Tuesday, 29 December 2020

Top posts of 2020

I’m well aware how lucky I am to have been living in New Zealand in 2020. Still, I’ve had multiple personal reasons why much of 2020 has been an annus horribilis anyway. This year has had some bright spots — the astounding speed of the vaccine development, the continuing decline of fossil fuels, the reappearance of Boba Fett and Luke Skywalker — but not enough to make up for the bad.

Now, at last, 2020 is hindsight. Good riddance.

This is not how the ancient Greeks discovered the shape of the earth.

Visits to Kiwi Hellenist this year have been evenly split between 2020 articles and older ones. So this year, this list is a combo: the old and the new.

  1. The Epic Cycle wasn’t as popular as you think (10 Feb.). Fans of the Trojan War legend tend to idolise the Epic Cycle. I pretty much agree. Unfortunately, the ancients didn’t feel the same way. They just didn’t read the poems. Most ancient understanding of them was filtered through prose versions, like myth encyclopaedias.
  2. An oldie: Why are there no Romans named ‘Quartus’? (Apr. 2018). I’m guessing people are still reading this piece because, it seems, it’s the only article on the internet that covers the question. Names like ‘Marcus’, ‘Quintus’, ‘Sextus’, and ‘Decimus’ come from month names (March, Quintilis, Sextilis, and December). The first four months of the early Roman calendar didn’t have number names, so there are no personal names corresponding to those numbers.
  3. Stripping myths down to a historical core (part 1) (29 Jun.). This piece was more popular than its follow-up in July — maybe because of its focus on the fact that Troy was never lost (despite what Schliemann wants you to think), or maybe because part 1 was headed by Total War: Troy imagery. There’s also a timeline here of the ‘battle of Bunarbashi’, as Rachel Davies has called it (and again, my piece seems to be the only place on the internet you’ll find such a thing). By the way, I was grateful to be allowed to contribute a condensed version of this — omitting Schliemann but including Euhemerus (from part 2) — to the SCS blog in October, under the title ‘Truth behind myth: video games and the recreation of the Trojan War’.
  4. Detecting the earth’s curvature (23 Jan.). I loved writing this one, so I’m glad people enjoyed it. The key points: (1) Ships going over the horizon are not how the earth’s shape was discovered. (2) The discovery came from astronomical observations. It was probably partly to do with the angle between the plane of the ecliptic and the plane of the celestial equator.
  5. Lucian’s parody of the book of Revelation (31 Aug.). Again, this was fun to write, partly because it seems so clear-cut to me, and partly because it seems to be poorly known, and poorly accepted. Modern observers are weirdly resistent to recognising the parody, even though it’s such a close parody, even though Lucian had every opportunity to read Christian texts, and even though it’s clear he was totally unaware of the traditional Jewish motifs that Revelation was drawing on.
One thing that happened this year was a major new game, Hades (not related to the Disney film above), which has been extremely well received. I’m pondering a piece on it for early next year.
  1. Another oldie: Who preserved Greek literature? (Part 1) (Dec. 2019). This was late in 2019, so it’s unsurprising that it continued to get some hits into 2020. It took me six months to write its sequel ...
  2. Who preserved Greek literature? (Part 2) (12 Jun.). Part 1 made the point that we don’t rely on mediaeval Arabic transmission at all for modern editions of ancient Greek texts, except in very sparse cases. Part 2 gives the true story. Both parts emphasise that the false narrative is strongly flavoured with prejudice against Byzantine Greeks and racism against modern Greeks. It strikes me that it seems to go hand-in-hand with the myth that modern Greeks aren’t descended from the ancient Greeks: Spencer Alexander McDaniel gives an excellent debunking of that myth here. Yes, I do think it’s that blatantly racist. The myth about modern Greek ethnicity isn’t a quirk of the 19th century. I’ve heard it in person from living classical scholars in Cambridge (‘They’re all Balkan immigrants anyway, aren’t they?’), and it’s a major component in scholars’ resistence to modern Greek pronunciation. Some of the responses to Spencer’s piece on Quora are spine-chilling.
  3. Two oldies occupy the 3rd and 2nd places. First, Shanties in Assassin's Creed: Odyssey (Jan. 2019), giving a transcription of the sea shanties sung by the ship’s crew in this hit game. I imagine the popularity of this article will start to decline once the next good Assassin’s Creed game comes out.
  4. The other oldie is Salt and salary: were Roman soldiers paid in salt? (Jan. 2017). This piece has been a hit for years — mostly, I’m guessing, because a couple of relevant Wikipedia articles have links to it. As of this year, it is now my most heavily visited piece of all time.

But that still wasn’t enough to beat this piece from 2020 —

(drum roll)

  1. How to make sense of ancient Greek colours (20 May). This topic is under a consistent barrage of misinformation, so I’m glad to help out in the huge task of repairing the damage. So much gets said on this topic by people who can’t be arsed to learn languages. Yes, the ancient Greeks could see blue. Yes, they had words for it. The misunderstanding comes from the fact that English suffers from disuse of terms for distinct parts of the blue side of the spectrum. English-speakers routinely assume that any well-designed colour terminology ought to have the same lack of distinctions.

The top three articles account for 54% of all visits.

And now, on to 2021. May the memories of 2020 fade quickly. But may we not forget those whose lives were lost because of politicians’ apathy.

Saturday, 19 December 2020

The Christmas stories: Matthew vs. Luke

Christmas isn’t adapted from pagan festivals, but that doesn’t mean the traditional stories of Jesus’ birth are a straightforward historical matter. The traditional Nativity story isn’t a replica of any ancient source. It’s a mash-up.

In this cinematic Nativity scene (The Nativity story, 2006) the traditional elements come from five separate sources — only two of them in the New Testament. The magi and the star are from Matthew 2.9–12, the shepherds from Luke 2.8–20, the cave from Protevangelium of James 18–21, the oxen from the Septuagint version of Habakkuk 3.2, and Mary’s blue robe from 5th century Byzantine art. (Also, Joseph and Mary are both played by Star wars actors ... but I guess that doesn’t count.)

