Seventeen Words About a Courier

• Anchor Bible Dictionary — “Colossians, Epistle to the,” “Colossae,” “Laodicea,” “Hierapolis,” “Elements, Elemental Spirits,” “Household Codes,” “Tychicus,” “Onesimus.”
• Dictionary of Paul and His Letters (IVP) — “Colossians, Letter to the,” “Elements / Elemental Spirits of the World,” “Households and Household Codes,” “Principalities and Powers,” “Fullness.”
• NIDNTTE — στοιχεῖον, πλήρωμα, θρησκεία, ταπεινοφροσύνη, χειρόγραφον, ἐμβατεύω.
• Dictionary of NT Use of the OT (Beale & Carson) — on Colossians, which is a short entry for a measurable reason: the letter contains no formal Old Testament citation at all.

Colossians is chapter-for-chapter and verse-for-verse identical across the editions checked here.

Derived by counting verse units in the MorphGNT file for Colossians against the chapter shape of the English verse-count files held on this shelf. There is no offset, and no verse on this page carries a second number.

Why a New Testament page still says this. The rule was written for the Psalms and the Twelve, where the Greek runs a chapter behind or a verse ahead and a silent page misleads. On a Greek letter it costs one paragraph and buys the reader the knowledge that somebody checked. The two places where a number on this page turns on a manuscript decision rather than on the count are 3:6 and 4:8, and both are text-critical rather than numerical — flagged in the codex and again in stage 5.

There is no source for it outside the letter’s own rebuttal. Every reconstruction — Jewish mysticism, a mystery cult, Middle Platonism, Phrygian folk religion, an early gnosis — is built by reading chapter 2 backwards. That is a legitimate scholarly move and it is not a measurement. This page prints the four warnings and the words in them, and leaves the label to the reader.

The one thing that can be said without inference is negative: the letter never uses αἵρεσις, never says false teacher, and never names an opponent. What it names are practices.

Colossae is named once in the Greek New Testament (1:2). Hierapolis is named once (4:13). Laodicea is named seven times, and five of the seven are in this letter.

Five mentions of Laodicea across four verses of a 95-verse letter, against a single mention of Colossae itself. This is written to a valley, not to a city, and it expects to travel.

What was not consulted. The excavation reports for Laodicea and Hierapolis, the inscriptional evidence for Jewish settlement in Phrygia, Cicero’s Asian correspondence, and Strabo on the Lycus earthquakes are all standard background here. No inscription, no papyrus, no excavation report and no classical author was consulted for this page. The four verses above are the evidence actually in hand.

Festival, new moon, Sabbath is a formula, and it is an Old Testament formula. Nine verses in the 64 Greek Old Testament files held here carry all three terms together:

Against every one of the nine, Colossians 2:16 measures a run of 1 — on this page’s Greek-side convention that means it shares single words and no two consecutive ones. The Greek Old Testament writes the three as a list of nouns in agreement; Colossians strings them with ἤ … ἤ … ἤ after ἐν μέρει. The idea is the calendar; the wording is not any of these verses.

⚠️ The gap that stood here, and what filled it. Until 30 August 2026 this paragraph read: Exodus, Leviticus, Numbers and Joshua are not held on this shelf, so Leviticus 23 and Numbers 28–29, the two calendar chapters a reader would reach for first, could not be screened at all. All four were fetched that day, with sixteen more CATSS files, and the shelf went from 44 files and 23,061 verse units to 64 files and 30,637. The two calendar chapters have now been screened, and here is what they give:

Every one of those twos is a conjunction and a preposition. The calendar chapters were the whole reason for the caveat, and they change nothing. Leviticus 23 — the chapter every reader reaches for — turns out not to contain the Greek word for new moon anywhere in it: it has festivals and it has sabbaths and no third member. The word arrives in the Pentateuch at Exodus 40:2 and 40:17 and at Numbers 10:10, 28:11 and 29:6. Not one of those five comes within two consecutive words of Colossians 2:16. The finding the caveat was hedging is confirmed, not overturned. Run python -m dr_tools coverage for the current list.

