CeleryKills

Homoioteleuton, Homoioarcton, and the Fragility of Copying

And now for something completely different, a gift for the real word nerds. There’s even almost a rhyme there. I came across this when talking about how much the Bible and other ancient text has lost over time, and the constant insistence that the Bible has not “virtually” changed since it’s first writing. My friend mentioned he had heard the proper words for the idea I was describing, that someone like Dan McClellan had used word he couldn’t pronounce and that were obviously not easy for Dan either. This got me intrigued.

Homoioteleuton (ὁμοιοτέλευτον) and homoioarcton (ὁμοιοαρχτον), as in HOY-moy-oh-TEH-lyoo-ton and HOY-moy-oh-ARK-ton, are a tongue twisting set of Greek words which sound like terms you’d hear in a seminar room with bad lighting and too many photocopies, but they’re simpler than they look.

I’ve handled enough messy transcripts, luckily nothing in Greek or Hebrew, where Latin and Old English were bad enough, to know they show up in places where eyes and memory start to blur. The first, homoioteleuton, is the slip that happens when lines end the same way and the copyist’s gaze jumps from one ending to another, skipping everything in between. The second, homoioarcton, is its mirror: the same kind of drift, but triggered at the beginning of words or lines that look alike. I tend to think of them as optical illusions in manuscript form, not mistakes of ignorance but of attention. The page itself becomes a kind of trap.

Homoioteleuton is the classic “same ending” error. In Greek manuscripts, when two successive lines or clauses terminate with similar syllables or forms, the eye returns to the second instance and the intervening text evaporates. It’s a failure of retention, not intention. Bruce Metzger treats it as one of the most common scribal phenomena in the transmission of biblical texts, particularly where repetitive phrasing is structural rather than stylistic (Metzger, The Text of the New Testament, 1992). I see it as a rhythm problem. The copyist gets into a cadence and the cadence collapses on itself.

A clear instance of homoioarcton, and one that feels almost uncomfortably mundane in its mechanics, appears in John 17:15. The longer reading preserves the full petition,

“οὐκ ἐρωτῶ ἵνα ἄρῃς αὐτοὺς ἐκ τοῦ κόσμου ἀλλʼ ἵνα τηρήσῃς αὐτοὺς ἐκ τοῦ πονηροῦ,”

rendered in English as, “I do not ask that you take them out of the world, but that you keep them from the evil one.”

In certain manuscript traditions, however, the phrase

“ἵνα ἄρῃς αὐτοὺς ἐκ τοῦ κόσμου”

is absent, leaving the compressed form,

“οὐκ ἐρωτῶ ἵνα τηρήσῃς αὐτοὺς ἐκ τοῦ πονηροῦ,”

translated simply as, “I do not ask that you keep them from the evil one.”

The initial sequence “ἵνα” appears twice in close proximity, and the eye slips from the first occurrence to the second, erasing the intervening clause. I think of this as a kind of textual narrowing, where the sentence loses its balance and becomes almost too efficient. The meaning changes course, not through reinterpretation but through omission, as if the text itself had blinked and resumed mid-thought. That kind of loss feels oddly familiar, like when I reread a draft and realize a whole line has vanished because two phrases started the same way and my attention never registered the transition.

Homoioarcton, by contrast, is about beginnings. Words or lines that share initial sequences invite the same kind of leap, but forward rather than backward. It’s less remarked on, though no less present in the textual record. When a scribe’s eye lands on a familiar starting pattern, the mind anticipates the continuation and skips the intervening stretch. Daniel Wallace notes that both phenomena can distort whole phrases without leaving obvious traces, especially in continuous script traditions where word boundaries are porous (Wallace, Textual Criticism, 2011). I find it unnerving how something so small reshapes meaning so decisively.

A concrete instance of homoioarcton appears in Philippians 2:6, where the longer reading preserves the full phrase

“ὃς ἐν μορφῇ θεοῦ ὑπάρχων,”

translated as “who, though he was in the form of God, existed.”

In some manuscript traditions, the clause is shortened, effectively dropping the intervening material that follows the initial “ὃς ἐν,” yielding a compressed reading that begins again at a later, similar syntactic opening.

The shorter form reads, in English, as something like “who… existed in the form of God,” with the intervening relational structure absent.