Previously I’ve written some polemical pieces rejecting the notion that Christmas has pagan origins (1 2 3). But there’s a fine line between polemic and apologetic —

  • Polemic: ‘Some party-poopers like to claim that Christmas is pagan, but they’re talking nonsense.’
  • Apologetic: ‘The reason the party-poopers are wrong is because the Bible is 100% literally true.’

Polemics can be fun, but I don’t want to act the part of an apologist. So this year, I’ll redress the balance.

First, I’d better repeat the main point of my earlier pieces. Modern Christmas customs aren’t pagan. What they are is ... modern. Secular Christmas customs in the English-speaking world are mostly Lutheran in origin (trees, presents, Santa, Advent wreaths and calendars) and were adopted in England and America in the 19th century. The only ancient bits are the bits that happen in church.

So let’s look at a genuinely ancient bit: the stories of Jesus’ birth.

Stories (plural)

Yes, stories. We have two primary sources on the Nativity, Matthew 1.18–2.23 and Luke 1.26–2.52, and they’re totally different. They actively contradict each another in certain respects. Two methodological principles:

  • We don’t take Christian traditions for granted. We don’t assume that Matthew and Luke are telling a single story which is 100% true, and that our job is to explain away the inconsistencies.
  • Conversely, we don’t adopt the New Atheist strategy of dismissing everything biblical out of hand. Good data or bad data, Matthew and Luke are still data.

I’d better add before we carry on that the overall thrust of my argument here is totally uncontroversial among biblical scholars. Here’s Raymond Brown, in his classic study of the Nativity stories (1993: 35–36):

Commentators of times past have harmonized these different details into a consecutive narrative ... But if originally there was one narrative, how did it ever become fragmented into the two different accounts we have now? ... This leads us to the observation that the two narratives are not only different — they are contrary to each other in a number of details. ... Indeed, close analysis of the infancy narratives makes it unlikely that either account is completely historical.

(Brown was a scholar who was also a Christian, for what that’s worth. The book even has an official imprimatur.)

Now, let’s put the stories side by side. Notice how disconnected they are:

Episode Matthew Luke
Reign of Herod or Archelaus: Angel visits Mary (the Annunciation).   1.26–38
Mary visits Elizabeth; birth of John the Baptist.   1.39–80
Reign of Herod: Engagement of Joseph and Mary; Joseph plans to end engagement because of pregnancy. 1.18–19  
Angel visits Joseph in dream, tells him to go ahead with marriage. 1.20–25  
(Fulfilment of prophecy: Isaiah 7.14.) 1.22–23  
Governorship of Quirinius: Quirinius conducts a census.   2.1–2
Joseph and Mary travel to Bethlehem.   2.3–5
Birth of Jesus; no room at the inn.   2.6–7
The shepherds hear the news and visit family in Bethlehem.   2.8–20
Circumcision of Jesus.   2.21
Family stops off in Jerusalem to present Jesus at temple; episode of Simeon and Anna.   2.22–38
Reign of Herod: Magi visit Herod in Jerusalem. 2.1–7  
(Fulfilment of prophecy: LXX Micah 5.1, LXX II Kings 5.2.) 2.5–6  
Magoi go to Bethlehem, offer gifts, then return home without visiting Herod. 2.8–12  
Angel visits Joseph in dream to warn him of Herod’s rage; family flees to Egypt. 2.13–15  
(Fulfilment of prophecy: Hosea 11.1.) 2.15  
Herod’s rage: massacre of the innocents. 2.16–18  
(Fulfilment of prophecy: LXX Jeremiah 38.15.) 2.17–18  
Reign of Archelaus: family returns from Egypt after Herod’s death. 2.19–21  
Family doesn’t return home to Judaea, for fear of Archelaus, but instead goes to Nazareth in Galilee (since Galilee was no longer under Jerusalem’s control). 2.22–23  
Governorship of Quirinius: Family carries on home to Nazareth.   2.39–40

The other two canonical gospels, Mark and John, add some information about Jesus’ home life, but nothing about his birth. They corroborate Luke’s story that Jesus’ hometown was Nazareth, in Galilee. (Or maybe Capernaum. But definitely not Bethlehem, as it is in Matthew.) Jesus has a mother, brothers, and sisters in Mark 3.31–32; Mark 6.3 gives him four named brothers, at least two sisters, and a mother named Mary; and depending on the textual variant you choose, either Jesus’ father is a builder, or Jesus himself is. John 7.3–5 refers to Jesus’ brothers, 7.41–42 addresses the inconsistency between Jesus’ home in Galilee and the fact that the Messiah is supposed to come from Bethlehem, and 19.25–26 refers to his mother and maternal aunt. A few other snippets in Paul’s epistles and in Josephus give some further independent testimony about Jesus’ siblings.

(Some apologists reject all of the references to siblings, wanting Mary to have perpetual virginity for some reason. The strategy is either to interpret the word ‘sibling’ metaphorically every time it appears, or make up a story that Joseph had a previous marriage, so that Jesus had some half-siblings. If you don’t start out taking perpetual virginity for granted in advance, the mental acrobatics look pretty silly.)

Anyway, while Jesus’ family life has some interest in its own right, these passages don’t have anything much to do with the Nativity stories. For the Nativity, it’s all about Matthew and Luke. And that isn’t encouraging: no element of either story shows up in the earliest Christian texts, that is, in Paul’s letters or in the earliest gospel, Mark. Right from the word go we’ve got every reason to think of the Nativity stories as two relatively late impositions on earlier traditions and beliefs.

Matthew vs. Luke

To summarise: Joseph’s dreams, the magi, the star, the massacre of the innocents, and the flight to Egypt appear in Matthew, and only in Matthew. John the Baptist’s birth, the census, the shepherds, and the presentation at the temple are in Luke, and only in Luke. These are not stories designed to reflect a single underlying series of events.

