Monday 11 October 2021

The dates of Jesus. 3. Christmas and Easter

§1–§3. The dates | §4–§5. Calendars and prophecy | §6. Christmas and Easter | §7–§8. Sources

Recap. Early sources on Jesus’ dates are vague. Around 200 CE Christians wanted more precise dates than they could get from the gospels of Matthew and Luke. Concocting precise dates involved an eclipse, and trying to make 500 years of regnal periods line up with a prophecy in Daniel.

This episode. A look at two modern theories on the origins of the traditional dates of Christmas and Easter. Also, a digression on the traditional dates of the equinoxes and solstices in the Julian calendar.

Andrea del Verrocchio and Leonardo da Vinci, The baptism of Christ (ca. 1475; Gallerie degli Uffizi, Florence)

§6. Christmas and Easter

The vague early datings of Jesus that we talked about in §2 don’t usually come up in connection with the origins of Christmas and Easter. But they ought to.

There are two modern theories about how Christmas came to be celebrated on 25 December:

  • HRT, the History of Religions Theory (or religionsgeschichtliche Hypothese): early Christians took pagan solstice and equinox festivals and made them the basis for their own festivals.
  • CT, the Calculation Theory (or Berechnungshypothese): early Christians used chronographical reckoning to put Christmas at the winter solstice, and Easter at the spring equinox, traditionally reckoned as falling on 25 December and 25 March in the Julian calendar.

HRT is the popular one. If you’ve heard that Christmas was the result of Christians stealing Saturnalia, or Mithras’ birthday, or Sol Invictus ... that’s HRT.

Susan Roll (2000) gives a good outline of the 20th century debate, which usually favoured HRT. Philipp Nothaft (2012) outlines some important developments since then: new research by Steven Hijmans and Hans Förster has blasted HRT to the eternal oblivion that it deserves. Nothaft himself supports CT (Nothaft 2013). Hijmans is agnostic about CT, and Förster opposes CT: Förster actually puts it that CT requires ancient Christians to have performed ‘breathtaking mental acrobatics’.

I’ve done some write-ups here on Christmas and Easter that push hard against HRT too, sometimes relying on Hijmans’ work. To put it briefly: the evidence that Christmas and Easter draw on pagan festivals is all either late, misrepresented, or fabricated.

  • Epiphanius’ Aion festival and the 354 Chronography’s Invictus festival are local, late, and poorly attested.
  • Aurelian didn’t make the sun god the chief god of the Roman pantheon: that was Elagabalus and the god Elagabal, and it only lasted four years.
  • The idea that Aurelian instituted the Invictus festival on 25 December 279 CE is a modern inference, and not well supported by ancient evidence (Hijmans 2009: 588–591).
  • Mithraism and Saturnalia are irrelevant, and Yule is mediaeval.
  • The notion that Christmas was originally on 6 January and later transferred to 25 December is based entirely on a 12th century scribal gloss (see Roll 1995: 150–152, 2000: 279–280).
  • There’s very little evidence of Mediterranean solstice or equinox festivals before the Christian ones came along. That is, Christmas and Easter are the archetypes for solstice and equinox festivals, not copies of pagan ones.

HRT hasn’t a shred of credibility and it never should have been taken seriously.

Does that mean I support CT, then? Well, no, not if the C stands for calculation. If it means concocting, then yes.

Does that mean I think early Christians engage in ‘breathtaking mental acrobatics’, as Förster puts it? Well, yes, of course. You did read §2–§5, didn’t you? The acrobatics are in plain sight.

Just look at how selective Clement and Tertullian have to be with their treatment of regnal periods to make them line up with Daniel’s ‘70 weeks’ prophecy. Africanus has to make up the idea that intercalary periods count as extra years, as well as glossing over several bits of Daniel, not least what he says about when the 490 year period begins. Modern editions and translations of Tertullian, even critical editions, frequently ‘correct’ his figures and details because they’re so flagrantly wrong.

And they do all of this in support of a date that was originally reached thanks to a chronological marker they all knew was wrong — the eclipse of 29 CE. Like I said, it’s creative adjustments all the way down.

The whole point of what I wrote in §2 is that, prior to Clement, people avoided calculating Jesus’ dates. They knew perfectly well that there was no hard evidence.

