Greek Fire
UnsolvedSummary
Greek fire was a Byzantine incendiary liquid, projected through tubes from warships, first recorded at Constantinople in the 670s. Its recipe was a state secret, and no recipe survives.
The sources describe what it did: the ships that carried it, the tubes, the fleets it burned, and the story one emperor prescribed for foreigners who asked for it. The two leading modern reconstructions both start from petroleum. Neither can be checked against a recipe.
What is documented
The first record. The main source is the chronicle of Theophanes the Confessor, written in the early ninth century. Under AD 671/2 he records that Constantine IV, told of a great Arab expedition against Constantinople, "built large biremes bearing cauldrons of fire and dromones equipped with siphons" and stationed them at the harbour of Caesarius, which his translators, Cyril Mango and Roger Scott, identify with the Theodosian harbour. Under AD 672/3 he describes seven years of seasonal Arab attacks on the city and the loss of the Arab fleet in a storm off Syllaion. Then he introduces Kallinikos, "an architect from Helioupolis in Syria", who fled to the Romans and "manufactured a naval fire with which he kindled the ships of the Arabs and burnt them with their crews".
The order of events. The siphon ships appear in the entry for the year before Kallinikos arrives. Mango and Scott note that Theophanes "appears to have overlooked" his own previous entry, and that the first use of the fire was allegedly against an Arab fleet that had attacked a coastal town of Lycia, in 671 according to the twelfth-century chronicler Michael the Syrian. J. R. Partington, in his 1960 history, noted that the Patriarch Nikephoros, describing the same siege, does not mention the fire. The siege's date is itself disputed: the Oxford historian Marek Jankowiak places it in 668.
Later record. In 717, with an Arab fleet at the city, Theophanes has Leo III send "the fire-bearing ships from the Acropolis" against part of it, after which the enemy "cowered with fear after experiencing the efficacious action of the liquid fire". In 812 the Bulgars took Mesembria and Debeltos and found "36 brass siphons and a considerable quantity of the liquid fire". Anna Komnene, writing the history of her father Alexios I, describes the fleet he built against Pisa after the crusaders took Jerusalem: brass or iron heads of lions and other animals on the prows, with the fire "directed against the enemy through tubes" passing through their open mouths. At Dyrrachium (Durrës), besieged by Bohemond in 1107 to 1108, she records the defenders burning a tunnelling party with pine resin "mixed with sulphur", blown through reed pipes and lit at the far end.
The secret. In a book of instruction for his son Romanos, completed around 951 to 952 and known as the De Administrando Imperio, Constantine VII told him what to say if foreigners ever demanded "the liquid fire which is discharged through tubes". The answer was that an angel had revealed it to Constantine the Great, with the charge that it be made only among Christians and only in the city they ruled. The son was to add the story of a military governor who handed some of it to foreigners for a bribe and was consumed by fire from heaven as he was about to enter a church. The book's editors describe the chapter as "a general directive on foreign policy from the emperor's own pen". It is a script for refusing requests, not a history. In the eleventh century the chronicler Kedrenos claimed that a descendant of Kallinikos named Lampros still made the fire. Partington judged Lampros "a purely fictitious person", lampron being a late Greek name for fire.
The names. Theophanes calls it naval fire, liquid fire and Roman fire. Constantine VII calls it liquid fire. Neither calls it Greek fire. Michael the Syrian, as Partington notes, called it naphtha.
Leading explanations
A petroleum base (the leading hypothesis, not established). Partington concluded that Greek fire was "a distilled petroleum fraction combined with other specified ingredients, but not containing saltpetre", the only theory he found consistent with the descriptions. In 1977 J. Haldon and M. Byrne proposed instead undistilled crude petroleum, kept in heated containers aboard ship; they doubted that Byzantine technicians could distil it. Both start from petroleum, which, as Partington pointed out, would float on water and keep burning. What was added, how it was prepared and how the siphon worked remain disputed. Partington read Anna Komnene's resin and sulphur as a description with the essential ingredient, petrol, deliberately left out; that is his interpretation, and her fire was blown through reed pipes, not pumped from a ship.