There are only four significant points of agreement:

  • the names of Jesus’ (mortal) parents are Joseph and Mary;
  • Joseph is descended from king David;
  • the birth takes place in Bethlehem;
  • by the end of the story the family is living in Nazareth.

But they differ wildly on the circumstances of these points.

Note. Brown 1993: 34–35 adds a few more points of agreement, but some are just incidental (e.g. that Jesus is born after Joseph and Mary begin living together), while others are tendentious: see below on the angels’ announcements.
  • Date. The timeline of who was in control of Judaea and Galilee at any given time is quite straightforward:
    • Herod the Great: king until his death, in either 4 BCE or 1 BCE.
    • Herod Archelaus: tetrarch of Judaea, but not Galilee, from Herod the Great’s death until being ousted by the Romans in 6 CE.
    • Quirinius: Roman governor from 6 CE onwards, when the entire region became part of the province of Syria.
    Each story dates itself perfectly clearly. In Matthew Jesus is born under Herod’s rule, and the family moves to Galilee in Archelaus’ time, since Galilee was no longer under Jerusalem’s control at that time. Luke’s story begins in the time of either Herod or Archelaus, and Jesus is born after Quirinius becomes governor. (The author of Luke may have been unaware of Archelaus’ existence: at 1.5 he calls Quirinius’ predecessor ‘king Herod’.) That is: in Matthew, Jesus is born in Herod’s reign, perhaps in 4 BCE; in Luke he’s born under Quirinius, in 6 CE.
  • Ancestry. Both gospels give patrilineal genealogies that make Joseph a descendant of king David (Matthew 1.1–17; Luke 3.23–38). But the genealogies are totally different. If they’re both true, then Joseph had two fathers.
  • Hometown. In Luke, the family lives at Nazareth in Galilee, and their trip to Bethlehem in Judaea is a there-and-back affair. In Matthew they live in Bethlehem, and only move to Galilee to evade Archelaus’ clutches.
  • What happens after the birth. In Matthew, after the birth, the family flees to Egypt to escape the anger of the king in Jerusalem. In Luke, the family goes straight home to Galilee, popping in to Jerusalem on their way.
  • The visit to Jerusalem. When the family presents their baby at the temple in Luke, the idea is clearly that they’re visiting Jerusalem on their way home. In Matthew’s timeline, this means the family is making a leisurely visit to Jerusalem under the nose of Herod the Great, exactly the place that the angel tells Joseph to run away from. The only way to harmonise these would be to have the family go to the temple in Jerusalem, then go back to Bethlehem for no reason, and only then get visited by the magi and flee to Egypt.
  • The angel(s). Matthew’s angel visits Joseph twice, in dreams, and visits Mary zero times; Luke’s angels visit Mary and the shepherds once each, in person, and Joseph zero times.
  • Prophecies. Matthew links the Nativity to five passages from the Hebrew Bible, and treats them as prophecies that are fulfilled by Jesus’ birth. Luke’s Nativity doesn’t feature any prophecies.
The routes taken by Jesus’ parents in Matthew (red) and Luke (blue).

In some of these points the two stories actively contradict each other: the date, and the family’s movements. In other respects there isn’t an explicit contradiction, but equally, there’s no reason to imagine they’re telling a single story without any overlap. Both stories have an angel or angels declare that the Holy Spirit will cause Mary to become pregnant, and that the child’s name will be Jesus. But they say it to different people, in different situations, and they say different things about the child’s future. If you don’t start with the assumption that every incident is true, the obvious conclusion is that Matthew and Luke both wanted to have an angel element, but they baked it into their stories in very different ways.

Internal problems in Luke

The census (1). Quirinius’ census (Luke 2.1–2) was real, but it wasn’t a census ‘of the whole world’. It was confined to Judaea. Judaea became part of the Roman province of Syria in 6 CE, so the new governor, Quirinius, conducted the census to establish taxation information. The census is described by Josephus (Jewish antiquities 18.1) and attested in an inscription from Beirut (CIL iii.6687, line 9). There’s no evidence of any historical census ‘of the whole world’, and such a thing would be totally pointless.

The census (2). Luke’s premise is that the census required Joseph to go to the hometown of one of his ancestors. Again, this is intrinsically implausible, and there’s no evidence of the Romans ever requiring such a crazy thing. Even if they did, which ancestor’s hometown are you meant to visit? Joseph’s supposed descent from David goes back forty generations, according to Luke 3.23–31! Was he to visit the hometown of every one of them?

The presentation at the temple. Yeah, this wasn’t a thing. The sacrifice of two turtle-doves or pigeons was real, associated with ritual purification no less than forty days after childbirth (Leviticus 12.2–8). But the sacrifice of firstborn animals, and redemption of the firstborn human child, was unrelated (Exodus 13.11–14). The combo in Luke must be designed to echo yet another incident in the Hebrew Bible, Hannah dedicating Samuel as a baby (1 Samuel 1.22–28) — except that Samuel was left to stay at the temple permanently. In other words this isn’t a standard procedure, it’s a mash-up of multiple unrelated bits of the Hebrew Bible.

The date. As I mentioned above, Luke 2.1–2 puts Jesus’ birth during Quirinius’ governorship, in 6 or 7 CE. But just one chapter later we get a totally different timeline. Luke 3.1–3 sets the baptism of Jesus in the 15th year of the reign of emperor Tiberius, and 3.23 tells us that he was 30 at the time. Now, Tiberius became emperor in September 14 CE: the 15th year of his reign was 28–29 CE. This means that as far as Luke 3 is concerned, Jesus was born in 3–2 BCE — at least seven years earlier than in Luke 2. This is a decent match for the timeline in Matthew, as it happens. But not for Luke’s own Nativity story. (Some scholars suspect that Luke’s Nativity story is a relatively late addition to the gospel, which would explain the inconsistency: more about this below.)

