← Articles & Videos

What Are Cherubim Doing at Eden, the Ark, and the Throne?

Sunlit path through a Mediterranean olive grove. Photo by Nikolaos D. Nomikos / Pexels.

The first cherubim in the Bible receive a posting order and no portrait. The man and woman are driven out of the garden, and then, in Genesis 3:22-24, God puts something on the east side of Eden to keep the road to the tree of life closed. The NET renders the Hebrew term as “angelic sentries” and drops the traditional transliteration into a footnote, which is already an interpretation rather than a description.

When he drove the man out, he placed on the eastern side of the orchard in Eden angelic sentries who used the flame of a whirling sword to guard the way to the tree of life. (Genesis 3:24, NET)

Count what the verse hands over. A direction. A weapon that turns. A job, stated with a verb that means to keep or guard. Now count what it withholds. Height, face, wings, number, voice, anything at all that a reader could sketch. The most famous creatures in the Bible’s opening chapters are introduced by their assignment and left physically blank.

My read of the whole body of cherubim texts is that this is not an accident of one verse. Scripture is steady about what these creatures do and stubbornly uneven about what they look like. The gold pair on the ark, the divine title about sitting above them, and the four living creatures Ezekiel sees by a Babylonian canal all put cherubim at the edge of God’s presence. They do not combine into one anatomy chart, and the two most common ways of filling in the blank, the winged infants of later European painting and the four-faced figures of Ezekiel 1, are both imports when they are applied to Genesis 3.

The word before the picture

The Hebrew singular is keruv and the plural is keruvim, which English borrowed whole, which is how older Bibles ended up printing “cherubims” with a doubled plural. Where the word came from is unsettled. The Encyclopedia of the Bible entry reports the etymology as uncertain and notes the long-standing proposal that the term is related to Akkadian karibu or kuribu. That proposal is worth knowing and it is not a result. A plausible loanword tells you the term may have traveled. It does not tell you what an Israelite scribe pictured when he used it.

So are cherubim angels? That depends entirely on what the word angel is doing in your vocabulary. In the Hebrew Bible, mal’akh is closer to a job title than a species: it means messenger, and a mal’akh shows up carrying something from God to a person. Benjamin Sommer’s overview of angels in the Hebrew Bible treats cherubim as a separate category, a class of heavenly being often described as composite and tied to God’s throne rather than to errands. That distinction holds up in the texts themselves. Cherubim are stationed, mounted, built in gold, and seen under the glory. They are not sent with news.

English-language habit does most of the damage here. If a reader’s mental file for “angel” is one undifferentiated drawer holding messengers, seraphim, cherubim, and whatever the Christmas pageant needed, then every distinction the Hebrew makes has already been erased before the reading starts.

Eden, the lid, and the title

Take the three main settings in order and the shared function is hard to miss. Eden closes with guards at the approach. Then Exodus 25:18-22 orders two cherubim of hammered gold at the two ends of what the NET calls the atonement lid, the kapporet, traditionally the mercy seat in English Bibles. Their wings spread upward and overshadow the lid. Their faces turn toward each other and toward it. Then the instructions do something the craftsmen could not have built: God says he will meet Moses there and speak from above the lid, between the two cherubim.

That is not decoration. It is an address. The pair marks the precise square of space where speech from God is promised, which is also the space Leviticus 16:2 tells Aaron he cannot casually enter, because God appears in the cloud over that same cover. The Day of Atonement instructions and the ark instructions are describing one restricted location from two directions, which is part of why the ritual geography of Leviticus 16 runs the way it does.

The same logic turns up in a standing title for God. When the ark is carried into battle in 1 Samuel 4:4, the NET calls the God of Israel the one who sits between the cherubim, and Psalm 99:1 has him enthroned above them while the earth shakes. The title assumes a reader who already knows the furniture. It never pauses to explain it.

Solomon’s inner sanctuary is where the assumption gets awkward. That pair stands ten cubits high with a ten-cubit wingspan, and they face the entrance rather than each other. Why would one sanctuary tradition arrange the same creatures two different ways? Set that beside the smaller pair on the lid, facing inward over a box, and one arrangement is a screen across an approach while the other is a canopy over a specific spot. The survey of the cherubim and their role on the ark lays out the competing functional readings that follow: guards, bearers of a chariot, the throne itself, or a protective buffer between God’s presence and everything else. Whether the ark pair is a seat, a footstool with sides, or a shield remains disputed, and I do not think the text of Exodus 25 decides it.

Here is what changes if the functional pattern is right. The cherubim stop being an angelic species the Bible forgot to describe properly and start being a signal about location. The road back to the tree of life is guarded. The lid of the ark is overshadowed. The title of God is spatial. Read that way, Genesis 3 is not an isolated flourish at the end of a sad chapter. It is the first instance of a pattern that the sanctuary texts later build in wood and gold.