Then everything changed when the Quartodecimans attacked.

Talley (1986) was right about this, at least: the Quartodeciman controversy was the beginning of a tidal wave of frantic Christian interest in chronography. The question was over which day to celebrate Easter. The initial dispute in the 150s ended amicably: Polycarp and Anicetus agreed to disagree, and the Anatolian and Roman churches went their own ways.

In the 190s things heated up. Synods were convened in several places around the Mediterranean to discuss the correct liturgical date to observe Jesus’ death and resurrection. The disagreement became bitter. The dispute created tremendous pressure to find accurate dates for Jesus’ death, and by extension his lifetime.

And lo and behold, straight afterwards we find Clement discussing dates that are precise to the exact day. We find people arguing over the eclipse theory and rejecting it. We find Tertullian doing somersaults to make Jesus’ dates line up with Daniel, and Africanus creating a new Christian chronography. In the mid-200s we find Anatolius working out a new form of the Metonic 19-year cycle.

The 190s, in a nutshell, is when Christians started paying really close scrutiny to chronographical precision. And, yes, getting it wrong. I don’t exactly blame them. If you pressure someone to produce results, but there’s no good evidence, don’t be too surprised if they come back with poor results.

The date of Jesus’ death and resurrection got pinned to the spring equinox, traditionally reckoned as 25 March in the Julian calendar. That’s the date Tertullian quotes, and it’s the date implied by Hippolytus’ paschal table (220s–230s CE). This wasn’t because of any pagan equinox festival — like I said, there weren’t any pagan equinox festivals to speak of — but a Christian innovation. The main motivation is that Jesus died at Passover, which is linked to the equinox, and the idea that Jesus was ‘the sun of righteousness’, a biblical phrase from Malachi 4.2. One 3rd century source, pseudo-Cyprian, explicitly links the equinox and Jesus’ death to Malachi’s phrase (De pascha computus 19). (Notice that this is decades before emperor Aurelian’s interest in the sun cult.)

Part of Hippolytus’ paschal table as printed in the Patrologia graeca (x.875–880). Original on the left; the Patrologia’s Latin translation on the right. Hippolytus codifies a 112-year Easter cycle. 223 CE, the year corresponding to 2 BCE in the 112-year cycle, is marked genesis Ch(ri)s(tou), ‘genesis of Christ’. 253 CE, corresponding to 29 CE, is marked pathos Ch(ri)s(tou), ‘suffering of Christ’, and specifies an Easter date of 25 March (πρὸ η' Κα. Απρει.). See Mosshammer 2008: 327–328; Nothaft 2011: 38–45.

Christmas comes into it as a secondary matter. In typological ancient Jewish thought, the prophets were imagined as dying on the same day they were born, living a whole number of years. When it comes to Jesus, though, there’s a discrepancy of exactly nine months between his birth date (winter solstice) and death date (equinox). Why would ancient Christians make his death date line up with conception, rather than with birth?

Throughout the 20th century, this was a serious problem for CT. Roll points out that there’s no evidence for a shift from emphasising birth to emphasising conception, and that’s an insuperable problem for CT (Roll 2000: 288). Nothaft regards it as ‘the only objection against CT that carries some considerable weight’ (2013: 256).

Take heart: there’s actually pretty good evidence if you look in the right place. Schmidt (2015: 548–552) finds it in Hippolytus’ use of the word genesis in the paschal table above. He shows that contemporary sources often use genesis to refer to conception, not birth.

Actually we can put the slippage between the two meanings even earlier. And guess when? Straight after the Quartodeciman synods of the 190s. Will the real Clement of Alexandria please stand up?

οὐδὲν δὲ οἶμαι ἐπὶ τούτοις χεῖρον καὶ τοὺς χρόνους τῶν Ῥωμαϊκῶν βασιλέων παραθέσθαι εἰς ἐπίδειξιν τῆς τοῦ σωτῆρος γενέσεως· ...
ἐγεννήθη δὲ ὁ κύριος ἡμῶν τῷ ὀγδόῳ καὶ εἰκοστῷ ἔτει ...
εἰσὶ δὲ οἱ περιεργότερον τῇ γενέσει τοῦ σωτῆρος ἡμῶν οὐ μόνον τὸ ἔτος, ἀλλὰ καὶ τὴν ἡμέραν ...

