The Census of Quirinius

"αὕτη ἀπογραφὴ πρώτη ἐγένετο ἡγεμονεύοντος τῆς Συρίας Κυρηνίου" -- Luke 2:2¹

"a headless inscription is at best a dangerous ally who may change sides at any moment." -- A. N. Sherwin-White, 1963²

Summary

Luke ties the birth of Jesus to two clocks.  One is Herod the Great, king when the story begins.  The other is a census "while Quirinius was governing Syria."  The only census under Quirinius that a source outside the Gospels dates is the census of Judea that Josephus places in AD 6 and 7, after Herod's son Archelaus was removed.  Herod died about nine years before that on the majority date, and about six years before it on the best-defended minority date.  So either Luke has the census wrong, or one of the proposed bridges between the two clocks is true.  This article tests each bridge against the sources.

Confidence words are used for one question at a time.  "Fixed" marks a date strongly supported by the chronological evidence, while leaving it open to revision if that evidence fails.  "Range" means the sources limit a date to a span.  "Reconstruction" means an inference from the surviving evidence rather than an event independently recorded as such.  "Possible" marks an option the evidence does not rule out; its positive support is assessed separately.  Agreement between two writers counts for more only when there is a reason to think they did not draw on each other.

1. What Luke claims.  Luke 1:5 sets the story "in the days of Herod king of Judea."  On a normal pregnancy and a continuous story, Jesus is born roughly fifteen months after that opening scene, but Luke never says Herod was still alive at the birth.  Matthew does say so.  Luke 2:1-2 says a decree of Augustus ordered "all the world" to be enrolled, and that "this enrollment was first" while Quirinius governed Syria.  Luke 3:1 and 3:23 make Jesus "about thirty" when he begins, after John begins in the fifteenth year of Tiberius, AD 28/29.  These markers point to a birth around the end of Herod's reign.  A birth in AD 6 would give an age between 21 and 23 during AD 28/29.  If Jesus began soon after John, that strains "about thirty."

2. Fixed:  Quirinius took a census of Judea in AD 6/7.  Josephus says Archelaus's land was added to the province of Syria when he was deposed, and that Quirinius came to assess property there.  He dates the end of the assessments to the thirty-seventh year after Actium, autumn AD 6 to autumn AD 7.  Cassius Dio, who arranges his history by consular years, puts the removal of Archelaus in AD 6.  Acts 5:37 shows that Luke knew a census linked to the revolt of Judas the Galilean.  One oddity in Josephus is real:  the high priest Joazar is appointed by Herod, removed by Archelaus, and then appears again in office in AD 6 with no reappointment recorded.  John Rhoads argues from this and other clues that Josephus misdated the census.  That case has not convinced most historians, and it rests on identifications no source states.

3. Range:  Herod died before AD 6 on both chronologies examined here.  The common date is spring 4 BC.  Andrew Steinmann and Rodger Young defend early 1 BC.  Steinmann and Young themselves keep AD 6 as Archelaus's last year, so the later date does not move the census.  It does change which earlier proposals could fit inside Herod's lifetime.

4. An earlier Syrian post for Quirinius is a disputed reconstruction.  Tacitus's short obituary mentions his consulship, his war on the Homonadenses "throughout Cilicia," and his service as adviser to Gaius Caesar.  It also leaves out his securely attested Syrian governorship of AD 6, so its silence about an earlier one carries little weight.  Josephus names Sentius Saturninus and then Quinctilius Varus as governors in Herod's last years.  The governor between about 4 BC and 1 BC is unknown.  Theodor Mommsen argued in 1883 that the unnamed man on the Tiburtine stone was Quirinius and that it records two Syrian legateships, the first about 3-2 BC.  Ronald Syme and others answered that the stone may commemorate someone else and that the Homonadensian war was fought from Galatia.  The question is still open.

5. The inscriptions add less than either side often claims.  The Lapis Tiburtinus has lost its name.  It says an unnamed man held Syria as legate iterum, "a second time."  Hermann Dessau, who assigned it to Quirinius following Mommsen, dated that second post to AD 6.  The Lapis Venetus names Quirinius as legate of Syria ordering a census of Apamea, but gives no year.  The two stones from Pisidian Antioch make Quirinius an honorary chief magistrate of that Roman colony.  That is a city office in Asia Minor, and it says nothing about a shared governorship of Syria.

6. Roman census practice makes some of Luke's details plausible and leaves others unattested.  Augustus counted Roman citizens in 28 BC, 8 BC and AD 14, and had provinces assessed at different dates.  No source earlier than Luke, or clearly independent of him, records one decree enrolling the whole empire at once;  the late writers who do, from Orosius in AD 417 to the tenth-century Suda, repeat Luke's words or story.  Augustus's own inventory of the empire listed the client kingdoms beside the provinces, and the people of Paphlagonia swore allegiance to him district by district in March of 3 BC, but neither is an enrollment of Herod's subjects.  In Egypt people registered at their legal home, and owners also declared where their property lay.  In the province of Arabia in AD 127, a Jewish woman, Babatha, declared her own lands at the provincial center of Rabbath-Moab with her guardian present.  No document makes descent alone a reason to register in a town.  A Roman-style census was imposed on subjects of a client ruler in Rough Cilicia in AD 36, enforced in the end by troops from Syria.  So a census inside a client kingdom is not unheard of.  No independent source records one in Herod's kingdom.

7. The church fathers add claims, not records.  Justin Martyr, about AD 155, says the birth can be checked in the census lists made under Quirinius, "your first procurator in Judea."  Tertullian, about AD 208, asserts that Augustus's census in the Roman archives is a "most faithful witness" to the birth, and in another place says censuses were held in Judea "by Sentius Saturninus."  Neither writer says he read the records, and the editor of Tertullian notes no non-Christian evidence that they were kept or available.  Tertullian's Saturninus is the one ancient detail that names a governor from Herod's lifetime.  It may preserve a tradition or may be a learned correction of Luke, and the text does not settle which.

8. Each harmonization needs at least one step that no independent source records.  The proposals are an earlier Syrian post for Quirinius, a special command or procuratorship under Saturninus, a phased census, an enrollment held by Herod, a misdated census in Josephus, and a reading of Luke 2:1-5 as a flash-forward to AD 6.  Some of these can be combined.  The translation "this census took place before Quirinius governed Syria" has a real grammatical parallel in the Greek of Jeremiah, but not an exact one, and it leaves unexplained why Luke named the later governor.

Verdict.  The most probable explanation is that Luke, or his source, linked the birth to the famous census of AD 6.  That is an error of date or a conflation, perhaps of a real earlier enrollment with the famous one.  The best rival joins an earlier Syrian post for Quirinius to an enrollment in Herod's last years.  It has some indirect support, but it needs several steps that are each uncertain.  The error explanation also has a cost:  it requires Luke to have set two of his own markers against each other.  A skeptic may fairly call Luke 2:2 a probable mistake.  A defender may fairly say that the unknown governor of 4 to 1 BC, the disputed Tiburtine stone and Tertullian's Saturninus leave a real opening, though no source fills it.

Part 1: What Luke says

The sentence

Luke 2:1-5 reads, in the standard critical Greek text:

Ἐγένετο δὲ ἐν ταῖς ἡμέραις ἐκείναις ἐξῆλθεν δόγμα παρὰ Καίσαρος Αὐγούστου ἀπογράφεσθαι πᾶσαν τὴν οἰκουμένην· αὕτη ἀπογραφὴ πρώτη ἐγένετο ἡγεμονεύοντος τῆς Συρίας Κυρηνίου· καὶ ἐπορεύοντο πάντες ἀπογράφεσθαι, ἕκαστος εἰς τὴν ἑαυτοῦ πόλιν.¹

In plain English:  "In those days a decree went out from Caesar Augustus that all the world should be enrolled.  This enrollment was first, while Quirinius was governing Syria.  And all went to be enrolled, each to his own town."  Joseph went from Nazareth in Galilee to Bethlehem in Judea "because he was of the house and family of David," to be enrolled "with Mary, who was engaged to him, being pregnant."¹

Four words carry the argument.

1. ἀπογραφή (apographē), "enrollment" or "registration." It is the word for writing a person or property into a list. Josephus uses it for the census of AD 6, beside ἀποτίμησις, "assessment" of property.³
2. πρώτη (prōtē), "first." It is an adjective agreeing with "enrollment." A. T. Robertson's grammar takes the noun and adjective together as the predicate: "this happened as a first enrollment."⁴ Part 6 tests the proposal that it means "before."
3. ἡγεμονεύοντος (hēgemoneuontos), "while governing." This participle comes from ἡγεμών, a general word for a leader or governor. Luke uses the same participle for Pontius Pilate, an equestrian prefect of Judea, and the related noun for the reign of the emperor Tiberius.¹ So the word tells us Quirinius was in charge of Syria in some sense. It does not tell us his rank.
4. Κυρηνίου (Kyrēniou), "Quirinius." This is the Greek form of the name of Publius Sulpicius Quirinius, the senator Josephus also names.

The text is secure as far as the editions show.  The SBL edition's apparatus compares it with other printed editions, and the only difference in this verse is an added article, "this the enrollment," in the Byzantine text.¹  That apparatus lists editions, not manuscripts.  Brown reports no manuscript evidence for reading Saturninus in place of Quirinius.⁵  The SBL editor also prints 2:2 in parentheses, as an aside within the sentence about the decree.

