What the text says. Luke opens the birth narrative with an imperial registration: a decree of Caesar Augustus "that all the world should be enrolled", and Joseph going up from Nazareth to Bethlehem "because he was of the house and family of David" (Luke 2:1, 2:4, ASV). A Roman census (Greek apographē) was not a head-count for curiosity; it was the assessment on which tax was levied, and provincial registration is well attested across the empire.
The document that shows a census could summon people home. A papyrus in the British Museum, catalogued P.Lond. 904, preserves an edict of Gaius Vibius Maximus, prefect of Egypt (in office 103-107), issued for the house-by-house registration of AD 104. It orders everyone living away from their own administrative district to go back: the house-to-house enrolment having begun, all who for any reason are outside their nome must be summoned to return to their own hearths, so that the customary business of registration can be completed and they can attend to the farming that belongs to them. That is a real Roman census order, from a Roman prefect, requiring travel home to register. It is the closest documentary parallel we have to the movement Luke describes, and it is a parallel, from a different province seventy years later, not a record of Luke's census.
What it does not settle, stated plainly. The edict sends a man back to the district where he is registered and holds property, and critics of Luke's account press exactly that distinction: Robert R. Cargill (University of Iowa) states the objection directly, residents "were not required to return to their ancestral homes, but to their present homes, so that both people and assets could be assessed." On that reading the papyrus supports return-to-your-own-domicile, which is not what Luke describes if Joseph's domicile was Nazareth. Defenders of Luke's account answer that the two categories frequently coincided, that a Judean's registration and any property interest could well lie in his family's town, and that Roman practice in a client kingdom was routinely adapted to local custom rather than imposed uniformly. Both readings are live in the literature; this file picks neither. What the papyrus establishes is narrower and worth having on its own: a census that ordered people to travel home to register is a documented Roman practice, not a thing invented by an evangelist.
The Quirinius problem, the real dating dispute, both sides. Luke dates the registration to when "Quirinius was governor of Syria" (Luke 2:2). The difficulty is a chronological one and it is not manufactured by sceptics: Matthew places Jesus' birth in the reign of Herod the Great (Matthew 2:1), and Herod died in 4 BC, while the census under Quirinius belongs to AD 6-7, a gap of about ten years. Josephus places it after Archelaus was banished and "Judea, which was now added to the province of Syria" came under direct Roman administration (Antiquities 17.13; 18.1.1), and dates its conclusion to "the thirty-seventh year of Cæsar's victory over Antony at Actium" (Antiquities 18.2.1), Actium was 31 BC, which is how that reckoning becomes AD 6-7. Josephus treats that census as a notorious first, the trigger for the revolt of Judas the Galilean.
The proposed answers, attributed, with no winner picked. (1) A grammatical reading: the Greek prōtē in Luke 2:2 is taken comparatively rather than as a superlative, so the verse means this registration happened before the one made when Quirinius was governor, a rendering argued by, among others, N. T. Wright. (2) A two-terms reading: Quirinius held some office or commission in Syria on an earlier occasion as well as in AD 6. (3) An earlier-governor reading: the enrolment began under another Syrian legate (Quintilius Varus governed Syria c. 7-4 BC) and was completed or is being dated by Luke under the better-known Quirinius. (4) The critical reading, argued by Cargill and many others: Luke has simply conflated two events, and the census is a narrative device to move the family to Bethlehem. Each of these has been argued by competent scholars and each has been answered; there is no consensus, and this file does not supply one. What can be said without dispute: the census of AD 6 under Quirinius is a solidly attested historical event, and the relation of Luke's registration to it is genuinely unresolved.
Sources: Luke 2:1-5; Josephus, Antiquities 17.13, 18.1.1 and 18.2.1; ISBE 1915, "Census" and "Quirinius"; Schürer, History of the Jewish People, Div. 1 Vol. 2, excursus on the census; the census edict of Gaius Vibius Maximus (P.Lond. 904, AD 104).
---