The Temple Sermon and the Idols That Cannot Speak

The lexica give פָּרִיץ as a violent man, a bandit, a housebreaker — from the root פרץ, to break through. מְעָרָה is a cave. So: a bandits’ cave, the place robbers go home to, not the place they rob. That distinction is worth keeping, because the English den of thieves has often been preached as though the temple were where the stealing happened.

The Greek is where the measurement gets interesting. The two words σπήλαιον λῇστῶν stand in exactly one of the 30,348 Greek Old Testament units screened here — Jeremiah 7:11 — and in exactly three of the 7,927 verses of the Greek New Testament: Matthew 21:13, Mark 11:17 and Luke 19:46. Four verses in the whole Greek Bible. By comparison the noun σπήλαιον on its own occurs fifty-five times in the Greek Old Testament files — mostly the cave of Machpelah in Genesis and David’s hiding places in Samuel — and λῃστ- occurs in fifteen New Testament verses. It is the pairing that is unique, not either word.

מְלֶכֶת הַשָּׁמַיִם stands in five verses of the whole Hebrew Bible, always with the preposition — לִמְלֶכֶת הַשָּׁמַיִם — and all five are in Jeremiah: 7:18, 44:17, 44:18, 44:19 and 44:25. The bare consonantal form מלכת stands in thirteen other verses of the Masoretic corpus, and it is not one word in them but several: going at 1 Samuel 30:21 and four more places, you reigned at 2 Samuel 16:8, reigning at 2 Kings 11:3 and 2 Chronicles 22:12, and the queen of Sheba at 1 Kings 10:4 and 10:10 and 2 Chronicles 9:3 and 9:9. The screen reads consonants and not vowels, so those last four are the same four letters as the queen of heaven, pointed differently. Every reference work identifies her with a goddess — Ishtar and Astarte are the usual candidates — and every reference work also notes that the consonants can be pointed as מְלֶאכֶת, the work of heaven, meaning the stars.

The Greek column has taken that second road once and the first road four times, inside one book, and this page can print the count. At Greek Jeremiah 7:18 the phrase is τῇ στρατιᾷ τοῦ οὐρανοῦ, the host of heaven. At Greek Jeremiah 51:17, 51:18, 51:19 and 51:25 — which are Hebrew and ESV 44:17, 44:18, 44:19 and 44:25 — it is τῇ βασιλίσσῃ τοῦ οὐρανοῦ, the queen of heaven. Four to one. The page reports the split. It does not say which is the translator’s first thought and which his second, because the files cannot tell it.

הַתֹּפֶת is a place in גֵּיא בֶן־הִנֹּם, the Valley of the Son of Hinnom, on the south side of Jerusalem, where children were burned. The Hebrew name stands in five verses: 2 Kings 23:10, Jeremiah 7:31, 7:32, 19:6 and 19:13.

The Greek column does not render the name; it spells it out. Ταφεθ stands in three verses of the Greek files held here — 2 Kings 23:10, Jeremiah 7:31 and Jeremiah 7:32 (twice in that one verse). At Greek Jeremiah 19:6 the same Hebrew word has become διάπτωσις, a falling to pieces, and the valley itself is πολυάνδριον, a place of many men — a mass grave. One Hebrew word, three Greek treatments, in one book.

And Gehenna is not here. The word γέεννα stands in twelve verses of the Greek New Testament — Matthew 5:22, 5:29, 5:30, 10:28, 18:9, 23:15, 23:33; Mark 9:43, 9:45, 9:47; Luke 12:5; James 3:6 — and in not one of the 30,348 Greek Old Testament units screened here. The valley that gave the word its name is named twice in this passage, and it is named φάραγξ υἱοῦ Ἑννομ both times. The road from this valley to that word runs through literature this page does not hold and did not consult.

7:12 and 7:14 send the crowd to look at Shiloh, where the tent had stood and where the ark was lost. The argument is architectural: God let the first house go, so the presence of a building is not the presence of God.