Luke's three clocks

Luke gives three time markers for the birth.  They have to be read together, and each carries its own assumptions.

1. Herod. Luke 1:5 opens "in the days of Herod king of Judea."¹ That dates the scene with Zechariah. Elizabeth conceives "after these days" (1:24), and the angel comes to Mary "in the sixth month" (1:26, 1:36). On a normal pregnancy, and reading the story as continuous, Jesus is born about fifteen months after the scene in 1:5. Luke gives no exact intervals, so fifteen months is an estimate, not a figure he supplies. Raymond Brown, whose 1977 commentary is a standard critical study of the infancy stories, draws the same conclusion: "This would fix the year of his birth at no later than 3 b.c."⁵ Brown used the 4 BC date for Herod's death. Luke does not say Herod was still alive when Jesus was born. On Luke's words alone, a birth up to about a year after Herod's death is not excluded.
2. Quirinius. The enrollment of 2:2.
3. Tiberius. John the Baptist begins "in the fifteenth year" of Tiberius (3:1). Jesus is "about thirty" when he begins (3:23).¹ Tiberius succeeded Augustus in August AD 14, so his fifteenth year is AD 28/29 on the usual count. Luke does not say how long after John Jesus began. The usual reading puts the two close together, because Luke tells them in one sequence. "About thirty" is also a rounded age, not an exact one.

The table gives the possible attained ages during AD 28/29 for a given birth year, with no year zero between 1 BC and AD 1.  The range includes both calendar years and the unknown birthday.⁶

If Jesus began a few years after John, every age rises by those years.  An AD 6 birth would still give an age of 27 or less for any start up to AD 33.  So the age marker does not set a sharp cutoff, but it fits a birth near the end of Herod's reign far better than one in AD 6.

Matthew gives an independent Herodian clock, but not Luke's other two.  He says Jesus was born "in the days of Herod the king," that Herod later killed the boys of Bethlehem "two years old and under," and that the family returned after Herod died, when Archelaus ruled Judea.¹  So Matthew, unlike Luke, puts the birth before Herod's death.  He does not mention a census or Jesus' age.

So the question is narrow.  Luke 1 and 3 point to the last years of Herod, or to the year or so after his death.  Matthew points to the years before it.  Luke 2:2 names the governor whose only independently dated census came about five to ten years later.  The early church's dates for the birth, which Part 5 sets out, agree with the first group, but Clement, for one, proves his year from Luke 3:1 and 3:23, so these dates are at least partly calculations from Luke, not a separate witness.⁷

Part 2: The fixed points

The census of AD 6, in Josephus

Josephus, writing his Jewish Antiquities in the early 90s, describes the end of Archelaus's rule.  In the tenth year of Archelaus, the leading Jews and Samaritans accused him before Augustus, who exiled him to Vienne in Gaul.³  His Jewish War says the ninth year.⁸  Then:

So Archelaus's country was laid to the province of Syria; and Cyrenius, one that had been consul, was sent by Cæsar to take account of people's effects in Syria, and to sell the house of Archelaus.⁹

Book 18 opens by describing Quirinius as a senator who had risen through the offices to the consulship, sent "to be a judge of that nation, and to take an account of their substance."  With him came Coponius, "a man of the equestrian order," to govern the Jews.  "Moreover, Cyrenius came himself into Judea, which was now added to the province of Syria, to take an account of their substance."⁹  The Jews at first took "the report of a taxation heinously," but most yielded at the urging of the high priest Joazar.  Judas, whom Josephus here calls a Gaulonite from Gamala and elsewhere the Galilean, joined a Pharisee named Saddok and called the assessment "no better than an introduction to slavery" and urged revolt.⁹

Josephus then dates the end of the work:  the taxings "were come to a conclusion, which were made in the thirty-seventh year of Cæsar's victory over Antony at Actium."⁹  Actium was fought on 2 September 31 BC.  The thirty-seventh year after it runs from autumn AD 6 to autumn AD 7.  A. N. Sherwin-White, the Oxford historian of Roman law whose 1963 lectures remain a standard treatment of Roman matters in the New Testament, puts it the same way:  "strictly A.D. autumn 6 to autumn 7."²

A second clock agrees.  Cassius Dio, a Roman senator writing about AD 220 who arranges his history by consular years, records under the consuls of AD 6 that "Herod of Palestine," whom the translator identifies as Archelaus, was accused by his brothers and banished beyond the Alps.¹⁰  Josephus returns to the census three more times, each time in connection with Judas or his descendants.³⁸⁸  Acts 5:37, written by Luke, refers to Judas the Galilean "in the days of the census," using the same word ἀπογραφή.¹  Acts ties Judas to "the census" but gives no year, so it confirms the link, not the date.  Dio's sources for AD 6 are not known.  Nothing suggests he read Josephus, and he follows Roman consular years rather than Jewish history, so his agreement is probably independent, though that cannot be proved.

Three details matter for Luke.

1. The census was treated as new. Josephus gives the reasons for the resistance. Judas called the assessment "no better than an introduction to slavery."⁹ In the War, Judas reproached the people for being willing to pay tribute to the Romans and to accept mortal masters after God.⁸ So the grievance was direct Roman taxation and Roman rule, which began only when Archelaus's land became a province. Brown calls Josephus's silence "an eloquent argument that there was no Roman census of Palestine before the census under Quirinius," and his argument is about a census imposed by Rome with a Roman tax.⁵ That silence weighs heavily against a Roman census and Roman tribute in Herod's kingdom. It weighs much less against an enrollment that Herod ran for his own taxes or for the oath to Caesar, which need not have provoked the same response. Josephus does record the oath of loyalty in Herod's last years, because more than 6,000 Pharisees refused it and were fined.³ So he noticed administrative acts in those years when they caused a religious conflict. A quiet registration would be less likely to reach him. Sherwin-White, discussing a different silence, warns that Josephus's "primary interest was in Judaea," so what he omits about Syria is "not decisive."² Detailed coverage of court intrigue does not show detailed coverage of routine tax records.
2. Galilee was not included. Only Archelaus's territory, Judea, Samaria and Idumea, became Roman. Josephus says that after it became a province, "Philip and Herod called Antipas" went on ruling their own tetrarchies.⁸ Nazareth was in Antipas's Galilee. Brown: "at that time Galilee was not under Quirinius’ direct supervision but was a tetrarchy ruled by Herod Antipas."⁵ Harold Hoehner, a defender of Luke's accuracy, makes the same point from the other side: "It is highly implausible that after the division of the kingdom, the residents of Herod Antipas’ territory (Nazareth) would go to Archelaus’ territory (Bethlehem) for a census for purposes of taxation."¹¹
3. Quirinius was a consular senator, Coponius the equestrian governor of Judea. Josephus calls Coponius Judea's ἐπίτροπος, procurator, "of the equestrian order."⁸ Quirinius was the senior man over the whole province. This bears on the claim, tested in Part 6, that Luke's word means "procurator."

Could Josephus have misdated it?

One part of Josephus's story is genuinely awkward.  Herod made Joazar high priest shortly before he died.³  Archelaus removed Joazar on his return from Rome, charging him with siding with the rebels.³  Yet in AD 6 Josephus has "the high priest Joazar" persuading the people to accept the census, and Quirinius then removes him.³  Josephus records no reappointment.

John Rhoads argued in 2011 that Josephus "misdated the census."  He holds that the Judas of AD 6 is the same man as a Judas active at Herod's death, that the Joazar data fit 4 BC better, and that Quirinius should be identified with Sabinus, Augustus's procurator of Syria in 4 BC.  On his reconstruction the census "began before Herod the Great’s death."¹²  He suggests it began about 5 BC and was "not completed until 3/2 bc."¹²  Rhoads concedes that "these three sets of data taken individually may not be ultimately persuasive."¹²

The Joazar problem is real evidence that Josephus's sources for these years were patchy.  It is weaker evidence for Rhoads's whole reconstruction.  A reappointment Josephus failed to record would also explain it.  His key step, that Sabinus and Quirinius are one man, is stated by no source:  Josephus names both and never links them.  Dio's AD 6 for Archelaus and Josephus's thirty-seventh year after Actium also point the same way.  The proposal is a reconstruction, and a costly one.

When Herod died

Josephus gives Herod a reign of thirty-seven years from his appointment by Rome and thirty-four from his capture of Jerusalem.³  He mentions an eclipse of the moon shortly before Herod's death and a Passover shortly after.³  On the usual reckoning, appointment in 40 BC and capture of Jerusalem in 37 BC, Herod died in spring 4 BC.

Herod's sons supply a check that does not depend on those starting points.  Philip died "in the twentieth year of the reign of Tiberius," AD 33/34, after ruling thirty-seven years, which counts back to 4 BC.³  Antipas's last coins are dated year 43, and he was removed in AD 39, which again counts back to about 4 BC.¹³  Archelaus was deposed in his ninth or tenth year, in AD 6.