The historical birth of Jesus

So, is there anything historical in either Nativity story?

One strategy could be to argue that, because they’re independent sources, the elements where they do agree ought to be the bits where they’re most likely to be historical. There are major problems, though.

For one thing, no trace of either Nativity story shows up in Paul or in Mark, as I mentioned earlier. There’s no reason to think any of it represents early traditions or beliefs about Jesus’ birth.

In addition, there are questions over the text of Luke. It has sometimes been suspected that Luke’s Nativity story is an interpolation, written decades after the rest of the gospel. That theory would explain why Luke’s Nativity has almost nothing in common with Matthew’s. It would explain why Luke 3 feels very introduction-ish: it opens with an elaborate introduction pinpointing in time the beginning of Jesus’ ministry, and the genealogy of Jesus doesn’t appear until chapter 3.

Well, maybe. Actually I’m not convinced. There are some striking thematic links between Mary’s and Simeon’s songs in Luke 1–2 and the rest of the gospel, which suggests that they’re integral. Also, I’ve tried some basic stylometric tests (without being an expert in stylometric analysis, mind), and my tests consistently group the three synoptic gospels together, and John separately. When it comes to the Nativity stories, though, the tests don’t see much difference. In terms of authorial style, the Nativities consistently come out pretty close to the rest of their respective gospels. By contrast, the same tests easily detect a passage which is known to be an interpolation: the episode of the woman caught in adultery in John 7.53–8.11.

Stylometric tests of the four canonical gospels conducted in RStudio with the package ‘stylo’. The text was prepared by stripping all accents, case, and punctuation; converting iota subscript to iota adscript; transliterating into the Roman alphabet using Beta Code; and dividing the text into chapters. Two passages known to be interpolations were separated from their chapters, John 7.53–8.11 and the long ending of Mark (shown here as Mk_16b). The tests shown here plot the 200 most frequent 3 word sequences (left) and 4 word sequences (right), and maximise the distance between each chapter using principal component analysis. Notice that the chapters of John (in red) are grouped separately from the three synoptic gospels, indicating differences in language, authorial style, or content.

Now, I’m no expert in stylometry. So I’ll just say that it isn’t self-evident to me that Luke’s Nativity story is an interpolation. That doesn’t mean there’s a shred of truth in it, mind.

Here’s my pick for the elements of the Nativity stories that accurately reflect historical reality:

  • The names of Jesus’ parents were Joseph and Mary;
  • Jesus grew up in Galilee (there’s actually some doubt about whether his home was in Nazareth, but it’s too much of a distraction: let’s just let that slide for now).

But the Bethlehem setting can’t stand. Both stories feel a pressing need to have the birth take place in Bethlehem, but the transparent falseness of Luke’s version — the census — indicates that Bethlehem’s role is an invention. Not a late invention: it does appear in two independent Nativity stories, after all, and it’s there in John 7.41–42 too:

But some asked, ‘Surely the Messiah does not come from Galilee, does he? Has not the scripture said that the Messiah is descended from David and comes from Bethlehem, the village where David lived?’

The reference to ‘scripture’ is to Micah 5.1, which Matthew quotes.

It looks like early Christians genuinely did wrestle with the fact that Jesus did not come from Bethlehem. Bethlehem was introduced into some of the stories very early on, as a way of addressing that fact. So three of the gospels solve the problem in three different ways. In Matthew Jesus’ family did indeed live in Bethlehem, but moved to Galilee (with Archelaus’ pursuit as a spurious excuse); in Luke they make a field-trip to Bethlehem (with Quirinius’ census as a spurious excuse); and in John Bethlehem isn’t involved at all, so we see people debating over the matter.

Of the three, John’s picture is probably the most true to reality.

The mash-up effect

The mash-ups that you see in Christmas cards, in Advent calendars, and in films like Ben Hur, The Nativity story, and The star, is not a new thing. It didn’t suddenly pick up speed with the advent of biblical literalism in the 19th century. The idea of combining Matthew and Luke into a single story, inconsistencies and all, goes back to the 2nd century, and a non-canonical text called the Protevangelium of James.

Another mash-up (The star, 2017): Matthew’s star, magi, and rage of Herod; Luke’s census and shepherds; Septuagint Habakkuk’s cluster of animals.

In James (see Elliott 1993: 46–67 for an accessible translation) the problem of the census taking place nearly a decade after Herod’s death is just ignored. But the author realised perfectly well that the presentation at the temple can’t possibly happen while Herod is raging. Instead, Herod murders the high priest Zacharias, John the Baptist’s father, and Simeon succeeds him, whereupon it is prophesied that Simeon will get to see the Messiah one day (James 23–24). Several elements are added, apparently out of thin air: a lengthy account of Mary’s conception, birth, and youth spent in the temple (James 1–8, 12); the birth takes place in a cave, and time stops at the moment of Jesus’ birth (18–19); there’s a midwife; and a woman named Salome tests Mary’s perpetual virginity by checking whether her hymen is intact after childbirth (19–20).

This may sound weird and quaint, but of such stuff are traditions formed. Jesus’ birth is still sometimes imagined as taking place in a cave, instead of a stable. The Protevangelium of James is where that trope comes from. James omits the prophecies that take up so much of Matthew’s story, and that does make me wonder whether the author had an earlier version of Matthew where the prophecies had not yet been inserted; but a respected biblical scholar has assured me that it’s unlikely that such a version of Matthew ever existed.

Anyway, the point is that the mash-up effect is key to how people approach the story of the Nativity. Story elements pile up around the key moment of Jesus’ birth, in much the same way that secular customs pile up around the Christmas festival. One thing leads to another. We can’t put much trust in the historicity of the Nativity stories, but it’s still very interesting to investigate the traces of how they grew and developed over time.