After this, I think, it would be no worse to set out the periods of the Roman kings for an account of the saviour’s genesis. ...
And our Lord was born in the 28th year (of Augustus) ...
There are those who not only research the year of our Saviour’s genesis but also the day ...
Clement of Alexandria, Stromateis, 145.1, 145.6

Clement is emphatic that what he’s dating is Jesus’ genesis. The word ἐγεννήθη in the second excerpt here is explicitly ‘he was born’, not ‘he was conceived’. But it isn’t as simple as that.

If we look at how he uses genesis elsewhere, it’s clear that it’s a way of equivocating between birth and conception. Sometimes Clement uses it to mean ‘birth’ (e.g. Strom. ‘the magi foretold Jesus’ genesis’, which can only mean his birth); sometimes gestation (Strom. ‘the womb ... which was created for the genesis of foetuses’); sometimes conception; and sometimes other metaphorical meanings.

There’s one passage where it’s especially clear that genesis is used to equivocate. In a discussion of sex within marriage, Clement mentions that Christians aren’t required to wash after sex as was commanded in the Hebrew Bible (Leviticus 15.16–18). He calls this post-coital washing ‘baptism’ — and bear in mind that infant baptism was standard by this time. The equivocation is between ‘baptism’ after conception, and baptism after birth. Then he goes on to use genesis to equivocate between both meanings.

82. ⁶ οὐδὲ μὴν τὸν ἀπὸ τῆς κατὰ συζυγίαν κοίτης ὁμοίως ὡς πάλαι βαπτίζεσθαι καὶ νῦν προστάσσει ἡ θεία διὰ κυρίου πρόνοια. οὐ γὰρ ἐπάναγκες παιδοποιίας ἀφίστησι τοὺς πιστεύοντας δι' ἑνὸς βαπτίσματος εἰς τὸ παντελὲς τῆς ὁμιλίας ἀπολούσας ὁ κύριος, ὁ καὶ τὰ πολλὰ Μωυσέως δι’ ἑνὸς περιλαβὼν βαπτίσματος.
83. ¹ ἄνωθεν οὖν ὁ νόμος τὴν ἀναγέννησιν ἡμῶν προφητεύων διὰ σαρκικῆς γενέσεως ἐπὶ τῇ γεννητικῇ <καταβολῇ> τοῦ σπέρματος προσέφερε τὸ βάπτισμα, οὐ βδελυσσόμενος ἀνθρώπου γένεσιν· ὃ γὰρ φαίνεται γεννηθεὶς ἄνθρωπος, τοῦτο δύναται ἡ τοῦ σπέρματος καταβολή. ² οὔκουν αἱ πολλαὶ συνουσίαι γόνιμοι, ἀλλ' ἡ τῆς μήτρας παραδοχὴ τὴν γένεσιν ὁμολογεῖ, ἐν τῷ τῆς φύσεως ἐργαστηρίῳ διαπλαττομένου τοῦ σπέρματος εἰς ἔμβρυον.

82. ⁶ But divine providence through the Lord does not command that (a man) should still nowadays be baptised after sex within marriage, as in olden times. For the Lord does not forbid believers the necessity of procreation: he has cleansed them of sex in one baptism for all time. In that way he has captured Moses’ many (baptisms) through one baptism.
83. ¹ From the beginning the law prophesied our rebirth through genesis in the flesh, and appointed baptism after the procreative <sowing> of seed, not treating genesis as abominable. For it is the sowing of seed that has the power to bring about what appears as a person after they are born (gennētheis). ² So it isn’t that having sex repeatedly is fertile: but it’s the womb’s receiving (of seed) that corresponds to genesis, as the seed is refashioned into an embryo in nature’s workshop.
Clement of Alexandria, Stromateis–83.2

Rebirth in the resurrection of the body is genesis, but so is conception. We don’t know the exact motivation for this shift from birth date to conception date, but Clement certainly documents the reality of the shift.

