Ed. note: Jake Gyllenhaal reads “Letterform, Islandform” by Joshua Cohen.

The letter I’d like to describe did not exist, it seemed, except in the dream I dreamt for three consecutive nights, December 2009. Coleridge smoked opium and hallucinated an entire poem, “In Xanadu did Kubla Khan / A stately pleasure-dome decree”: whereas I, not a poet and trying to wean myself off Xanax and Vicodin and Percocet, had difficulty retrieving from my rest even a coherent letter.

I went to bed early all three evenings at my parents’ in Jersey, the anti-Xanadu. Three evenings of uneasy slumber and yet upon waking this was all I could recall: an elongated Hebrew lamed (ל), a distended Arabic lam (ل), the rough form of a fishing-hook (lengthened, stretched; though the Semitic Ur letterform is thought to derive from the shape of a shepherd’s staff or cattleprod), a finger curled to beckon, a kinked tongue, a carpet hung from the railing of a balcony and beaten of its dust until the remnant’s pattern was pure black outline (but this is embellishment now) . . . the dream was of this letter only, without color, just a character inked or perhaps even incised in black upon the nothingness that is not black itself, just sleep.

To describe a letter that already exists — the letter “j,” say — without writing it or saying it, is as difficult as describing a piece of music or plastic art: Rather, how to form the lines and how to pronounce the sound and name of that letter can be described without being demonstrated just as accurately, or just as inaccurately, as can the melodies, harmonies, and timbres of a symphony, or the shapes of a nonfigurative sculpture or painting.

But to describe a letter that doesn’t exist is a task seemingly more difficult than describing fake music, which Thomas Mann did well in not a few of his books, or describing fake art, which Marcel Proust, who regularly went to bed early, excelled at. I know of no writer who has, even unsuccessfully, described an imaginary letter whether in sound or image. However, I emphasize that my own experience was visual only. Before proceeding I have to admit I have no idea how my dreamletter would sound if pronounced (though something tells me it would be closer to a vowel than a consonant, and certainly it lacks the “lateral approximant” “l” sound of the lamed/lam).

In thinking of my dreamletter on the mornings after, I thought not of its identity as a letter (it couldn’t have “an identity”), but about its “meaning”: the meaning of its appearance. Dreams can be interpreted as representations of fears, but dreamletters cannot represent fear or any other emotion or thing — only letters and words actually existing can function in that way as symbols. Indeed, the only content of my recurrent dream besides the shape of the letter — pineally burning, nearly gashed into my forehead, that’s how close it felt in retrospect, how deeply substantial — can be said to be the thought, the feverishly intellectual thought, that “this is a false letter, without significance, at most it’s a self-reference, at most it’s a picture of a picture,” and so my dream announced, silently, its own meaninglessness, interpreting itself as unfit for interpretation.

First thing upon waking after the second night’s dream I tried drawing the letter, but never captured what I could so clearly remember envisioning (which is not the same as being able to clearly envision it, to recall it “photographically”).

My first attempt has a strange spermy loop at bottom that my hand forced me into but that was not in my dream. Also, I find it all too sinuous (I should mention that half of my failure is due to haziness; the other half is that I can’t draw).

My second attempt, after the third night, locates that loop at the top of the shape, not the bottom. Unlike the previous day’s, this sketch’s lines are too rigid, too severe: squared off.

I’d been exploring toponyms — specifically, how islands came to be named after their shapes in the days before dirigible, helicopter, and airplane flyovers: aerial photography. How did the premodern inhabitants of an island called Snake Island know that it was shaped like a snake without being able to fly above it, or without seeing pictures taken from altitude? Perhaps they stood on mountains, but not every island named for its shape has a mountain peak at center from which the entirety of the island can be surveyed. Is it possible that one of the skills possessed by earlier man was the ability to walk the perimeter of a landmass and, from that walk alone, to develop a mental picture of the total form of his route? If so, then this pedocartography is a talent since lost and remains unrecorded in every language.