Measured: Σηλωμ stands twenty-two times across twenty-one rows of the Greek Old Testament files held here (1 Samuel 1:24 carries it twice). Ten of the twenty-two are in 1 Samuel and 1 Kings — the sanctuary and its fall. Six are the personal name Shelah and have nothing to do with the place: five in Genesis 38 and 46, one at 1 Chronicles 4:21. Four are in Jeremiah: 7:12, 7:14, and 33:6 and 33:9 (which are Hebrew and ESV 26:6 and 26:9, where Jeremiah is put on trial for saying this house will be like Shiloh). One is Judges 18:31 and one is Psalm 77:60 (ESV 78:60), he forsook the dwelling at Shiloh.

Not verifiable from local corpora: the archaeology of Khirbet Seilun, the destruction layer and its date. Nothing on this disk speaks to it and this page makes no claim about it.

It is not a Greek word. It is the Hebrew כַּוָּנִים written in Greek letters and given a Greek accusative plural ending — a transliteration, not a translation. Measured: χαυών stands in exactly two verses of the 30,348 Greek Old Testament units screened here — Jeremiah 7:18 and Jeremiah 51:19 — which are precisely the two verses where the Hebrew has כַּוָּנִים.

Where a text transliterates, it is carrying a word it is not rendering. The domestic detail is the point of the Hebrew line: children fetch the wood, fathers light the fire, women knead the dough. It is a family, working together, on a task. That is what the sermon objects to — not a foreign cult in a corner but a household craft.

Not verifiable from local corpora: what a כַּוָּן looked like. Baked-clay moulds of a female figure are regularly cited in the commentaries; no such artefact is on this disk and this page does not describe one.

Jeremiah 10:11 is Aramaic and the rest of the book is not — the standard identification, which this page has not tested against an Aramaic lexicon. כִּדְנָה תֵּאמְרוּן לְהוֹם — thus you shall say to them — and then a sentence about gods who did not make heaven and earth, in the language a Judean would have used to say it to a Babylonian. The Masoretes fenced it off: it stands alone in a section of its own — between the open marker {פ} that closes 10:10 and the closed marker {ס} that closes 10:11.

Measured on the Hebrew screen it reaches two, tie band one, and the single match is a homograph accident: לא עבדו is Aramaic here for they did not make and Hebrew at Malachi 3:18 for who did not serve him. The screen reads consonants. It does not know what language it is looking at. A page that reported that hit without checking would have printed a parallel that does not exist, and this page prints the near miss on purpose.

Not verifiable from local corpora: whether this line is a proverb, a gloss, a marginal note that migrated into the text, or an Aramaic sentence Jeremiah composed. The Elephantine papyri and the Aramaic of Daniel and Ezra are the comparanda everyone reaches for; the Aramaic sections of Daniel and Ezra ARE on this disk inside the Masoretic file, but no dialect comparison was run and none is claimed here.

It is a satire on manufacture. Every line is about a process: a tree is cut from the forest, a craftsman works it, silver and gold are hammered on, nails and hammers fasten it so it will not totter, blue and purple are put on it, and then somebody has to carry it because it cannot walk.

Measured: the Greek row for 10:5 begins at the Hebrew’s fifth word, so the scarecrow and the clause and they cannot speak are both absent from the Greek line. וְלֹא יְדַבֵּרוּ stands in four verses of the Hebrew Bible — Jeremiah 10:5, Zephaniah 3:13, Psalm 115:5 and Psalm 135:16 — and three of the four are held in both columns. This one is not.

The Greek word for the idols in this chapter is worth a count. εἴδωλον appears in three verses of the whole Greek Jeremiah (9:13, 14:22, 16:19) and in none of chapter 10; γλυπτ-, carved thing, in five (8:19, 10:14, 27:38, 28:17, 28:52); and μάταιος, worthless, in six (2:5, 8:19, 10:3, 10:15, 23:16, 28:18). The satire calls them worthless things and carved things, and almost never idols.

