How the echoes work

1What an echo is, and why it matters

An echo is a place where Revelation’s wording shares words with an earlier scripture closely enough that a reader who knew the earlier passage would hear it.

John almost never tells you he is borrowing. Revelation never says “it is written” (that was checked), and it has no footnotes. He builds his sentences out of earlier ones, so the only way to see the borrowing is to put the wording side by side. That is all an echo is: a recorded, checkable overlap of words, with the older passage named and the closeness stated.

Why it matters

1. The book explains itself. The figures people argue about most — the one like a son of man, the lamb, the animal, Babylon, the two witnesses — are introduced in words that already meant something in the scriptures John’s first readers knew. Following the words back lets the older text speak before any later system does.
2. It limits interpretation from both sides. An echo points to one particular older passage, with its own surroundings. A reading that goes against those surroundings has to say why. A reading that fits them has something to stand on.
3. It is the evidence the translation’s own rules call for. The translation is made from the Greek, with no English versions consulted, and no later framework (dispensationalist, preterist, futurist or any other) may decide a word. Scripture read against scripture is what is left to decide with, and echoes are where that happens on the page.
4. It makes the work checkable. Every marker can be opened, the two passages compared, and the closeness judged by the reader. Someone who disagrees can see exactly what they disagree with.
5. It shows where the text is silent. A verse with no echo has no older passage behind it that we have found. Anything said about that verse has to come from somewhere else, and the rest of this page is partly about how that “somewhere else” gets labelled.

2The process, step by step

Step 1 — Choose the sources, and the rules for them

Three texts are used, and each is open: the Nestle 1904 Greek New Testament for Revelation itself (public domain, with its morphology released under CC0); Swete’s Septuagint, the Greek translation of the Hebrew scriptures that John’s readers would have heard (a digital edition under CC BY-SA 4.0); and the Westminster Leningrad Codex, the Hebrew itself, through the OpenScriptures project (public domain). Sources whose licences restrict use were left out on purpose, and no English translation is consulted, for wording or for checking.

The Septuagint is not enough on its own. Where John follows the Hebrew and the Greek translation goes another way, only the Hebrew shows it. At 18:22 he has Jeremiah’s millstones; the Greek translation of that verse has perfume in their place. That echo is invisible to a program that only reads Greek, which is why the Hebrew is always checked by hand.

Step 2 — Let a program propose candidates, with no theology in it

A small program compares Revelation with the Septuagint, verse by verse, and lists places where the two share distinctive words. It knows nothing about doctrine. How it matches:

• Every word of Revelation comes with its dictionary form (lemma). The Septuagint has none, so each of its words is looked up in a table built from the New Testament; forms the New Testament never uses are matched by stem and marked as guesses.
• Articles, pronouns, conjunctions, prepositions and particles never count as shared words on their own.
• A word counts only as much as it is rarer than everyday vocabulary. “Say”, “all”, “earth” and “Lord” add nothing, and every candidate must share at least one rare word.
• The shared words must sit close together in both texts, not scattered across a long verse.
• A run of consecutive shared words earns a bonus, so a phrase such as μετὰ τῶν νεφελῶν (“with the clouds”) counts as more than its parts.

In its most recent run it scored 452,851 pairs of verses, and listed 13,875 candidates scoring 5 or more, a median of 29 for each verse of Revelation. It is deliberately generous: it would rather hand a person too many candidates than hide one.

Step 3 — Grade the program against the notes already made

The program checks itself. Of the 186 places where an echo note was already marked, it had the right source as its single best candidate for 58, in its top five for 96, in its top ten for 106, and somewhere in the list for 163. So it finds a little over half in its top five, and it misses echoes that rest on the Hebrew, on a name, or on a single distinctive word. That is why it only proposes.

Step 4 — A person reads every candidate in the original

Each candidate is read in the Greek and, where the Hebrew may differ, in the Hebrew. The questions are the same for each: Are the shared words distinctive, or stock vocabulary that would match any list of gold or wine? Is it the same sentence, a name, or only a shared image? Is it a common formula (“thus says the Lord”) that scores well by accident? A candidate is marked, rejected, or held back as looser than the standard. The 195 echo notes are what survived; the list the program produced was far longer.

Step 5 — Read the verses the program did not flag

A program that only sees shared vocabulary has a blind spot, so a second pass was made by hand. Every one of the 292 verses that then carried no echo marker was read and checked in Swete, in Nestle 1904 or in the Leningrad Codex, which added 51 more notes. The same lesson came from a separate scan of the seven letters (chapters 2–3) against every verse of the New Testament outside Revelation, made for the purple layer. It missed two links that were later found by hand, the name Judah at 2:9 and the phrase “named a Jew”, because every word they share (Ioudaios, lego) is so common that the program discounts it. A program’s silence is not evidence that nothing is there.