Digression: solstices and equinoxes in the Julian calendar

The winter solstice and spring equinox were traditionally reckoned as falling on 25 December and 25 March in the Julian calendar: for early examples see Columella, De re rustica 9.14.1, 9.14.12; Pliny, Natural history 18.220–221 (both 1st century CE; note that the Loeb translations linked here both mistranslate the Latin dates).

Columella and Pliny use some strikingly similar language, and that indicates a common source. Columella states that the winter solstice ‘is completed roughly around 25 December, at the 8th degree of Capricorn’ (fere conficitur circa viii calend. Ianuarii in octava parte Capricorni); Pliny states that all four equinoxes and solstices fall ‘at the 8th degree of their (zodiacal) signs, the winter solstice in Capricorn on roughly 25 December’ (in octavis partibus signorum, bruma Capricorni a. d. viii kal. Ian. fere). Their most likely source is Sosigenes, whom Pliny credits as the designer of the Julian calendar. The repetition of in octava/-is parte/-ibus and, still more, the non-technical word fere, indicates that Sosigenes’ work was in Latin.

Pliny gives exact figures for the intervals between each equinox and solstice. Ptolemy, Almagest 3.4, quotes some of the same figures, and attributes them to Hipparchus. Combining the three sources, we arrive at:

  • Winter solstice: 25 December (Pliny, Columella)
  • Spring equinox: 90.125 days after winter solstice (Pliny) = 25 March (Columella)
  • Summer solstice: 94.5 days after spring equinox (Pliny, Ptolemy) = 27/28 June
  • Autumn equinox: 92.5 days after summer solstice (Ptolemy; lacuna in Pliny) = 88.125 days before winter solstice (Pliny) = 28 September

Even in Sosigenes’ time the winter solstice and spring equinox didn’t fall on these dates. The 365¼-day cycle of the Julian calendar approximates the solar year fairly well, but it isn’t exact, and it slips out of synch by one day every 130 years. Sosigenes’ use of fere ‘roughly’ indicates that he was aware of the discrepancy, but also that the placement of the equinoxes and solstices was already a codified standard in his time.

The size of the discrepancy shows that the standard dates and intervals go back earlier than Hipparchus. The only period when all four equinoxes and solstices fell on the exact dates given above, retrojecting the 365¼-day cycle, was from 429 BCE to 298 BCE. The traditional relationship with the 365¼-day cycle was almost certainly codified in that period.

That’s more than a century before Hipparchus was active. In addition, we know Hipparchus put the equinoxes and solstices at the start of their respective zodiacal signs; Columella and Pliny both state that they fell halfway through their zodiacal signs at the 8° mark, as in Babylonian astronomy. So Sosigenes, their immediate source, was evidently not drawing on Hipparchus but on something earlier. Eudoxus or another astronomer of his time is a much better fit than Hipparchus, both for the timeframe and for the detail of the 8° mark.

The last episode will be a compilation of the sources, given in the original languages and in translation, with notes and bibliography. It will be large.

§1–§3. The dates | §4–§5. Calendars and prophecy | §6. Christmas and Easter | §7–§8. Sources

The dates of Jesus. 2. Calendars and prophecy

§1–§3. The dates | §4–§5. Calendars and prophecy | §6. Christmas and Easter | §7–§8. Sources

Recap. Early sources on Jesus’ dates are vague. The gospels of Matthew and Luke were their only primary sources. Around 200 CE Christians wanted more precise dates, so they needed some external point of reference. That point of reference — a solar eclipse in 29 CE — turned out to be no good, but even so, the dates quoted after 200 CE are much more precise.

This episode. I was planning to get to the dates of Christmas and Easter today, but it’s just too long. We’ve got to lay the groundwork first. Grab a cup of tea (or something stronger) and we’ll dive into the chaos of ancient calendar systems and biblical prophecy.

§4. Calendars and calendar-era systems

When ancient sources give calendar dates, it’s often a problem to understand what they intend to say. The Roman world had no universal system for saying which year an event happened. If you were in Rome itself, you might refer to the consuls for that year — but most of our sources didn’t live in Rome.

