Al-Jazari's castle water clock

Al-Jazari's castle clock opens doors, drops balls from two falcons' beaks and lights roundels at night, multiplying time signals within one machine ². Behind the spectacle are water, a float and an operator resetting the apparatus, not magic or perpetual motion ³. Its contribution is a calibrated seasonal-hour clock within an older tradition, not invention of time or the first computer.

The first chapter of the book's first category describes a water clock indicating solar hours, with ten sections within the chapter ⁴. A display faces the observer while interconnected equipment operates behind it; Hill distinguishes the falling reservoir float from the outflowing water, two sources driving different parts ⁵.

It is not the elephant clock, double-piston pump or peacock washer. The book separates clocks, water raising and service vessels, so features cannot simply migrate between designs ⁶. “Castle clock” names the known design, not an excavated stone clock assigned to a particular fortress.

The text starts by acknowledging a division attributed to Archimedes, then an anonymous maker's instrument and other attempts; tested divisions did not perform as intended ⁷. Erasing those predecessors cannot establish that the first drop ever used for timing came from this clock.

Nor does a transmitted attribution authenticate Archimedes's authorship: TDV identifies this reference as Pseudo-Archimedes ⁶. The engineer's reported source and modern authentication of its author are different matters. Recognizing inheritance helps identify a particular improvement rather than denying every novelty.

TDV places the commissioned book under Artuqid patronage at Amid and dates completion to 1205 ⁶. Hill's introduction derives 1204 or 1206 from variant service chronologies ⁸. Our1204/1206 range therefore represents differing documentary dates, not three years of construction or a verified inauguration.

We read selected passages of Hill's1974 translation and commentary, not an autograph or a fresh collation of every manuscript ⁹. Preserved instructions are not daily operating logs. Our illustration is a1354 manuscript painting, neither contemporary with composition nor the distinct 1315 copy represented by the elephant-clock image ¹⁰.

Sunni frame: TDV calls the individual an Islamic scholar, documenting Muslim identity; the declared Sunni presentation frame is not newly established testimony to his personal school ⁶. A name or employer's affiliation cannot substitute for individual evidence.

The book attributes description, experiments and adjustments to al-Jazari; TDV places his improvements within a longer engineering tradition ^{11 6}. The Boston painting's artist is anonymous. A copyist-painter is not the engineer-author, and a drawing cannot name every person who made components ¹⁰.

First, reconsider “hour.” The design divides daylight into twelve portions and darkness into twelve, although their lengths change seasonally; a seasonal hour is not always sixty minutes ¹². In our invented example, a15-hour day gives 75-minute portions and a9-hour night 45-minute portions. These are not claimed as a historical date or location.

Behind the screen, a reservoir float descends with the water and transmits movement through ropes and pulleys ¹³. But falling water level in an unregulated tank changes the outflow conditions. A separate chamber has another float and conical plug: rising water closes the inlet, falling water permits replenishment, maintaining its level approximately ¹⁴.

These are not one float performing every task. One provides motion through descent; the other regulates the feeding chamber. A further regulator changes outlet position for the required duration ¹⁵. The second float can be understood as hydraulic feedback, but modern control terminology does not make the machine a computer program or the first regulator in history.

Seasonal hours have a numerical consequence. Emptying the same hypothetical total volume in 15 hours and then9 requires the second average flow to be 15/9 times the first. In our paper example,600 volume units divided by 15 gives 40 per hour, while division by 9 gives about 66.67. Those quantities and units are ours, not apparatus specifications or measured historical performance.

A circular scale alone does not ensure the required ratios. The author reports failed geometric divisions, then measures water discharged in a fixed hour and establishes scale positions through calculation and experiment before engraving them ¹¹. Geometry serves calibration rather than replacing it. The operator chooses the relevant setting; the machine does not independently sense tomorrow's daylight duration.

The mechanism moves a cart carrying a crescent indicator, sequentially releases doors and sends balls towards the falcons. A ball changes a falcon's balance, making it tip and then recover after release ⁵. A separate water route drives a wheel and pegs moving percussionists' arms; Hill explains wind sound through an air vessel and flute, not a free-choice robotic musician ⁵.

At night, illuminated roundels indicate the passing divisions, while the operator sets the moon's position and an opening representing its phase ¹⁶. Displaying the sky is not independently computing its motions. The observer sees the result of mechanics and human-supplied initial information, not continuous astronomical observation.

