Sanctuary treasuries and temple finance

What is attested

Two different things are called treasuries in Greek sanctuaries, and confusing them distorts everything downstream. A thesauros can be a building — a small, temple-fronted structure put up inside a major sanctuary by a single city to hold and display that city's dedications, of which rows survive at Delphi and Olympia. A thesauros can also be a stone offertory box, sunk and lockable, into which visitors dropped coin. And the sanctuary's 'treasure' in the wider sense is the accumulated stock of dedicated objects and money held in the name of the god, kept in temple rooms and administered by named officials. Only the third of these generates the documents that make sanctuary finance visible.

Those documents are of three kinds. Inventories list objects held, item by item, and were produced when one board of officials handed over to the next. Accounts record income and expenditure over a term: rents, interest, sales, wages, purchases. Building accounts record a construction project's finance in detail. All three were routinely inscribed on stone and set up publicly, and it is this publication habit, not the record-keeping itself, that is exceptional and that determines what survives. The inscribed texts refer to underlying records kept on wood and papyrus, which implies that record-keeping was far more widespread than inscription; only some communities, in some periods, converted their records into stone.

At Athens, the treasurers of Athena produced annual inventories of the objects in the rooms of the Parthenon and in the Erechtheion, in separate series for separate rooms, running from the 430s to around 300 BCE. What they list is startlingly miscellaneous: gold crowns and silver vessels weighed to the drachma, but also worn, broken and effectively worthless objects retained because they had been dedicated. The Kallias decrees regulate the handling of sacred money and the treasuries of the Other Gods, and a stele of the board of accountants, the logistai, records loans made from the treasuries of Athena Polias, Athena Nike and the Other Gods to the Athenian state across the first decade of the Peloponnesian War, with interest calculated. This is the clearest surviving demonstration that a Greek state could and did borrow at scale from its own gods, on a reckoning that was afterwards published.

At Delos, the board of hieropoioi administered Apollo's property during the island's period of independence and left a long series of inscribed accounts. Apollo owned farms on Delos, on Rheneia, and, in the later third century, on Mykonos, leased out under written terms with rent payable in cash; the god's capital was lent at interest to individuals and to cities; the accounts also record the sanctuary's routine spending. At Delphi, after the temple of Apollo was destroyed in the 370s BCE, an international board of naopoioi managed the reconstruction under Amphictyonic authority, and its accounts — collected in the second volume of the Delphic corpus, edited by Bousquet — record contributions, contracts, transport costs and credit arrangements across decades of work.

The sample is the thing to keep in view. Athens and Delos dominate the evidence for reasons that have nothing to do with Greek religion in general: both had unusually strong habits of public inscription, both were excavated early and thoroughly, and both had political reasons to publish accounts. Delphi and Epidauros add major building dossiers. Beyond that handful, the great majority of Greek sanctuaries — including most sanctuaries of most cities, and effectively all rural and deme-level shrines — have left no accounts at all. What we possess is a small number of very deep soundings, not a survey.

How the rite worked

Custody was annual, collegiate, and publicly audited. At Athens a board of treasurers of Athena held office for a year and transferred the treasure to their successors by an itemised handover; the inventory inscription is the record of that transfer, which is why its logic is custodial rather than economic. Objects are grouped by the room they sit in, described well enough to be identified again, and weighed where they are precious metal. Condition is recorded — items are noted as broken, incomplete or unweighed — because the outgoing board needed to be discharged of them. Boards were subject to the standard end-of-office scrutiny, and separate boards of accountants reckoned what had passed through.

Income came from several streams, in proportions that varied enormously by sanctuary: dedications from individuals and states; first-fruits levies; rents from land, houses and workshops owned by the god; interest on loans; the contents of offertory boxes; fees and fines specified in sacred regulations; and, at sanctuaries attached to imperial systems, transfers from tribute. Expenditure covered sacrifice and festival costs, wages, repairs, construction, and the manufacture of new plate. Loans ran in both directions: sanctuaries lent to states and to private borrowers, and states borrowed from sanctuaries, with the loans recorded as debts and interest reckoned even where repayment was slow or never completed.

Dedicated objects were not permanently frozen. Sanctuaries periodically melted down accumulated silver and gold plate and reworked it into new, standardised objects — which explains both why inventories track weight so obsessively and why so little dedicated metalwork survives archaeologically compared with what the inscriptions list. The decision to convert dedications was itself an act requiring authorisation and record.

Major building projects were financed through dedicated boards with their own accounting. At Delphi the naopoioi drew on Amphictyonic contributions and on credit, and let contracts for quarrying, transport and cutting; at Epidauros a comparable dossier records the same kinds of transaction for the Asklepieion. These accounts are the closest thing antiquity offers to itemised project costing, and they are also the reason we know unit prices for labour and materials in specific places and years — figures that should not be generalised beyond the site and decade that produced them.