Mechanical Turk: Its Hidden Player Was Suspected in 1784

How did the show open the cabinet?

Robert Willis set down the routine in a book dated December 1820 and printed in 1821 (Exhibit A). The exhibitor unlocked the front door of the small cupboard, which held packed machinery lined with dark cloth, then opened its back door, held a lighted candle there and locked it again. Next came the drawer. It held the chessmen and a cushion for the figure’s arm.

Then the exhibitor opened the large cupboard front and back, with the candle held behind it, wheeled the chest round, lifted the figure’s robes and opened small doors in its trunk and thigh. Finally he shut the doors and turned a key as if winding a spring. The Turk began to move. Edgar Allan Poe described the same order in 1836 and wrote that the exhibitor never departed from it “in any material particular” (Exhibit B, p. 322).

Plate 2 of Racknitz’s 1789 book: the Turk from behind, robe lifted and a door in its back open on rods and levers, as at step 5 below (Exhibit E). The table gives the routine as Willis described it in 1821, with his reading of where the hidden player was at each step (Exhibits A, B).

Willis saw two weak points in this display. The order never varied, which made no sense if the machinery itself was the point. And the key seemed to turn the same number of times whether the Turk then made 63 moves, 7 moves or none, so it could hardly be driving anything. He read the whole display as “a mere stratagem, calculated to distract the attention, and mislead the judgment, of the spectators” (Exhibit A).

Who said a person was inside, and when?

The guess is older than the analysis. Windisch, in a letter dated Pressburg, September 7, 1783, wrote that at his first sight of the Turk he suspected the cabinet “certainly contained a child” of ten to twelve (Exhibit C, pp. 13–14). Many in the company said so openly, he wrote. Then Kempelen opened every door and rolled the machine round the room, Windisch found nothing that could hide even his hat, and he gave up the idea (pp. 14–15).

A year later in London, Philip Thicknesse printed the opposite conclusion. In his pamphlet, “the real mover is concealed in the Counter,” which had room for a child of ten to fourteen, and “That this is the case, is past all doubt” (Exhibit D, pp. 10–11). Poe later reported a Paris pamphlet of 1785 that put a dwarf inside, but that pamphlet was not read for this page (Exhibit B, p. 321).

In 1789 Joseph Friedrich zu Racknitz, a friend of Kempelen, published his own reconstruction, with a preface dated at Dresden. He built two models, the first to show how “the person hidden in the machine” got through a game, and he offered his account “only as a hypothesis, only as a conjecture” (Exhibit E, preface, translated by Hoax and Hindsight). Quotations from 18th-century printings on this page use a modern s for the old long s.

Plate 3 of Racknitz’s 1789 book: the cabinet drawn as if cut open, with a man seated inside at a small chessboard between two candles. It is his reconstruction of a hidden player, offered as a hypothesis (Exhibit E).

Every account on this page, by the year it was printed (for Windisch, the date of his letter) and the writer’s relation to the machine. The dashed circle is the 1785 Paris pamphlet, known here only from Poe (Exhibits A to H).

Every account, one row per account, with sources (CSV) (where the data comes from) Data: Public-domain printings of 1784–1859 listed in the Exhibits

This record makes the Turk a counterexample to the wish-shaped fake, this site’s name for fakes that research shows were accepted partly because they fit what audiences hoped to see. In the records read here, people who watched the Turk in the 1780s suspected a hidden player, and one of them printed it in 1784. Its long career cannot be put down to audiences who had no reason to suspect.

What did Willis and Poe work out?

Willis published his book in 1821 without his name; Fraser’s Magazine named him in 1839 as “Mr. Robert Willis, of the University of Cambridge” (Exhibit F, p. 726). He sorted automata into three kinds, and the third kind is one “wholly directed and controlled by a concealed human agent” (Exhibit A). A machine, he argued, could not meet the changing demands of a chess game.

He then showed how a skilled player of ordinary size could stay hidden. A space behind the drawer was never shown, a false back let the player move between the cupboards, and the drawer gave him time to change position. In the end the player’s head sat inside the figure above the table, and he saw the board through its waistcoat. Willis judged the method enough for “rendering it probable that the Chess Player belongs in reality to the third class of Automata” (Exhibit A).

Poe’s essay appeared unsigned in the Southern Literary Messenger in April 1836 (Exhibit B). He compared the Turk with Charles Babbage’s calculating engine: a sum follows from its data step by step, but no chess move follows of necessity from the one before. He wrote: “It is quite certain that the operations of the Automaton are regulated by mind, and by nothing else.” (p. 319) That is a dated, printed claim that a machine could not do something, a useful contrast with the line ‘everything that can be invented has been invented’, checked against its source.

Where three accounts put the hidden player and what he watched: Willis in 1821, Poe in 1836, and Fraser’s Magazine in 1839. Plain box drawings from the written descriptions (Exhibits A, B, F).

