Bildad’s Last Word Was Job’s

• Anchor Bible Dictionary — “Job, Book of,” “Retribution,” “Wisdom Literature,” “Hexapla,” “Septuagint.”
• Dictionary of the OT: Wisdom, Poetry & Writings (IVP) — “Job 4–27,” “Retribution,” “Speech Cycles,” “Septuagint of Job.”
• NIDOTTE / TDOT — צָדַק (ṣādaq, be in the right), זָכָה (zāḵāh, be clean), אֱנוֹשׁ (ʾənōš, mortal), חֵלֶק (ḥēleq, portion).
• On the Greek — NETS Iob (tr. Cox) and Ziegler’s Göttingen edition, for the shape of Old Greek Job and for which lines Origen supplied. Neither is on this disk; the Greek printed here is Rahlfs, which carries no Aristarchian signs — see Stage 5.

• David J. A. Clines, Job 21–37 (WBC) — on chapters 21–25 and the third-cycle problem.
• C. L. Seow, Job 1–21 (Illuminations) — for chapters 20 and 21.
• Carol A. Newsom, The Book of Job: A Contest of Moral Imaginations — on the dialogue as a contest of voices rather than a debate with a winner.

Bildad’s second line is prayed every day in synagogues. עֹשֶׂה שָׁלוֹם בִּמְרוֹמָיו — he makes peace in his high places (25:2) — is the phrase that closes the Kaddish and stands at the head of the hymn Oseh Shalom. The exact two-word sequence occurs twice in the Hebrew Bible, here and at Isaiah 45:7, where the ESV renders those same two words I make well-being and the sentence continues and create calamity. The liturgy took the half-verse; the book of Job gave it to the friend who is about to stop speaking. And it is not in the Greek column at all — Rahlfs reads ὁ ποιῶν τὴν σύμπασαν ἐν ὑψίστῳ, who makes the whole universe in the highest, with no word for peace.

Greek Job speaks a dialect of its own. The word this translator uses for a human being at the crucial moments is βροτός — an epic, Homeric word for mortal, the kind of vocabulary that belongs to hexameter poetry rather than to a Bible translation. Across the 30,348 Greek Old Testament verse units a run screen sees here, βροτός occurs seventeen times, and all seventeen are in Job. It occurs nowhere in the Greek New Testament. Whoever put Job into Greek was writing for readers who knew Greek poetry.

Chapter 23 is short one verse in the Greek and long one rendering. CATSS/Rahlfs Job 23 carries seventeen rows against seventeen Hebrew verses, but the numbers do not line up: there is no row numbered 23:14, and 23:15 appears twice — as 23:15 and as 23:15a — in two quite different renderings of the same Hebrew line (διὰ τοῦτο ἐπ᾽ αὐτῷ ἐσπούδακα and ἐπὶ τούτῳ ἀπὸ προσώπου αὐτοῦ κατασπουδασθῶ). That doubling is the classic signature of Hexaplaric supplementation — a later rendering set beside the older one — but Rahlfs carries no asterisks, so this page cannot say from files on this disk which row is Old Greek. It reports what the file shows and stops there.

• Job 9:2 → Job 25:4. Read them in that order once and the page’s whole argument is visible. Job: truly I know that it is so; how then can a man be in the right before God? Bildad, sixteen chapters later, verbatim.
• Deuteronomy 19:14; 27:17; Deuteronomy 24:12–13, 17 — the boundary and pledge laws behind Job’s catalogue in chapter 24. Treated at length on the 22:1–26:14 study.
• Isaiah 45:7 beside Job 25:2 — the only other עֹשֶׂה שָׁלוֹם in the Hebrew Bible, and the one that adds and create calamity.
• 11QtgJob (11Q10) (Qumran — material evidence, a witness and not Scripture) — the Aramaic Targum covers this stretch of the dialogue. It is not held on this disk and was not consulted; it is named so that a reader knows a pre-Christian non-Greek witness exists.
• Origen’s Hexapla (third century — a scholarly edition, not Scripture) — the framework within which the 23:15/23:15a doublet is normally explained.