שֶׁקֶר — a lie. It is the word for the deceptive words of 7:4 and 7:8, for the false swearing of 7:9, for the lying pen of the scribes at 8:8, for the lying tongue of 9:2 and 9:4 (ESV 9:3 and 9:5), and for the metal-worker’s cast image at 10:14, where שֶׁקֶר נִסְכּוֹ means his molten image is a lie.

The Greek scatters it: λόγοις ψευδέσιν at 7:4 and 7:8, ψευδῆ at 9:4 (ESV 9:5) and 10:14, and at 8:8 something else entirely — the Hebrew’s lying pen of the scribes becomes σχοῖνος ψευδὴς γραμματεῦσιν, a false measuring-line for the scribes. A pen in one column, a surveyor’s cord in the other. This page prints both and adjudicates neither.

Jeremiah 7:32 and Jeremiah 9:24 (ESV 9:25) both reach six consecutive Greek word forms with Hebrews 8:8, tie band one apiece. The six are ἰδοὺ ἡμέραι ἔρχονται λέγει κύριος καί. Band one means neither verse reaches six with any other New Testament verse.

And Hebrews 8:8 is quoting Jeremiah 31:31. In the Greek Jeremiah that verse is numbered 38:31, and it reads ἰδοὺ ἡμέραι ἔρχονται φησὶν κύριος — says in a different verb. Hebrews reads λέγει, so the run against the verse it is actually quoting stops dead at three. Against seven other rows of Greek Jeremiah that read λέγει κύριος καί — 7:32, 9:24, 16:14, 19:6, 23:5, 23:7 and 28:52 — it reaches six.

So the opening clause of the New Testament’s longest Old Testament quotation matches seven verses it is not quoting more closely than the verse it is. And the rest of that quotation is not close at all — it is identical. Hebrews 8:9, 8:10 and 8:11 reach twenty-three, thirteen and twenty-three consecutive word forms with Greek Jeremiah 38:32, 38:33 and 38:34, band one apiece, and those are the three deepest runs in the whole 1,298-row Greek Jeremiah. Only the first line of the quotation drifts, and it drifts onto this page. The page reports the arithmetic and does not say which text acted on which.

Matthew 21:13, Mark 11:17 and Luke 19:46 each measure THREE against the CATSS Greek of Jeremiah 7:11, and the three words are ὁ οἶκός μου. Those three words are the opening of the Isaiah 56:7 half of the saying: my house shall be called a house of prayer.

The Jeremiah half is σπήλαιον λῇστῶν, two words, and in the Greek Jeremiah those two words stand at the front of the verse while in the gospel sentence they stand at the end. A contiguous run cannot jump the gap between them. The measurement therefore reports the Isaiah phrase and not the Jeremiah phrase, on a quotation of Jeremiah.

This is the clearest case on the shelf of a number failing to see a citation that is obviously there. The right response is not to inflate the number. It is to say what the two words are worth on their own: they occur in four verses of the entire Greek Bible, and three of the four are the Gospels quoting the fourth.

Four rows of Jeremiah 7 reach five with Acts 15:17, tie band one each: 7:10, 7:11, 7:14 and 7:30. The five are ἐπικέκληται τὸ ὄνομά μου ἐπί — upon whom my name has been called.

Acts 15:17 is not quoting Jeremiah. It is James at the Jerusalem council quoting Amos 9:12 — and Greek Amos 9:12 carries the same five words. Measured: the phrase stands in five Greek Old Testament verses (Amos 9:12 and those four in Jeremiah 7) and in exactly one New Testament verse. In Hebrew the corresponding נִקְרָא־שְׁמִי עָלָיו stands in six verses, all six in Jeremiah: 7:10, 7:11, 7:14, 7:30, 32:34 and 34:15. The Hebrew phrase is a Jeremiah signature; the Greek phrase is shared with Amos; and the New Testament verse that carries it is reading Amos. Four fives that look like contact with the temple sermon, and are not.