  • Brown, R. E. 1993 [1977]. The birth of the Messiah. A commentary on the infancy narratives in the gospels of Matthew and Luke, updated edition. Doubleday.
  • Elliott, J. K. 1993. The apocryphal New Testament. A collection of apocryphal Christian literature in an English translation. Clarendon Press. (Protevangelium of James at 46–67)

Thursday, 29 October 2020

Cannabis use in ancient Greece and Rome

On Friday we should know the result of a referendum held in New Zealand to decide whether cannabis should be legalised. The referendum was held in conjunction with the 2020 general election. Around 82.5% of eligible voters participated, but we’re still waiting on the counting of special votes -- nearly 17% of all votes cast.

It’s always been daft to restrict such a useful plant, especially when its demonised recreational use is much more harmless than either tobacco or alcohol. (It was only ever banned in the first place for racist motives.) In spite of that, polling suggests the referendum is unlikely to pass. In the meantime, let’s have a look at its use in the ancient Greco-Roman world.

Hemp seeds and leaves: in Greco-Roman antiquity, it was mainly just the seeds that were consumed. Cannabis wasn’t smoked until the modern era. (Source: PNG all, CC 4.0 BY-NC)

If you bring up the subject of cannabis in antiquity, be prepared to have Herodotus quoted at you every single time, usually inaccurately. Here’s Wikipedia on the subject:

The oldest written record of cannabis usage is the Greek historian Herodotus’s reference to the central Eurasian Scythians taking cannabis steam baths. His (c. 440 BCE) Histories records, ‘The Scythians, as I said, take some of this hemp-seed [presumably, flowers], and, creeping under the felt coverings, throw it upon the red-hot stones; immediately it smokes, and gives out such a vapour as no Greek vapour-bath can exceed; the Scyths, delighted, shout for joy.’ Classical Greeks and Romans also used cannabis.
‘Cannabis’, Wikipedia, Oct. 2020

OK, here’s what has gone right in this report:

  • Herodotus does indeed report on Scythians burning hemp seeds (not flowers) while under a cloth and getting high off the fumes.

... and that’s it for the positives. Everything else here is terrible.

  • Herodotus isn’t talking about ‘central Eurasia’: that would imply Kazakhstan or thereabouts. He’s talking about the region inland from Greek colonies on the Black Sea, that is, Ukraine.
  • ‘Steam baths’ is a bit misleading. He does refer to the hemp seeds on hot stones giving off an ἀτμίς ‘vapour’, but this is in the context of talking about the Scythians’ purification rituals. It’s about religious cleanliness, not physical cleanliness.
  • The Histories date to the 420s BCE, not 440. A minor point, but still.
  • The article cites its source as ... the entirety of the Histories. That’s about 260,000 words to search through, or 500–600 pages. Good luck finding the passage! (It’s 4.73–75, by the way.)
  • ‘Presumably flowers’: no, that’s pure imagination. Herodotus very explicitly refers to seeds. Nearly all human consumption in antiquity was of the seeds.
  • ‘Scyths’? You’d never guess that 21st century translations exist, would you.
  • ‘Shout for joy’ is a mistranslation: ὠρύονται means ‘howl’ (like dogs).
  • The last sentence, ‘Classical Greeks and Romans also used cannabis’ (as a mind-altering drug), is given without citing any evidence. There’s a good reason for that. It’s because there is no evidence.

Yes, that’s right: there’s no evidence of anyone using cannabis to get high in the ancient Greco-Roman world. I’m afraid an awful lot of the argumentation you find from people like Carl Ruck and D. C. A. Hillman is wishful thinking, with little interest in petty concerns like evidence. (They usually focus on harder or more dangerous drugs anyway — opium, ethylene, ergot.)

For the Greeks and Romans, there’s no evidence for the deliberate use of cannabis as a mind-altering substance. The main reason is because (a) they were mainly interested in the seeds, and (b) the idea of smoking the buds wasn’t invented until the modern era. There’s one ancient report that eating too many seeds can affect the mind — but only as a side-effect of a common food item. There’s no suggestion of deliberate use as a psychoactive agent.

Chlöe Swarbrick, a New Zealand politician who has been a vocal supporter of the 2020 cannabis referendum. Up until the October 2020 election Swarbrick held the Drug Law Reform portfolio for the Green Party; at the time of writing she is (provisionally) the MP designate for the electorate of Auckland Central. (Source: New Zealand Herald)

Let’s refine how we frame this. The Wikipedia claim, ‘Greeks and Romans used cannabis’, has a very different meaning depending on whether we’re talking about smoking joints, or weaving cloth out of hemp. We can distinguish four basic categories of how people use psychoactive plants:

  • uses not relating to psychoactive effects (like food, or industrial goods);
  • medicinal use;
  • recreational use;
  • religious use.

Take alcohol, for example. The ancient Greco-Roman world gives us copious evidence of alcohol use in all four of these categories. For opium, we find ancient sources talking about medical uses frequently (and hardly ever about any other use). What’s the situation with cannabis?

Well, we have good evidence of cannabis use in the first category. People ate hemp seeds, either by themselves (shelled, of course), or as an ingredient in other dishes. There were heaps of industrial uses, including rope-making, textiles, hunting nets, and shipbuilding (for details see Brunner 1973: 348–349, 351–354). Hemp is an incredibly useful plant, and it’s insane that its use has become so restricted in the modern era.

The seeds are mentioned as an ingredient in sweet dishes in particular. A fragment of the comic playwright Ephippus mentions them in a rambling list of delicacies (fr. 13 Kock).

and after dinner ...
mnous [meaning unknown], wheat-cakes, fruit, nuts,
milk, hemp seeds, cockles, juice,
and Zeus’ brain.

(No, we don’t take this as evidence that people ‘ate Zeus’ brain’ symbolically or anything like that. It’s a comedy.)