There are many Snake Islands: a curl off Boston Harbor, a straighter Snake Island of the Philippines. Other Snake Islands are named Snake Island because there are snakes there, but these two, and a few others, are said to be named for their islandform: their serpentine coasts.

The Dry Tortugas of the Florida Keys were named for the many turtles sighted there by Ponce de León, but history has it that Columbus called the Caribbean island of Tortuga what he called it because it was shaped like a turtle, or tortoise — humpbacked. Yuantouzhu means Turtle Head Isle. Kelyfos Island also translates to Turtle Island.

Crocodile Island, off Boracay, also of the Philippines, is shaped like a crocodile. Muuido, of South Korea, means dancer’s dress and looks like a dancer’s dress. Shark Island, offering views of the Sydney Opera House, resembles a shark (there’s at least one other Shark Island, in Thailand). Elephantine, in the Nile, looks like an elephant’s tusk. Gato Island, a cat (skeptical tourists are told, “a sitting cat”). Tongpan Island, a barrel. Udo Island is a cow lying down or, in an alternate account, the head of a cow. Tobago exports tobacco, but is also said to be shaped like the smoke from Trinidad’s pipe (though Columbus named Trinidad for the Trinity). Anguilla is a slippery eel. Dolphin’s Nose, of India, is not an island but a massive jutting rock.

Naming islands strikes me as different from naming rock formations after what they resemble — e.g., the Mitten Buttes, Horseshoe Mesa — because while it seems a natural imaginative leap to imbue giant risen inanimate stone with animate qualities, it seems quite unnatural to imbue the earth underfoot, the native earth, with equivalent personality. Rather, I have the sense that indigenous peoples can never regard their own land as, for example, a llama or tree, whereas a distant hill or another tribe’s islet, precisely because it is conceivable as external, might be so understood.

Conquistadors experienced not just geography but whole cultures at a similar remove and so named what they saw without attempting to understand it. Sighted from the safety of a crow’s nest or prow, the foreign was always one thing — “the foreign” — not many things; the exotic is singular for a reason: ignorance. The New World was really, in terms of enduring civilizations, older than the Old; each India has its Indians. There is no Native American word for “America” in toto.

The first aerial photograph was taken in 1858; the Frenchman’s name was Gaspard-Félix Tournachon, aka “Nadar.” The image was a view of the village of Petit-Bicêtre, taken from the basket of a hot-air balloon tethered to float at an altitude of 80 meters. The balloon’s basket contained a darkroom. Tournachon’s pictures haven’t survived, however, so the oldest extant aerial photographs are James Wallace Black’s of Boston from 1860, also taken from a balloon roped to earth.

Images captured by automatically timed exposures from rockets, kites, and carrier pigeons — “bird’s-eye views” — were popular throughout the 1880s and ’90s, while the first widely publicized images taken from an aircraft were (silent) movies courtesy of Wilbur Wright, captured on an exhibition trip to Rome in 1909 (European governments were quick to grasp the military promise of flight photography and cinema).

Sherman Fairchild of Oneonta, New York, invented the flash camera in 1915, and five years later pioneered the aerial imaging of Manhattan, creating the most perfect map of that borough’s imperfect grid by assembling a series of overlapping photos (Manhattan’s earliest substantive aerial imaging had been accomplished by British photojournalist James Hare, 1906).

Mannahatta, in Lenape dialect, was most likely pure description: “many hills.” The island has since been described, by a Victorian travel guide, as a “sole-fish”; paintings of the Depression depicted it as an ironing board and a trowel; not a few poems have transmuted it into a sweat drop or tear.

Sitting on the subway one afternoon, I looked up from my book to the system map and saw it — I saw my letter. The lower bulge of the Battery, the upper winnow of Harlems and Inwood.

Manhattan, incognoscible first letter of an alphabet dreamt but as yet undiscovered.

From the book ATTENTION: Dispatches from a Land of Distraction by Joshua Cohen. Copyright © 2018 by Joshua Cohen. To be published by Random House, a division of Penguin Random House LLC.