W. E. Filmer in 1966, and Andrew Steinmann with Rodger Young more recently, argue for early 1 BC.  Their case is that Josephus's consular dates for Herod's appointment and conquest are each a year too early, that Appian and Dio put Herod's appointment after the treaty of Misenum in 39 BC, that Josephus did not count reigns by the inclusive method the 4 BC date needs, and that the days between the start of the month and the Passover are too few for Herod's funeral.  To explain the sons' coins they argue that all three sons backdated their reigns to 6 BC, during Herod's lifetime.¹³  That is a real argument with real evidence, such as Herod's statement that he had shared royal power with his son Antipater, and it has convinced a minority.  It also needs three separate rulers to have adopted the same backdated starting point.  Timothy Barnes answered Filmer in 1968, and Steinmann and Young themselves call 4 BC the "Consensus view."¹¹¹³  Josephus's eclipse cannot settle the dispute, because each date has one.  NASA's catalogue lists a partial eclipse on 13 March 4 BC, with about a third of the Moon's diameter inside the Earth's full shadow, a total eclipse on 10 January 1 BC, and two total eclipses in 5 BC.¹⁴

For the AD 6 census by itself the choice between 4 and 1 BC matters less than it seems.  Steinmann and Young themselves give Archelaus's "Last year" as "AD 6 according to Dio Cassius."¹³  The census followed Archelaus, not Herod.  So:

1. Herod died in spring 4 BC. Luke's Herodian setting puts the birth by roughly 3 BC, and Matthew's before spring 4 BC. The census is at least eight years later.
2. Herod died in early 1 BC. The birth is by roughly AD 1 on Luke's setting, and before 1 BC on Matthew's. The census is still at least five years later.

The later date suits Luke's "about thirty," and it fits the early church's 3/2 BC.  It does not bring the AD 6 census any closer.  What it does change is which earlier office for Quirinius could fall inside Herod's lifetime, which matters for proposal B in Part 6.

Timeline

The rows on the Syrian governors come from Josephus and other ancient evidence as gathered by Sherwin-White and Brown.²⁵  Two rows deserve a warning.  The dates for Saturninus and Varus are ranges, not fixed years.  Josephus gives no years for them, so every list fits his narrative to other evidence, and lists differ by about a year:  Evans gives Saturninus 9-6 BC, Carrier's list 10-7 BC.¹⁵¹⁶  That is a difference of reconstruction, not an error on either side.  Carrier's list also fills the 4 to 1 BC slot with Calpurnius Piso, following Syme's reading of the Tiburtine stone.  That too is a reconstruction:  no text names the governor of those years.  The 4 to 1 BC row is a true gap:  no source names the governor.

Part 3: Quirinius's career and the inscriptions

What the literary sources say

Tacitus gives Quirinius an obituary when he died in AD 21.  He was not from the old patrician Sulpicii but from the town of Lanuvium.  "Impiger militiae," a tireless soldier, he won the consulship under Augustus, then triumphal honors for storming the forts of the Homonadenses in Cilicia, and was made "rector," adviser, to Gaius Caesar when Gaius held Armenia.¹⁷¹⁸  The consulship was in 12 BC, or 742 AUC in Dessau's commentary.¹⁹  Strabo describes the Homonadensian campaign without a date.²⁰²  Suetonius remembers him as a rich, childless old consular who had his ex-wife Lepida condemned.²¹

Tacitus's notice is a selective obituary, not a list of offices.  It leaves out Quirinius's governorship of Syria in AD 6 and his census of Judea, both attested by Josephus.  The Venetian stone independently names his Syrian office and a census of Apamea, without a year.  So Tacitus's silence about an earlier Syrian post is little or no evidence against one.  He picked a few points for a short character sketch.

The campaign is a separate question.  Tacitus puts the Homonadensian forts "per Ciliciam," throughout Cilicia.  Theodor Mommsen argued in 1883 that Cilicia then answered to the legate of Syria, so Quirinius must have fought the war as legate of Syria, and he dated that first Syrian post to 3-2 BC.²²  Ronald Syme argued in 1934 that the war was fought from the province of Galatia, which borders the Homonadenses.  Sherwin-White accepts Syme:  "So the best evidence for an earlier legateship in Syria collapses."²  Syme's study was not read for this article, so its argument is known here only through Sherwin-White and Brown.  Neither side has a text naming the province from which Quirinius fought.

Josephus names the governors of Herod's last years.  Saturninus and a procurator named Volumnius are both "Caesar's governors," and both are said to be "in charge of Syria" in the dispute with the Nabateans.³  At the trial of Herod's sons at Berytus "the governors" presided:  "Saturninus and the legates with Pedanius, and with them Volumnius the procurator."⁸  Then Varus came "as successor to Saturninus in the rule of Syria."³  This is real evidence for one part of the defenders' case.  Josephus shows several Roman officials of different ranks acting together in Syria, and he calls a procurator one of the "governors."  It names none of them Quirinius.

The Lapis Tiburtinus

Object.  A marble fragment found near Tibur, modern Tivoli, in 1764, and now in the Vatican Museums.⁵  The top, with the man's name, is lost.

Text.  Hermann Dessau's standard edition prints, with square brackets for the editor's restorations:

[...r]egem, qua redacta in [potestatem imp. Caesaris] Augusti populique Romani senatu[s dis immortalibus] supplicationes binas ob res prosp[ere gestas et] ipsi ornamenta triumph[alia decrevit]; pro consul Asiam provinciam op[tinuit; legatus pr. pr.] divi Augusti iterum Syriam et Ph[oenicen optinuit].¹⁹

That is:  "...king; for which, brought into the power of Augustus and the Roman people, the senate decreed two thanksgivings to the immortal gods, and to him the triumphal ornaments.  As proconsul he held the province of Asia.  As legate of the deified Augustus he held Syria and Phoenicia iterum, a second time."

Date.  After AD 14, since Augustus is "deified."

Identification.  Theodor Mommsen and William Ramsay gave it to Quirinius.  Ramsay in 1898 claimed "practical certainty."²³  Syme argued for L. Calpurnius Piso, who, as Richard Carrier summarizes, governed Asia and twice defeated Thracian kings.¹⁶  Brown summarizes:  "The thesis that this is a reference to Quirinius is a pure guess with as many dissenters as adherents."⁵  Sherwin-White adds two problems:  the stone mentions a king, and "there was no king of the Homonadenses";  and "It is also not known that Quirinius was ever proconsul of Asia."²  Dessau's commentary had already offered an answer to the first:  the king is Amyntas of Galatia, whom the Homonadenses killed, and the people subdued is the Homonadenses.¹⁹

Inference.  The stone does not name its man, so everything depends on two separate judgments:  who he was, and what iterum qualifies.  Mommsen, whom Dessau cites for the attribution, read the man as Quirinius and iterum as a second governorship of Syria itself.  His reason is a point of Latin usage:  a governor is said to hold a post "again" only "cum eadem provincia ab eodem homine pluries" is governed, when the same man governs the same province more than once.²²  Dessau's note on iterum reads simply "A. p. Chr. 6," AD 6.¹⁹  That dates the second Syrian term.  It does not deny a first one.  The other reading takes iterum with legatus, "legate for the second time," with Syria as the second legateship and the first elsewhere.  Sherwin-White's balance is fair:  it "could, but need not, mean that the nameless person was twice legate of Syria."²  If the man is Quirinius and Mommsen's reading is right, the stone is positive evidence for an earlier Syrian term.  If either judgment fails, it is evidence for nothing about Quirinius.

The Lapis Venetus

Object.  The epitaph of Quintus Aemilius Secundus.  Its upper part was copied in Venice in 1674 and later lost, and the text was long thought a forgery.  The lower part was found in Venice in 1880, which vindicated the copy.  Mommsen argued that the stone came from Berytus, modern Beirut, the only Roman colony in Syria under Augustus.²⁴⁵

Text.

Q. Aemilius Q. f. Pal. Secundus [in] castris divi Aug. s[ub] P. Sulpicio Quirinio le[gato] C[a]esaris Syriae honoribus decoratus, pr[a]efect. cohort. Aug. I, pr[a]efect. cohort. II classicae; idem iussu Quirini censum egi Apamenae civitatis mil[lium] homin. civium CXVII; idem missu Quirini adversus Ituraeos in Libano monte castellum eorum cepi...¹⁹

That is:  "Quintus Aemilius Secundus, son of Quintus, of the Palatine tribe, in the service of the deified Augustus under Publius Sulpicius Quirinius, legate of Caesar in Syria, decorated with honors, prefect of the first Augustan cohort and of the second naval cohort.  Also, by order of Quirinius, I held the census of the community of Apamea, 117,000 citizen persons.  Also, sent by Quirinius against the Ituraeans on Mount Lebanon, I took their fort."

Date.  After AD 14 for the stone.  The census itself is undated.

Identification.  Certain.  Quirinius is named, as legate of Syria.

Inference.  It shows that Quirinius, as legate of Syria, had a census taken of a Syrian city.  It does not say when.  Dessau's note connects it with the provincial census of Luke and Josephus, and Mommsen lists it under Quirinius's legateship of AD 6.²²  Sherwin-White:  it "confirms the legateship of Syria and the census, but provides no clue as to date."  He adds that this "is likely to have been at least the second census taken in Syria under Augustus. Possibly its predecessor was taken by Sentius Saturninus c. 9 B.c."²  That last sentence is a guess, and Sherwin-White labels it one.  The Apamea census is a census of a city in the province of Syria, not of Judea, so it does not itself fix the year AD 6 either.  Its natural home is the known legateship of AD 6.  If Quirinius held an earlier Syrian post, the stone could belong to that one instead, so it neither proves nor disproves an earlier term.  Titus Kennedy prefers an 8 BC assessment run by Quirinius as a military commander under Saturninus, linking it with Augustus's citizen count.²⁵  The stone supports military involvement in an assessment, but supplies neither that date nor the proposed division of authority.  Its census of Apamea is local; treating it as part of a wider program does not establish which program or who was counted.

