For most of its history the Avesta was not a book. It was memorised and recited by priests, and the tradition held that the sacred word had power in its sound. Writing it came late and deliberately: in the Sasanian period, probably the fifth or sixth century CE, scribes devised the Avestan alphabet, an unusually large set of some fifty-odd letters built out of the Pahlavi script and designed to record vowels, length and fine phonetic distinctions exactly as recited. It is one of the most phonetically precise scripts of the ancient world, and it exists because the point was to fix pronunciation rather than meaning.
Even so, the surviving manuscripts are astonishingly late. The oldest, a Visperad manuscript in Copenhagen, is dated to the thirteenth century, and the oldest securely dated copies of the Yasna and Videvdad are from 1323, some eighteen centuries or more after the Gathas were composed, and seven after the alphabet was invented. Alongside the Avestan text runs the Zand, the Middle Persian translation and commentary made under the Sasanians. Zand is not Avesta; the two are distinct in status, and the European habit of calling the scripture "Zend-Avesta" came from a misreading of that relationship.
Europe knew almost nothing of the text until Anquetil-Duperron enlisted as a soldier of the French East India Company in 1755, spent years at Surat persuading Parsi priests to teach him and to part with manuscripts, and published a three-volume translation in 1771. William Jones promptly denounced it as a forgery, arguing that the work was too tedious to be genuine scripture and that Anquetil had not known enough Persian to read it. Anquetil was vindicated in the 1790s once Germanic scholars examined the manuscripts. The Avesta is still recited today by Parsi priests in India and Irani priests in Iran, in a language none of them speaks.