Pliny the Elder, writing in the 1st century CE, does mention that the stalk and branches, too, could be used as a vegetable (olera; NH 19.175). But not an appealing one. An epigram in the Greek anthology, by the 1st century BCE poet Automedon, compares them to old cabbage (11.325):

Yesterday I dined on a goat’s foot, and a ten-day-old
     quince-coloured cabbage stalk, like cannabis.
I won’t mention the person who invited me. He’s sharp-tempered,
     and I’m scared he might invite me back again.

For the second category, medicinal use, we have a variety of evidence: for full details see Butrica (2008). The Byzantine Geoponica describes the use of cannabis as an insect repellent, and a preparation of cannabis ash and honey as a remedy for ulcers (Geoponica 13.11.4, 16.15.2). The 1st century CE medical writer Dioscorides, after mentioning its industrial use and its unpleasant smell, states (On medical material 3.148)

It has ... a round-shaped seed which is edible, and when eaten in excess diminishes sexual potency. The juice of a fresh seed is suited to treating earaches, administered by eardrops.

Notice that he’s talking only about the seeds. We see the same tendency in other medical sources. No one tried burning or smoking cannabis buds, which grow on seedless plants.

In November 2018 Helen Clark, Prime Minister of New Zealand from 1999 to 2008, spoke out in favour of the 2020 cannabis referendum, and campaigned actively in 2020. Since 2009, Clark has been on the United Nations Global Commission on Drug Policy, among several other United Nations roles. In August 2020 she was appointed Chair of the Commision. (Source: YouTube)

The most suggestive snippet comes to us from Galen, the greatest medical writer of antiquity. He repeats Dioscorides’ claim that excessive consumption reduces sexual potency, and that the juice of the fresh seed can be used to treat earache (On the combination and effect of single medicaments §5, xii p. 8 Kühn). Elsewhere he states that the seeds cause stomach aches and head aches. But straight after that he adds something about their intoxicating effect — the only ancient source other than Herodotus to do so (On the effects of foods 1.34, vi p. 550 Kühn):

Even so, some people eat (the seed), toasting it with other sweets. By ‘sweets’ I refer to things eaten after dinner to induce enjoyment of drinking. The seed has a moderate warming effect, and so, when a large amount is taken in a short space, it overcomes the head, filling it with a warm and drug-like vapour.

Galen didn’t know THC by name, of course. And note that he’s still only talking about using the seeds, not buds or other parts of the plant.

Some possible hints appear in Oreibasius’ Synopsis for Eunapius, written in the 4th century CE. He says cannabis prevents flatulence (4.21), but he also includes it in a list of things that ‘harm the head’ (4.20). Elsewhere Oreibasius says cannabis has a warming effect, as does Galen ... but bear in mind that ‘warmth’ isn’t a metaphor, but a technical term in ancient Greek medicine. This list isn’t made with any psychoactive effects in mind (4.31):

Things that have a warming effect: boiled wheat and bread made from it; marsh water; oats; fenugreek; juniper berries; sweet dates; sweet apples; sesame; hedge mustard (which also has a drying effect); cannabis seed; sweet grapes ...

We haven’t addressed the fourth category: the religious angle. There’s no evidence for religious use in the Greco-Roman world. But there is plenty of archaeological evidence to confirm Herodotus’ report of Scythian ritual burning of cannabis seeds. There have been finds of cultivated cannabis seeds in Ukraine at Kaminske (Pashkevich 1999: 600) and Nemirov, and in the Kuban between the Black Sea and the Caspian Sea (Corcella 2007: 635).

Archaeological finds matching the physical details of Herodotus’ description of Scythian religious practice, from Pazyryk, mound 2 (5th–3rd century BCE). Bottom right, a pot containing hemp seeds; top right, a censer with stones for heating the seeds; left, the frame to hold up the 45 cm tent over the censer. (Source: Artamonov 1965: 108)

Pride of place, though, goes to a find at Pazyryk, in Siberia, close to the Mongolian border. There in 1947 Soviet archaeologists found, in a burial mound probably dating to the 5th–3rd centuries BCE, cannabis seeds along with equipment suited to exactly the ritual that Herodotus describes. This is a long, long way from any place that Herodotus visited — 4000 km from Ukraine — but Scythian culture was so widespread that it is generally accepted as reasonable to infer a comparable ritual among Siberian Scythians.

Cannabis took a long time to take off. People didn’t smoke it until the modern era. Prior to that, the psychoactive nature of the plant wasn’t universally used: it wasn’t even universally known. Like all crops, cannabis had to spread.

Even in Herodotus, the phrasing of his report is sometimes taken to imply that cannabis wasn’t all that familiar to his readers: it may have been a newcomer. One archaeologist puts the arrival of cannabis in the Greek world in the Hellenistic period, that is, more than a century after Herodotus (Kroll 2000: 67). I’d say its presence in Ephippus’ list of after-dinner treats puts it a bit earlier than that. But it took a long time for cannabis to be understood. For that reason alone it’s no surprise that recreational and religious use are rare, or even non-existent, in the evidence we have from ancient cultures.


  • Artamonov, M. I. 1965. ‘Frozen tombs of the Scythians.’ Scientific American 212.5: 100–109.
  • Brunner, T. F. 1973. ‘Marijuana in ancient Greece and Rome? The literary evidence.’ Bulletin of the history of medicine 47.4: 344–355.
  • Butrica, J. L. 2008. ‘The medical use of cannabis among the Greeks and Romans.’ Journal of cannabis therapeutics 2.2: 51–70.
  • Corcella, A. 2007. ‘Book IV.’ In: Asheri, D.; Lloyd, A.; Corcella, A. A commentary on Herodotus books I–IV. Oxford UP. 543–721.
  • Kroll, H. 2000. ‘Agriculture and arboriculture in mainland Greece at the beginning of the first millennium B.C.’ Pallas 52: 61–68.
  • Merlin, M. D. 2003. ‘Archaeological evidence for the tradition of psychoactive plant use in the Old World.’ Economic botany 57.3: 295–323.
  • Pashkevich, G. 1999. ‘New evidence for plant exploitation by the Scythian tribes during the Early Iron Age in the Ukraine.’ Acta palaeobotanica Suppl. 2: 597–601.
  • Zohary, D.; Hopf, M.; Weiss, E. 2012. Domestication of plants in the Old World, 4th edition. Oxford UP.