The stones from Pisidian Antioch

Object.  Two statue bases found by Ramsay's team in 1912 and 1913 near Pisidian Antioch, a Roman colony in central Asia Minor.  They honor a local man, Gaius Caristanius Fronto.

Text.  Dessau's edition of the first:

C. Carista[nio] C. f. Ser. Front[oni] Caesiano Iul[io] praef. fab., pon[tif.], sacerdoti, prae[f.] P. Sulpici Quirini IIv[iri], praefecto M. Servili. Huic primo omnium publice d. d. statua posita est.²⁶

That is:  "To Gaius Caristanius Fronto Caesianus Iulius, son of Gaius, of the Sergian tribe, prefect of engineers, pontifex, priest, prefect for Publius Sulpicius Quirinius the duumvir, prefect for Marcus Servilius.  To him first of all a statue was set up at public cost by decree of the councillors."

Date.  Not stated.  Ramsay dated the first statue "probably 8 B.C."²⁷  Dessau suggested Quirinius may have been duumvir "when he was accompanying Gaius Caesar in the East," about 1 BC to AD 4.²⁶  Sherwin-White allows either the Homonadensian war or the mission with Gaius.²

Identification.  Certain.

Inference.  A Roman colony's chief magistrates were a pair of duoviri.  Sherwin-White calls Quirinius's post here "an honorary municipal office," and Ramsay's own caption calls Quirinius "honorary duumvir."²⁷²  Caristanius was the praefectus, the local deputy, who acted for Quirinius when the colony gave him the office, and he did the same for M. Servilius.  Dessau's note calls Servilius "hic quoque sine dubio duovir Antiochiae," "he too without doubt duumvir of Antioch."²⁶  So the stone shows Quirinius was a patron of one colony in Asia Minor.

Kennedy treats the duumvirate as supplementary support for shared authority with Saturninus.²⁵  The word does describe one of a pair of joint magistrates, but of the colony, not of a province.  The stone does not mention Syria, Judea or Saturninus.  It is not evidence for a shared governorship.  His proposed division of authority may still be true, but its support would have to come from somewhere else.  Kennedy himself declines to identify the unnamed Tiburtine official confidently with Quirinius.²⁵  His defense therefore should not be confused with Mommsen's two-governorship case.

What the stones and books add up to

None of these sources securely dates a Syrian office of Quirinius before AD 6.  The Tiburtinus is the one item that could be positive evidence for it, and only if two uncertain readings both hold:  that the man is Quirinius, and that iterum means a second Syrian term.  The Venetus names a Syrian post but leaves its year open.  So an earlier term is a live reconstruction with some indirect support, not a demonstrated fact and not an empty guess.

Part 4: What Roman census practice can and cannot show

"All the world"

Augustus did take censuses, and he boasted of them.  In his own Res Gestae he lists three lustra, counts of Roman citizens:  4,063,000 in 28 BC, 4,233,000 in 8 BC, and 4,937,000 in AD 14.²⁸  These were counts of citizens, the old Republican census revived.  Provincials who were not citizens were not in them.  Nothing in Luke or Acts suggests that Joseph, a Galilean villager, was a Roman citizen.  Kennedy links Luke's enrollment with the 8 BC count and the Egyptian evidence for returning home.²⁵  Those are two different bodies of evidence: the citizen totals establish a count across the empire, while the papyri establish provincial registrations.  An assessment of Herod's subjects alongside the citizen census is possible, but neither source records that extension.  Kennedy's preferred birth date, 8 or 7 BC, remains a reconstruction.²⁵

Provincial assessments were a separate thing, held province by province at different times.  The Gauls were assessed in 27 BC and again in 12 BC by Drusus, and a later Gallic census caused unrest.²⁹³⁰  In Egypt the papyri show a census in AD 11/12 and a register in AD 4/5.  Roger Bagnall and Bruce Frier, in the standard study of the Egyptian returns, reconstruct four Augustan censuses at seven-year intervals, 11/10 BC, 4/3 BC, AD 4/5 and AD 11/12, "for three of which (all except that of year 27) there is some direct evidence";  Augustus's year 27 is 4/3 BC.  They warn that "There are as yet too few points of evidence to make this reconstruction more than suggestive."³¹  Syria had a census under Quirinius, attested by the Apamea inscription.

No source earlier than Luke, or clearly independent of him, records one decree ordering all inhabitants of the empire to be enrolled at once.  Later writers do record one, and a defender may fairly ask why they do not count.  The answer lies in their dates and their wording.

1. Orosius, AD 417. In the year of Christ's birth, he says, Augustus "censum agi singularum ubique prouinciarum et censeri omnes homines iussit," ordered a census of every province everywhere and the registration of all men, and Christ was "Romano censui statim adscriptus ut natus est," entered in the Roman census as soon as he was born.³² That is Luke's story retold as theology.
2. Cassiodorus, about AD 507-511. In a letter written for King Theoderic about a boundary dispute between two landowners, he says that "Augusti siquidem temporibus orbis Romanus agris divisus censuque descriptus est," in the time of Augustus the Roman world was divided into plots and recorded by census, so that no one's holding would be uncertain.³³ His subject is land survey, not the Nativity. But orbis and describere are the words of the Latin Bible at Luke 2:1, "ut describeretur universus orbis," and he wrote five centuries after Augustus.
3. The Suda, a Byzantine encyclopedia of the tenth century. Augustus, it says, chose twenty men of the best life and character and sent them "ἐπὶ πᾶσαν τὴν γῆν τῶν ὑπηκόων," over all the land of his subjects, to register people and property, and "αὕτη ἡ ἀπογραφὴ πρώτη ἐγένετο," this census was the first.³⁴ The twenty commissioners and the property assessment are not in Luke. The last clause is Luke 2:2 word for word. Its translator, David Whitehead, judges that "The source of this Suda paragraph is indeterminable."

These texts show that an Augustan census of the whole empire was widely believed from the fifth century on.  They cannot show that it happened, because none is within four centuries of Augustus and each carries Luke's words or Luke's story.  The Suda's commissioners are the one detail that might come from a lost non-Christian source, and that source cannot be traced.

Kennedy identifies such a census with the citizen count, but the distinction above prevents that identification from settling Luke's case.  Sherwin-White:  "Critics hasten to remark, correctly, that there never was a single census of the whole Roman empire."²  The further disagreement is over what Luke meant.

1. The defenders' reading. Sherwin-White and Hoehner read Luke 2:1 as a summary of policy. Augustus did assess the provinces, and he did issue general edicts explaining local acts. On this reading the census of Judea was one local application of an empire-wide program. Sherwin-White: "His whole statement means that the general policy of Augustus was carried out piecemeal in Judaea in A.D. 6 by Quirinius."² Note where that leaves the date: in AD 6.
2. The critics' reading. Emil Schürer, the nineteenth-century historian whose survey of Jewish history in this period, revised in the 1970s, is still the standard reference, argued that Luke generalized from local censuses, as he did with the famine under Claudius, which Acts 11:28 says came "on the whole world."³⁵¹

Both readings are fair to the Greek.  Neither moves the census of Judea before AD 6.

Censuses inside client kingdoms

Herod was a "client king," an ally who ruled and taxed his own kingdom under Rome's oversight.  Brown states the standard view:  "A client king who paid tribute to Rome, Herod had his own taxes and tax collectors; and there is no evidence that the Romans collected taxes based on a census within his realm".⁵  Defenders answer with five pieces of evidence.  Each is real.  The question is what each shows.