Joachim Patinir, The baptism of Christ (ca. 1515; Vienna, Kunsthistorisches Museum)

There were tidy calendar-era systems that simply counted from a particular event. Like the Seleucid year, counting from the reign of Seleucus I Nicator, king of Persia and Syria from 311 BCE on, which remained in use into the modern era (Samuel 1972: 246). Anatolia and Syria had various other calendar era systems too (Bickerman 1980: 62–79). But our sources don’t use these. Hebrew reckoning from the year of the Creation wasn’t used until the Mediaeval period. Counting years since Rome’s founding, ‘AUC’, is mostly just a modern fad (Bickerman 1980: 77–78).

The 354 Chronography is the only one of our sources to use ‘AUC’ (and only in some sections). Eusebius is the only one to make full use of Greek Olympiads. These are both pretty late.

Our sources rely mainly on regnal years — Ptolemaic kings, Judaean kings, Roman emperors — with occasional references to the number of years since Adam, or Abraham, or the Babylonian exile. But complexity breeds inaccuracy.

For example, some Christian calculations put the regnal year ‘Augustus 1’ in 44 BCE, counting inclusively from Julius Caesar’s death. Eusebius puts it in 43 BCE. Clement of Alexandria puts it in 30 BCE, following the fall of the Ptolemies. In real life, the emperors counted by tribunician years — and Augustus didn’t acquire permanent tribunician power until 23 BCE.

And when does the year begin? Not on the 1st of January! That would be too simple. Regnal years varied depending on where you were:

In Syria, for example, the second year of the new emperor began on the next 1 October after his accession, that is, at the next New Year of the calendar of Antioch ... In Egypt, the second regnal year began on 29 August after the accession, that is, the Alexandrian New Year ...
Bickerman 1980: 66

So the first regnal year would generally be a ‘short’ year. The tribunician year is an exception: tribunates began on 10 December, and emperors normally treated the 10 December following their accession as the start of their first tribunician year. And the Olympiad year began in midsummer.

So, for example, it seems the first two years of Tiberius’ reign would be expressed as follows. Tiberius became emperor on 14 August 14 CE.

Roman reckoning (consulship) consulship of Pompeius and Appuleius: begins 1 Jan. 14 CE consulship of Drusus and Norbanus: begins 1 Jan. 15 CE
Roman reckoning (tribunate) Tiberius 1: begins 10 Dec. 14 CE Tiberius 2: begins 10 Dec. 15 CE
Alexandrian reckoning Tiberius 1: begins 14 Aug. 14 CE Tiberius 2: begins 29 Aug. 14 CE
Syrian reckoning Tiberius 1: begins 14 Aug. 14 CE Tiberius 2: begins 1 Oct. 14 CE
Olympiads Ol. 198,3: begins June 14 CE Ol. 198,4: begins June 15 CE
Note. My thanks to Prof. Tim Parkin for pointing me towards reliable information on when regnal years began.

Note that the Alexandrian ‘Tiberius 1’ would be very anomalous, lasting just two weeks, if their practice at the time was the same as in later years. ‘Tiberius 2’ would have begun before Alexandria even found out that there was a new emperor. So there’s uncertainty over even this much.

These complexities, and the inherent complexity of regnal year systems, create errors and internal inconsistencies. Clement and Tertullian quote regnal periods that are wildly inconsistent with reality. Tertullian puts Jesus’ birth in Augustus 41, 28 years after Cleopatra’s death — but those are two different years: Augustus 41 ought to be 4 BCE, 28 years after Cleopatra’s death was 2 BCE. Pseudo-Hippolytus and the Acts of Pilate put Jesus’ death in Tiberius 18 or 19 (regnal year), and the consulship of ‘Rufus and Rubellio’ (consular year) — but Tiberius 18 and 19 were 32 and 33 CE, and the consulship of Fufius and Rubellius (their actual names) was 29 CE. (See §3 for a possible explanation of this last inconsistency.)

Our sources don’t have a single, coherent chronology to work with. They’re playing a game of telephone, and the rules change every turn.

The beginning of Jesus’ ministry in Jesus of Nazareth (dir. Franco Zeffirelli, 1977; Jesus is played by a blue-eyed Englishman, naturally)

§5. Daniel’s 70 ‘weeks’

Many Christian writers give dates for Jesus that aren’t based on historigraphical testimony — as we saw in §2, no testimony exists, except for the gospels — but on prophecy in the Hebrew Bible.

