Every text, definition and note on this site is in the public domain. This page records exactly what was used and how it was processed, so you can check it or reuse it.
All poems are taken from Project Gutenberg plain-text editions, which are transcribed and proofread by volunteers. The Project Gutenberg header, footer and license text were removed before processing. The poems themselves are public domain and carry no restriction; the Project Gutenberg trademark license applies only to files that keep their branding, which these do not.
Ebook numbers link to Project Gutenberg. "Corrections" counts the typographical corrections applied at build time, each recorded with its source in the pipeline's corrections file.
| Work | Poet | Text | Original beside it | Commentary | Corrections |
|---|
Which texts count as public domain. The site follows United States law: everything published before 1931 is public domain there whatever the author's dates. That covers every text here. It includes the early books of poets who lived long after: Robert Frost's A Boy's Will (1913) and North of Boston (1914), Eliot's Prufrock (1917), Sassoon's Counter-Attack (1918) and H.D.'s Sea Garden (1916). Later collections by those poets are still in copyright and are not used. Readers outside the United States should note that most other countries protect an author's work for seventy years after death, so twelve poets here are still in copyright there: Robert Graves (died 1985), Ezra Pound and Marianne Moore (1972), Langston Hughes, John Masefield, Siegfried Sassoon and Jean Toomer (1967), T. S. Eliot (1965), Robert Frost and William Carlos Williams (1963), H.D. (1961) and Robert W. Service (1958). The rest of the library is public domain everywhere.
Where a public domain annotated edition exists, its notes appear beside the stanzas in the reader, anchored to the line they gloss. Editors and dates are shown at the foot of each annotated section. Hand-aligned editions: E. H. Coleridge's Byron, Sykes's Ancient Mariner with Coleridge's own gloss, Robertson's Keats, Poe's essay on The Raven. Harvested from the editions' own note sections. Among them: Buckley's notes to Pope's Iliad, Mickle's to the Lusiads, Bellows's to the Poetic Edda, Pollard's Herrick, Rolfe's Gray and Warren's Faerie Queene. Also the school editions of Scott's Lady of the Lake and Longfellow's Evangeline, Bridges's Hopkins, Burns's own notes, Way's Nibelungenlied, FitzGerald's Rubáiyát, and the Purves glosses to Chaucer. Those four hand-aligned editions in full:
Glossary entries come from Webster's Unabridged Dictionary (1913), Project Gutenberg #29765. Only the first sense of each entry is used, quotations are trimmed, and the dictionary's own tags such as Obs., Archaic and Poetic are preserved. A word enters a work's glossary when it is rare in ordinary English, measured by how seldom it appears across all Webster definitions. It also enters when Webster flags it as obsolete, archaic or poetic.
Names of people, gods and places in the reader's word panel come from Smith's three dictionaries: Greek and Roman Biography and Mythology (1870), Greek and Roman Geography (1854) and Greek and Roman Antiquities (1890). They are public domain books digitised by the Perseus Digital Library, whose encoded edition is CC BY-SA 4.0. Only the names that occur in the library are kept, and each entry is credited on the panel. Names Smith does not have fall back to a Wikipedia summary, CC BY-SA, credited and linked.
The full account of every measure, with the rules as tables and the statistics, is on the methods page.
Rhyme sounds and syllable counts use the CMU Pronouncing Dictionary, a public domain resource that records American English. A rhyme sound is the part of a word from its last stressed vowel to the end. Words the dictionary lacks are keyed by spelling analogy to words with the same ending. Syllable boundaries in the display come from hyphenation patterns and vowel groups, adjusted to the dictionary's count.
Because the dictionary is American, some pairs it reports as non-rhyming still rhyme in British English (again / men, been / seen). The drifted-rhyme tables hide the purely dictionary-driven mismatches by default.
Poets of the period read words such as every, heaven, power and hour as fewer syllables than a modern dictionary counts. When a line runs longer than its meter expects, the pipeline applies these elisions before counting. When a line runs short, it expands a spelled -ed ending to its own syllable. Every count shown is the reading that fits the meter best, not the only possible reading.
Words of two or more syllables have fixed stresses in the dictionary, and those are what the game scores. One-syllable words take their stress from the sentence, so they are free. The model answer places them on the meter's beat, but you are not marked down for reading them differently.
A rhyme scheme is stated as a fact of the form only where the standard handbooks settle it and the text bears it out. The handbooks are Saintsbury's Historical Manual of English Prosody (1910) and Gummere's Handbook of Poetics (1885), both public domain. So: ottava rima for Don Juan, Isabella, Rose's Ariosto and Fairfax's Tasso; the English sonnet for Shakespeare; the Petrarchan sonnet for Sonnets from the Portuguese; the Spenserian stanza for Childe Harold and The Faerie Queene; the Raven's ABCBBB; the ballad quatrain for the Mariner; FitzGerald's AABA. For every other work the scheme is detected from the dictionary sounds of the end words, and it can miss slant rhymes. A dash means "no exact rhyme found," not "unrhymed by design." The pipeline checks each stated scheme against the detected sounds and reports the agreement in the data. It is typically 71 to 96 percent; the shortfall is slant rhyme, eye rhyme and words outside the dictionary.
Meter labels come from the same handbooks and from the editions themselves: Pope's and Dryden's couplets, Milton's blank verse, Longfellow's unrhymed tercets, Gummere's alliterative Beowulf. A work without a settled meter is labelled by its dominant syllable count and marked "mostly".