1. Augustus's anger in about 8/7 BC. After Herod attacked the Nabateans without leave, Augustus wrote that "whereas of old he had used him as his friend, he should now use him as his subject."⁹³ Brown notes that the evidence is that Augustus did not carry out the threat.⁵ Herod kept minting his own coins and ruling as king until he died.¹¹
2. The loyalty oath. Late in Herod's reign "all the people of the Jews gave assurance of their good-will to Cæsar, and to the king's government," by oath, and more than 6,000 Pharisees refused.⁹³ An oath to the emperor taken inside a client kingdom shows Rome's reach. It is an oath, not a census. Paul Barnett proposed that Luke's apographē was a registration for this oath; Brown replies that "there is no evidence of such a registration."⁵ Two facts bear on that reply. First, the oath was administered, not just proclaimed. Josephus can count the refusers, and the king fined them, "βασιλέως ζημιώσαντος χρήμασιν," a fine that Pheroras's wife then paid.³ Someone knew who had sworn. That is a record of a kind, though of one party's refusers, not a list of every household. Second, a dated parallel survives from Asia Minor. On 6 March of 3 BC, at Gangra in Paphlagonia, a land ruled by its own kings until about 6 BC, an oath of loyalty to Augustus and his family was sworn by the inhabitants and by the Romans trading among them. The stone adds that "κατὰ τὰ αὐτὰ ὤμοσαν καὶ οἱ ἐ[ν τῆι χώραι] πάντες," all those in the land swore in the same terms, in the shrines of Augustus district by district, at his altars.³⁶ Wilhelm Dittenberger, who edited the stone, rejected the first editor's reading of its date formula but reached the same year: "Sic quoque ad annum 3 a. Chr. deferimur," this way too we arrive at 3 BC.³⁷ So Rome did have whole regions swear allegiance, place by place, in Herod's last years.
3. Augustus set the tribute of Herod's successors. When he divided the kingdom in 4 BC, Augustus knew what each part yielded, 200 talents a year from Antipas's land, 100 from Philip's and 600 from Archelaus's, and he remitted a quarter of the Samaritans' tribute.³ So Rome knew the revenue of Herod's districts. Records of revenue are not the same as a Roman census. Augustus kept such figures for the whole empire, kingdoms included. The summary of the state's resources that he wrote out in his own hand, read to the senate after his death, listed "quantum civium sociorumque in armis, quot classes, regna, provinciae, tributa aut vectigalia," in the Church-Brodribb translation "the number of citizens and allies under arms, of the fleets, subject kingdoms, provinces, taxes, direct and indirect."³⁸³⁹ Suetonius adds that Augustus treated the allied kings "as integral parts of the empire."⁴⁰ This is the best secular support for the defenders' reading of Luke 2:1 as imperial policy: Augustus's own picture of "all the world" did include the client kingdoms. It does not show how he got his figures. A king's report of his own troops and tribute would fill such a list as well as a census would. So the list proves Roman interest in Herod's resources, not a Roman count of his people.
4. Apamea. Hoehner cites the Apamea census as a Roman census of an autonomous city.¹¹ Apamea was a city inside the province of Syria, under the legate; Dessau's note treats the Apamea census as part of the census of the province.¹⁹ It is evidence for a census inside the province, not inside a client kingdom.
5. The Clitae, AD 36. Tacitus says the Clitae, a people "subject to the Cappadocian Archelaus," fled into the Taurus mountains "because they were compelled in Roman fashion to render an account of their revenue and submit to tribute." They held off "the king's unwarlike troops" until the legate of Syria sent a Roman officer with 4,000 legionaries to put them down.⁴¹⁴² The geography needs care. Cappadocia itself had become a province after the old king Archelaus died, about AD 17.⁴³ But Augustus had given that king Rough Cilicia as well, because Rome preferred kings in that rugged region.⁴⁴ Brown calls the ruler of AD 36 "Prince Archelaus the Younger" and places his subjects in the Cilician part of the old kingdom.⁵ So this is a client ruler in Rough Cilicia, outside the province. Tacitus does not say who ordered the census. The form was Roman, "nostrum in modum," and Roman troops from Syria ended the revolt only after the king's own forces failed. Brown is right that the passage "does not say that the Romans imposed this census." It is still a real parallel: a census on the Roman model inside a client territory, with the Syrian legate stepping in when it failed.

Ernest L. Martin built the fullest form of the oath proposal in item 2 on Josephus's oath and the Paphlagonian stone.  He argues that "the “census” mentioned by Luke could well have been the “oath” referred to by Josephus, as well as Orosius, and the Paphlagonian inscription," an empire-wide registration of loyalty in 3 BC for the title pater patriae, "father of the country," which Augustus received early in 2 BC.  The registration itself he infers from the title:  "For the totality of the citizenry to approve the bestowal of the Pater Patriae must have involved an Empire-wide accounting."⁴⁵  The sources do not bear that step.  Augustus says only that "senatus et equester ordo populusque Romanus universus appellavit me patrem patriae," the senate, the equestrian order and the whole Roman people named him father of the fatherland.²⁸  Suetonius tells how it was done:  first the commons "by a deputation sent to Antium," then a crowd at the theatre, then the senate "not by a decree or by acclamation, but through Valerius Messala."⁴⁰  That is acclamation by Romans at Rome, with no registration of the provinces.  Josephus gives his oath no year.  Martin puts it "about twelve to fifteen months before the death of Herod" and so in 3 BC, which needs a death of Herod in 1 BC.  Orosius, as shown above, retells Luke.  So the oath proposal now has a real, exactly dated provincial parallel and a Judean oath whose refusers were counted.  It still lacks any record that Herod's people were enrolled, and the link to pater patriae fails on Suetonius's account.

So a Roman-style enrollment of Herod's kingdom, ordered by Herod, perhaps at Augustus's wish, in his last troubled years, is plausible in kind.  No independent record shows that it happened.

Going to "his own town"

Luke says all went "each to his own town," and that Joseph went to Bethlehem "because he was of the house and family of David."¹

Most of the documentary evidence for where people registered comes from Egypt.  In AD 104 the prefect Gaius Vibius Maximus ordered:

it is essential that all persons who for any reason whatsoever are absent from their nomes be summoned to return to their own hearths, in order that they may perform the customary business of registration and apply themselves to the cultivation which concerns them.⁴⁶

The "hearths" are the household and district where a person was registered, and the edict gives farming as one reason to go back.  Bagnall and Frier describe the rule as filing at one's legal or "fiscal domicile," and add that "persons resident in one place owning property in another com- monly file declarations in the places where they own property."³¹  The Roman jurist Ulpian, writing in the early third century, gives the same rule for land:  a man who has land in another city must declare it in the city where the land is.⁴⁷

Michael Kruger, relaying Sabine Huebner's 2019 book, draws the defenders' conclusion:  "there were times when an individual had to register in his hometown—namely, when he owned property in that town and was temporarily living elsewhere."  He adds that Joseph "likely owned a family plot" in Bethlehem.⁴⁸  The first half of that is documented.  The second half is a guess.  Luke gives descent, not property, as the reason for the trip, and he calls Nazareth "their own town" when the family goes home (2:39).¹  Brown puts the critics' case:  the Egyptian edict "offers little support for sending Joseph from Nazareth where he permanently resided (2:39) to Bethlehem where clearly he had no property or wealth".⁵  "Clearly" is too strong.  Luke says nothing about property either way.  But no ancient document sends a person to register in an ancestor's town, and Bagnall and Frier warn that discussion of Luke on this point "suffers from a lack of realism about the circumstances of the Egyp- tian countryside."³¹

The nearest documents from a province next to Judea point the same way.  In the census of the province of Arabia in AD 127, declarations were drawn up "in the city of Rabbath-of-Moab," a provincial center, by landowners of the village of Maoza who were "domiciled in my own private property in the said Maoza."⁴⁹  Babatha's own return, P.Yadin 16, begins:  "I, Babtha daughter of Simon, of Mahoza in the Zoarene [district] of the Petra administrative region, domiciled in my own private property in the Mahoza, register what I possess."⁵⁰  These people traveled to an administrative center to declare property in their home district.  The documents are more than a century later than Luke, and Bagnall and Frier warn that even within Egypt it is "hardly necessary to imagine that there was a single, universal procedure applied in all cases with juridical precision."³¹  So these papers illustrate provincial practice without proving how an Augustan or Herodian census worked.

What follows for Joseph is narrower than either side usually says.  Descent alone is not documented anywhere as a reason to register in a town, and Luke gives descent, not property, as the reason for the trip.  A link that is documented, such as land, a legal domicile or household membership in Bethlehem, would make the trip intelligible, and Luke neither states nor excludes one.  If the census was Quirinius's of AD 6, Galilee was outside it, so a Galilean would be drawn in only by some tie to Archelaus's former territory.  If it was an enrollment of Herod's whole kingdom, a man registering at his legal home in Nazareth would not need to travel at all.  Either way the journey needs a tie Luke does not mention, or a rule of registering by descent that no document attests.

Did Mary have to go?

1. Women were counted. Egyptian household returns list wives, daughters and freedwomen by name and age. One from AD 189 is filed by a house-owner, Achilles, for the tenants of his house, among them Heracleia, the wife of one tenant, aged 40.⁴⁶ Ulpian says that in Syria "men from fourteen and women from twelve, up to their sixty-fifth year, are liable to the poll tax."⁴⁷ That rule is two centuries later than Luke, and when it began is unknown.
2. The householder filed. The same returns are filed by one person for the whole household. Being listed did not require being present.
3. Luke does not say Mary had to be enrolled. He says Joseph went "to be enrolled with Mary." The Greek can mean that both were enrolled or that she went with him. Hoehner's claim that "Every person needed to appear to be questioned" rests on Ethelbert Stauffer, not on a document he quotes.¹¹
4. A woman could declare for herself. Babatha registered her own property in AD 127, with her husband present as guardian.⁵⁰ Her declaration lists property but not her own age, unlike a man's from the same census. Naphtali Lewis, the editor of the Greek Babatha papers, says the idea that a woman was excused from declaring her person "must be regarded as at best speculative."⁴⁹

Evidence that women were registered, or could declare property, is not evidence that a pregnant woman had to travel.  Luke does not claim she had to.  Accompanying her husband by choice is just as consistent with his words.

