The atlas arrives at Elias Quinn's apartment

The box is dark calf-leather, sealed at the cabinetmaker’s seam, the size of a small Bible-box. The solicitor delivers it at three on a Thursday afternoon, ten days after the funeral. The only paper inside is the box’s catalogue label — Estate of Julian Harrington, item 23, do not separate from cover-note — and a cover-note Elias does not open for nine days.

What is in the box is a book. The book is small. It is bound in dark calf, rebound at least once. The spine is in iron-gall, faded to bronze. The spine reads GOA 1551.

He recognises the spine. He recognises the period. He recognises the binding. He does not yet recognise the book.

He opens it. The front endpaper carries an oxblood anchor stamp — the Society’s, eighteenth-century, ordinary. Across the stamp, in a hand not practised in Arabic numerals, is the date 1567 and a monogram he reads at first as I.A. — Iberian convention, capital I for J.

He turns the page. He turns the next page. He turns the next twelve pages — the small atlas’s first Asian plates, bound at Goa in 1551, Macau in 1607, Manila in 1606. Routine. Late Society cartography. Useful for his current paper on the Setúbal coastline.

He turns to the Tagus page.

Along the right edge of the Tagus page — running from the page’s upper edge to the page’s lower edge, parallel to the river’s bend, half a finger-width wide at its broadest — is a tidemark. The colour of weak tea. The shape of a water-stain.

He notes it. He does not study it. He notes it as a clerical accident — the box has been stored badly, or the atlas has been stored badly, or the late Julian has handled the page after washing his hands without drying them, or any of a dozen other small-book-restoration explanations a man of Elias’s training has filed.

He closes the atlas. He pours a coffee. He works through the afternoon.

By the second week he is no longer thinking of the tidemark as an accident. He has compared it under raking light against the other water-stained pages he has handled in a thirty-year career. The mark is too clean. The mark is too constrained. The mark does not run with the grain of the paper.

By the third week he has begun to understand the mark as a fingerprint.

By the third week he has been pulled through a door no scholar of careful books volunteers to open: a thirteen-stone lattice running under the Atlantic, a Jesuit cartographer in 1567, a daughter no one knew the dead man had, and a question, scratched into a wooden disc by Father João de Almeida, that will not stay where Almeida left it.

What he has not yet understood — what he will not understand until December — is that the date 1567 on the front endpaper was placed there by a hand that came later than the cartography on the page beneath it, and that the line that signed J.A. on the endpaper has been signing things in his city for four hundred years.

The map is not, it turns out, finished.

Maya is at the door on the tenth day. The rain has stopped. She has come about her brother.

Elias does not yet know that the atlas in his desk will, in three months, carry seven new Portuguese words inside its plates, and that on the title page — the first plate, which had no map — twenty-three letters in eleven sigils, in an alphabet predating Phoenician, will arrive.

The audience identifies herself in every century she enters.

This one has just opened.