Said Twice

• Anchor Bible Dictionary — “Proverbs, Book of,” “Wisdom Literature,” “Hezekiah.”
• Dictionary of the OT: Wisdom, Poetry & Writings (IVP) — “Proverbs, Book of,” “Sayings Collections,” “Septuagint.”
• NIDOTTE — תּוֹעֵבָה, תּוֹרָה, עָתַק.
• Commentaries consulted for orientation — Fox, Proverbs 10–31 (Anchor Yale); Waltke, Proverbs 15–31 (NICOT); Clifford; Longman.

Every handbook says Proverbs repeats itself. What none of them prints is how deep any particular repeat runs, or how many verses in a given stretch have no repeat at all. Those are countable, and they are what this page counts.

Counted over the whole book, 302 of the 915 Masoretic verses of Proverbs — 33.0% — share no two consecutive words with any other verse in the Hebrew Bible once their own chapter is set aside. In 27:1–29:27 the figure is 23 of 82, 28.0%. This passage is slightly more connected than the book it sits in, not less.

Zero here means one specific thing. On the Hebrew screen single words are not counted, so a verse reported at zero shares no two consecutive words with any other verse. It may share — and generally does share — a great many individual words. This is the convention the Obadiah page on this shelf used, and it is not the convention used on the Greek side.

The CATSS Greek Proverbs holds 939 rows. Forty-eight of them carry a letter after the verse number and answer no Masoretic verse. Six of the forty-eight fall inside the material answering Masoretic 25–29, and three of those six are in these three chapters: 34:20a, 34:21a and 35:17a.

Set against the Masoretic side, the same stretch has the opposite problem in the earlier chapters:

Read at the labels the file actually uses, nothing is missing here. The three extra rows are the whole of the difference in count — and one row, 34:27, differs in content without differing in count.

Proverbs 27:1 in Greek ends οὐ γὰρ γινώσκεις τί τέξεται ἡ ἐπιοῦσα. Sirach 8:18 ends οὐ γὰρ γινώσκεις τί τέξεται.

Five consecutive words, and across the 30,348 Greek Old Testament verse units screened here the tie band is one: no other row comes as close to Proverbs 34:1 as Sirach does. Both files sit on the same disk. The page reports the overlap and says nothing about which way it runs, or when, or whether both are drawing on a proverb older than either.

What this is not. It is not evidence that Sirach used Proverbs, and it is not evidence that a Greek translator of Proverbs used Sirach. A shared five-word idiom in two Greek wisdom books is exactly the kind of contact that can arise from either direction or from neither.

παίδευε υἱόν καὶ ἀγαπήσει σε καὶ δώσει κόσμον τῇ σῇ ψυχῇ · οὐ μὴ ὑπακούσῃς ἔθνει παρανόμῳ

Every distinctive phrase in the row stands somewhere else in this same Greek book. Its long clause is the row at 36:17; its closing pair is the close of the first line of 36:18, the very next row after that one. So the unnumbered row stands twenty-eight rows before the two rows it corresponds to, and it corresponds to them in sequence.

What the page will not say. It will not say the row is a doublet of 36:17–18, nor that it was moved, nor that a scribe wrote it twice. Those are claims about direction and about process, and nothing measured here can decide them. What is measured is that the language is shared and where.

The noun βδέλυγμα stands in eleven rows of the Greek Proverbs: 11:1, 11:20, 12:22, 15:8, 15:9, 15:26, 16:12, 20:23, 21:27, 34:20a and 36:27.

Ten of the eleven sit at a number whose Masoretic verse reads תּוֹעֵבָה. The eleventh is 34:20a, which has no Masoretic verse at all.

On the Hebrew side תּוֹעֵבָה stands 21 times in Proverbs, three of them in this passage: once at 28:9 and twice at 29:27, the last verse of the collection. The Greek row at 28:9 answers it with the verb ἐβδέλυκται rather than the noun, which is why 35:9 is not in the list of eleven.

One row of that table is deeper than it measures. Proverbs 27:12 and 22:3 are the same seven-word proverb. They differ in three prefixed conjunctions and one spelling — פְּתָאיִם here against וּפְתָיִים there — and the ESV prints the two verses word for word alike. The screen counts FORMS, so a prefixed waw cuts the run and the pair measures three. It is a whole-verse doublet the measurement cannot see, and the three named above are the three that reach five or more.

Twenty-four Masoretic verse numbers in Proverbs have no Greek row anywhere in the file. Screened against the rest of the Hebrew Bible with their own chapter set aside, exactly two of those twenty-four reach four or more: 19:1 at five and 20:16 at seven. Both are twins of verses in this passage. The other twenty-two reach three, two or zero.

Two of twenty-four is a small number and the page treats it as one. The absence of these rows was not selected for by the twin relation, and the co-occurrence is reported as a co-occurrence. The obvious story — that someone rendering the book passed over a saying he had already rendered — is a claim about process and about direction, and nothing measured here can test it. It is not asserted.

Proverbs 35:9 (Masoretic 28:9) and Isaiah 59:1 share six consecutive words, tie band one.