Part 5: The church fathers

Justin Martyr

Justin, a Christian teacher at Rome writing about AD 155, addresses the emperor:

Now there is a village in the land of the Jews, thirty-five stadia from Jerusalem, in which Jesus Christ was born, as you can ascertain also from the registers of the taxing made under Cyrenius, your first procurator in Judæa.⁵¹

His Greek calls Quirinius "τοῦ ὑμετέρου ἐν Ἰουδαίᾳ πρώτου γενομένου ἐπιτρόπου," your first epitropos in Judea.⁵²  Elsewhere he dates the birth "one hundred and fifty years ago under Cyrenius,"⁵¹ and in the Dialogue with Trypho he retells Luke:  "on the occasion of the first census which was taken in Judæa, under Cyrenius, he went up from Nazareth, where he lived, to Bethlehem, to which he belonged, to be enrolled".⁵³⁵⁴

Sabine Huebner, an ancient historian at Basel who works on the papyri, proposes that "the Gospel of Luke made reference to Qurinius not as governor but as procurator," under Saturninus, whom she dates as governor of Syria "from 9 to 6 BCE."⁵⁵⁵⁶  Michael Kruger, summarizing her, rests part of the case on Justin's word:  "Justin doesn’t call Quirinius a “governor” but a “procurator” ( epitropos )—a different office entirely."⁵⁷  Huebner's own argument on this point could not be read, so what follows tests Kruger's form of it.  Justin's title can be read four ways, and the text alone does not choose among them.

1. An echo of AD 6. Justin calls Quirinius Rome's first procurator "in Judea." Judea first received a Roman governor in AD 6, when Coponius came as epitropos alongside Quirinius. A writer who knew that Roman rule of Judea began with Quirinius's census could easily give him the title of that new regime. This is the simplest reading, but it is an inference about Justin's knowledge, not something he says.
2. An analogy with Pilate. Brown suggests Justin called him procurator "probably on the analogy of Pilate,"⁵ whom Justin's readers knew as Judea's governor.
3. Loose usage. Greek writers used several titles for Roman officials, and Luke's own word family covers both an emperor's reign and a prefect's office. A second-century apologist need not have used the technical title.
4. A separate office. The title preserves a tradition that Quirinius held a lesser post, as Huebner proposes. No other source gives him one, and a consular senator serving as a procurator would be unusual.

Two further points limit what Justin can show.  His Dialogue retells Luke 2:2-4 closely:  "first census," "under Cyrenius," "from Nazareth" to Bethlehem.⁵³  So the basic story very probably comes from Luke.  That does not prove that every extra detail, such as the title or the appeal to registers, also comes from Luke, since neither is in Luke.  Their source is unknown.  His other date, "one hundred and fifty years ago under Cyrenius," is a round number that fits either a birth under Herod or one in AD 6.⁵¹

The appeal to records is an invitation, not a report of reading them.  "You can ascertain" invites the emperor to check.  It does not say that Justin, or anyone he names, had checked.  A writer confident that the records existed could have made the same appeal without seeing them.  It establishes a public claim to checkable evidence, without showing that anyone checked it or recording an official response.

Tertullian

Tertullian, writing at Carthage about AD 208 against Marcion, who denied that Christ was born at all, argues that Jesus' mother and brothers in Luke 8 were real relatives:

Sed et census constat actos sub Augusto nunc in Iudaea per Sentium Saturninum, apud quos genus eius inquirere potuissent.⁵⁸

Ernest Evans's translation:  "Also it is well known that a census had just been taken in Judaea by Sentius Saturninus, and they might have inquired of his ancestry in those records."¹⁵  Evans's note:  Saturninus was "legate of Syria 9-6 B.C." and "Only Tertullian connects him with the Judaean census of Luke 2:1."¹⁵

Tertullian says more about records in two other places in the same book, and both have to be read with this one.

1. Book 4, chapter 7. Arguing that Jesus could not have taught in a synagogue as an unknown, he asks how he could have been admitted when no one knew his tribe, his people, his house, "de censu denique Augusti, quem testem fidelissimum dominicae nativitatis Romana archiva custodiunt."⁵⁸ Evans translates: "or even of Caesar's census, which the Roman registry still has in keeping," "a most faithful witness to our Lord's nativity."¹⁵ Here Tertullian plainly asserts that the Roman archives keep Augustus's census as evidence of the birth.
2. Book 4, chapter 36. The Jews were so carefully divided into tribes and families that no one's descent could easily be unknown, "at least from the recent census of Augustus, of which perhaps the records were still on display."¹⁵ The Latin is plural, "de recentibus Augustianis censibus," "the recent Augustan censuses," and "still on display" refers to Jesus' own day.⁵⁸

Evans adds a caution at chapter 7:  "There seems to be no non-Christian evidence that they were preserved in Rome or would be available to inquirers."¹⁵

So Kruger's summary, relayed by CMI, that Tertullian "too says anyone can check out the records of the census Joseph participated in," is half right.⁵⁷  Tertullian does assert, in the present tense, that the Roman archives hold the census.  He does not say that he or anyone else consulted them, and he does not invite readers to look.  Justin had made a similar appeal half a century earlier, so Tertullian's assertion may rest on that Christian tradition rather than on an inspection.  That is possible, not shown.  An assertion of records is a claim to weigh, not a record.  In chapter 19 the people who "might have inquired," potuissent, are Jesus' contemporaries who doubted his family.

Four further points concern the Saturninus passage.

1. Plural. Census … actos is plural, "censuses were held." Brown: "The fact that Tertullian speaks of cen¬ suses (plural) is often overlooked".⁵ A plural does not exclude one census relevant to the birth among several. It does not identify a particular decree or explain how Saturninus's enrollment relates to Luke's Quirinius.
2. A real governor, and a name Luke does not give. Saturninus was a genuine legate of Syria in Herod's last years, and no earlier Christian text names him. So Tertullian, or his source, knew something about Augustan Syria. That is extra testimony, and it deserves weight on its own merits. Sherwin-White thinks "he must be repeating a version which aimed, already in antiquity, at removing the contradiction posed by Luke," and notes the "absence of any evidence that formal lists of past governors were published or preserved in provinces."² A. J. B. Higgins likewise: "This may be no more than a deliberate correction of Luke".⁵⁹ Neither critic shows the correction happened. They show it would explain the name.
3. His birth date does not fit Saturninus. In Against the Jews, Tertullian dates the birth to "the forty-first year of the empire of Augustus, when he has been reigning for xx and viii years after the death of Cleopatra."⁶⁰⁶¹ The twenty-eighth year after Cleopatra's death in 30 BC is 3/2 BC. Saturninus had left Syria by about 6 BC, when Varus succeeded him. So Tertullian's two statements cannot both be precise. This does not show which one is the learned addition. The 3/2 BC figure matches Clement's, and Clement proves his year from Luke 3:1 and 3:23, so it is very probably a calculation from Luke.⁷ The Saturninus statement comes from somewhere else. A mismatch between a computed year and a named governor is what one would expect if either was borrowed, and it cannot decide between them.
4. A different name is not a joint office. CMI's synthesis is that "Quirinius wasn’t the governor during the birth of Jesus but the procurator who executed the census under Saturninus."⁵⁷ Tertullian names Saturninus and does not mention Quirinius. Justin names Quirinius and does not mention Saturninus. Neither mentions a joint office. The joint office is a modern proposal that would let both be right. It is not something either writer reports, though neither writer rules it out.

Clement of Alexandria

About AD 200, Clement writes that "our Lord was born in the twenty-eighth year, when first the census was ordered to be taken in the reign of Augustus," and cites Luke.⁷⁶²  Clement's twenty-eighth year, counted from Egypt's annexation in 30 BC, is the same 3/2 BC as Tertullian's.  His "first" and "census" are Luke's words.

What the fathers add

These are authentic early texts, and Justin's is within about 150 years of the events.  Being early and authentic does not make them independent.  Justin's census story tracks Luke, though his title and his appeal to records are not from Luke.  Tertullian's Saturninus is the one detail that no earlier Christian text supplies, and the one that names a governor of Herod's lifetime.  It could preserve a memory of an enrollment under Saturninus.  It could be an intelligent second-century correction.  The text does not decide between them, and that does not make them equally likely.  The correction reading needs a reader who knew the governors of Herod's reign, for example from Josephus.  The tradition reading needs information whose transmission cannot now be traced.  Both are possible, and neither is shown.

Part 6: The proposed solutions, tested

Can πρώτη mean "before"?

This proposal goes back to the seventeenth century and was revived by the Dominican scholar M.-J. Lagrange in 1911.¹¹  It has two forms.

1. The comparative form. Lagrange, Nigel Turner and Wayne Brindle read "This census was before the census which Quirinius, governor of Syria, made."⁶³ Later Greek often used πρῶτος, "first," where classical Greek used the comparative πρότερος, "former," and πρῶτος could then take a genitive meaning "than." Robertson's grammar gives the standard examples: John 1:15, πρῶτός μου ἦν, "he was before me," and John 15:18, ἐμὲ πρῶτον ὑμῶν, "me before you."⁴ This much is uncontroversial.
2. The adverbial form. A. J. B. Higgins in 1969, followed by Hoehner, read πρώτη as governing the whole participle phrase directly: "This census took place before Quirinius was governor of Syria."⁵⁹ Hoehner: "This gives good sense to the passage at hand."¹¹

The objections are of three kinds.

