Projects

Our Temple Was Made Without Hands

A meta-historical novel · in progress

Keating records that in Ireland long ago a college of antiquaries convened regularly to feast and examine the history of the kingdom. After dinner the antiquities were brought before the assembly and read aloud, and examined strictly, “least any Falshoods or Interpolations should have crept in.” Scholars have always dreamed, sometimes to the point of mania, of a mechanism like that: a guild charged with policing the transmission of the textual inheritance from one generation to the next.

The book begins with such an institution. A Society of Minstrels, chartered in England in 1604 under James I, its members heralds, antiquaries, poets, librarians and textual scholars, secretly regulating the documentary history of Britain. Its minutes pass from secretary to secretary for three centuries, each adding to the record. In 1881 F. J. Furnivall recovers the manuscript and publishes it—but he had it from the dispersal of John Payne Collier’s library, and its authenticity is questioned at once. Greg thinks it genuine. Others attack it. Its readings are collated, its provenance reconstructed, its interpolations debated.

Yet none of this happened.

That is the premise: a critical-variorum edition of a document that never existed. The apparatus is the novel. Its editorial voices are reconstructed from real textual critics—Aubrey, Percy, Pope, Malone, Furnivall, Eleanor Marx, Greg—and set against one another in a manufactured dispute three hundred years long, in order to pay them homage, to introduce the reader to exploits that deserve to be better known, and to poke fun at them.

The frame runs from 1604 to the years around the First World War: from Ben Jonson, Camden, and Robert Cotton’s forgeries in the service of the Overbury poisoning, to John Carter and Graham Pollard’s Enquiry and the unravelling of Thomas J. Wise. The literary debts are to Burton’s Anatomy, Pope’s Dunciad Variorum, Scott’s The Antiquary, Trollope, Thackeray, Dorothy L. Sayers, and Conrad’s Under Western Eyes.

The question underneath it is whether a chain of editorial custody is sufficient warrant for belief—and what happens when you inspect the collations yourself instead of trusting the authority they confer.

The Pseudodoxology Podcast Network

Podcast · patreon.com/wydna

Published by wydna, which describes the work as creating the radical science of tomorrow. Its subjects are the histories this site keeps circling—among them the method of the antiquaries of the seventeenth to nineteenth centuries, which is the ground the book and the essays are worked out on.

Episodes and archive at patreon.com/wydna.

bergcatalog

Python, TEI P5 · digitising the Berg Collection card catalogue

The card catalogue of the Berg Collection at the New York Public Library, scanned as six volumes of photographed cards, seven by three to a page. The scans are 580 DPI but bitonal, and the typewriter strokes have bled together; OCR tops out around 82% token recall on a good card, with the errors landing on exactly the proper names and dates a catalogue exists to record.

So the pipeline transcribes with a vision model, uses OCR as an independent auditor rather than as the transcript, and keeps every card image as ground truth. The archival format is TEI P5, which turns out to fit the problem unusually well, because TEI already has machinery for a transcription you cannot fully trust: unclear with a confidence and a responsible party, gap for what cannot be read at all—never a guess.

Every card records not only what it says but who read it and how far to trust them. Correcting a card in the browser rewrites its provenance from machine to human and strips the hedges the model added, so the pool of unverified cards shrinks as people use it.

Currently built: 735 cards across 35 pages, all transcribed, at a measured cost of about a penny and a half per card. Six volumes are on disk—4,914 pages and something over a hundred thousand cards.

handpress

Racket · github.com/wydna-group/handpress

A simulation of hand-press composition, and of the New Bibliography run back over the result.

Feed it a copy-text. It casts the copy off into pages, sets it in type as a named compositor with his own spelling habits, justifies each line by Moxon’s rules, imposes the pages in formes, prints, corrects at press, and distributes the type back into the case. Then it analyses its own output the way a bibliographer would—compositor attribution by spelling test, skeleton formes recovered from damaged running titles, casting-off recovered from crowded and gaping pages—and reports how much of what actually happened it recovered.

That last part is the point. A real book never knows the truth about its own printing; the simulator does, so the analysis can be graded against it. It usually does worse than the literature implies.

This site

Racket, Pollen · single source to web and print

Every page here, and every essay, is written once in a semantic markup and compiled to both a web page and a PDF. Note numbers, cross-references between notes, citations and the bibliography are computed by the compiler rather than maintained by hand—which is the whole reason for building it, since the previous arrangement kept two versions of an essay that had to be renumbered against each other by hand.

There is one citation formatter for both outputs, so a work cited in the PDF and the same work cited on the page cannot disagree. The reader chooses where the notes appear: at the foot, in the text, beside it in a pane, or not at all.

The design follows The Diary of Lady Willoughby (Chiswick Press, 1844), one of the first books of the Caslon revival, whose ruled marginal column—it holds dates there, and notes here—turned out to be the right answer to a question this site had anyway.