Points of Interest

From Book Learning to Witnessing

Ismail al-Jazari is called the father of robotics, the inventor of the crankshaft and the builder of the first programmable robot. None of those titles survives a careful reading. What survives is odder and more useful: a court craftsman about whom almost nothing is known, and a machine book that names its sources, says where they fell short, and tells the reader how to build everything in it.

The myth

The titles arrive before the facts. Ismail al-Jazari, born 1136, died 1206, is the father of robotics. He invented the crankshaft, and so stands behind every engine since. He built the first programmable robot: a boat of mechanical musicians whose drummers could be reset to play a different rhythm.

Serious outlets tell it in good faith. A 2020 profile by Jorge Elices in National Geographic's history magazine gives the birth year and a birthplace, credits the book with a crankshaft, and describes the boat as "regarded by many to be the first programmable 'robot' in history". Each claim has something real underneath it. Al-Jazari did describe a pump driven through a crank. He did build figures that poured, struck and offered towels. There is a shaft with pegs on it under the musicians.

What does not hold up is the shape: the lone inventor, first at everything. His own book argues against it on nearly every page.

One preface, and nothing else

Donald R. Hill, the engineer who published the annotated English translation in 1974, wrote the entry on al-Jazari for the Dictionary of Scientific Biography. It opens by saying that everything known of his life is contained in the introduction to his book. Hill gives no birth year, only that he flourished in Diyar Bakr in 1206.

The introduction says this much. In Hill's translation, as quoted by the engineer Avi Golan in his long-running commentary on the book, al-Jazari was in the service of the Artuqid king Nasir al-Din Mahmud, after serving that king's father and brother, for twenty-five years, "the first of them year 577", which is 1181. A 2008 Muslim Heritage essay by Salim al-Hassani gives 570, or 1174, instead. The book was finished in the year 602, early in 1206, and the oldest surviving copy, a manuscript in the Topkapı Palace library, was made within months.

The birth year of 1136 appears everywhere and, so far as this piece could find, rests on nothing. Golan: "I have not found any historical source, and I fear that this is a complete fabrication." His death is an inference too. Golan reports that the early colophon adds, after his name, the formula used for the dead. The name al-Jazari points to the Jazira, the land between the upper Tigris and Euphrates, perhaps to the town now called Cizre; whether he was born there is not recorded. The court he served sat at Amid, the modern Diyarbakır. Accounts that place him in Baghdad, as one widely copied article does, have no support in the preface.

The book exists because a king asked for it. As Medievalists.net quotes the preface, Nasir al-Din told him: "I wish you to compose for me a book which assembles what you have created separately, and brings together a selection of individual items and pictures." Constantin Canavas, in a 2017 study of the treatise, reads the commission as a wish to record marvellous but fragile devices in a form that would last.

A craftsman who cites

Hill counts fifty chapters in six categories: water clocks and candle clocks, vessels and figures for drinking parties, pitchers and basins for washing, fountains and automatic flutes, machines for raising water, and a miscellany that includes a palace door and its locks. There are 173 drawings.

What sets it apart is the voice. Al-Jazari describes his own training as a movement from reading to doing. "I progressed, by practicing it, from the stage of book learning to that of witnessing," runs Hill's translation, as Golan quotes it. And he opens with a complaint about the literature, quoted by National Geographic: "I found that some of the earlier scholars and sages had made devices and had described what they had made. They had not considered them completely nor had they followed the correct path for all of them".

Then he names them. Of his great clock he writes, "I followed the method of the excellent Archimedes in distributing the twelve signs of the Zodiac"; Hill's entry notes that the Arabic treatise he drew on is the work of a pseudo-Archimedes. Golan's summary of the chapter is that al-Jazari measured the water regulator attributed to Archimedes, found it insufficient, and then explained how his first attempts to fix it failed.

On fountains he is blunter: "I did not follow the system of the Banū Mūsā, may God have mercy upon them, who in earlier times distinguished themselves in the matters covered by these subjects." The Banu Musa were three brothers of ninth-century Baghdad. Al-Jazari's objection, in the passages Golan reproduces, is practical: their fountains changed shape with each turn of a vane, "but this is too short an interval for the change to appear". He does not say he built their version. And where their text seemed muddled he allows for a copyist's error: "I do not know whence this confusion [came], from the original or from the copy."

For the automatic flutes he lists three sources: "a well-known paper by Apollonius, the Indian carpenter"; an old instrument "about which I found no written report, but a drawing"; and a paper written in Baghdad in 517 of the Islamic calendar, 1123 or 1124, by Hibat Allah al-Asturlabi, who, he grants, "makes a real innovation". Who Apollonius was is unsettled. Hill's entry calls him Apollonius of Byzantium; a surviving Arabic treatise on a fluting machine names its author as Apollonius the carpenter and geometer, and in Arabic script "Indian" and "geometer" differ by a single letter. On candle clocks al-Jazari admits he had nothing to go on: "I have never come across a work by anyone on candle-clocks".

This is not how a lone inventor writes. It is how someone writes who thinks of himself as late in a long line.

The elephant, the castle and the peacock

The machines are court pieces. The first chapter's clock, usually called the castle clock, has two rows of twelve doors, a crescent that travels across them through the day, falcons that lean forward to drop bronze balls into vases, a zodiac dial, and a band of drummers, trumpeters and a cymbalist. Hill records that it was rebuilt at the Science Museum in London for the World of Islam Festival of 1976, and that the reconstruction worked.

The elephant clock keeps time with a pierced bowl floating in a tank inside the animal. The bowl takes half an hour to fill and sink; as it goes down it pulls a cord, a ball drops, a serpent tips, a scribe's pen moves, and the driver strikes the elephant. A sentence circulates online explaining that the elephant stands for India, the dragons for China, and that "the turban represents Islamic culture", and reading the clock as a statement of multiculturalism. Golan, who traced it no further than a reference article, is flat about it: "The quote certainly is not by al-Jazari".