1. Syntax. Richard Carrier makes three separate claims. First, a genitive participle with its own subject is always a genitive absolute, a free-standing "while Quirinius was governing" phrase, and so cannot be the thing compared. Second, the verb ἐγένετο stands between "first" and the phrase, so the phrase cannot depend on "first." Third, πρῶτος never really means "before," only "first."¹⁶ Higgins points to a parallel that Lagrange had found, the Greek of Jeremiah 29:2, numbered 36:2 in the Septuagint: ὕστερον ἐξελθόντος Ιεχονιου τοῦ βασιλέως, "after king Jeconiah had gone out."⁶⁴ The Hebrew there is a preposition with an infinitive, "after the going out of Jeconiah."⁶⁵ The translator rendered that preposition with the comparative adverb "later" and a genitive participle with its subject. That is good evidence that a comparative can take such a phrase as the thing compared, which answers Carrier's first claim in its "always" form. It is not proof, because the Greek can also be read as an adverb, "later," beside a free genitive absolute, "when Jeconiah had gone out." With an aorist participle both readings mean the same thing. The parallel does not answer the other two claims. In Jeremiah the comparative stands right next to its phrase, and it is a different word. In Luke "first" is a feminine adjective agreeing with "census," with ἐγένετο between it and the phrase. On the third claim, Robertson's examples show that πρῶτος with a genitive can mean "before," as at John 1:15, but there the genitive is a single pronoun. No one in the literature read for this article cites a sentence in which πρῶτος takes a participle phrase in this way. A search of the tagged Septuagint and Greek New Testament made for this article, about 760,000 words, finds the same gap. Of 485 forms of πρῶτος and πρότερος, 14 have a genitive participle within the next six words, and none uses the participle phrase as the thing that comes later, "before X did Y."⁶⁶ The nearest in form are dating formulas such as 1 Esdras 6:23, "ἔτους πρώτου βασιλεύοντος Κύρου," "in the first year of Cyrus's reign," where "first" is an ordinal and the participle phrase dates the event.⁶⁴ That is how the ordinary translation reads Luke 2:2. Two corpora are not all of Greek literature, so a parallel may still exist elsewhere, but the books closest to Luke's own style do not supply one. Higgins himself thought the participle made Lagrange's comparative version "difficult to justify," and Brindle thought Higgins's adverbial version failed because John 15:18 uses the neuter πρῶτον, not the feminine πρώτη.⁵⁹⁶³ Each version of the proposal is doubted by a defender of the other. Higgins also calls the ordinary rendering, as George Ogg gives it, "the admittedly more grammatical rendering (by classical standards)."⁵⁹
2. Luke's own usage. When Luke means "before" with a following event, he writes πρίν or πρὸ τοῦ, as at Luke 2:26 and 2:21, two verses from this one.⁴¹⁶ Sherwin-White said the proposal "could only be accepted if supported by a parallel in Luke himself."²
3. Sense. Why date an event by a governor who had not yet arrived, instead of the one in office? Higgins conceded "A possible objection to this interpretation would be Luke's silence about the well-known census held by Quirinius," and answered that it was too famous to need naming.⁵⁹ Brindle pressed the same point against Higgins: his reading "fails to answer why Quirinius is mentioned at all."⁶³ On the comparative form there is an answer: Luke distinguishes the birth census from the famous one. But then the sentence means "this census came before the one under Quirinius," and a reader would expect the famous census to be named as a census, not by its governor alone.

Grade:  grammatically possible, with a real but inexact parallel;  strained in Luke's word order and context;  and even if right it only removes Quirinius from the birth.  It still needs an enrollment in Herod's lifetime that no independent source records.

The other proposals

The table sets out each proposal by its own defenders' claims.  "Positive evidence" means evidence for the extra step itself, not evidence that the step is compatible with the record.

The sources for the rows are the parts above:  Josephus for the governors and the AD 6 terms,³⁸ Brown for Stauffer, Barnett and the Archelaus reading,⁵⁵ Higgins for the special command,⁵⁹ Hoehner and Brindle for the Herodian enrollment,¹¹⁶³ Steinmann and Young for the late date of Herod,¹³ Rhoads for the misdating,¹² and Armitage for the flash-forward.⁶⁷

Armitage's reading deserves a fair statement, because it is the one proposal that accepts every date as it stands.  He reads 2:1-5 as a digression, "presumably concerned with definitively establishing Joseph’s Davidic credentials," and 2:6 as a return to the birth story.⁶⁷  He takes Mary's pregnancy in 2:5 as a reminder of her earlier state, not her state on the journey.  He claims only that the reading is "feasible," "without necessarily asserting that this option has a higher probability than others," and he admits it "would thus require that the passage was misunderstood from an exceptionally early stage."⁶⁷  The plain sense of 2:6, ἐν τῷ εἶναι αὐτοὺς ἐκεῖ, "while they were there," ties the birth to the stay in Bethlehem that the census caused.¹  The reading is possible, but it asks the reader to set aside the most natural link in the paragraph.

Three cautions apply to the table.

1. Compatible is not the same as supported. B, C and E are each compatible with the gaps in the record. Listing three ways the record might have been is not three reasons to think it was. They are also not all rivals. An ordinary earlier governorship and a lesser commission are different reconstructions of Quirinius's office. E describes who ran the enrollment, and it can combine with either: Herod could have run an enrollment while Quirinius held a Syrian post or a special commission. A combined account removes some objections, but the chance of all its required steps cannot exceed the chance of any one. Their uncertainties may overlap, so their probabilities cannot simply be multiplied as if the steps were independent.
2. The gaps are real. The governor of 4 to 1 BC is unknown. The date of the Homonadensian war is unknown. Josephus could omit a quiet registration run by Herod. The defenders are right that these gaps exist and that Josephus's silence is evidence, not proof.
3. The gaps do not all fall where they are needed. Mommsen's 3-2 BC slot follows Herod's death on the majority date. Its earlier part could overlap Luke's approximate birth range after a 4 BC death; 2 BC falls beyond the ordinary continuous-story estimate. The whole slot conflicts with Matthew on the 4 BC chronology. Combined with a 1 BC death, it fits both Gospels, and that combination, G with B, is the most complete form of the harmonization. It is also two disputed reconstructions resting on each other.

Weighing the explanations

The skeptic's best case is short.  Josephus describes one census under Quirinius, after Archelaus, and treats it as a new thing that set off a revolt.  Acts links that census with its revolt.  Luke 2:2 names the same governor and the same kind of event.  The most economical reading is that Luke, or the tradition he used, attached the birth to the census preserved in Josephus and Acts.

The defender's best case is also real.  Luke refers to Judas's census (Acts 5:37), sets his opening under Herod (1:5), and makes Jesus about thirty when he begins, after John begins under Tiberius (3:1, 3:23).  A writer holding those markers together has a reason to mean an enrollment earlier than AD 6.  The word "first" may suggest he knew more than one census.  Where Luke's naming of officials can be checked, as with Lysanias in Luke 3:1, it often holds up;  the article on ancient historiography sets out that record.  That record is indirect evidence for this verse, not proof.  It raises the cost of the error explanation without showing that this particular statement is right.  Tertullian's Saturninus and the Tiburtine stone add a little more.

Jack Finegan, whose Handbook of Biblical Chronology is a standard reference, states the defenders' position with unusual candor.  He sees no reason why Quirinius "could not have been associated with Saturninus in such a project," perhaps in 6 or 5 BC, and adds:  "That Quirinius actually took this census is still only concretely affirmed by Lk 2:2; under the circumstances as we have reconstructed them the affirmation is not unlikely."⁶⁸

So there are three live accounts, not two.  Luke conflated the birth with the famous census.  Luke meant an earlier enrollment, named its governor correctly, and Quirinius held an office in or over Syria then that no independent source records.  Or Luke meant an earlier enrollment and named the wrong governor.  The error and conflation accounts are reconstructions too:  no source says how Luke formed his belief.

The error account is preferred here for three reasons.  The one securely dated link between Quirinius and a census in Judea is AD 6, and Luke knew it.  A Judean census fits the moment Judea came under direct Roman rule, while an earlier one in a client kingdom has no independent record.  And the positive support for the rival rests on two uncertain readings of one nameless stone, a disputed inference about where the Homonadensian war was run from, and a different governor's name in a writer two centuries later.  Against that, the error account must explain why Luke would place the birth under Herod and at about thirty while tying it to AD 6.  A writer who knew of a real registration in Herod's time, and knew only one famous census by name, could make exactly that mistake.  That explanation is plausible, but it is not recorded either.  Brown's summary is accurate for the census, though too confident about how Luke came to write it:  "Luke seems to be inaccurate in associating that birth with the one and only census of Judea (not of Galilee) conducted in a.d. 6-7 under Quirinius."⁵

