
Herodotus · Book VII
Thermopylae
Θερμοπύλαι
7,068 words · 1069 unique lemmas
Herodotus > How a Coalition Holds > Thermopylae
Thermopylae is the most analyzed tactical defeat in Western military history. Its strategic significance is almost the inverse of what popular memory suggests: the Greeks lose the battle, but Leonidas's decision to remain and fight to the last converts a military failure into the coalition's most durable political asset. The pass is not held — it is spent, deliberately, on something more important than terrain.
On the third day at Thermopylae, when Ephialtes has already shown the Persians the mountain path that will encircle the Greek position, Leonidas dismisses the allied contingents and retains only the 300 Spartans, the 700 Thespians, and the 400 Thebans. The Thespians stay voluntarily; their city will be destroyed either way. The Thebans, Herodotus suggests, stay because Leonidas holds them as hostages against Theban treachery. The Spartans stay because Spartan law forbids retreat. What Herodotus is describing is a force held together not by shared purpose but by three entirely different institutional constraints — and that, precisely, is the coalition in miniature.
The strategic question at Thermopylae is not whether the pass can be held indefinitely — it cannot — but how long it can be held and at what cost to the Persian army's timetable. Leonidas's force of roughly seven thousand men holds Xerxes' army for two days, inflicting disproportionate casualties in a confined space where Persian numbers are irrelevant. The Persian elite infantry, the Immortals, are committed twice and repulsed twice. This is significant not for the casualties alone but for what it signals to every Greek state watching: Persian infantry, in the right terrain, can be stopped.
Ephialtes' betrayal on the second night forces the decision. When a local man guides the Persians' Immortals around the mountain path to hit the Greek rear, the position becomes untenable. Here the Greek coalition's structure becomes visible as a liability and an asset simultaneously. As a liability: local knowledge — of mountain paths, of vulnerable flanks — belongs to local people whose loyalty to the coalition is conditional. Ephialtes is not a Persian spy; he is a Greek who saw an opportunity and took it. Every coalition faces this problem: the more territory you need local cooperation to defend, the more exposure you have to local defection.
As an asset: Leonidas's decision to stay, fight, and die publicly solves the coalition's central political problem. The alliance of thirty-one city-states is fragile partly because no one has demonstrated that its terms include genuine sacrifice. A Spartan king dying at a pass in central Greece, with the bones of his Spartans around him, is a proof of commitment that no diplomatic communiqué can replicate. Sparta has now, irrevocably, invested blood in Greek survival. Athens and the remaining coalition members can calibrate their own commitment accordingly.
Herodotus records Dieneces' remark, told before battle, that if the Persians' arrows darken the sun, the Greeks will simply fight in the shade. The context matters: Dieneces is not making a morale speech. He is solving a tactical problem out loud. The Persian archery advantage disappears in close, confined fighting. Thermopylae's geography converts Persian scale into Persian irrelevance for as long as the pass holds.
Cross-Civilizational Connection
Parallel: The Zhanguoce records several instances of small-state defenders using a sacrifice — of troops, of territory, of a general's life — to signal commitment and buy coalition credibility. The Yan chapters (Zhanguoce 29–31) document how Yan's survival against Qi in 284-279 BC depended on exactly the same mechanism: Tian Dan's defense of Jimo was strategically irrational in the narrow sense but politically essential for preventing total collapse. Sunzi's discussion of 'death ground' in Chapter 11 is directly relevant: troops on ground from which there is no escape fight differently, and leaders who choose death ground are making a statement about their seriousness that opponents and allies both register.
Difference: In the Warring States system, a commander dying at a pass would more likely be interpreted as a failure of strategic judgment — the texts, and Sunzi specifically, consistently valorize the commander who achieves victory through maneuver, not sacrifice. Chapter 3 of the Sunzi ('Attack by Stratagem') explicitly ranks 'fighting and winning in actual battle' as the lowest form of excellence. Leonidas's death is celebrated in the Greek tradition precisely because it exemplifies a value — the courage to stand and die — that the Chinese strategic canon treats with deep suspicion. The traditions are not simply different; they are inverted on this point.
Limit: The comparison between Thermopylae and Warring States pass defense has geographic limits that matter. The Thermopylae corridor was genuinely singular in 480 BC — the only viable route for a large army from northern Greece southward that could be held by small numbers. The Warring States passes (Hangu Pass protecting Qin most notably) operated differently because the strategic system around them included multiple flanking options, naval dimensions, and the possibility of buying allied betrayal at scale. Ephialtes' mountain path had no Warring States equivalent in terms of decisive consequence.
Thermopylae teaches that coalition cohesion is not primarily a diplomatic problem. It is a credibility problem. The thirty-one Greek states held together through Salamis and Plataea not because they trusted each other — they manifestly did not — but because after Thermopylae they had evidence that at least some of them were serious. Leonidas did not hold the pass. He bought something more durable than terrain: proof of intent. In any coalition facing a hegemonic threat, that proof is worth more than the territory it costs.
See Also
- sunzi-bingfa/11-the-nine-situations — Sunzi's 'death ground' — where soldiers fight with maximum intensity because retreat is impossible — is precisely the condition Leonidas creates deliberately. The chapter's analysis of how terrain shapes fighting spirit is the sharpest Chinese parallel to the Thermopylae decision.
- sunzi-bingfa/10-terrain — Sunzi's taxonomy of terrain types, and his insistence that a commander must understand both friendly and enemy terrain, makes explicit the analytical failure Ephialtes exploits: the Greeks had not fully reconnoitered the mountain paths behind their position.
- zhanguoce/29-yan-1 — The first Yan chapter opens with Su Qin's analysis of Yan's strategic position as a small state surrounded by powerful neighbors — the same problem structure Leonidas is solving at Thermopylae, using commitment and sacrifice to signal that the coalition is real.
- sunzi-bingfa/03-attack-by-stratagem — Sunzi's ranking of victory forms — 'to win without fighting is best' — stands in productive tension with Thermopylae's logic. The Greeks at the pass win nothing militarily; the value is entirely in the signal. Sunzi's framework does not have a clean slot for this, which is itself illuminating.
Edition & Source
- Author
- Ἡρόδοτος Herodotus
- Greek Text
- Perseus Digital Library
- Translation
- G.C. Macaulay (1890)