The documented transmission here is textual and pictorial: the later Boston folio preserves the facade, while TDV identifies the 1974 English translation and 1979 Arabic edition ^{10 6}. Many copies would not automatically mean many working clocks; a reader can copy a drawing without a reservoir or workshop.

These selections give no adoption count for towns or households. A verifiable path of transmission is better than an invented diffusion map. A book's arrival at a place does not prove installation of every machine it describes.

The author's named references and anonymous instrument place him within attempts to regulate water outflow ⁷. The observable refinement concerns testing, rejecting, adjusting and integrating regulation with a compound time display ¹⁷.

We do not add descent from the elephant clock because both have falcons and balls, or from the pump because both use water. A shared component is not a document of priority or direct transmission. Nor is the musicians' shaft counted as another completed invention merely because it can be drawn separately.

Hill's commentary reconstructs equipment layout but distinguishes inferred positions and dimensions and explains why some manuscript drawing placements cannot be built exactly as shown ⁵. An interpreted layout is not a photograph inside surviving original hardware.

TDV reports modern reconstruction of clocks from the book, without making every model this particular castle clock ⁶. Our evidence does not establish direct descent of every European clock or subsequent control system. Educational comparison and documented genealogy have different boundaries.

The museum describes doors, falcons, balls, illuminated roundels and musicians repeatedly signalling time ². Multiple interfaces make a moment visible and audible, but do not prove who attended or that every person could interpret or hear the signals equally.

The instructions expose human work: returning water, resetting parts and establishing positions for the next cycle ¹⁶. “Automatic” must not erase the operator. Automation is not absence of care; these instructions do not establish the attendant's wages or measured water losses over months.

This is evidence for calibration in progress, not only the finished spectacle: attractive divisions fail, a solar-declination attempt proves insufficient, and experiment locates acceptable marks ¹¹. Book learning and material practice interact within the author's own account.

That supplies no modern uncertainty figure or actual maintenance log. Nor does a modern interpreted dimension necessarily match every manuscript. The teachable contribution links regulation, calibration and representation while retaining evidential limits; it does not need a world-first contest to merit attention.

Sunni frame: Declared presentation accompanies individual Muslim evidence, not a newly invented personal-school declaration ⁶.

Design attribution: The book describes a particular clock and its author's trials while acknowledging predecessors. We record improvement and integration, not the first clock or all feedback ¹¹.

Textual limit: A translation footnote warns that the opening account's ninth-night-hour performance differs from detailed later sections; the ending describes midnight and dawn ¹⁸. We do not silently repair conflicting instructions or infer perfect sound synchrony from a static painting.

Readers can tabulate a hypothetical daylight duration divided into twelve and ask what happens if the flow remains set for yesterday. That is our modern exercise explaining a need visible in the text, not a dimensional reconstruction ¹⁹.

Three tasks remain distinct: regulating water level, choosing an appropriate flow, and displaying elapsed time. They connect but are not one name for artificial intelligence. Our child activity uses no flame, weights, water tank or loose balls: paper and arithmetic suffice to compare seasonal hours and fixed minutes.

Boston14.533 is catalogued as an anonymous 1354 painting: zodiac arch above two door rows, falcons, vessels and musicians ¹⁰. It represents later book transmission, not an autograph or certainty about every component's actual material.

Commons identifies a faithful reproduction of public-domain two-dimensional art with the Public Domain Mark; we credit the museum, Google Art Project and Commons ²⁰. Hill's modern text has separate copyright: we read and explain in our own words, rather than treating all printed illustrations as public domain.

The painting combines astronomical forms, figures and music with visible time; the catalogue emphasizes multiple ways of presenting the same information ². Visitors can appreciate a display without assuming every ornament performs essential calculation or every figure chooses its actions.

The solitary-genius image can conceal people not named by our sources: attendant, copyist and anonymous artist. The instructions specify operator labour and the image metadata leaves the painter unnamed ^{16 10}. Acknowledging known and unknown contributors makes the account fuller without removing its author's documented contribution.

On paper, divide a hypothetical 900-minute day into 12 boxes, then a540-minute night into 12. The results are 75 and 45 minutes, not60 in both cases. This invented example explains the text's twelve seasonal divisions of day and night ¹². Colour successive boxes instead of building a tank or lighting a lamp. Then ask who returns them to the start: in the described clock, a person resets the apparatus ¹⁶.

• Selected Hill translation and engineering commentary ²¹.
• Official MFA-via-Google catalogue and 1354 image metadata/rights ²².
• Sadettin Ökten on identity, engineering context and editions ⁶.