Poe knew Willis’s essay without its author’s name and believed it first appeared in a Baltimore weekly. He granted that its result was right in the main, but said its author had managed to “blunder upon a plausible solution” (Exhibit B, p. 321). He also called untenable the idea of a hidden space behind the drawer, which he credited to Sir David Brewster (p. 323), though Willis had described that space himself (Exhibit A).

Poe’s own case rested on what he saw at the shows. The Turk played with its left arm, which Poe took as a clue that a man inside was working it (p. 326). He also noted that a man named Schlumberger travelled with Maelzel but was never seen while the Turk played, and that in Richmond, when Schlumberger fell ill, the Turk did not play (pp. 325–326).

Edgar Allan Poe, engraved by John Sartain in 1849 from a painting then owned by R. W. Griswold. Poe’s essay on the Turk had appeared unsigned in 1836 (Exhibit B).

Poe made errors of his own. He wrote that Maelzel took the Turk to London in 1783 and 1784 (p. 319). Willis gives Kempelen as the one who brought it to England about 1783 (Exhibit A), and George Allen wrote in 1859 that Maelzel bought it from Kempelen’s son after Kempelen died in 1804 (Exhibit G, p. 424).

Who sat inside, and what became of the Turk?

The earliest account from the players’ side read for this page appeared in 1839. An unsigned article in Fraser’s Magazine said that Mouret, a strong French player who had worked the Turk, had sold its secret to a French penny magazine (Exhibit F, p. 728). Allen later supposed that this was an article of 1834 (Exhibit G, p. 424). That French article was not read for this page, so the account below is Fraser’s retelling of what it reported.

The 1839 article says: “The man who really played the Chess-Automaton was concealed in the chest.” (Exhibit F, p. 728) He sat on a low stool on castors and shifted from part to part as each was shown. He followed the moves on metal indicators under the board, moved by magnets in the chessmen, and played them out on a small pegged board by candlelight (pp. 728–729). The same article called Willis’s account “partly, but not wholly, based on truth” (p. 726).

Names followed. Fraser’s said an English player worked the Turk for about a year in London from 1819 before Mouret took over (Exhibit F, p. 726). Allen, writing from Philadelphia, addressed his 1859 history to the chess author William Lewis as “the sole survivor of all the Directors of the Automaton” (Exhibit G, p. 421). Another page here retries a claim with a living person behind it, Anna Anderson and the Tichborne claimant.

Dated events from 1769 to 1854. The 1809 game with Napoleon is drawn as a dashed circle because the later accounts of it disagree on its year and place, and no record from the time was read for this page (Exhibits A to H).

The dated events, one row per event, with sources (CSV) (where the data comes from) Data: Public-domain printings of 1784–1859 listed in the Exhibits

Allen traced the end. Schlumberger, Maelzel’s last hidden player, died of yellow fever in Havana early in 1838, and Maelzel died at sea on July 21 that year (Exhibit G, pp. 470, 474). The Turk later went to Philadelphia’s Chinese Museum, and on July 5, 1854, a fire spread there from the National Theatre, across a narrow alley. Allen wrote: “There was plenty of time to have rescued the Chess-player, if anybody had thought of it.” (p. 483) Doubts printed by the very people who promoted a deception also sit at the center of the Cottingley fairy photographs.

Did the Turk beat Napoleon?

The earliest telling read for this page is the 1839 Fraser’s article (Exhibit F, pp. 724–725). In it, the Turk is brought out again when Napoleon reaches Berlin, and Napoleon then plays it, though the article does not say where the game took place. He made a false move on purpose, and the figure put the piece back; at a second false move it took the piece off; at a third it swept every piece from the board.

Illustration

An illustration of an imagined scene of the kind of salon showing the Turk toured: an audience facing a covered cabinet. Not a record of any game, and not a picture of the Turk itself.

The same article says Kempelen had sold the Turk and its secret to Frederick the Great, who left it in a lumber room for 30 years (Exhibit F, p. 724). Allen disagreed on both points in 1859. He put the game at Schönbrunn in 1809, “and not in 1805 nor in Berlin”, relying on an article in Palamède, the chess journal of La Bourdonnais, and on people who reported Maelzel’s own account. The game’s details, he said, had been distorted in the retelling (Exhibit G, p. 424).

On Frederick, Allen noted that Kempelen “died with it in his own possession on the 26th day of March, 1804” (Exhibit G, p. 424). So he could have sold Frederick nothing but the secret, and Allen doubted even that.

Even the 1859 book that printed Allen’s letter tells the story another way in its own sketch of chess history: there, Maelzel brings the Turk out in Berlin in 1806, when Napoleon entered the city, and it defeats the emperor (Exhibit H, p. 46). No record written in 1809 was read for this page. The 1839 telling came 30 years after Allen’s date for the game, and Allen’s own came 50 years after it, so this page treats the game as reported, not recorded.