The peacock basin was for washing before prayer. Water runs from the bird's beak; as the used water collects below, floats rise and two small doors open in turn, one figure offering soap and the next a towel. In Golan's account a living servant fills it and works a hidden lever first.

Many of the paintings are no longer with their texts. Canavas notes that about half the illustrations of some copies were cut out by collectors. The elephant clock in the Metropolitan Museum and the peacock basin in the Cleveland Museum of Art are single leaves, both Syrian and both dated 1315.

The crank and the pegs

The crankshaft claim rests on the water-raising chapters. In one, an animal turns a geared shaft whose crank swings a scoop. In the fifth, a paddle wheel drives, through gears, a slotted rod that pushes two pistons in opposed cylinders. Hill's entry calls the pump "a significant contribution" and singles out its "true suction pipes rather than drowned suction". The historian Ahmad Y. al-Hassan, on his own site, claims more: "In 1206 the crank-connecting rod system was fully developed in two of al-Jazari's water raising machines." Neither says he invented the crank. Cranks are attested in Han China; the Banu Musa used partial ones; and a relief on a third-century sarcophagus at Hierapolis, published by Tullia Ritti, Klaus Grewe and Paul Kessener in 2007, shows a waterwheel driving stone saws through a crank and connecting rod, nine centuries earlier. Hill's entry never uses the word "first" of the crank. It never uses the word "robot" at all.

The programmable robot is newer than it sounds. In its current form it traces to Noel Sharkey, a professor of robotics at Sheffield, in a 2007 University of Sheffield note and a New Scientist article of the same summer. Sharkey wrote that the musicians' motion "appeared to depend on" a pegged cylinder, and that with extra holes drilled they could be given new rhythms "simply by rearranging the pegs". He built a model drummer to show it. That is an argument about what the mechanism allows, not a report of what al-Jazari says he did. Golan's verdict: "We will never know." Sharkey's own article goes on to find an older candidate, a self-moving device by Hero of Alexandria whose path was set by cord wound around its axle and a peg.

Why the truth is the better story

The myth offers a man who was first. Firsts are fragile; they last until someone finds a sarcophagus.

The record offers something sturdier. Here is a working craftsman at a provincial court, with no known teacher, no known birthplace and no certain dates, who was told by his employer to write down what he had made, and who chose to write it so that it could be made again. He says which old designs he read and why he left them. He admits what he had never seen. Near the end, in Hill's translation as Golan quotes it, he owns up to a limit: "I have omitted to mention many devices which I invented, for fear of obscurity or ambiguity." Clarity came before credit.

Hill's judgment, in his Studies in Medieval Islamic Technology as quoted by al-Hassani, was this: "It is impossible to over-emphasize the importance of al-Jazari's work in the history of engineering." His reason was the documentation: no other text before modern times offers "a comparable wealth of instructions for the design, manufacture and assembly of machines". Canavas adds the cooler note that no influence on later technical writing has been traced. The book may have changed nothing. It remains the clearest view we have into a medieval workshop, written by a man who is otherwise only a name and a preface.

On sources. The statement that all biographical knowledge comes from the introduction, the absence of a birth year, the fifty chapters, six categories and 173 drawings, the pseudo-Archimedes identification, "Apollonius of Byzantium", the pump ("a significant contribution", "true suction pipes rather than drowned suction") and the 1976 Science Museum reconstruction are from Donald R. Hill's entry on al-Jazari in the Complete Dictionary of Scientific Biography, read at Encyclopedia.com. Hill's translation (The Book of Knowledge of Ingenious Mechanical Devices, Reidel, 1974) was not consulted directly; every quotation of al-Jazari is given as reproduced by a named intermediary. From Avi Golan's commentary at aljazaribook.com: "the first of them year 577", the book-learning sentence, the Archimedes sentence and Golan's summary of the regulator tests, the three Banu Musa passages, the three flute sources, the candle-clock admission, the omitted-devices sentence, the colophon inference, the peacock basin's servant, and Golan's own remarks on 1136, the elephant sentence and the boat. The Banu Musa sentence is also quoted, with Hill's page 157, by Gunalan Nadarajan at Muslim Heritage (2008). The "earlier scholars and sages" passage, the birth claim and the programmable-robot phrase are from Jorge Elices, National Geographic, 30 July 2020. The king's commission is as quoted by Medievalists.net (29 July 2025), which does not state its translation. Hill's "impossible to over-emphasize" and "comparable wealth" phrases, and the 570/1174 date, are from Salim al-Hassani's Muslim Heritage essay of 30 January 2008; the conflict with 577 is unresolved here. Canavas is Constantin Canavas, in Le livre technique avant le XXe siècle (CNRS Éditions, 2017). The Topkapı manuscript (Ahmet III 3472), the elephant clock's mechanism, the "turban" sentence, early cranks and the Hierapolis relief were taken from Wikipedia entries and their citations; the Ritti, Grewe and Kessener paper was not read, and sources disagree on the exact days of the book's completion and the first copy. Al-Hassan's sentence is from history-science-technology.com. Sharkey's words are from his New Scientist article of July 2007 as reposted on the Atlas Games forum; "Baghdad" is his placement. The Sheffield note, New Scientist's site, Britannica and the Metropolitan Museum's object page could not be reached; the Met leaf's date is from search results, the Cleveland leaf's from the museum's page. The title of the Apollonius treatise is from Golan and a Qatar Digital Library record seen only in search results. The "Indian"/"geometer" observation is this publication's own, and no scholar's ruling on it was found.