τὸ οὖς αὐτοῦ τοῦ μὴ εἰσακοῦσαι

Across all 85 rows of this passage screened against the 30,348 Greek Old Testament verse units held here, that is the deepest run there is. In Isaiah it is God’s ear and it is not too heavy to hear; here it is the man’s ear and he turns it away from the law. Same six words, opposite subjects. Nothing about direction follows.

The full distribution of the 85 rows against the screened Greek Old Testament, self excluded — exact counts, not cumulative: 4 rows top out at one, 31 at two, 38 at three, 7 at four, 4 at five, and 1 at six. (The New Testament figures in the next panel are cumulative, in the shelf’s usual wording.)

Across the 85 CATSS rows against 7,927 verses of the Greek New Testament: 73 reach two, 20 reach three, and nothing reaches four.

For scale, the previous page on this shelf measured Proverbs 25:22 against Romans 12:20 at ten, and Proverbs 3:12 against Hebrews 12:6 at eleven. In these three chapters the ceiling is three, and the deepest threes are formulas: τὸν οἶκον τοῦ at 34:10 with a band of 14, εἰς τὴν ζωήν at 34:27 with a band of 12.

The page prints its own counter-examples.

These zeros are the other convention. A named pair comparison has no unigram fallback and no tie band; 0 there means the two verses share no word form at all after normalisation. It does not mean the thought is unrelated. Luke 14:11 and Proverbs 29:23 say something very close in Greek that has no word in common, which is precisely what a run screen is built to be unable to see.

No Hebrew or Greek critical apparatus was consulted. No Qumran or Cairo Geniza fragment of Proverbs, no Syriac, no Vulgate, no Aquila, Symmachus or Theodotion, no Sahidic. Nothing bearing on whether the three lettered rows stand in every Greek witness or only in the tradition Rahlfs printed — which is a question about manuscripts, and the shelf holds no manuscripts.

That last gap matters more here than usual. The central finding of this page is about a row that exists in one edited Greek text. If 35:17a is weakly attested, the finding shrinks to a fact about Rahlfs. The page cannot tell you either way, and it should not be read as though it could.

1. That the Greek file ‘drops’ Proverbs 19:1 and 20:16 because they were already rendered at 28:6 and 27:13. That is the natural story and it may well be right. It is also a claim about the direction of work and about a person’s intention, and nothing on this disk can test it. What is measurable is the co-occurrence: two of the twenty-four absent numbers are the two most deeply repeated of the twenty-four, and both of their partners are in this passage. That is what is printed.

2. That 34:27 ‘replaces’ Masoretic 27:27. Withdrawn as worded. What is measured is that the row at that number carries other material and that no row in the file carries the goats’-milk verse. Whether something was removed, or never rendered, or stood somewhere the file does not preserve, is not decidable here.

3. That the bracket at 28:12 and 28:28 is an inclusio. Kept, but demoted. The two verses share two words, one of which carries a conjunction in one of them, and the run screen scores both at zero against the entire Hebrew Bible. It is a reading supported by a lexical repetition, not a measured structure, and the structure diagram labels the whole passage a sequence for that reason.

The content alignment of the eighty-two rows was checked by eye, not by lexicon. Each Greek row was read beside the English of the Masoretic verse of the same number, and one row — 34:27 — was judged to carry other material. Several others render very freely: 34:16 turns 27:16 into a saying about the north wind, 34:19 negates 27:19 (as faces are not like faces), 36:25 renders 29:25 twice inside one row. Those were judged to be free renderings of the same verse rather than different material, and a different reader could draw the line elsewhere.

The Hebrew figures are form counts. No Hebrew lemmatiser is held here, so ‘shares no two consecutive words’ means no two consecutive forms. The deepest twin in the passage illustrates the cost precisely: 27:13 and 20:16 measure seven on the qere and five on the ketiv, because 20:16 has ketiv נכרים against qere נׇכְרִיָּה. The shelf reads the qere, so seven is the published figure and five is printed beside it. The ESV itself splits at exactly that point — “an adulteress” at 27:13, “foreigners” at 20:16 — so the English columns of the two verses do not look like twins even though the consonants very nearly are.

And the corpus is not the Septuagint. It is the 64 CATSS files held here. Run screens see 30,348 of the 30,637 verse units on disk, because the Odes are excluded: Ode 9 is Luke 1 and Ode 13 is Luke 2 copied into a Greek Old Testament manuscript, and ten more Odes are second copies of passages already in the corpus, which would double a tie band for no textual reason. Every ‘band of one’ on this page is a band of one across those 30,348.

25:1 says these sayings were copied. Three chapters later the counting agrees with the heading in the plainest possible way: the material repeats, and the two files do not repeat the same things.

The Hebrew says one proverb twice at 20:16 and 27:13, seven words apiece. The Greek does not have 20:16 at all, and says a different proverb twice instead, once with a number and once without. Neither file is tidier than the other. Both are collections that grew.

And the most quoted line in the passage is the loneliest one in it. “Iron sharpens iron” shares no two consecutive words with any other verse in the Hebrew Bible. It has been repeated more than any other verse here, and the book itself never repeats it once.

If you have read this far, tell us what you would count next. The screens are exhaustive and deterministic, and a question you can state precisely is a question this shelf can answer.