Jeremiah 7:18 reaches four with Acts 7:42, tie band one — τῇ στρατιᾷ τοῦ οὐρανοῦ, the host of heaven, in Stephen’s speech.

The Hebrew of Jeremiah 7:18 does not say host of heaven. It says queen of heaven. The phrase τῇ στρατιᾷ τοῦ οὐρανοῦ stands in five Greek Old Testament verses — 2 Chronicles 33:3, 33:5, Jeremiah 7:18, Jeremiah 19:13 and Zephaniah 1:5 — and in exactly one New Testament verse, Acts 7:42. Had the Greek Jeremiah here read queen, as the same book does four times in chapter 51, the run would be zero. A run created by a rendering is still a run, and it is still worth printing — but only with the rendering printed beside it.

Deepest: Jeremiah 7:16 → Jeremiah 11:14, eleven consecutive Hebrew words, tie band one — the ban on intercession, twice. Jeremiah 9:8 (ESV 9:9) → Jeremiah 5:29, ten, band one — nine with the section markers stripped. Jeremiah 10:12 → Jeremiah 51:15, nine, band one. Jeremiah 7:32 → 16:14 and 19:6, nine, band two. Jeremiah 7:1 → 11:1, 18:1, 30:1, eight, band three. Jeremiah 7:34 → 16:9, 25:10, 33:11, eight, band three. Jeremiah 8:11 → 6:14, eight, band one. Jeremiah 8:2 → 25:33, eight, band one. Jeremiah 7:23 → Leviticus 26:12, seven, band one — the covenant formula, and the only run of seven or more in these chapters that leaves the book of Jeremiah.

The doublet at 10:12–16. These five verses stand again at Jeremiah 51:15–19. In Hebrew the five pairs measure nine, six, seven, seven and four. In Greek — where the second member is numbered 28:15–19, not 51:15–19 — the same five pairs measure five, thirteen, nine, three and ten — and the ranking is nearly inverted: the Hebrew’s deepest pair is the Greek’s second shallowest. Two Greek renderings of the same Hebrew, in one book, differing where the Hebrew does not.

Zero. Four verses in these chapters share no two consecutive words with any other verse in the 23,204-unit Masoretic corpus: 8:20, 10:8, 10:17 and 10:20. The Hebrew-side screen runs with the unigram fallback OFF, so 0 means no adjacent pair, not no words. One of the four, 10:8, is also one of the four verses the Greek column does not carry.

Revelation 18:23 and Jeremiah 7:34. Both say the voice of bridegroom and bride. Measured, the pair reaches two. Revelation 18:23’s deepest contact anywhere in the 1,298-row Greek Jeremiah is four, at Jeremiah 9:23 (ESV 9:24), and it is four words of connective tissue. The nearest thing to a real contact is three, at Greek Jeremiah 40:11 (= ESV 33:11).

Revelation 15:4 and Jeremiah 10:7. Revelation asks who shall not fear, Lord? and Jeremiah 10:7 asks who would not fear you, King of the nations? The Greek column has no row for Jeremiah 10:7, so there is nothing to measure. Revelation 15:4’s deepest contact in the whole Greek Jeremiah is four, and it lands on 9:25 (ESV 9:26), a verse about circumcision.

Paul on boasting. 1 Corinthians 1:31 says as it is written and measures two against Greek Jeremiah 9:23 (ESV 9:24). 2 Corinthians 10:17 measures one. 1 Corinthians marks its sentence as a quotation and 2 Corinthians does not; neither is visible to a contiguous-run screen.

Stephen on the uncircumcised heart. Acts 7:51 measures one against Jeremiah 9:25 (ESV 9:26), because Stephen’s καρδίαις is not the Greek Jeremiah’s καρδίας.

And the formula that looks like a link. Jeremiah 10:11 reaches five with Luke 16:17, Acts 4:24, Acts 14:15 and Revelation 14:7 — τὸν οὐρανὸν καὶ τὴν γῆν. Tie band four. Four New Testament verses at the same depth is what a stock phrase looks like on this screen, and it is printed here so that the band-one sixes above are not read as more than they are.

