Vitruvius
On Architecture, Book 10 — machines, and how they multiply a man
De architectura, liber X · Written in the 20s BCE and dedicated to Augustus, by a man who had served as a military engineer under Caesar.
Cranes, hoists, water-screws, force-pumps, water-organs, an odometer that drops a pebble into a bronze bowl for every Roman mile, and then the siege engines. The one surviving classical account of how ancient machines were actually built, by an engineer who had built them and who keeps telling you which parts break.
How to read this page — what a person checked, and what the model wrote
The notes beside this text are here so you can read it straight through without stopping to look things up: what the passage assumes you already know, who is named without introduction, when the author is joking, what a social rule cost to break, and where the translator quietly tidied something up.
A language model wrote them, and no specialist has reviewed them. That makes them a companion rather than an authority — useful the way a well-read friend is useful, and wrong now and then in the same way. If you are writing something with a bibliography, cite the translation, named at the foot of this page, and let a note point you at what to check.
The division of labour on this page:
- Checked by a person: the translation, unaltered from the printed book; the source text, from the edition named below; every map dot for a named place, verified by hand against a gazetteer; every quotation, supplied from a hand-built list.
- The model's work: every note, headnote, gloss and piece of background; scene-location pins, which are its judgement and say so; any claim about tone, register or intent — those are arguments rather than facts, and you are free to disagree with them.
Quotations are the one thing the model is never allowed to propose — it may only explain one it has been handed from a hand-verified list, which is why you will see few of them. Spotting an allusion is the most valuable thing a commentary does and the easiest thing to invent convincingly, so that restraint is deliberate and is not going to change.
Spotted something wrong? Tell me — a reader who catches an error is the fastest way this gets better.
Morris Hicky Morgan (1859-1910) of Harvard produced the standard English Vitruvius; he died before it was finished and it was seen through the press by colleagues in 1914. It is a scholar's translation with an engineer's diagrams behind it, and its great virtue is that Morgan decided what each technical term meant and then used one English word for it consistently, which no earlier translator had done. The cost of that decision is that it hides the ambiguity: Vitruvius' own vocabulary is not consistent, and where he is genuinely unclear about a mechanism Morgan's steady English can make him sound settled. Roman units are usually kept — cubits, feet, digits — rather than converted, which is the right choice and occasionally a demanding one.