Saltpetre (disputed). An older theory made Greek fire an early gunpowder, with saltpetre as the key ingredient. Partington rejected it. Bert S. Hall, introducing the 1999 reprint of Partington's book, reports that Emmerich Pászthory (1986) revived it for the later hand-held weapons, the "fire lances", which he argued must have used an oxidiser, "most likely saltpeter". Hall concluded that Partington's refutation "has also been left behind in later work".
Water-triggered mixtures (rejected). Other theories had water set the fire off: quicklime, which heats when wet, or calcium phosphide, which gives off a self-igniting gas. Partington dismissed both. Quicklime "would not be a suitable material at all" and is never mentioned as a component; crude calcium phosphide "would sink rapidly in water".
What the popular version gets wrong
"Kallinikos brought the secret to Constantinople, and it saved the city from the great siege of 674 to 678." Theophanes, the source of the story, has siphon ships at Constantinople the year before Kallinikos appears and, as Partington read him, places his arrival at the end of the seven years of fighting. He took the account to mean that the fire was invented in Constantinople rather than brought from Syria. The first use may have been off Lycia, Nikephoros' account of the siege does not mention the fire, and the date of the siege is disputed. The version in which a refugee's secret saves the capital is tidier than any of its sources.
"Kallinikos was a Jewish refugee." Theophanes gives his city and his trade; Constantine VII gives his city. Neither gives his religion. Partington, in 1960, thought it "more than likely that he was a Jew", on no evidence he states.
"It caught fire by itself and could not be put out with water." The self-igniting part belongs to the quicklime and calcium phosphide theories, which Partington rejected. Anna Komnene's fire at Dyrrachium was lit at the end of the pipe. That a fire kept burning on water is what burning petroleum does.
Current status
Unsolved. What Greek fire did is well recorded. What it was made of is not: no recipe survives. The leading reconstructions share a petroleum base, and Britannica calls it "evidently a petroleum-based mixture" whose precise composition "remains unknown to this day". The additives, the preparation and the working of the siphon are disputed, and Hall, in 1999, thought it "doubtful that the last word has been written on this subject". A surviving recipe, or a chemical residue that could be tied to a Byzantine siphon, would change the status. Neither is known to exist.
Sources
- Theophanes the Confessor. The Chronicle of Theophanes Confessor: Byzantine and Near Eastern History, AD 284 to 813. Translated by Cyril Mango and Roger Scott. Oxford: Oxford University Press, 1997. Entries for AM 6164, 6165, 6209 and 6305, with the translators' notes.
- Constantine Porphyrogenitus. De Administrando Imperio. Greek text edited by Gy. Moravcsik, English translation by R. J. H. Jenkins. Washington, DC: Dumbarton Oaks, 1967. Chapters 13 and 48, and the editors' introduction.
- Anna Comnena. The Alexiad. Translated by Elizabeth A. S. Dawes. London, 1928.
- Partington, J. R. (1960). A History of Greek Fire and Gunpowder. Cambridge: W. Heffer and Sons. Reprinted Baltimore: Johns Hopkins University Press, 1999.
- Hall, B. S. (1999). "Introduction, 1999." In the Johns Hopkins reprint of Partington, above. The source here for the work of Haldon and Byrne and of Pászthory.
- Haldon, J. and Byrne, M. (1977). "A Possible Solution to the Problem of Greek Fire." Byzantinische Zeitschrift 70 (1), 91 to 99. Known here through Hall (1999); not seen directly.
- Encyclopaedia Britannica. "Greek fire." Online edition, consulted September 2026.
- University of Oxford, Faculty of History. Profile of Dr Marek Jankowiak, consulted September 2026.
Last reviewed: September 2026. Records are provisional. Where the evidence changes, the entry changes. Found an error? Tell us.