This passage is one of the places where the textual evidence matters most, and this page has none of it. Seven of the 107 verses are absent from the Greek column, four of them in a ten-verse stretch. The scrolls from Qumran Cave 4 — 4QJer^a, 4QJer^b, 4QJer^c, 4QJer^d — are the evidence that bears directly on whether a shorter Hebrew Jeremiah existed. None of them was collated. No photograph, transcription or apparatus was opened. Nor was Ziegler’s Göttingen edition of the Greek Jeremiah, nor BHS, nor BHQ.

Everything this page says about the shorter Greek is therefore a report on one digitised file and nothing more. The CATSS files are a transcription of Rahlfs’ Septuaginta. Where this page says the Greek column has no row for 10:7, that is a statement about Jer1.mlxx on this disk. It is not a statement about the Septuagint, about any manuscript, or about what the translator had in front of him.

The draft of this page said that Jeremiah 10:17 shares no two consecutive words with anything, in either language. That is false and it has been cut.

What is true, and what the page now says: on the Hebrew screen 10:17 reaches 0 against the 23,204-unit Masoretic corpus, and on the Greek screen it reaches 1 with a tie band of 3,300 against the 7,927-verse Greek New Testament — the only one of the ninety-nine rows in this passage that does not reach two there. But screened against the whole 30,348-unit Greek Old Testament it reaches three, tie band one, at Ezekiel 26:11 (τὴν ὑπόστασίν σου). The wider claim was appealing and the narrower one is correct, and the difference is the corpus the screen was pointed at. Every zero and every one on this page carries the corpus it was measured against, for exactly this reason.

• Commentaries not opened for this page: Holladay (Hermeneia), Lundbom (Anchor Yale), McKane (ICC), Carroll (OTL), Thompson (NICOT), Craigie, Kelley and Drinkard (WBC), Brueggemann, Fretheim, Allen (OTL), Stulman, Miller (NIB), Jones (NCB), Keown/Scalise/Smothers (WBC).
• On the two editions of Jeremiah: Janzen’s Studies in the Text of Jeremiah, Tov’s work on the two literary editions, Bogaert, Stipp, Piovanelli, Shead. This is the literature that would decide the question this page keeps refusing to answer, and none of it was read for this page.
• On the Greek Jeremiah specifically: Ziegler’s Göttingen Ieremias, Thackeray on the two translators of Greek Jeremiah, Tov’s The Septuagint Translation of Jeremiah and Baruch. The doublet measurement in stage 4 — 10:12–16 against 51:15–19 in both columns — touches that question directly and stops short of it deliberately.
• On Jeremiah 10:11: nothing on the Aramaic of the verse, its dialect, or its relation to the Elephantine corpus was consulted. The verse is reported as Aramaic, fenced by section markers, and measured. Nothing else is claimed about it.
• On the temple sermon and Jeremiah 26: the relation between Jeremiah 7 and Jeremiah 26 — the same sermon and its consequences — is one of the standard questions in the literature. This page measures the Shiloh vocabulary they share and does not discuss the relation.
• Archaeology: Shiloh, the Hinnom valley, the Ketef Hinnom burials. No site report, photograph or excavation record is on this disk.

The conflated row. The CATSS file carries Hebrew 7:27 and 7:28 as one row labelled Jer 7:27/28. Every Greek figure for those two verses is therefore a figure for the pair, not for either one, and no separate number exists for them.

Section markers are counted as words. The Masoretic file held here prints {ס} and {פ} inside the verse text, and the frozen normaliser tokenises them — at the end of thirty of this passage’s 107 verses, and of 3,447 verses across the whole Masoretic corpus. Two figures on this page do include a marker, and here is the whole difference. Jeremiah 9:8 (ESV 9:9) against Jeremiah 5:29 reads TEN with the markers counted and NINE with them stripped; Jeremiah 8:23 (ESV 9:1) against Lamentations reads THREE at tie band two with them counted and TWO at tie band fifty-three with them stripped. The screen was run both ways and no other figure on this page changes. normalize.py was not edited.