The repetition of the initial sequence “ὃς ἐν” creates a visual and cognitive bridge the scribe crosses too quickly, landing on the second occurrence and leaving the first half of the clause behind. I keep circling back to how ordinary that looks on the page, how nothing announces itself as wrong, and yet the sentence has quietly lost its scaffolding. It is the kind of error that feels almost like a skipped heartbeat, where the text continues but something essential about its cadence is gone, replaced by a cleaner but thinner line of thought.

The irony, and I do not use that lightly, is how closely this resembles mutation in genetics. A copyist’s error propagates through transmission the way a nucleotide mispairing propagates through replication. Not identical mechanisms, but a parallel logic: repetition invites slippage, and slippage becomes inheritance. The manuscript tradition behaves like a living system, accumulating variations that are neither purely random nor fully controlled. It does not take much; just a repeated ending, a similar opening… and the structure shifts.

In the Hebrew Bible, homoioteleuton shows up in places where formulaic endings are common. The Psalms are especially vulnerable. Consider Psalm 145, an acrostic. The Masoretic Text lacks a line for the letter nun, a gap widely attributed to scribal omission. Some scholars link such gaps to homoioteleuton, where repetition of similar endings in adjacent verses could have caused accidental skipping (Tov, Textual Criticism of the Hebrew Bible, 2012). I’ve stared at that psalm and felt the absence like a missing step in a staircase.

A more precise Old Testament example sits inside the textual instability of Psalm 40:6–8, especially when set against its quotation in Hebrews 10:5–7. The Masoretic Text reads, in Hebrew,

“אָזְנַיִם כָּרִיתָ לִּי” (’oznayyim karita li),

literally, “ears you have dug for me.”

The Septuagint, however, reflects a different underlying reading:

“σῶμα δὲ κατηρτίσω μοι,”

translated, “but a body you have prepared for me.”

The divergence is not merely interpretive; it points to a textual alteration in which the Hebrew phraseology may have been simplified or reshaped through a loss of intervening material. In one hypothetical reconstruction, homoioteleuton explains how the longer form could collapse:

a scribe’s eye moves from a verbal ending like “-ti” in “כָּרִיתָ” to a later, similar ending, dropping what lies between.

The step from “אָזְנַיִם כָּרִיתָ לִּי” to a shortened or altered form is not visible in a single manuscript, but the effect accumulates across transmission layers. I find myself uneasy here, because the Hebrew is concise enough already, and still it is not immune to being pared down into something else entirely.

The pattern reappears in the broader Psalter tradition, where repeated syntactic endings make the text especially vulnerable. In Psalm 100:3, the phrase

“הוּא עָשָׂנוּ וְלוֹ אֲנַחְנוּ”

“hu ʿasanu velo anakhnu”,

“it is he who made us, and we are his,”

sits within a structure that echoes across nearby lines. A plausible homoioteleuton scenario emerges when a scribe’s eye skips from one instance of the pronominal suffix “-נוּ” (-nu, “us”) to another, omitting the intervening clause in some transmitted witnesses. The result is a truncated reading that loses the relational balance of the verse, collapsing agency and possession into a thinner statement. I keep noticing how these endings, small as they are, act like magnets for the eye. They draw attention forward and away from what should remain anchored in place, and the text pays for that movement by losing part of its own continuity.

The New Testament offers more explicit cases. In 1 John 5:7–8, the so-called Comma Johanneum controversy involves textual expansion and omission across manuscript families. While the debate is complex, homoioteleuton is invoked as one plausible mechanism for the shorter reading: similar phrases about “the Spirit” and “the water” could have led to the accidental dropping of intervening material (Metzger, Textual Commentary, 1994). The result is not just a variant reading, but a doctrinal fault line, which is its own kind of irony.

Homoioarcton appears with less fanfare but similar consequences. In Mark 10:47, some manuscripts omit “Son of David” following “Jesus,” and one explanation is that the repeated initial sound patterns in nearby phrases created conditions for omission. The copyist’s eye lands on a familiar starting cue and the intervening phrase collapses. I’ve seen this in modern transcription too, when a repeated clause at the start of successive sentences causes a missing line in a draft. The ancient and the contemporary are not as far apart as we like to imagine.