• Bildad’s third speech is 44 Hebrew words — the shortest in the dialogue by more than a factor of two. Every figure in this line counts the whole chapter unit, the narrator’s ‘then X answered and said’ included; Bildad’s own words are 40 of the 44. The next shortest unit is Job 26 at 101. Across the three rounds, in Hebrew words: Eliphaz 355 → 262 → 211; Bildad 167 → 143 → 44; Zophar 149 → 209 → nothing.
• ⚠️ So the ‘collapse’ is not monotonic, and the page says so. Zophar’s second speech is 40% longer than his first, not shorter. Two speakers shrink steadily; the third gets louder and then disappears. A tidy curve would have been easier to write and would not have been true.
• 80% of Bildad’s 44 words are forms already used earlier in Job — only nine of the word-forms in chapter 25 are new to the book. Compare chapter 20 at 44%, chapter 22 at 58%, chapter 23 at 59%, chapter 24 at 47%. Measured against all 1,065 six-verse windows in Job, 25:1–6 ranks 23rd; twenty-one of the twenty-two windows above it lie in chapters 33–42, where the stock of used words is far larger, and the twenty-second is Job 25:3–26:2 at 80.0%, four of whose six verses are Bildad’s own. Nothing in chapters 1–24 reaches it — the highest is Job 2:1–6 at 72%.
• The deepest run in the passage is 5, at 25:4, band 1, against Job 9:2. Against the friends’ own earlier statements of the same doctrine it measures 1 (Job 4:17) and 2 (Job 15:14).
• 17 rows reach a run of 3+ with a tie band of 1. Fifteen of them point back into Job. The two exceptions are 21:5 (Micah 7:16) and 21:17 (Proverbs 24:20).
• 57 of the 141 verses share no two consecutive words with any other verse in the Hebrew Bible. Per chapter: 20 — 15 of 29; 21 — 15 of 34; 22 — 10 of 30; 23 — 6 of 17; 24 — 11 of 25; 25 — 0 of 6.
• ⚠️ One run the eye sees and the screen does not. At 21:16 Job writes עֲצַת and at 22:18 Eliphaz writes וַעֲצַת — the same noun with a conjunction. The screen therefore returns 3, not 4, because it compares whole consonantal forms and folds nothing on the Hebrew side. Runs are a floor, not a ceiling.
• Greek-to-Hebrew word ratio, 20:1–25:6: 1.59, against 1.62 for the whole book. No signal. These chapters are neither compressed nor expanded relative to Job as a whole in the printed Greek — and see Stage 5 for why that measurement cannot speak to the famous shortness of the Old Greek.

• Job 9:2 → Job 25:4. Five consecutive words, tie band 1, and the only two places in the Hebrew Bible where the sequence occurs. In Greek the same pair runs five as well (πῶς γὰρ ἔσται δίκαιος βροτός), so this is one of the few links here that survives translation intact.
• Job 20:29 → Job 27:13. Four consecutive words in Hebrew, band 1; seven in Greek. Zophar’s last sentence and a sentence of Job’s four chapters later. The overlap is measured; who is quoting whom is not, and this page does not guess.
• Job 21:14 → Job 22:17 and Job 21:16 → Job 22:18. Three words each, band 1 each. Eliphaz’s third speech reuses two clauses from Job’s preceding one. Both links vanish in Greek — the pair screens measure 0 and 2 there.
• Job 15:15 → Job 25:5. Three words, band 1: are not clean in his eyes. Eliphaz said it of the heavens; Bildad says it of the stars. One noun changed.
• Job 31:19 → Job 24:7. Four words, band 1: without clothing, and no covering. Job uses the phrase in chapter 24 of what the wicked do to the poor, and in chapter 31 in the oath where he swears he never did it.
• ⚠️ No New Testament quotation of Job 20–25. Screened mechanically. All 141 CATSS rows against all 7,927 verses of the Greek New Testament: 126 reach two, 46 reach three, four reach four, and nothing reaches five. The four fours are printed below so that they cannot be mistaken for citations.

• Job 20:7 ↔ Mark 6:49 — run 4, band 1. οἱ δὲ ἰδόντες αὐτόν. In Job, those who had seen the wicked man ask where is he?; in Mark, the disciples see Jesus on the water and think he is a ghost. A tie band of 1 means the run is unique in the corpus. It does not mean the run is meaningful.
• Job 21:10 ↔ Revelation 12:2 — run 4, band 1. ἐν γαστρὶ ἔχουσα καί. In Job it is a cow that does not miscarry; in Revelation it is the woman clothed with the sun, crying out in labour. Four identical words, band 1, and no relation whatever.
• Job 22:26 ↔ Matthew 14:19 / Mark 6:41 / Mark 7:34 / Luke 9:16 — run 4, band 4. ἀναβλέψας εἰς τὸν οὐρανόν. This one at least describes the same gesture — and the band of 4 is exactly why it should be read as a stock phrase for looking up to heaven rather than as contact.
• Job 23:2 ↔ Luke 6:6 — run 4, band 1. καὶ ἡ χεὶρ αὐτοῦ. Job: his hand is heavy on my groaning. Luke: the man in the synagogue whose right hand was withered. Nothing.
• Held against those four, the γεννητός link is the interesting case — and the screen cannot see it. Job 25:4 and Matthew 11:11 share a rare adjective and its noun, in different cases. Every run measurement between them is zero. This is the clearest example on the page of why a run figure is evidence of contact and never evidence of its absence.