3:18–4:1 is the shortest of the New Testament household codes, and its proportions are lopsided.

Four verses out of nine, and 56 of the unit’s 116 words — just under half — go to one group. And 4:9 says one of the two men carrying this letter is Onesimus, who is one of you — the same name as the runaway in Philemon. The page reports the name and the count; it does not assert that the letter was shaped around him. The counts above are counts of normalised word tokens in the MorphGNT file — the same tokens the run screen counts — with apparatus sigla and punctuation excluded.

4:7–9 is a courier note, which in the ancient world was not an afterthought. A letter had no envelope, no postmark and no independent authentication; the person who handed it over was the authentication, and was expected to supply the news the letter did not contain. That is exactly what 4:7 says Tychicus is for: he will tell you all about my activities.

The same two verses are the deepest verbal contact this letter has with anything. Whatever else that means, it means the courier paragraph is the part two letters would most naturally share if they travelled together. The page states that as the obvious reading and does not claim to have established it.

What was not consulted. No papyrus letter, no study of ancient letter-carrying, no work on the amanuensis. The observation above is a commonplace of the handbooks and is offered as one.

Three of the eight are in one verse, 2:23 — which is also one of the two or three hardest sentences in the Pauline corpus to construe. Four more of the same kind sit further on: αἰσχρολογίαν (3:8), μομφήν (3:13), ἀθυμῶσιν (3:21) and παρηγορία (4:11). Four, not five: ἐρεθίζετε (3:21) looks like one of them and is not — the same verb stands at 2 Corinthians 9:2 as ἠρέθισεν, which a search on the printed stem does not reach. A letter whose central chapter argues in vocabulary the rest of the New Testament never uses is a letter a run screen will find almost nothing in — and that is exactly what happened.

Counted over 7,927 verses of the Greek New Testament as digitized by MorphGNT, on normalised word forms and stems. A form search is not a lemma search: a rarer form of a commoner verb would not be caught by it. Each of the eight above was checked on its stem, not only on the form printed.

The tradition vocabulary is on both sides of the argument, two verses apart.

The noun παράδοσις occurs thirteen times in the Greek New Testament. Eight of the thirteen are the Gospels’ tradition of the elders, which Jesus opposes; the rest are Pauline, and there the same word is used positively of what was handed on (1 Corinthians 11:2, 2 Thessalonians 2:15 and 3:6) and negatively here and at Galatians 1:14. The letter is not against handing things on. It is against handing on the wrong things.

Sixty-six rows, screened book by book. One book answers and the rest do not.

Seventeen against six. That is not a difference of degree. Three rows of Colossians 2–4 reach eight words or more into another book, and all three are Ephesians: 4:8 at seventeen, 3:6 and 4:7 at ten. No other book in the New Testament yields a single row above six.

Colossians 4:8 is seventeen words long. Ephesians 6:22 is seventeen words long. After normalisation they are the same seventeen words in the same order.

Taken as a single block, Colossians 4:7–8 and Ephesians 6:21–22 share nineteen consecutive words: the run crosses the verse division (… ἐν κυρίῳ, ὃν ἔπεμψα …) because Colossians ends 4:7 with the same two words Ephesians ends 6:21 with.

⚠️ All three numbers depend on editorial decisions, and here they are.

None of the three collapses to nothing, and the seventeen stays the deepest row on any reading. But a page that printed 17 / 10 / 10 without this table would be reporting the SBLGNT’s editorial judgement as if it were the manuscript tradition.

Every verse of the Greek New Testament was screened against every verse in a different book. Here is what stands above twelve.

Now drop the Gospels and Acts and run the same list again.

A five-word gap, and everything on the far side of it is the envelope. The deepest thing two New Testament letters share, once the openings and closings are set aside, is a sentence about a man being sent to tell them how the writer is. Nothing in either letter’s theology comes anywhere near it.