Runs are word forms, never lemmas. Two forms of the same word do not match. That is why Stephen’s καρδίαις measures one against καρδίας, and why the word order of 1 Corinthians 1:31 holds its run against Jeremiah 9:23 to two. The number is not a judgement about whether a text is being quoted.

The English column was not verified against a held ESV text. There is no ESV file for Jeremiah on this disk and no Crossway API key in this project. The ESV above was typed and its verse alignment rests on verse counts, chapter openings. The Hebrew and Greek columns were pulled programmatically from the corpus files.

Three places, named plainly.

1. The structure diagram is a Hebrew-side artefact. It is built from one edition’s section markers. The Greek files carry none, so the shape shown is not a shape either ancient reader of the Greek would have seen. And six of the sixteen rungs group two or more scribal sections for legibility.

2. The six-word runs into Hebrews 8:8 are formulaic by construction. ἰδοὺ ἡμέραι ἔρχονται stands in twenty Greek Old Testament verses, thirteen of them in Jeremiah. The band of one is a fact about the New Testament, where the phrase occurs once. It is not evidence that Hebrews had Jeremiah 7 or 9 in view, and the page does not suggest it.

3. The passage is quoted selectively. Sixteen rows out of 107 verses. The rows were chosen because they carry a measurement, which means the chapters’ long stretches of ordinary indictment — most of 8:1–10, most of 9:9–21 — are represented by their screen figures and not by their text. A reader who wants the argument of the sermon should read all four chapters; this page is about what can be counted in them.

The sermon’s first move is to take a chant away from the people who are chanting it. The temple of the Lord, the temple of the Lord, the temple of the Lord — three times in Hebrew, twice in Greek, and nowhere else in the Bible — and the answer is: go and look at Shiloh.

The argument is not that the building is worthless. It is that the building has a history, and the history includes a time God let it go. What the sermon asks for instead is in 7:5–6, and it is not liturgical: justice between a man and his neighbour, the sojourner, the fatherless and the widow not oppressed, innocent blood not shed.

Jeremiah 10 is funny, and it is worth saying so. The tree is cut down, the craftsman works it, the silver and gold go on, and then somebody nails it to the floor so it will not fall over. The joke is in the direction of the verbs: everything is done TO the god, and nothing is done BY it.

Against that stands 10:12–16, and it is the same list of verbs pointed the other way: he made the earth, he established the world, he stretched out the heavens. The measurement of that unit is one of this page’s clean results — those five verses stand again at Jeremiah 51:15–19 at nine, six, seven, seven and four consecutive Hebrew words. Somebody thought it was worth saying twice.

The deepest thing these four chapters do in the Greek New Testament is share six words of a stock formula with a verse that is quoting a different chapter. The most important thing they do — supply Jesus with a phrase for what the temple had become — measures three, and the three belong to Isaiah.

That is not a defect in the tool. It is the tool doing exactly what it does: counting adjacent identical word forms. What it cannot do is recognise a quotation, and this page has been careful to keep the counting and the recognising apart. The numbers above are checkable by anyone with the same files. The reading of them is not, and it is not being smuggled in.

Why are seven of these 107 verses missing from the Greek, and why are four of the seven inside ten verses of the idol satire?

There are two families of answer — the translator’s Hebrew was shorter, or the translator abridged — and choosing between them requires the Qumran evidence and the critical editions named in stage 5, none of which was consulted. This page can tell you exactly which verses are absent and exactly how the surviving ones are ordered. It cannot tell you why, and it is not going to guess. What it will say is that the absent verses are not random: 10:6, 10:7 and 10:10 are the three lines in the satire that praise the Lord directly, and 10:8 is one of only four verses in these chapters that share no adjacent word-pair with anything in the Hebrew Bible. That is a pattern. Whether it is a meaningful one is the question, and it is open.