Does Ezekiel settle what a cherub looks like?

This is where most confident answers come from, and it is the place to slow down. Ezekiel 1 describes four living creatures with four faces each, multiple wings, human hands, burning movement, and wheels. Nine chapters later the prophet makes the identification explicit in Ezekiel 10:15-22.

These were the living creatures that I saw at the Kebar River underneath the God of Israel; I knew that they were cherubim. (Ezekiel 10:20, NET)

That sentence is doing real work, and it settles less than it appears to. It tells you that what Ezekiel saw belongs to the category his tradition called cherubim. It does not tell you that every cherub in Scripture is built this way, for two reasons the text supplies on its own. The first is genre. Ezekiel reports a vision in relentlessly comparative language, a likeness of this, something with the appearance of that, and a reader who converts likeness-language into a specification sheet has changed what kind of writing it is. The second is Exodus. The craftsmen building the ark were given no faces to sculpt, no wheels, and no count of wings beyond the pair that overshadows the lid. If Ezekiel’s creatures were the standard form, those instructions are missing almost everything.

A fair objection forms right here. Maybe the silence is not ignorance but assumption, and the original audience knew exactly what a cherub looked like because such figures were everywhere in their visual world. That objection is strong, and it may well be right about the original audience. It still leaves the modern reader outside the room. The Encyclopedia of the Bible entry cited above reports Josephus saying that no one knew what the cherubim looked like, which I am passing along secondhand rather than from Josephus directly. If that report is accurate, the knowledge was already gone by the first century.

The winged toddlers are easier to date. Those come from a Greco-Roman artistic vocabulary that later European painting absorbed and attached to the biblical word, and there is a rabbinic reading, credited to Rabbi Abbahu, that connected the term to a word for child. Raanan Eichler’s survey of what kind of creatures the cherubim are walks through that history and finds the child etymology unconvincing. It is late, it is a proposal about a word rather than a report about a text, and nothing in Genesis, Exodus, Samuel, Kings, or Ezekiel supports it. Two further collapses are worth refusing while the topic is open. Revelation’s four living creatures resemble Ezekiel’s closely, but Revelation does not call them cherubim. And Ezekiel 28 uses cherub language inside an oracle against the king of Tyre, which is a taunt against a human ruler, not a biography of Satan and not a physical description to be reverse-engineered into Genesis 3.

The disagreement worth keeping

The most common reconstruction for the ark and temple statues is a winged sphinx, a creature with a lion’s body, wings, and a human head. The case is genuinely good. Composite winged guardians are well attested in the ancient Levant and Mesopotamia, thrones flanked by such creatures are a known royal form, and that reading makes the phrase about God sitting above the cherubim concrete: the creatures are the sides of a seat with no visible occupant.

Eichler argues against that for the sanctuary statues specifically. His reading of the ark and temple cherubim is that they are upright winged humans, and the argument runs off measurements rather than iconographic parallels: the height of Solomon’s pair, the span of their wings, the way the wings reach over and shelter the cover. A crouching sphinx is a difficult shape to stretch into those proportions. Whether that argument carries the day is exactly the live question, and presenting it as settled in either direction would be dishonest about where the discussion stands.

The ancient parallels illuminate without proving. Lamassu colossi, sphinx thrones, and karibu figures show that Israel’s neighbors populated sacred thresholds with composite creatures, and that context makes the biblical language less alien. It does not establish that Israel copied one particular foreign creature, that every biblical cherub shared a single look, or that a statue in a Judean sanctuary matched an Assyrian gate figure. Similar function across a region is not the same as identical form.

So where does that leave a reader who wants an answer? With a boundary, which is the thing the texts actually agree on. I lean toward composite form for Ezekiel’s creatures, because the vision says so in detail, and I think the ark statues are genuinely undetermined. What would move me is a securely provenanced image from an Israelite or Judean sanctuary context that can be tied to the Hebrew term rather than matched to it by resemblance. Until something like that turns up, confidence about cherub anatomy is a preference wearing the clothes of a conclusion.

Which is why the New Testament’s one explicit mention is so restrained. Hebrews 9:5 places the cherubim of glory over the atonement lid and then declines to go into detail. The writer had the same opportunity every artist since has taken, and passed.

Go back and read Genesis 3:24 one more time with the blank left blank. The chapter does not tell you what stood east of the garden. It tells you that the way was guarded, and the same concern about controlled approach reappears in the instructions for a gold lid and in a title about a throne. Whatever a cherub looked like, the text is consistent about what their presence means: you have arrived at a place where access is controlled and someone else decides the terms. The question the Bible keeps answering is not what they were. It is where you are standing when you see them.

← Back to the feed