Four places where a reader expects a match and the measurement gives one or two.

The Isaiah row is the instructive one, because the relationship is obviously real. Greek Isaiah 29:13 ends διδάσκοντες ἐντάλματα ἀνθρώπων καὶ διδασκαλίας. Colossians 2:22 has τὰ ἐντάλματα καὶ διδασκαλίας τῶν ἀνθρώπων — the same three content words, reordered, with an article and a καί moved. A contiguous-run screen cannot see through a reordering, and here it does not. Matthew 15:9 and Mark 7:7 quote the same Isaiah clause and this row reaches one word against each of them. The connection is there; the instrument is blind to it; both facts are printed.

Why these are on the page. The seventeen-word result is only worth anything if the reader can see what the same instrument does with the passages that feel parallel. It gives them two. That is the calibration, and a page that reported only its best number would be hiding it.

The unit everyone sets beside Ephesians 5:22–6:9 — and its deepest row is six.

Eight rows, and the total of the deepest runs across all eight is thirty. The deepest single row in the whole household code is six; one verse in chapter 4 reaches seventeen. The two codes address the same six groups in the same order and say much the same thing — and they do it in different words. Whatever relation these letters stand in, it is not that one was copied out of the other paragraph by paragraph.

The result is the flattest on this shelf, and a third more corpus did not change it. Nothing reaches five.

⚠️ Two corrections in that table, and both are printed rather than swapped in silence. The first column used to read 39 at three-or-more and 9 at four. Recounted on the same 44 files it is 40 and 10: the row that was missing from the four-count is Colossians 3:23, which the page had already discussed under the household code and had left out of this census. The second is the corpus itself — 44 files to 64 on 30 August 2026. Three rows moved up from two to three (2:19, 3:7 and 4:6), and every one of the three found its new partner in a book that was not on the shelf before: Leviticus and Judith at 2:19, 3 Maccabees at 3:7, Judith again at 4:6. Judith and Maccabees are witnesses, not Scripture. Nothing reached five in either column.

And all ten of the fours are stock phrases, not quotations.

⚠️ Two tie bands in that table were corrected on 2026-09-24. Until then the table gave Colossians 2:8 a band of 1 (Sirach 35:22 alone) and Colossians 3:8 a band of 1 (1 Kingdoms 2:3 alone). Re-run on the 63 files screened, 2:8 also reaches four against Wisdom 1:14, on different words, and 3:8’s phrase also stands in Exodus 23:13 and Numbers 32:24 — all three partners arrived with the 30 August 2026 fetch. No row moved above four, and every one of the ten is still a stock phrase.

This is what a letter with no formal Old Testament citation looks like when you measure it. Colossians never writes as it is written, never names a prophet, and never introduces a quotation. The screen agrees with the commentaries — which is worth reporting precisely because it is the boring answer.

⚠️ And the scope of that negative — which is now wider than it was. This paragraph used to read: Twenty CATSS files are not on this shelf, among them Exodus, Leviticus, Numbers, Joshua, Wisdom and Baruch. Leviticus 23 and Numbers 28–29 are the calendar chapters behind 2:16 and they were not searched. Wisdom is regularly brought into the discussion of 1:15–20 and 2:3, and it was not searched either. The sentence ‘nothing reaches five’ is true of the 44 files held, and this page will not state it more broadly than that. All twenty were fetched on 30 August 2026. Every book that paragraph named has now been screened, and “nothing reaches five” is true of the 64 files and the 30,637 verse units — a claim a third larger than the one it replaces. Wisdom in particular was searched: Colossians 2:8 now ties with it (Wisdom 1:14, four words) and 2:23 finds its band there, and nothing in the letter reaches five against it. Wisdom, Judith, Tobit and the books of Maccabees are witnesses, not Scripture, and this page labels them so every time it names one. ⚠️ What is still out of scope: no Greek book beyond the CATSS set, no papyri, no inscriptions, and no versions.