Monday, 14 September 2020

The classical elements

Earth; Fire; Air; Water. Long ago, Empedocles theorised that four elements lived together in harmony. Then everything changed when Democritus attacked. Only Aristotle could stop him. But when the world needed it most, modern chemistry was still several thousand years from being invented ...

Empedocles’ poems don’t survive. But Aristotle’s Physics and On coming to be and passing away do. As a result, a number of people have got it into their heads that Aristotle was reponsible for the idea of the classical elements. Here’s a snippet from a 2002 lecture by a Florida State physics professor about ‘Aristotelian’ physics:

  • everything on Earth made of (mixture of) four elements: earth, water, air, fire
  • every element has a “natural place”:
    • earth at center of Earth,
    • water above earth,
    • air above water,
    • fire above air;
  • celestial bodies (stars, planets, Moon) made from fifth element, “ether”, which also fills space between them; ether is perfect, incorruptible, weightless; ...

This slide has ended up being cited by the Wikipedia article on ‘Aristotelian physics’ as the supreme authority on the classical elements.

Now, it isn’t completely made up — in the limited sense that some people, including even some specialists, do claim some of these things.

For example Werner Jaeger (1948), Aristotle. Fundamentals of the history of his development, Oxford, 143ff. One report on a lost Aristotelean work refers to the elements earth, water, and air, then refers to the fourth as Latin ardor. Jaeger confidently translates the word as ‘aether’ (149).

But you won’t find any of it in Aristotle himself.

  • Aristotle never says that all matter is made of (a mixture of) earth, water, air, and fire.
  • He doesn’t talk of elements having a ‘natural place’.
  • He doesn’t add aether as a ‘fifth element’.

The thing about aether is probably the biggest one, because I have seen it repeated in so many places. So let’s just repeat. Aristotle doesn’t treat aether as a fifth element.

Here’s what Aristotle does say.

  • He uses Empedocles’ four elements as categories of material qualities, rather than substances.
  • He also refers to Democritus’ model of matter, which is based on atoms, not elements, and which allows an unlimited variety of fundamental substances.
  • He accepts and rejects aspects of both models. He suggests that matter is differentiated by the presence or absence of various properties, rather than the substances it’s made of; and he’s willing to go along with Anaxagoras’ idea that every homogeneous material is an element, to the extent that each homogeneous material corresponds to a unique set of properties.
  • He talks of materials having a natural motion depending on their properties: hot things (including fire and air) tend to rise, that is, move away from the centre of the earth, while cold things (including earth and water) tend to move towards the centre.
  • And, in a discussion that has nothing to do with elements, he does say that the sky uniquely has a circular natural motion — the sky as a single thing, that is, not individual heavenly bodies — and he borrows the ancient name ‘aether’ to refer to this unique motion.

Now a lot of this is wrong of course. Atoms are real; natural motion isn’t a thing, but buoyancy is; it’s the earth that’s rotating, not the sky. But that’s no excuse for misrepresenting Aristotle. His theory of matter is about four physical properties, not about four substances. And nowhere does he declare a canon of five elements, with aether as the fifth.

He even calls Empedocles’ model self-contradictory. Empedocles’ elements are supposed to be distinct, fundamental substances; yet Water evaporates into Air, condensation from Air produces Water.

So Empedocles evidently contradicts the observed phenomena, and contradicts himself too. For he says that none of the elements can emerge from another, but that everything is made of these; but at the same time, once he gathers all of nature (except Strife) into the One, he says that each of the elements is derived from the One.
The early Greek flat-earth cosmology as found in Homer, centuries before Aristotle. Thick, foggy Air is at ground level; above it is the clear Aether, where it is hard to breathe and there are no clouds. Elements played no role in this cosmology.


A second-hand report of one of Aristotle’s lost works makes it clear that he thought of heavenly bodies as ‘fiery’, and not made of a fifth element.

Therefore, because the Fire of the sun is similar to those fires that exist in the bodies of animate creatures, it must be that the sun is also animate ... So since the origin of some creatures lies in Earth, others in Water, and others in Air, Aristotle thinks it absurd to imagine that no animal is generated in that element which is most suited to generating animate things.

Cicero (our source) goes on to talk about how the ‘stars occupy the aetherial region, which is extremely thin’. Various modern scholars have taken that as meaning that Aristotle, too, linked aether to the usual four elements. But, again, it’s indirect. And even Cicero doesn’t cast aether as the fifth member of a set of five. He refers to ‘the aetherial region’ (aetherium locum) as a location, not a substance. He talks of stars being generated there: that is, Cicero is actually thinking of ‘the aether’ as a place occupied by Fire, but only very sparsely, until it generates stars.

Aether isn’t a distinct element: it’s a hangover from archaic flat-earth cosmologies. For example in Homer, thick misty Air (aēr) is at ground level, and clear, bright Aether (aithēr) is above it, filling the heavens. On mountain-tops the air is thinner, and you find yourself above the clouds: that’s supposedly where you get close to the aether.

The elements in Plato

Aristotle isn’t our earliest source on Empedocles’ elements. There’s an earlier report in Plato’s Timaeus. And it’s Plato, I think, that has prompted some modern readers to think in terms of a group of five.

The passage is hard to interpret (like most of the Timaeus): Plato could be misconstrued as saying that celestial bodies are made of a fifth element. What he really seems to be talking about is the shape of the universe as a whole, not the material that stars are made of.

Aether does come up a bit later, in passing. But Plato makes it very clear that it isn’t a separate element. He treats aether as a variety of Air, in the same way that flames, firelight, and glowing embers are all varieties of Fire (Timaeus 58c-d).

