Following the Paper Trail

Every entry on Where Tales Began links to this page. It explains what the dates, labels and panels in an entry mean, and the rules that every claim on the site follows. The short version: each claim about a date, a version or an origin is tied to a named source, and a claim is only as strong as the record behind it.

First-ink dates and the word "earliest"

The site dates records, not origins. A first-ink date is the date of the earliest written record of a story or custom that can be checked. It marks where the paper evidence starts. A story may have been told, or a custom practiced, long before anyone wrote it down, and the record cannot say for how long. For that reason entries say "the earliest surviving record" and do not say "the origin".

"Earliest" is used only inside a stated scope: a tale type, a language, a region or a kind of record. The Cinderella story of Ye Xian, written down before 863 by Duan Chengshi in the Youyang Zazu, is a good example. An older story about Rhodopis, recorded by Strabo, is often brought into the comparison, but it lacks the heroine mistreated by a stepmother and stepsisters. Whether Ye Xian counts as the earliest Cinderella depends on how the tale type is defined, so an entry states its definition before it uses the word.

The Earliest ink panel and the checked-through date

Each entry has an Earliest ink panel near its end. It names the earliest written record of the entry's subject: the document, its date and a link to the source.

Under the panel is a line of the form Sources checked through YYYY-MM-DD. The date is the cut-off for the sources behind the entry. Editions, finds and studies that appeared after it are not reflected until the next scheduled re-check, and the date changes when that re-check is done.

Versions are labeled and kept apart

The same story often exists in several versions, and the versions differ. Every version in an entry is named with its text, its author or collector, its edition and its year. The Grimms' 1812 and 1857 editions of a tale are treated as two texts, and so are the Grimms' version and Perrault's. Quotations appear as block quotes labeled with the edition and year, and a detail found in one version is never credited to another.

Similar is not the same as connected

Two stories from different places can look alike without one coming from the other. When versions resemble each other, an entry compares them and stops there. Maps of versions show where and when each one was recorded, with no lines between the places.

A direction is drawn only when published research supports a textual relation between two versions. In those charts a solid line with its source marks an established relation between texts, and a dashed line with its named source marks a hypothesis. A claim that a story passed from one place to another appears in the text only with the study that supports it, named in the same sentence.

Published theories of a common origin are presented by name, with their evidence, the objections to them and how settled the question is: for example, that one scholar proposed a link in a given year, that another disagreed, and that there is no consensus. A hypothesis is not written up as a conclusion. A shared tale-type number marks a shared plot, not a proven common ancestor.

Which source settles which question

Public domain is a rights status, not a measure of reliability. Which source settles a question depends on the kind of statement being made.

What an ancient author says about the beginning of a custom is treated as material under study, not as the answer. If a classical writer says that a rite began with a god, the entry reports it as "the writer records that…". All of these sources rank above general encyclopedias and other web pages. When a popular account and the scholarship disagree, the entry says so and follows the scholarship. When scholars disagree with one another, their positions are set side by side without a ruling.

What is recalculated, and what that proves

Some parts of an entry are calculated rather than copied, so that anyone can redo them:

• Star charts are drawn by program from published coordinates, such as the IAU's constellation boundaries and its list of star names, so the positions can be reproduced.
• Movable feasts are recalculated from their rules: Easter from the computus, the church's method for finding its date, and lunisolar dates by calendar conversion.
• Timelines and comparison tables take their dates from the same data as the text, so a date cannot say one thing in a table and another in a paragraph.
• Tale-type numbers follow the Aarne–Thompson–Uther index (Uther, 2004). Only the numbers and type names are cited, because the index is a copyrighted publication.

A recalculation confirms the arithmetic and the drawing, and nothing more. Whether the input data are right depends on the sources they come from.

Texts, translations and rights

Every quotation is checked on four separate points: the original work, the translator, the edition and the extent of use. A public-domain original does not make a translation public domain, because a translator holds a separate copyright.

Where a public-domain translation exists, it is used, and the translator, edition and year are named. Where none does, a short passage appears in the site's own translation, labeled (Translation: Where Tales Began), and the original is quoted from a public-domain edition in the original language. Only short passages are translated this way. A text that is free to read for checking is not, by that fact alone, free to reprint; texts whose terms are unclear serve as research sources only.

Numbers come with their dataset

Counts and proportions, of cultures, people or versions that share something, are not given without data. A share or a count names its dataset, the range of the sample and its year: for example, the 345 variants of Cinderella and related tales that Marian Roalfe Cox gathered in 1893. Where no such dataset exists, an entry lists named cases instead.

Re-checks

New texts are found, manuscripts are re-dated and studies are revised, so entries are checked again on a schedule:

• The site's seven core reference entries, the long ones that other entries point to, are re-checked once a year.
• Other entries that make claims about the earliest record, about transmission or about a disputed origin are re-checked every 24 months.
• Entries on movable feasts get a fresh calculation and a re-check eight weeks before the feast each year.
• Lost Books entries on work still in progress, such as the reading of the Herculaneum scrolls, are updated once a year, and seasonal stargazing entries get the current year's dates.
• Passages that only retell a story, with no claim about dates or origins, change only through corrections.

After a re-check the checked-through date is updated. If the re-check changes an entry's earliest ink or one of its judgments, the change is also recorded as a correction.

When something is wrong

Errors can be reported to contact@wheretalesbegan.com. How corrections are made and listed is set out on Fixes to the Record.