In Daniel 9.20–27 the Messiah’s coming is linked to a period of 70 ‘weeks’ (or ‘hebdomads’) of years, that is to say 490 years, starting from Cyrus’ order for the restoration of Jerusalem. That order was given in 538 BCE. Daniel divides these 70 weeks into a period of 7 weeks (9.25), 62 weeks (9.25–26), and 1 week cut in half (9.27, echoed at 12.7–12: ‘a time, two times, and half a time’, that is, 3½ days or half a week). At the end of the 62 week period the Messiah will be ‘cut off’, and enemies will destroy Jerusalem and the temple. Things will remain that way until the end of the 70 weeks. It’s all very ambiguous, but Christian writers were incorporating aspects of this into their ideas about history by the time of Revelation (11.7–11: a period of 3½ days).

Daniel’s prophecy had already become inconvenient in the first century BCE. The Messiah was supposed to be ‘cut off’ after 7 + 62 weeks, that is, 481 years; and that points to 57 BCE, about a century after Daniel was written. But 57 BCE came, and nothing much happened. Daniel was left looking a bit silly. There was pressure to fudge the details.

We can see a reflection of one case of fudging in Josephus. The Judaean king Aristoboulus I took the throne in 104 BCE, and died the following year. But Josephus states that his accession was 481 years after the end of the exile (Jewish antiquities 13.301). This must be a result of someone trying to identify Aristoboulus as the Messiah; treating his brief reign as ending after the 62 week mark (434 years reckoned from 537 BCE instead of 538), and having Daniel’s 70 weeks begin a ‘week’ early, at the start of the exile. The result would be that Aristoboulus comes at the 69 week mark (481 years).

If this sounds like breathtaking mental acrobatics, my advice is: yes it is, and it’s best not to think about it too hard. See Eshel 2005: 107 for a similar case of chronographical creativity in one of the Dead Sea scrolls.

Jesus’ death was nearly a century after Daniel’s 69 week mark, and Titus’ destruction of Jerusalem in 70 CE was well over a century late. So ancient Christian chronographers had to get even more creative. Clement provides two different timelines, hedging his bets. Tertullian gives regnal years for the Achaemenid, Ptolemaic, and Roman rulers from Darius II up to Vitellius, but he gets many of them significantly wrong — 25 years off in one case (Ptolemy Euergetes II) — and he leaves things out. In particular, he deletes two significant rulers, Ptolemy VI Philometor and the Roman emperor Claudius. Even after all that, his final calculation is still off by 30 years.

Julius Africanus cuts through the red tape of regnal years and plunges straight into calculating a conversion rate between solar years and lunar months for the entire period. To make it work, he skips the bit about the Messiah being ‘cut off’ at the 62 week mark, and he assumes that intercalary periods count as bits of extra years. For some reason.

For from Nehemiah, who was sent by Artaxerxes to resettle Jerusalem in the 115th year of the Persian empire ... until this time, which was the second year of the 202nd Olympiad, the 16th year of the reign of Tiberius Caesar, there is a total of 475 years. This represents 490 Hebrew years, since they number their years according to the lunar month, which is commonly said to be 29½ days. For the cycle of the solar year is 365¼ days, and the twelve-month lunar cycle is 11¼ days less. For this reason, both the Greeks and the Jews insert three intercalary months every eight years. For 11¼ multiplied by 8 makes a period of three months. Therefore, 475 years come to 59 eight-year periods, remainder three; since there are three intercalary months in an octaeteris, this adds up to 15 years. Added to the 475 years, they make 70 hebdomads.
Julius Africanus, Chronographiae F93 Wallraff (tr. Wallraff)

It’s creative adjustments all the way down. Daniel’s 70 weeks were already a creative extension of Jeremiah, who prophesied 70 years for the Babylonian exile. In reality it was 48 years. (Jeremiah 25.11–12, 29.10; Daniel 9.2 cites Jeremiah explicitly.)

In Part 3 we’ll move on to the modern debate over how Christmas and Easter developed in early Christianity.

§1–§3. The dates | §4–§5. Calendars and prophecy | §6. Christmas and Easter | §7–§8. Sources