Step 6 — Apply the green bar

The first of the five note layers, green, is held to a strict standard:

• A close verbal echo, identifiable from the wording itself, not from a cross-reference table. A thematic resemblance does not earn green.
• The Hebrew scriptures only. Echoes of the Septuagint’s other books (Wisdom, Sirach, Maccabees, the Greek additions to Esther and so on) are not marked.
• A name counts only when the name is doing work in the passage, set beside something or picked up by the text. Otherwise its meaning is a coincidence, not an argument.
• Every echo note says how close the connection is when the wording does not make it obvious — a named person or event, or a shared image with little shared wording, is said to be exactly that.

The standard has been applied to its own candidates. Nimrod was tested as an echo behind the name Nikolaos, which suggests “conquers the people”; but Genesis calls Nimrod a gibbor and the Greek has gigas kynegos, so neither half of the word is there, and it was cut.

Step 7 — Sort every note into a layer

Each kind of claim gets its own colour, so a reader always knows what sort of statement they are looking at. The count is of notes as they stand today.

The first four are reporting layers: they say what the words are, what they were taken to mean, or what a decision was. Gold is the only layer that argues, and it is labelled so.

Step 8 — Place the marker on the exact words

The marker sits on the words the note is about, not on the whole verse. Two notes can share a phrase (an echo and a translation decision at 1:6, for instance), and one word of a phrase can be handed to one note and the rest to the other. Echoes are numbered once through the book, in reading order, so the same note keeps its number wherever it returns. In the margin, and inline on a narrow screen, each marker shows its layer’s colour.

Step 9 — Count carefully, and check the counts

A note once counted verses where it said words. Asked whether a note’s “thirteen times” was a mistake, a re-count found it was: the word occurs 15 times in 13 verses, because one verse has it three times. Every such figure in the published notes was then re-checked by word against verse; three published notes were wrong, and were corrected. A small tool now prints the words and the verses side by side for any figure a note wants to claim, and an audit of the gold notes, the front matter and the checkable claims in the others is part of the record. Errors found are corrected, and a list of the ones still open is kept.

Step 10 — Review, and publication

Jon makes every editorial call. Reporting notes may be published for him to read on the site; gold readings are his to approve, and until he has read one it carries a line saying so (10 readings carry that line today). Every reader page says that the work is a second draft and that each reader can decide for themselves what the scriptures say and whether Jon’s arguments hold.

3One echo worked through

Revelation 1:7 shows each step on a single verse.

1. The program. It ranked Zechariah 12:10 first for this verse (score 10.6, shared words kopto “beat”, with the phrase above) and Daniel 7:13 second (score 8.2, shared words orao, erchomai, nephele, with the phrase μετὰ τῶν νεφελῶν).
2. A person reads them. Both are distinctive, not stock phrases; both are in the same verse, and both are marked.
3. The Hebrew settles something the Greek alone cannot. Revelation says meta, “with”. The Old Greek of Daniel says “on”; Theodotion’s Greek and the Aramaic say “with”. So John’s wording here follows the Aramaic and Theodotion’s version, not the Old Greek, and only the comparison shows it.
4. The green bar. Shared phrase, shared sentence, one named older passage: green. The note says it takes the wording almost intact, and that in Daniel the phrase means a figure in human shape, set against the animal figures earlier in the vision.
5. It comes back. The same Daniel passage is marked in four verses of Revelation (1:7, 1:13, 11:15, 14:14). That is one of the more connected echoes in the book; the table below lists the others.

What the echo does not say is what John means by it, or who the figure is. That is a reading, and it would be a gold note.

4The numbers

Which books the echoes come from

Counted from the passages named in each echo note’s title. A note that names two books is counted under both, and the verse count is the number of verses of Revelation carrying an echo from that book.

The most connected echoes

Some echoes are marked more than once, because John goes back to the same passage. These are the ones marked in the most verses.

Where the echoes fall, chapter by chapter

About half the verses carry an echo and about half do not. Two things can be true of a verse with no echo: John did not borrow identifiable wording there, or he did and it has not been found yet. The program only sees Greek, so the second is always possible.

Everything else

Besides the echoes, the notes include 25 cultural notes, 23 translation decisions, 18 notes on wording shared with the later writings, and 57 readings.

5Is it objective?