• David J. A. Clines, Job 21–37 (WBC) — the standard treatment of the third-cycle problem, of chapter 24’s difficulties, and of the proposals to reassign 27:13–23 to Zophar. Not on this disk. Named, not consulted.
• C. L. Seow, Job 1–21 (Illuminations) — for Zophar’s speech and Job’s reply. Not consulted.
• Carol A. Newsom, The Book of Job — on the dialogue ending by exhaustion rather than by defeat. Not consulted.
• NETS Iob (tr. Cox) and Ziegler’s Göttingen Iob — the only sources that would let this page say which lines of the Greek are Origen’s. Neither is held here.

• It cannot separate Old Greek from Origen’s supplement. The Greek printed here is Rahlfs, via CATSS, and Rahlfs carries no Aristarchian signs. Old Greek Job is famously about a sixth shorter than the Hebrew; this page’s measured Greek-to-Hebrew ratio of 1.59 says nothing about that, because it is measuring the completed text. Where a neighbouring study gives asterisk counts, those come from Ziegler and NETS — not from files on this disk, and not from this page.
• It cannot say who quoted whom. Job 9:2 and Job 25:4 are identical for five words. That is a fact about the text. Whether the author gave Bildad Job’s line deliberately, whether both draw on a stock formula now lost, or whether the transmission produced it, is not decidable from anything measurable here, and the page never says ‘Bildad quotes Job’ as a claim about the author’s intent — only as a description of two verses.
• The 80% figure has a known bias and the page has tried to control for it, not remove it. Later text always reuses more, because more has been said; and short samples are noisier than long ones. The control was to rank 25:1–6 among all 1,065 six-verse windows in the book and to report that twenty-one of the twenty-two windows above it lie in chapters 33–42 — the twenty-second being Job 25:3–26:2, which is mostly Bildad’s own six verses read again. That is a real control and it is not a perfect one — a longer speech at the same point in the book might have scored lower simply by being longer.
• It did not check the third-cycle reconstructions. Many scholars reassign parts of chapters 24–27 to Bildad or Zophar, partly on the strength of the very 20:29 / 27:13 overlap this page measures. No Hebrew manuscript supports any of those rearrangements, and this page reads the received text as it stands. It does not claim that reading is right; it claims only that it is what was measured.
• The Qumran Aramaic Targum was not consulted. 11QtgJob covers this stretch and is one of the very few pre-Christian witnesses to Job outside the Greek. It is not held on this disk. Any conclusion here about the Hebrew of these chapters is a conclusion about the Masoretic text alone.
• The Greek Old Testament scope, stated once. The corpus on disk is 64 CATSS files holding 30,637 verse units, of which 30,348 are what a run or lexeme screen actually sees, the Odes being excluded as a liturgical anthology of material counted elsewhere (Odes 9 is Luke 1, Odes 13 is Luke 2, and ten more repeat Exodus 15, Deuteronomy 32, 1 Samuel 2, Habakkuk 3, Isaiah 26, Jonah 2, Daniel 3 twice, Isaiah 5 and Isaiah 38; the Prayer of Manasseh repeats nothing, the morning hymn only a few borrowed lines, and both go out with the rest). Every Greek count on this page is against the 30,348. Until 2026-09-24 this item began ‘One tooling infelicity’ and warned that the tools’ own scope line said “35 files”; the tools were corrected on 2026-09-20 and now print the 63 files screened of the 64 held. It also said the morning hymn repeats nothing: measured on 2026-09-24, eight of its 46 lines share eight or more consecutive words with Psalms 89, 118, 142 and 144 in the Greek numbering (English 90, 119, 143 and 145) or with Daniel 3.

• The argument did not lose. It ran out of sentences. Six verses, forty-four words, four fifths of them already used. Nobody in this book is refuted; the friends simply stop having anything to put in a sentence, and Scripture keeps the transcript at full length rather than tidying it into a victory.
• Bildad’s closing doctrine was Job’s opening complaint. How can a man be in the right before God? — Job asked it in chapter 9 as a man who has already lost, and Bildad returns it in chapter 25 as a rule. The same true sentence is a confession in one mouth and a weapon in the other. That is not a difference of theology. It is a difference of position.
• Quoting somebody accurately is not the same as hearing them. Eliphaz gets Job’s words right at 22:17–18 — three words each, exact — and draws the opposite conclusion from them. Accuracy of citation is no evidence at all of having listened.
• One concrete response. Before the next time you answer someone in trouble, ask what fraction of what you are about to say they have already heard from you. Bildad’s figure was four fifths. If yours is close, the honest thing may be the thing this book does next — stop talking.