Objections and responses

1. "Luke is an excellent historian elsewhere, so he should get the benefit of the doubt here." Luke's record elsewhere is real indirect evidence. It raises the prior chance that a statement of his is right, and it is a good reason to look hard for a solution, as this article has. Its limits are just as real. A good record on officials' titles does not show that he had a source for this census, and a good historian can still misdate one event. A rule that a reliable author can never be wrong cannot be tested. The record is weighed above, and the verdict here is limited to Luke 2:2.
2. "Josephus is the one who is wrong. He confused two censuses." This is a real proposal, made by W. Lodder in 1930 and by John Rhoads in 2011, and the Joazar notices give it a foothold. Against it, Josephus's date for the AD 6 census rests on three linked facts: Archelaus's deposition, which Dio dates in the same year; the thirty-seventh year after Actium; and the revolt of Judas, which Luke himself places "in the days of the census." Sherwin-White, answering Lodder, wrote: "This kind of reasoning, without supporting evidence, destroys the value of all sources."² Rhoads's key step, that Sabinus was Quirinius, is stated by no source.
3. "Argument from silence: Josephus doesn't mention everything." True, and Sherwin-White makes the same point. Silence counts for more when the writer covers the subject in detail and the event would have mattered to him. Josephus treats the Roman census of AD 6 as a shock because it brought Roman tribute. His silence is strong evidence against a Roman census with Roman tax in Herod's kingdom. It is weaker evidence against a quiet enrollment run by Herod, because nothing shows Josephus had good records of routine royal administration.
4. "The word ἡγεμονεύοντος just means 'in charge,' so Quirinius could have held any office." The word is general, and Luke uses it of Pilate. That removes an objection; it does not supply evidence. No independent source puts Quirinius in charge of anything in Syria before AD 6. Nor does the word mean "procurator" in particular: Brown notes it was used for both legatus and procurator.⁵
5. "Justin and Tertullian dared their readers to check the archives." Justin invites the emperor to check. Tertullian asserts that the Roman archives hold Augustus's census as a witness to the birth, but he does not invite anyone to look, and his "might have inquired" refers to Jesus' contemporaries. Neither says he saw a record. Justin had made his appeal half a century before Tertullian, and Evans notes no non-Christian evidence that such records were kept or open to inquirers. The appeals show confidence. They do not show research.
6. "Tertullian names the right governor for a birth before 4 BC, so he must have had a source." He names a real governor of about 9 to 6 BC, and no earlier Christian text does. That is a point in favor of a tradition. His birth date, 3/2 BC, was probably computed from Luke 3, and it falls after Saturninus left. The mismatch is what one would expect if either statement was borrowed or corrected, so it cannot decide which. The name could be a remembered tradition or a learned correction drawn from Josephus's account of Herod's last years. The second needs less, but neither is shown.
7. "The Apamea census proves Quirinius took a census under Saturninus." The stone calls Quirinius "legate of Caesar in Syria" and gives no year. His known legateship is AD 6. For an earlier date the stone would need an earlier legateship, which is the point in dispute. It is compatible with one and evidence for none.
8. "The Antioch inscriptions call him duumvir, a joint magistrate, so he shared power with Saturninus." The duumvirate is the chief magistracy of the colony of Antioch in Pisidia, given to him as an honor, with a local deputy doing the work. It is a city office in Asia Minor and is unconnected to Syria.
9. "Luke says 'first,' so he knew of two censuses." He may have. "First" also reads naturally as "the first census of Judea," which the AD 6 census was; that is Brown's and Ogg's reading.⁵⁹⁵ The word cannot decide the question.
10. "No census would send Joseph to Bethlehem, so the whole story is invented." That overreaches the other way. People in Egypt and Arabia did declare property where it lay and traveled to provincial centers to do so, and Luke says nothing about whether Joseph had property or a legal home in Bethlehem. The evidence shows only that no document confirms registration by descent alone. The census detail can be wrong in date while the family's presence in Bethlehem is a separate question, which belongs to the Nativity Accounts overview and its other children.
11. "Herod died in 1 BC, which solves the problem." It solves a different problem, Luke's "about thirty," and it would let Mommsen's 3-2 BC Syrian post fall inside Herod's life. Archelaus's deposition and the AD 6 census stay where they are on that chronology, as its own defenders state. So a 1 BC death helps the harmonization only together with an earlier office for Quirinius.
12. "This is a Christian site; it should defend inerrancy here." The site's rule is to follow the evidence and state weaknesses plainly. On this verse the evidence favors a mistake in Luke's dating of the census, though not overwhelmingly. A reader who holds to inerrancy can point to the real gaps in the record and to the indirect support for an earlier Syrian post, but should not claim that any source fills them.

Issues and unanswered questions

1. Huebner's own argument was only partly read. Her central sentence on p. 47, that Luke "made reference to Qurinius not as governor but as procurator," is known from Hans Förster's verbatim quotation. Her reasons, her reading of Luke's participle, her census mechanics and her notes on pp. 138-164 were not reached. The arguments tested here under her name are Kruger's form of them.
2. Kennedy's census discussion was read directly, pp. 22-28 of the publisher's excerpt. The rest of his book was not read for this article.
3. Several standard works are relayed, not read. Joseph Fitzmyer's commentary on Luke, the Vermes-Millar revision of Schürer, Ronald Syme's studies of the Tiburtine stone and of Galatia, George Ogg's 1968 article, Brook Pearson's 1999 article in Catholic Biblical Quarterly, Mark D. Smith's 2000 article in the same journal, John Thorley's 1979 note in Greece and Rome, Nigel Turner's grammar, Ethelbert Stauffer's book, Timothy Barnes's 1968 reply to Filmer, and Steinmann's 2009 article in Novum Testamentum appear here as other writers report them. Pearson and Smith in particular are cited by Armitage on earlier Herodian censuses with Roman features and on property links. Any of them could hold evidence that changes a grade here.
4. Babatha's census return of AD 127, P.Yadin 16, was read only through Hannah Cotton's translation of its opening and Naphtali Lewis's notes on a sister document. The full edition was not obtained.
5. The date and base of Quirinius's Homonadensian war are unsettled. If Cilicia answered to the legate of Syria, Mommsen's case for an earlier Syrian post gains; if the war was run from Galatia, as Syme argued, it loses its best support. The date of the Antioch duumvirate is also open: Ramsay dated the statue to about 8 BC, and Dessau in 1916 suggested about 1 BC to AD 4. Adolf Deissmann, citing F. Bleckmann's 1920 study, took the stones to put Quirinius's "first Syrian governorship" earlier: "now it is perhaps to be dated 11-8 B.c." His translator's note adds that "If Dessau’s dating is correct these inscriptions cease to have any bearing on the date of Quirinius’ first governorship of Syria."⁶⁹ Neither Bleckmann's nor Dessau's 1920 paper was read.
6. Finegan's revised Handbook (1998) is reported by a Bible-study website to date a Quirinius governorship to 3-2 BC, pp. 302-306, combined with a 1 BC death of Herod. That is Mommsen's slot joined to the late Herod date, the most complete form of the harmonization. Only the 1964 edition was read. Eric's research note that "Tertullian says Quirinius was governor from 3 to 2 BC" traces to that website, which attributes the date to Finegan, not Tertullian.
7. The governor of Syria from about 4 to 1 BC is unknown. A new inscription or coin naming him could strengthen proposal B, or close it.
8. A Herodian enrollment remains possible. A document showing any enrollment of Herod's subjects, for any purpose, in 8 to 4 BC would move proposal E from "plausible in kind" to "attested," though it would still leave Luke's naming of Quirinius to explain.
9. Acts 5:36-37 raises a related problem, the order of Theudas and Judas, which Brown also counts against Luke.⁵ It needs its own treatment.
10. The grammar search covers two bodies of text. The search for πρῶτος with a participle phrase meaning "before" covered the Septuagint and the New Testament. The Thesaurus Linguae Graecae, which holds most surviving Greek literature, was not searched. A clear parallel there would raise the grade of the "before" translation.
11. Two of Martin's sources were not read. He also cites Orosius 7.2 and the Armenian history of Moses of Khorene for an oath and census in 3 BC. Both are centuries later than Luke, and neither was read for this article.

Verdict

1. Authenticity of the text: secure in the editions compared. The critical editions in the SBL apparatus differ in Luke 2:2 only by an article, and Brown reports no manuscript evidence for reading "Saturninus." That is not a full inventory of manuscripts, but no reading that removes Quirinius is known.
2. Translation: the ordinary reading, "this was the first enrollment, while Quirinius was governing Syria," is the natural one. "Before Quirinius governed" is grammatically possible, with a real but inexact parallel in the Greek of Jeremiah. It is not impossible, and it is strained in Luke's sentence.
3. Dates: strongly supported that Quirinius was legate of Syria in AD 6 and that his census of Judea ended between autumn AD 6 and autumn AD 7. Herod died in spring 4 BC on the majority view and early 1 BC on the best minority view. On the normal-pregnancy and continuous-story reading, Luke's Herodian marker points to a birth by roughly a year after Herod's death, and Matthew's to a birth before it. On either chronology examined here, that leaves a gap of roughly five years or more to the AD 6 census.
4. Quirinius's career: no independent source securely dates an office of his in Syria or Judea before AD 6. The Lapis Tiburtinus could record a first Syrian term, as Mommsen argued, but it names no one, and its iterum has two readings. An earlier term is a disputed reconstruction with some indirect support.
5. Roman practice: the Rough Cilician case of AD 36 attests a census on the Roman model inside a client territory. Registration at one's legal home, and declaration of property where it lay, is attested in Egypt and Arabia. The documents examined here do not attest registration by descent alone.
6. Source access: the decisive primary texts, Josephus, Tacitus, the inscriptions, Justin and Tertullian, were read in their own languages. Huebner was read only through a reviewer's quotation; Kennedy's census discussion was read directly in the publisher's excerpt.
7. Historical accuracy of Luke 2:2: probably mistaken in date. The census Luke names is probably the AD 6 census, roughly five years or more after the birth suggested by his other markers. The rival that keeps Luke correct, an earlier Syrian office for Quirinius plus an enrollment in Herod's last years, perhaps with a 1 BC death of Herod, is a coherent reconstruction with some indirect evidence. It needs several disputed steps without secure independent confirmation. The error account needs Luke to have joined a birth tradition near Herod's reign to the known census, which is also unrecorded but simpler. The balance favors error, with moderate confidence.

Sources