Translation adds another layer of distortion. A translator working from a manuscript already affected by homoioteleuton or homoioarcton inherits the gap and often smooths it over unconsciously. Jerome’s Latin Vulgate, for example, reflects variants where shorter Greek readings may have arisen from such omissions, and later translators treat those readings as text rather than accident (Smalley, The Study of the Bible, 1983). The error migrates across languages, disguised as choice. I find myself asking whether a translation ever fully escapes the conditions of its source, or whether it simply inherits the same optical traps in a different format.

One Greek example that shows how a small scribal omission ripples into translation is Romans 8:1. The longer Greek reading preserves the clause

“μὴ κατὰ σάρκα περιπατοῦσιν,”

 “mē kata sarka peripatousin”,

literally “not walking according to the flesh,”

while the shorter reading drops it, leaving

“Οὐδὲν ἄρα νῦν κατάκριμα,”

“Ouden ara nyn katakrima”,

“there is therefore now no condemnation.”

Jerome’s Latin Vulgate follows the shorter form, “Nihil ergo nunc damnationis est,” “there is therefore now no condemnation,” as if the conditional frame never existed. I keep noticing how translation does not repair the gap; it simply walks across it.

Another Greek case sits in Matthew 6:13, where some manuscripts omit

“ὅτι σου ἐστιν ἡ βασιλεία,”

“hoti sou estin hē basileia”,

“for yours is the kingdom.”

The shorter Greek form produces the Latin

“ne nos inducas in tentationem,”

“do not lead us into temptation”,

without the doxological closure. The ending disappears at the level of the source text, and the translation inherits that absence as if it were structurally intended. There is a quiet flattening here, not dramatic, just a clause that no longer returns.

The Hebrew tradition shows the same mechanism within the Masoretic Text of Psalm 23:1. The verse reads

“יְהוָה רֹעִי לֹא אֶחְסָר,”

“YHWH ro‘i lo ’eḥsar”,

literally “the Lord is my shepherd, I shall not lack.”

A hypothetical homoioteleuton scenario within the repetition of similar verbal endings could collapse the line into the shorter

“יְהוָה רֹעִי,”

“YHWH ro‘I”,

“the Lord is my shepherd.”

A translator working from that reduced form would carry forward only the opening assertion. I find myself uneasy with how easily the outcome vanishes, leaving the sentence grammatically intact but conceptually incomplete.

In the Septuagint tradition, Psalm 23:1 resists that collapse. It reads

“Κύριος ποιμαίνει με, καὶ οὐδέν με ὑστερήσει,”

“Kyrios poimainei me, kai ouden me hystērēsei”,

“the Lord shepherds me, and I shall lack nothing.”

The fuller Greek preserves the balanced movement between statement and consequence, a structure that feels almost resistant to erosion. I keep comparing the two versions and noticing how the presence or absence of a single clause changes the entire weight of the line.

The Dead Sea Scrolls offer a different kind of evidence in Psalm 145, where 11QPsᵃ preserves a fuller acrostic sequence that includes the verse for nun missing in the Masoretic Text. The reconstructed line,

“נֶאֱמָן יְהוָה בְּכָל־דְּבָרָיו,”

“ne’eman YHWH be’khol-devarav”,

“faithful is the Lord in all his words,”

restores continuity where the later tradition shows a gap. The Masoretic omission is often read as a probable case of homoioteleuton, where similar endings in adjacent lines draw the eye forward and erase what sits between. The contrast between witnesses feels less like disagreement and more like a record of how fragile repetition can be.

The phenomenon does not belong exclusively to biblical transmission, even if that is where it is most studied. Classical literature carries the same traces, often in places where repetition is structural rather than accidental. In Homer’s Iliad 2.15, a line describing Zeus sending a deceptive dream contains a repeated verbal frame that, in some manuscript traditions, appears contracted, likely through homoioteleuton:

“ἔκ τε Διὸς προΐησιν κακὸν ὄνειρον,”

“ek te Dios proiēsin kakon oneiron”,

 “and from Zeus he sends a baneful dream,”

while a shorter transmitted form drops the intervening participial structure and reads only

“ἔκ τε Διὸς ὄνειρον,”

“ek te Dios oneiron”,

“and from Zeus a dream.”

The loss is subtle, but the causal chain is gone. I keep thinking about how easily narrative momentum can erase its own connective tissue.

