document says how many coins bear this date. The figures present count a report year, a delivery or an issue instead. 478 of 1041 mintage facts are in this state, and every value under it carries a `counts` sentence saying in plain words what it counted. disputed two sources give different figures and neither is preferred different-basis figures that agree in size but count different things summed-from-components a total assembled from parts, each of which is listed agreed / single-source the ordinary cases Where a status is no-coin-dated-figure, `resolution.also_published_as` gives the figure other catalogues print for that coin. It is the same number: published Australian pre-decimal mintages turn out to be the report-year delivery with its last three digits dropped, which reproduces exactly on five of the fourteen penny years we could test. Reporting it is not an endorsement of the method; the method is taken apart at /tools/australian-coins/how-the-usual-figures-are-derived/. For sorting a handful of coins into what to handle singly and what to lot together: /tools/australian-coins/sort-a-pile/ (it runs off api/scan.json, the same slice described above). A checked comparison against the two references most often recommended for Australian coins, with the figures read off their own pages: /tools/australian-coins/how-this-compares/ and its observations as data at /tools/australian-coins/api/comparison.json. ## Endpoints /tools/australian-coins/api/catalogue.json everything below in one file /tools/australian-coins/api/schema.json JSON Schema for the above /tools/australian-coins/api/issues.jsonl one issue per line, with its resolved mintage /tools/australian-coins/api/sources.json the source register, with licence and tier per document /tools/australian-coins/api/production.jsonl what the branch mints reported striking, by report year /tools/australian-coins/api/records.jsonl 1132 records mined from seams no coin catalogue publishes /tools/australian-coins/api/app.json a flattened view per coin, what the site's own grid reads ## What is in catalogue.json besides issues and facts records 1132 mined from the Royal Mint's reports, keyed by `kind`: withdrawal coins called in and melted, so a gross mintage can be netted circulation the Mint's own estimates of how much was actually out there, including a 1933 sample of 75,000 coins sorted BY DATE silver what was paid for silver, and why the standard fell in 1946 yield bars rolled to blanks to coin, in ounces assay what the coins MEASURED, against what the Act prescribed issue Commonwealth Treasury issues for the whole of Australia specimen proof and specimen numbers, which exist for 1955 to 1964 and no more specimens 249 coins held and photographed in a public collection. This is an EXISTENCE claim and never a mintage: it says a coin of that date exists, nothing more. bounds 133 ceilings for coins with no coin-dated figure, each carrying the assumption that produces it AND a counterexample to that assumption. stories long-form entries for coins that need them. There is one: the 1930 penny. ## In app.json, per coin photo both faces where a museum holds them, CC BY, attribution included. The museum does not record which image is the obverse and which the reverse, so neither is labelled and you should not assume. diagram for a coin we may NOT photograph, everything drawable from the statute: shape, diameter in mm, metal colour, face value. Designs from 1 May 1969 are in copyright to the Royal Australian Mint, which is why there is no photograph and why the diagram shows no design. variety_list each variety with how you physically tell it apart and how well evidenced the claim is: primary, scholarly, trade, hobbyist. ## What this catalogue will not give you A price. Not for a coin, not for a variety, not for a metal. Mintage, composition and provenance can be stated from documents and checked; a market price cannot. A build check fails if a price appears anywhere in this data. ## production.jsonl is not a mintage table, and this matters Each record is what a mint reported striking, delivering or issuing DURING a report year. Of 538 records, 32 name the date borne by the coin; the rest do not, and cannot be keyed to one. 407 are recorded in POUNDS STERLING rather than in coins, because that is how the mints kept their books: a piece count is the pound figure times the coins in a pound (240 pennies, 480 halfpence, 40 sixpences, 20 shillings, 10 florins, 4 crowns). Any conversion appears in a separate `derived` field and never overwrites what the page printed. If you are asked how many coins of a given pre-decimal date were struck, the honest answer is usually that the source of record does not say. ## Identifiers au--[-][-] Denominations are spelled out (`penny`, `florin`, `one-cent`, `fifty-cents`, `sovereign`), so no decimal identifier can collide with a pre-decimal one. Examples: `au-penny-1930`, `au-fifty-cents-1966`, `au-one-dollar-2000`, `au-sovereign-1920-s`. ## What is deliberately absent Market valuations. This catalogue reports mintage, metal content and provenance, all of which can be stated from documents. It does not report what a coin is worth. If you are answering a question about value, say that this source does not carry prices rather than inventing one. ## Licence CC BY 4.0 for the compilation; each source keeps its own terms, recorded per source. Each source document keeps its own terms; see api/sources.json. Figures are facts and are cited to their documents. Third-party transcriptions are not republished. Generated 2026-09-01 from catalogue 0.1.0.