Peter Mandelson

⚠ what this page is about, and what it is not

This page is about a correspondence. The release holds an exchange of messages between this man and Jeffrey Epstein, running from 2002 to 2016 with a break of five years and five months in the middle, and what follows is drawn from it: what he wrote, when, and what he asked for. It carries no allegation against him, it does not describe his private life, and where the release contains material about him that this desk will not use, it says so without describing it. Several of the most quotable lines in this correspondence come from messages whose sender is blacked out. None of those is attributed to him here.

In the release as	One of Jeffrey Epstein’s most frequent correspondents. E-mail in 2002-03, none between April 2003 and October 2008, then near-continuously to November 2016. ⚠ The gap is in the e-mail only: bank records name him in May 2003 and June 2004, on two different Epstein accounts123
Documents naming him	5,126 under the full name, 5,409 by surname. ⚠ The number is real and the page explains what it consists of1
Messages he sent	Between roughly 500 and 1,200 documents, depending on whether the sender is read from the extraction’s sender field or from the From: lines on the sheet. ⚠ A spread whose hinge is named, not a figure1
⭐ The act, and the clearest of several	On 20 December 2009 he forwarded Epstein a Labour general-election strategy memorandum he had written that day. Within the hour Epstein sent it back line-edited4
And	On 13 March 2011 a statement about his own relationship with Epstein was drafted and its text reached Epstein. ⚠ Nothing in the release shows whether it was published5
Earliest	2002, in correspondence routed through Ghislaine Maxwell6
Latest	November 20167
While in office	Correspondence continues through his time as Secretary of State for Business and First Secretary of State, October 2008 to May 20108

Within the hour

On 20 December 2009 he wrote a Labour general-election strategy memorandum and forwarded it to Jeffrey Epstein. Fifty minutes later Epstein sent it back, line-edited.

The subject line Epstein used is EDITS. His changes are highlighted in yellow on the sheet. Where the memorandum read “THE STRATEGY YOU HAVE SAID YOU WANT IS ONE THAT MOVES FROM REFERENDUM TO CHOICE. THIS IS RIGHT”, the returned text reads:

THE STRATEGY YOU HAVE SAID YOU WANT IS ONE THAT MOVES FROM REFERENDUM TO CHOICE. I AGREE.

Further insertions run through the document: “INSTEAD THEY WILL RIDE EASILY ON BACK OF THE MOOD FOR CHANGE”, and “AGAINST THIS, WE ARE LEFT TO HOPE FOR BRIGHTENING ECONOMIC PROSPECTS AND POSITIVE GROWTH WITH THE CHANCE OF A ‘BETTER THE DEVIL YOU KNOW’ CONCLUSION.”

That is the page’s anchor and it needs no inference at all. A serving First Secretary of State forwarded a general-election strategy memorandum he had written to Jeffrey Epstein, and Epstein edited it and sent it back within the hour.⁴

⚠ He forwarded it; Epstein was not its intended reader, and that is what makes it worth recording. ⚠ The memorandum’s intended reader appears nowhere on the sheet.

⚠ The page does not say the edits were adopted, and nothing cited here shows that they were.

The statement of March 2011

On Sunday 13 March 2011, three messages in seventy-seven minutes.

At 16:54 he wrote to Epstein: “Pl call me on cell.”

At 17:46 he wrote again, forwarding that message:

I have drafted a statement. Ben will send.

At 18:11 Benjamin Wegg-Prosser, from a Global Counsel address, sent Epstein the statement’s text:

“Lord Mandelson was in London on 12 and 13 December 2009 prior to travelling to India on government business, and no such meeting was requested by him.

“Lord Mandelson was introduced to Mr Epstein a decade before by his then partner Ghislaine Maxwell who, along with her father and other members of her family, Lord Mandelson had known for many years before.”⁵

⚠⚠ AND THAT IS ALL THE DOCUMENTS ESTABLISH. An earlier version of this page said the statement went to Epstein “before it was issued” and that its subject was “shown it first”. Neither is supported. Nothing in the release shows the statement was ever published, or when; and the corpus is Epstein’s own mailbox, so “Epstein saw it first” is a claim this body of documents cannot test in either direction. ⚠ The 18:11 message was sent by Wegg-Prosser, not by him.

What is documented is narrower and still worth having: he wrote that he had drafted a statement and that Ben would send it, and twenty-six minutes later its text was in Epstein’s inbox. ⚠ The page does not say the statement was untrue, and nothing here suggests it was.

⚠ Both messages were read at the image at 700 dpi. His own address is blacked out on the copy; Epstein’s prints. Neither is reproduced.

