Terni Lapilli · Food for Thought
A three-pebble game left its grid scratched across an empire, and left almost none of its rules behind.
Somewhere on the paving of the Roman Forum, on the steps of a basilica, on some stretch of public stone, a person with an idle moment and a sharp point scratched a small grid of lines and began to play. We know people played there because grids scratched into that stone survive. What we do not fully know is what game they were playing.
The name that has come down to us is terni lapilli, literally “three pebbles at a time.” The tic-tac-toe that schoolchildren draw today is usually presented as its descendant, and Wikipedia's entry on that game states the lineage plainly: an early variation was played in the Roman Empire around the first century BCE, called terni lapilli, in which “instead of having any number of pieces, each player had only three; thus, they had to move them around to empty spaces to keep playing.”1 That single clause, move them around to empty spaces, is the hinge of the whole story. It is the feature that separates the Roman game from the modern one, and it is the feature this build restores.
Three pebbles apiece, and once they are down you must keep moving. That is not tic-tac-toe. That is a different, and older, idea.
The archaeological signature of the game is unusually democratic. The same Wikipedia account notes flatly that “the game's grid markings have been found chalked all over Rome.”1 The word is loose, since chalk would not outlast two thousand years; what survives are grids scratched into the stone itself. Not carved into a temple, not painted in a villa, but scratched and worn into public stone by anyone who passed. A board that costs nothing to make and nothing to lose is a board that leaves fingerprints everywhere.
And yet ubiquity is not the same as documentation. The Romans who cut these grids did not cut the rules beside them. No surviving ancient text sets out, move by move, how terni lapilli was won. This is the central and slightly vertiginous fact any honest history of the game must begin with: the boards are common, the rules are conjectural. What we reconstruct today is an inference from a name, a mechanic (“three pebbles”), a shape (a grid of nine), and a family of much better-attested relatives.1
The nine-point board with diagonals, as this build draws it. Its movement rule also lets each point in the middle of a side step diagonally to the two neighbouring mid-side points, a link not drawn here. Whether the ancient game was drawn or played exactly this way is not certain.
The relative that carries the most surviving evidence is three men's morris, which the same Wikipedia account calls “closely related,” and which may be a sibling of terni lapilli or the same game under another name.1 Here the documentary trail is warmer. The poet Ovid, in the third book of his Ars Amatoria, around the turn of the era, describes what may well be it: “A table has three pieces on either side; the winner must get all the pieces in a straight line.”3 Three pieces a side; a line to win. That is the skeleton of the game, glimpsed from inside its own century.
The morris tradition also gives us the deep prehistory that gets attached, sometimes too confidently, to the Roman game. Boards resembling morris have been reported from Kurna in Egypt, which one older authority dated to around 1400 BCE.3 But this is exactly where a careful history must slow down. The scholar Friedrich Berger argued that Coptic crosses found alongside those Egyptian boards make the early date “doubtful,” and concluded bluntly that “certainly they cannot be dated.”3 A board scratched into ancient stone tells you someone played there. It very rarely tells you when.
A scratched board is a fingerprint without a date stamp. It proves presence, not chronology.
Set the relatives side by side and the shape of the confusion becomes clear. Modern tic-tac-toe keeps the grid and the goal but throws away the scarcity: you drop marks until the board fills. Three men's morris and terni lapilli keep the scarcity, three pieces, no more, and therefore must add motion, because a game with six pieces on nine cells cannot simply keep placing. Nine men's morris scales the same idea up to a larger board and adds “mills,” lines of three that let a player capture an opposing piece. They are all recognisably the same instinct: get your pieces into a line.13
| Game | Pieces / side | Motion? | Earliest cited trace |
|---|---|---|---|
| Tic-tac-toe | at most 5 marks (first player) or 4 (second) | no, fill the grid | “noughts and crosses” in print, 1858 |
| Terni lapilli | 3 | yes, slide to empties | Roman, ~1st c. BCE |
| Three men's morris | 3 | yes, place then move | Ovid, turn of the era |
| Nine men's morris | 9 | yes, place, move, fly | Roman era, at least |
The conflation of terni lapilli with three men's morris is not a mistake so much as an honest limit. With three pieces a side, a grid of nine, and a line to win, the two descriptions may simply be two names for one popular street game. The morris literature preserved a rulebook;2 the Roman pebble game preserved only its nickname and its scratches. Where a modern reconstruction fills the gap between them, it is borrowing the morris rules to clothe the Roman skeleton, a reasonable move, provided we say so out loud.4
The engine in this project does not try to settle any of this. Its own header calls the game “Terni Lapilli (a.k.a. Rota)” and describes it as “the Roman three-in-a-row WITH movement.”4 Rather than pretend to a certainty the evidence cannot support, it makes a set of explicit, playable choices: a 3×3 grid of nine cells; three pieces each; an alternating placement phase of six stones; then a movement phase in which a piece slides to an adjacent empty cell, orthogonally or diagonally; and three in any row, column or diagonal wins in either phase. A side left with no legal slide loses, and 120 slides without a line end the game as a draw.4 It also adopts the balancing convention recorded for the adjacent-move form of three men's morris, where barring the centre on the first placement removes the first player's winning strategy: the opening stone may not go on the centre, a rule the code annotates as “the standard balancing rule that defuses a trivial forced win.”24
None of those are claims about what a specific Roman in 40 BCE actually did. They are a coherent, defensible modern realisation of a game we know mostly by its absence. The honest thing is to separate the two registers: this is what the build allows, distinct from this is what the ancient world is documented to have done. The first is precise. The second is a name scratched all over Rome, and a shrug.
Reasonably solid: the name terni lapilli, three pieces per player, a grid goal of three in a line, boards common in Roman public space, and a close kinship with three men's morris as described by Ovid.1
Genuinely uncertain: the exact movement rule (adjacent-only vs. any empty cell), whether diagonals counted, the opening restrictions, and the precise dating of the oldest boards. These are reconstructions, and this build's version is one reasonable reading among several.
Even the name repays a moment's attention. Terni lapilli is not a title so much as a description of the equipment: terni, “three each” or “three at a time,” and lapilli, “little stones” or pebbles. The Roman game was named for its pieces, and its pieces were whatever small stones lay to hand. A name like that is itself a clue to how the game was played and by whom: you do not name a game after loose pebbles if it lived on carved ivory boards in wealthy houses.1
The build carries a second name in its header: it calls the game “Terni Lapilli (a.k.a. Rota),” and describes the nine cells as the round Rota wheel, “8 points on a ring + cell 4 at the centre.” On screen that “ring” is the square through the eight outer points, crossed by four lines through the centre: the same board shown above.4 Rota, Latin for “wheel,” is the name given to an allegedly Roman-era variant played on a circular board, codified only in a 1916 article by the Latin scholar Elmer Truesdell Merrill.2 This is worth flagging as another instance of the field's productive uncertainty: the square grid, the wheel, and the terni-lapilli name may describe one game, or two, or a family that blurred together in practice. What the build does, calling its board a wheel while drawing it as a square and scoring the square grid's rows, columns and diagonals, is a reasonable modern synthesis, and it should be read as one, not as settled fact.
So the honest summary of the origin story is a layered one: a real Roman game, genuinely popular, genuinely everywhere, whose name and equipment we know, whose exact rules we do not, and whose modern form is stitched together from its better-documented cousins. That is not a weakness of the history. It is the history, and a build that plays its reconstruction openly, rule by rule, is being honest about the game it revives.