Partly. The observation is. The interpretation is not. The layers exist so that you can tell which one you are looking at.

What anyone can check

• The words. Both texts are open, and the marker names the older passage. Two lists of Greek or Hebrew words either overlap or they do not.
• The counts. How many times a word occurs, and in how many verses, is a count. This is why the figures are checked by word and by verse, and why errors, when they are found, can be stated precisely and fixed.
• The standard. The green bar is written down above and is applied the same way to every candidate, including the ones that would be convenient for a favoured reading. Nimrod was cut by it.
• The starting point. The candidates come from a program with no doctrine in it. The method starts from the text and not from a theory about what the text should say.

What is a judgment

• Which candidates to mark. Out of thousands the program proposed, 195 were kept. That is a human selection. Another reader could keep a different set; the criteria are written down so they can.
• How close is close enough. The bar is strict, but where a connection sits just above or below it is a call, and the note says how close it judged it.
• Which part of the older passage is being evoked. An echo of Daniel 7:13 could be heard as the whole vision or as one line. The note says which it takes.
• What the echo means. Two passages sharing words is a fact. Why John shares them is a reading.

6Where opinion reaches past the text

Because every kind of statement is labelled, a statement that goes beyond what the text shows shows up as being in the wrong place. That is the use of the whole structure. There are four tools:

1. The layer itself. Green, blue, grey and purple notes report. If a claim is argued, it is in a gold note, and a gold note opens by declaring that it is interpretation and shows how far it reaches.
2. Marked sections. Where a reading is Jon’s own view and not simply what the texts give, the section is headed as his. There are currently 134 such headings in the readings.
3. Red flag on any claim that is not found anywhere in scripture. A flag does not mean the claim is wrong. It means the text does not say it, so the reader should know it is a step outside the text, and the note gives the nearest texts on each side. There are 65 flags across 25 readings today.
4. The unread line on a reading Jon has not yet read through, so a reader knows the wording may change.

What a flag looks like

These are real flags from the readings, shortened:

• Red flag That the number is Nero’s. The book hands the reader a method and declines to give the name.
• Red flag That the city of the seven hills is a present institution. The book names the city by its hills and its rule over kings; it names no church and gives no date.
• Red flag That Matthew 23 describes a pope. It is spoken to the scribes and Pharisees of Jesus’ own day, and no verse applies it to a later office.
• Red flag That the eight kinds in 21:8 name what a person is, as against what he has done. The Greek forms do not say it; an agent noun can describe a deed.

In each, the reading is his and is stated as his; the flag marks the exact point at which it goes past what scripture says, so that the reader can follow the scriptures up to that line and decide for themselves whether to cross it.

A test you can run

Take any gold note. Read the translation and its green markers alone, then read the note. Whatever the note adds beyond what the words and the echoes support is the interpretation. If it is not found in scripture it should carry a flag, and if it is, the nearest texts are cited so you can check the step yourself.

7What the method cannot do

• It cannot tell you what an echo means. It records the overlap and says how close it is. Meaning is a reading.
• It cannot prove John intended an echo. Two passages can share rare words by coincidence, and a score is evidence of shared vocabulary, not of an echo. A person judges; the judgment can be wrong.
• It misses what it cannot see. The program reads Greek only; echoes of the Hebrew that the Greek translation does not share, echoes of a name, and echoes built on one distinctive word are found by hand or not at all. The program’s own grade, above, shows how many it misses.
• It can be wrong in its sources. Swete’s text was digitised by machine and corrected, so a few words may be misspelled. Psalms and Jeremiah follow the Septuagint’s numbering, which differs from the Hebrew, and the notes say so where it matters.
• It does not settle the disputed questions. Chapter 20 is deliberately under-noted: the note there says what the text says and what it leaves open, and adjudicates nothing.
• It has had errors, and will again. They have been found by re-counting, by asking whether a figure was right, and by audit, and each has been fixed in public notes. Work that says it is a second draft means it.

8Check one yourself

1. Open chapter 1 and click the green marker on he is coming with the clouds at 1:7. The note names Daniel 7:13 and says how close it is.
2. Open the Greek text of Revelation 1:7 and of Daniel 7:13 and compare the words for yourself. The sources are the Nestle 1904 Greek New Testament, Swete’s Septuagint and the Hebrew and Aramaic of the Westminster Leningrad Codex.
3. Ask whether the shared words are distinctive and whether the note’s closeness line is fair.
4. If you disagree, say so. jon.pjcomputing@gmail.com. A disagreement about an echo is the most useful kind, because it can be settled by the words.