Colossians and Philemon share almost their whole cast and almost none of their wording.

Nine of the eleven personal names in Philemon stand in Colossians — and the deepest verbal run between Colossians 2–4 and any verse of Philemon is three words. That pairing is the cleanest demonstration on this page of what the instrument does and does not see. Two documents can share a household, a courier and a prison and share no sentences at all. Which is also the warning to read back across the Ephesians result: seventeen shared words prove seventeen shared words, and the step from there to a claim about who wrote what is a step the measurement does not take.

Colossians is disputed, and this page has measured the easiest thing to measure rather than the thing actually at issue.

The scholarly case against Pauline authorship rests mainly on style — long compound sentences, a high rate of genitive chains, the piling of synonyms, an unusually large hapax count for the letter’s length — and on theology, chiefly a realised eschatology in 2:12 and 3:1 where the undisputed letters keep the resurrection future. None of that was computed here. No sentence-length distribution, no particle frequency, no genitive-chain count, no hapax-per-thousand-words figure. Those are the numbers the debate runs on and this page does not have them.

What it has is the Ephesians overlap, and that cuts both ways. A seventeen-word identity in a courier note is compatible with one writer sending two letters by one man, and equally compatible with a later writer copying a paragraph he had in front of him. Both hypotheses predict exactly this number. The measurement narrows nothing about authorship; it only says where the two letters touch.

And note the shape of where they touch. If a second writer were working from Ephesians, one would expect the theology to be closest and the incidentals to drift. The opposite is measured here: the doctrinal paragraphs of Colossians 2 reach two words against the whole New Testament, and the errand reaches seventeen. ⚠️ That is an observation, not an argument — the same pattern is what you would get if two genuine letters were dictated on the same day and handed to the same man.

No Greek critical apparatus. The SBLGNT is one edited text among several; its apparatus sigla were stripped before measuring, so every figure on this page is a figure about the SBLGNT reading and not about the manuscript tradition. The three variants that move numbers here are named in stage 4 with their consequences (4:8, 3:6, 4:7). Others were not weighed at all — among them 2:2 (the mystery of God, Christ, which has one of the longest variant lists in the New Testament), 2:7 (ἐν αὐτῇ / ἐν εὐχαριστίᾳ), 2:18 (ἃ ἑόρακεν against ἃ μὴ ἑόρακεν — a negative that reverses the sense of the clause), and 3:16 (τῷ θεῷ against τῷ κυρίῳ). The 2:18 variant in particular would change how that codex row reads, and this page prints only one side of it.

Twenty CATSS files were not on the shelf when this page was first published, and they are now. This paragraph used to name Exodus, Leviticus, Numbers, Joshua, Wisdom and Baruch as absent, and to say that two of those absences sat directly under this passage — Leviticus 23 and Numbers 28–29 for the calendar of 2:16, and Wisdom for the language of 2:3. All twenty were fetched on 30 August 2026, and every Old Testament negative on this page is now a negative on 64 files and 30,637 verse units. Three of this page’s statements moved as a result and each is printed with its history where it stands: the festival triad went from eight verses to nine (Judith 8:6), the four-count census was corrected from nine rows to ten, and the flat negative on χειρόγραφον was overturned outright — the word stands six times in Tobit, and Tobit uses it for a cancelled debt-note. Tobit and Judith are witnesses, not Scripture.

No papyri, no inscriptions, no patristic citations, no versions. Nothing from the Latin, Syriac or Coptic. Nothing from Qumran, and that is a real gap for 2:18, where the Qumran angelic-liturgy material is regularly brought into the discussion of worship of angels. Nothing from the mystery-cult inscriptions that carry ἐμβατεύω, which is the single piece of external evidence most often cited on that word.

No commentary was consulted for any measurement. Where this page says “every commentary prints these together” it is reporting a commonplace, not a survey, and should be read as one.