Beside the translations sit the originals. Latin: the Aeneid, the Metamorphoses, De Rerum Natura, Juvenal's Satires, Horace's Odes, and Catullus from the Burton edition. Greek: the Iliad, the Odyssey and Hesiod's Works and Days. Italian: Dante and Ariosto. Portuguese: Camões. German: Faust and the Nibelungenlied. Finnish: the Kalevala. Old English: Beowulf (Harrison and Sharp, 1883). Middle English: Skeat's Canterbury Tales beside the modernised text. The Gutenberg originals are public domain transcriptions. The Greek and Latin classics come from the Perseus Digital Library's TEI files: the texts are public domain, the Perseus editions CC BY-SA 4.0, credited on each page. Still missing: Tasso's Italian, Pushkin's Russian, the Old French Roland, the Old Norse Edda, the Persian Rubáiyát and the Sanskrit Mahabharata, none of which Project Gutenberg has.
Where the reader offers the original beside the translation, the original is a Project Gutenberg text aligned by book or canto number. Originals are shown for reading only; nothing in the analysis or games measures them, because the pronouncing dictionary and tagger are English tools.
Portraits and plates are public domain images from Wikimedia Commons, checked file by file for licence. Artist, date and source are recorded in the image credits files and shown in the gallery. Some photographs of sculptures and manuscripts are CC0 rather than public domain proper; both permit reuse without conditions. Every poet in the library has an image. For about thirty of them no likeness survives at all, and what stands in is not a portrait but something from the poet’s own country and century — a painting, the title page of their own book, a leaf of manuscript. Those are captioned for what they are, and never as a face: Barbour has Bannockburn from the Scotichronicon, Peele the title page of David and Fair Bethsabe, Charles Wolfe the burial of Sir John Moore that his one famous poem describes. Beowulf, Roland, the Edda, the Nibelungenlied and the Cid stand behind their manuscripts. For Vyasa it is a seventeenth-century Mewar painting of Ganesha taking down the Mahabharata.
The plates inside the reader are from the illustrated editions. Doré's Dante (1861), Paradise Lost (1866), The Raven (1884), the Ancient Mariner (1876), Orlando Furioso (1879) and the Idylls (1868). Flaxman's outlines to the Iliad and Odyssey (1793). Blake's own plates for the Songs. Delacroix's lithographs for Faust (1828). Gallen-Kallela's Kalevala. And the Ellesmere, Cotton Vitellius, Digby and Hundeshagen manuscripts. A plate sits at the line it illustrates where the caption quotes or numbers the line; otherwise it is placed by order through the book. Sets that Commons holds only under CC BY-SA (Crane's Faerie Queene Book I, Doré's Elaine and Guinevere) are left out. Illustrations can be switched off in the reader.
A username and password are the only things stored. Passwords are hashed. Scores are kept in your browser as well as on the server, so the games work without an account and without a connection.
What is on this site is under three kinds of terms. Public domain: the poems, translations and originals from Project Gutenberg; Webster's 1913; Smith's dictionaries; the notes from the nineteenth-century editions; the illustrations; Wikidata's dates and places (CC0). The Gutenberg trademark licence applies only to files that keep Gutenberg's branding, which these do not. These can be reused, sold or reprinted by anyone without conditions. CC BY-SA 4.0: the biographies adapted from Wikipedia, and the Perseus Digital Library's encodings of the Greek and Latin texts and of Smith's dictionaries, credited where they appear. The files derived from them stay under that licence: the biographies file, the classical-names files, the Greek and Latin originals. The licence allows commercial use so long as the credit stays and any adapted version of those files is shared on the same terms. The site's own work: the analyses, measures, games, courses, tests and code are the author's. The processed data files (one JSON per work, plus glossaries and measures) may be downloaded and reused for study and research with a credit to Laurel.
Laurel is made by one person, the publisher of Contubernales Books, a small press. There is no company behind it and no staff. The texts are Project Gutenberg's, the definitions are Webster's, and the notes belong to the editors named on each page; the parsing, the measures, the games and the courses are his own work, and every part of the method is set out on the methods page. The site is free and always will be; nothing on it is a sample of something for sale, and no text is held back. If you want to support it there is a donation page, which pays for the server and for adding more books.
Every text is swept for editorial residue and compared word by word against a second Gutenberg edition. The text quality page lists what those checks found for each work, with links to the stanza. Nothing is changed automatically.
Every stanza in the reader has a report action for errors in the text. For anything else use these; a person reads every report before anything changes.
A description you are welcome to paste as it stands into a library guide, a course page or a syllabus. The figures are counted from the site and were last checked in September 2026.
Laurel (laurelpoetry.com) is a free reading room for public domain English poetry: 416 books by 375 poets, about 1.2 million lines, every poem set with its rhymes lettered, its stresses marked and its hard words explained from Webster's 1913. More than 18,000 editors' notes from annotated editions published before 1931 sit at the lines they gloss, and seventeen translated works carry the original language beside the English. Every text names the printing it reproduces, every poem has a permanent citable address, and the method of analysis is published on the site with its measured error rate. No account is needed to read. Maintained by Contubernales Books, a small press.
To cite the site: Laurel, Contubernales Books, 2026, laurelpoetry.com. To cite a poem, use the Cite as line at the foot of its page, which names the printing.