The context is that Plato is trying to combine two theories of matter: Empedocles’ model of four elements, and Democritus’ atomic theory. Plato also injects his own fondness for ideal abstractions. He proposes that each element is made of a single type of atom, and the four atom types have the shapes of the ‘Platonic solids’. There are five possible Platonic solids, so according to Plato’s perfectionist logic, that implies there must be some fifth thing in the material world that corresponds to the fifth solid, the dodecahedron.

The ‘Platonic solids’: the regular tetrahedron (4 sides), cube (6), octahedron (8), dodecahedron (12), and icosahedron (20). These solids have a number of unique properties that no other solid can have. Each has equilateral faces all the same shape (equilateral triangles, squares, or pentagons); every vertex coincides with the surface of a circumscribed sphere; the centre of every face coincides with the surface of an inscribed sphere.

Of these solids, Plato decides the cube is the most ‘immobile’ one, so that’s Earth. For the others, Fire is the lightest and smallest, made of tetrahedron-shaped atoms; Water is the largest of the triangle-faced solids, the icosahedron; Air is in between, the octahedron.

And, since there was still one more structure — the fifth — God applied it to the whole, and decorated it with life.

‘The whole’ must mean the cosmos, the sky. But not the substance of the sky. In the last bit, ‘decorated it with life’, the Greek word used, diazōgraphōn, is probably meant to evoke the twelve constellations of the zodiac around the ecliptic. The zodiac isn’t arranged in a dodecahedral shape, but perhaps the theme of twelve-ness was enough to sustain Plato’s wordplay.

For Plato, it seems, the dodecahedron is the shape of the cosmos. It looks like the idea is that a dodecahedron is the shape that most closely approximates a sphere — the sphere of the sky.

Surprising but true. Intuitively, you’d imagine an icosahedron would be closer to spherical, because it has more polygons. It turns out that pentagon-shaped faces are more of an advantage. The dodecahedron is better at filling a sphere. If you’re an ancient Greek and you want to calculate the volume of a sphere, but you don’t know the value of pi, and so you resort to the method of exhaustion that was popular in classical Greek geometry, then a dodecahedron provides closer bounds than any other solid.

Solid Number of faces Corresponding phenomenon (Tim. 53c-56e) Volume as proportion of sphere
tetrahedron 4 Fire 12.3%
cube 6 Earth 36.7%
octahedron 8 Air 31.8%
dodecahedron 12 universe 66.5%
icosahedron 20 Water 60.5%
A cube, all of its corners touching a sphere from the inside, fills a bit over a third of the sphere; a dodecahedron fills just under two thirds.

So, in Plato’s model, the dodecahedron is the shape of the universe; the other four solids are the shapes of the four kinds of atoms. Don’t go taking this seriously, of course! Plato liked maths, but he was definitely no empiricist. This whole imaginary concoction is driven by his obsession with the idea that real phenomena are manifestations of ideal abstractions.

Aristotle’s theory of matter

Aristotle is more logical, albeit with the same limitations on his knowledge. Given that he knew nothing about atoms and elements as we understand them nowadays, he still makes a reasonable amount of sense, so far as he goes.

For Empedocles says that there are four physical (elements); or a total of six, including those (elements) that cause motion (i.e. Love and Strife). Anaxagoras and Leucippus and Democritus, on the other hand, say there are infinitely many. The first (Anaxagoras) treats homogeneous things as elements, like Bone, Flesh, and Marrow, and every substance where a portion is synonymous with the whole. Democritus and Leucippus, however, say that these (homogeneous) things consist of indivisible particles, and that they are infinite in number and in their shapes; while substances differ from one another depending on the elements they consist of and their composition and arrangement.

Anaxagoras’ followers apparently take a stance opposed to that of Empedocles' followers. For Empedocles says that there are four elements — Fire, Water, Air, and Earth — and that these are the substances that are simple, rather than flesh and bone and other homogeneous substances. But Anaxagoras’ followers say that it is these homogeneous substances that are simple elements, while Earth, Fire, Water, and Air are compounds.

Aristotle rejects Democritus’ atomic theory, but he also rejects Empedocles. He accepts the four elements as empirical phenomena, but he rejects the idea that everything is made of them. He doesn’t reject Anaxagoras’ ideas about homogeneous substances: it seems that’s where his sympathies lie.

For Aristotle, Empedocles’ elements are emergent. They’re derived from combinations of properties: hot, cold, dry, and wet. Apply these properties to raw matter — what he calls hylē (literally ‘wood’) — and you get stuff that approximates the four elements. If something is hot and dry then it’s fiery, if it’s cold and wet then it’s watery. The properties come first.

  Hot Cold
Wet Air Water
Dry Fire Earth

He quickly gets into trouble, though. Aristotle questions whether these binary properties denote the presence vs. absence of a quality: so hot = ‘with heat present’, cold = ‘without heat’, and so on.

But someone might well be puzzled about a simple absence, and whether it is one of two terms in an opposition. For example, is it that Earth and other massive matter denotes the absence of a quality (heat), while Fire and rising substances denote the presence of that quality? Or is it that Earth also denotes a presence, while the absence denotes raw substance — the hylē of both Fire and Earth alike? And is the hylē of each of them different? For then they could not come to be out of one another, that is, out of their opposites. For the oppositions are manifested in these things: Fire, Earth, Water, Air.

This is where Aristotle’s lack of knowledge gets him tied in knots: this passage makes little sense to a modern reader. It’s obvious to us, with an understanding of chemical reactions and thermodynamics, that these are actually very bad questions. We know that physical properties are determined by heat and pressure, not by irreducible characteristics like whether ‘wetness’ is present.

This business of thinking about presence and absence of qualities misleads Aristotle before he can even begin. I think it’s still a huge improvement on Plato, with his fantasies about four species of polyhedral atom. But it isn’t like you’ll actually learn anything about chemistry from Aristotle.