The chapter people come for is the chapter with nothing in it. Colossians 2:14 measures two with a tie band of 226. 2:16 measures two with a band of 28. 2:23 measures two with a band of 42. A reader arriving for the record of debt nailed to the cross will find that this page has almost nothing to add to it, and that is reported plainly rather than padded. The measurement was run; it came back empty; the empty result is the honest one.

The seventeen is the page’s strongest number and its easiest to over-read. It is a fact about two edited verses. It does not establish dependence in either direction, it does not establish common authorship, it does not establish separate authorship, and it does not establish that the two letters travelled together — that last is the natural reading and it is an inference, not a measurement. Direction of influence is not asserted anywhere on this page.

Three words in this passage are left open, and readers who want them closed will be disappointed. στοιχεῖα — elemental spirits or elementary principles; the word carries both and the letter does not disambiguate. θρησκείᾳ τῶν ἀγγέλων — worship offered to angels or worship performed by them; the genitive does not decide it. ἐμβατεύων — entering, or being initiated, or investigating; the ESV itself footnotes the uncertainty. On all three this page reports occurrence counts and stops.

And one whole verse of this passage cannot be studied at all. 4:16 names a letter from Laodicea that no one has. Whether it is Ephesians, Philemon, a lost letter or something else is not measurable from any corpus on this shelf, and the page declines to guess — even though the Ephesians result above is exactly the sort of number a guess would like to lean on.

• James D. G. Dunn, The Epistles to the Colossians and to Philemon (NIGTC).
• Douglas J. Moo, The Letters to the Colossians and to Philemon (PNTC).
• Peter T. O’Brien, Colossians, Philemon (WBC 44).
• Eduard Lohse, Colossians and Philemon (Hermeneia) — the standard statement of the case against authenticity.
• Clinton E. Arnold, The Colossian Syncretism (WUNT 2/77) — on 2:18 and the inscriptional evidence for ἐμβατεύω.
• Morna D. Hooker, “Were There False Teachers in Colossae?” — the classic argument that the reconstruction outruns the evidence.
• None of these were consulted for the measurements on this page, which were run on the text files named in the footer.

Set the Gospels aside, where whole paragraphs repeat. Set the addresses and the closing graces aside, which every letter carries. Then ask what is the longest stretch of Greek that two different New Testament letters have word for word in common. The answer is seventeen words, it has a tie band of one, and this is it:

“I have sent him to you for this very purpose, that you may know how we are and that he may encourage your hearts”Colossians 4:8 = Ephesians 6:22 · seventeen consecutive Greek words, tie band 1

Not the fullness of deity. Not the record of debt. A note about a man carrying the mail. The next deepest thing two letters share is twelve words, and it is grace to you and peace from God our Father — with a tie band of five, because six letters open that way.

There is a reading of that which is worth sitting with. The parts of these letters that are most exactly repeated are the parts about people: who is coming, who sends greetings, who is in prison, who needs to be told how the writer is. The doctrine is written fresh each time. The care is written the same way twice. ⚠️ That is a reflection and not a finding, and the page marks it as one.

1. Read Colossians 2 counting the warnings. 2:4, 2:8, 2:16, 2:18 — four times, let no one … you. Then notice that the letter never says what the thing is called. Ask why a writer would argue that hard against something he never names.

2. Read Colossians 4:7–9 and Ephesians 6:21–22 side by side in English. You will see the overlap without knowing any Greek — and you will also see the one thing Colossians adds: and fellow servant, plus a second man, Onesimus, who is one of you.

3. Look up Colossians 3:6 in your own Bible. If it is short, your translation has followed the manuscripts that omit upon the sons of disobedience. If it is long, it has followed the ones that keep it. Five of the ten words this page measures there are inside that decision, and your footnote is telling you so.

Whole-passage study, 66 verses screened, chapters 2, 3 and 4 all represented in the codex above. Greek New Testament and ESV. Colossians 1:15–20 has its own study on this shelf.