A similar pattern appears in Plato’s Phaedrus 245c, where a longer philosophical clause about the soul’s motion is preserved in fuller manuscripts:

“ἡ ψυχὴ πᾶσα ἀθάνατος,”

“hē psychē pasa athanatos”,

“every soul is immortal,”

followed by a detailed explanatory sequence. In abbreviated transmission, the repeated syntactic opening “ἡ ψυχὴ” invites a skip, and the explanatory material disappears, leaving a compressed statement. The Latin tradition, when it follows the shorter reading, renders it as “anima semper immortalis est,” “the soul is always immortal”, a sentence that feels complete but no longer carries Plato’s layered movement. I notice how philosophy, too, is not immune to this kind of thinning, as if abstraction made it more vulnerable, not less.

In Latin legal texts, the Digest of Justinian shows homoioarcton in action where repeated openings of clauses begin with similar legal formulas. In one passage on obligations, the fuller reading preserves two parallel clauses beginning with “si quis,”  “if anyone,” but some copies omit the intervening condition, collapsing the structure into a single provision. The shortened form reads as if only one condition existed, whereas the longer version clearly distinguishes two. The error is not cosmetic; it alters the architecture of the argument. I keep returning to how law, of all things, depends on precise sequencing, and still it can be undone by a repeated opening.

Scribal transmission in the Epic of Gilgamesh offers another case, particularly in the Akkadian tablets where formulaic endings are frequent. In Tablet XI, a repeated closing phrase like

“ina šamê u erṣeti,”

“in heaven and earth,”

appears in adjacent lines, and in some witnesses the intervening narrative is absent, likely due to homoioteleuton. The shorter reading flattens the sequence of actions, removing the transition between Gilgamesh’s questions and Utnapishtim’s response. It feels almost like a breath that was not recorded, only the inhale and exhale remaining.

Even in medieval Irish annals, where prose is dense and repetitive, similar errors surface. In the Annals of Ulster, entries beginning with the repeated formula “Mors…” mors, “death,” are sometimes transmitted with omissions that skip intervening names or events. A fuller entry might read

“Mors Aedha mic Neill…,”

“the death of Aed son of Niall,”

while a shortened version preserves only “Mors Aedha…,” losing the patronymic chain. The repetition of the initial word creates a visual rhythm that invites the eye to jump forward. I find it striking how even annalistic precision, which should resist this kind of drift, still carries the same vulnerability.

There’s a kind of quiet violence in these phenomena. They do not announce themselves. They erase by resemblance. In the long arc of biblical transmission, homoioteleuton and homoioarcton function like friction points, places where repetition becomes loss. The page is not neutral. It has contours that invite the eye to fail.

And yet, I am drawn to the fact that these errors are predictable. They depend on structure, on pattern. That predictability makes them almost elegant. They show how fragile continuity is, how easily a text can lose a part of itself without anyone noticing until centuries later. I’ve come to think of them not as flaws in the manuscript tradition but as evidence of how deeply human the process is. The copyist is not a machine. The eye drifts. The mind anticipates. The text mutates.

Maybe that’s the point. Meaning is not stable. It is replicated, and in replication it shifts. Homoioteleuton and homoioarcton just make the shift visible.

References

• Metzger, Bruce M. The Text of the New Testament. Oxford University Press, 1992
• Metzger, Bruce M. A Textual Commentary on the Greek New Testament. United Bible Societies, 1994.
• Tov, Emanuel. Textual Criticism of the Hebrew Bible. Fortress Press, 2012.
• Wallace, Daniel B. Textual Criticism. Dallas Theological Seminary, 2011.
• Smalley, Beryl. The Study of the Bible in the Middle Ages. University of Notre Dame Press, 1983.
• Homer. Iliad. Critical edition traditions (e.g., Oxford Classical Texts), c. 8th century BCE (textual transmission).
• Plato. Phaedrus. Critical edition traditions (e.g., Oxford Classical Texts), c. 4th century BCE (textual transmission).
• Justinian. Digest of Roman Law (Digest). c. 533 CE, transmitted in medieval manuscript traditions.
• Epic of Gilgamesh. Akkadian tablet tradition, c. 2nd millennium BCE (standard edition: Andrew R. George, 2003).
• Annals of Ulster. Medieval Irish chronicle tradition, c. 5th–16th century CE (standard edition: Seán Mac Airt and Gearóid Mac Niocaill, 1983).