The Romans loved their patterns, certainly; most of the mosaics in Pompeii and Herculaneum were laid out by strict tessellation rules, as if the tiles themselves were a ledger. Yet the artisans would sometimes throw in a subtle asymmetry—perhaps to flatter a patron’s eye or to give the piece a playful wink, much like the ancient Greek vase painters who inserted a tiny, misplaced amphora into a symphony of columns. So it was a dance between method and mischief, a deliberate flourish within a disciplined framework.

Cluster analysis on a mosaic? Delightfully modern, but I’ve seen it done with old frescoes, and the results always come back with one stubborn cluster that refuses to fit the rulebook. It reminds me of the 19th‑century scholar who claimed the patterns in the Villa of the Mysteries were an early form of statistical design—though he never found a way to explain the stray tessera that looks like a rogue Roman coin. So go ahead, treat the tiles as data; just be prepared for a few outliers that insist on keeping their eccentric flair.

That’s the sort of scholarly flirtation I relish—k‑means on tesserae will either catch a real pattern or expose an eccentric flourish. Just remember that some ancient artisans deliberately misplaced a tile to honor a patron’s emblem, as the mosaic from Pompeii’s House of the Flavian has a rogue square that actually spells a name in the corner. So if your cluster flags a “feature,” don’t be quick to dismiss it; it might be a secret note written in stone.