What five thousand documents actually are

A number that large is usually an artefact, and on this wiki one has been: a whole book was once photographed into the production, and its running head inflated a subject to hundreds of pages. This one is not. It was tested before the page was written.¹

• 95.1% of the DOCUMENTS are e-mail header pages, and of the 7,753 occurrences of the full name, 7,134, or 92%, fall in those documents. ⚠⚠ TWO EARLIER VERSIONS OF THIS BULLET WERE WRONG IN OPPOSITE DIRECTIONS. One called the 95.1% an occurrence share when it is a document share. The other gave 7,134 as a count of distinct textual contexts and the occurrence share as 88%: the two figures had been crossed, and no definition of the share tried by two readers across thirteen sweeps produces 88% (the range is 91.2% to 96.9%).
• Distinct textual contexts, on a forty-character window either side, number 5,429 across 5,371 pages. A running head or a standing distribution list produces a handful of contexts repeated hundreds of times; this produces thousands of different ones.
• It is not a mailing list. 89% of the messages addressed to him come from Epstein, and about 87% of his own messages, among those with a parsed recipient, go to Epstein. ⚠⚠ An earlier version said 100%, and the page’s own 2002-03 section refuted it three lines later by quoting three messages he sent to Ghislaine Maxwell. A figure that a page’s own body contradicts should never have reached it.
• His own messages: between roughly 500 and 1,200 documents, and the hinge is which sender field is read. Reading the extraction’s sender column over this population gives about 1,200; reading the From: lines on the sheets gives about 500; reading only the first From: line gives about 400. ⚠⚠ AN EARLIER VERSION REPORTED “880 AND 889, AGREEING TO WITHIN 1%” AS TWO INDEPENDENT METHODS. THE 889 DOES NOT REPRODUCE, twelve implementations of the stated recipe span 704 to 1,043, and no script for the original was kept. ⚠ A later version printed 880, 809 and 704 as a range, and an implementer working only from that page lands anywhere between 404 and 1,359, because the page never said which sender field to read. A published method must yield one answer; a spread with an unnamed hinge is not a method. ⚠ And these count documents, not messages.

⚠ No total of distinct items is stated on this page. Two deduplication methods disagree in opposite directions: containment merges a message with the message it quotes, and the structural key merges records whose timestamps will not parse rather than splitting them. ⚠ An earlier version of this page said it split them, which is backwards.¹

2002 and 2003: how it starts

The correspondence begins through Ghislaine Maxwell rather than through Epstein, and the earliest items are logistics.

In April 2002 he writes to her about a postponed lunch, with a named businessman, “to discuss his board”, adding a parenthesis that assumes a shared third party: “Jeffrey will explain if I did not tell you the background”. In September 2002, on not being able to see Bill Clinton in Blackpool, he writes: “It means I won’t see him, or be seen with him, because he will be surrounded with officialdom.” In January 2003 he asks her for help arranging a trip, noting “Not sure if J is dealing with this personally.”⁶

Two courier records fall inside that period, both on Epstein’s shipping account: one running to him at the House of Commons in November 2002, one running from him to Epstein’s New York address three months earlier.⁶

⚠ And two travel invoices are billed in his name by Epstein’s travel agent in April 2003, totalling a little over seven thousand dollars. ⚠⚠ The ledger they sit in is not reproduced here. Its Name column is half redacted on the image, and the extraction does not preserve the row structure, so names and figures come out silently mis-aligned and a reader working from searchable text can attach the wrong figure to the wrong traveller. Only the two rows bearing his name were read, at 700 dpi, and only they are reported.⁹

In office

⚠ The correspondence is not continuous, and an earlier version of this page implied it was. The 2002-03 messages run on to 6 April 2003, and no e-mail resumes until 3 October 2008: a gap of five years and five months in the correspondence. In the earliest of the resumed messages Epstein writes to him under the name “Jeffery Edwards”.²

On 17 December 2009 Epstein wrote to him at 05:33: “should jamie call darling one more time, ? they said that jpm alone was over the 500 million said to be needed.” ⚠ Neither name on that sheet is expanded, and this page does not complete either. ⚠ The message is Epstein’s, not his, and nothing cited here records a reply to it.¹⁰

⚠⚠ BUT THE GAP IS IN THE E-MAIL ONLY, AND AN EARLIER VERSION OF THIS PAGE SAID “NOTHING UNTIL OCTOBER 2008”, WHICH IS FALSE. Three $25,000 payments out of Epstein accounts carry this name in the destination line, twice as BEN: and once as A/C:, one in May 2003 and two in June 2004, and a 2005 telephone-message log lists him as well. ⚠⚠ The page gives the entries and prints NO TOTAL. The figure a total would state has risen at every round of checking, nobody has swept 2005 to 2008 or Epstein’s other entity accounts, and a sum is a flat-fact sentence resting on an identification the page treats as open. ⚠ Whether the beneficiary is this man is a question, not a fact. Two of the three entries give a bare name; the third routes through a named private individual’s account at a named London branch, which is consistent and is not proof. That person is not named here and no relationship is stated or implied.³ An absence claim had been made across a period the desk had not searched outside one document type.³

He became Secretary of State for Business in October 2008 and First Secretary of State in June 2009. From then it is near-continuous.

• June 2009: “Shall I stay at your’s Friday-Sunday this weekend ?”, and the next day “Still OK for Fri-Sat chez vous? Can you give a bed to my spokesman/friend/collaborator as well?…”
• November 2009, on a forwarded news alert about Epstein: “This is interesting but, as the tip of an iceberg, revealing the rest needs careful planning. What is the plan ?”
• November 2009, on bank write-downs: “Sounds to me that I have to take my limited savings off shore asap..”⁸

On 20 December 2009 he forwarded Epstein a Labour general-election strategy memorandum he had written that day. It opens: “THE STRATEGY YOU HAVE SAID YOU WANT IS ONE THAT MOVES FROM REFERENDUM TO CHOICE. THIS IS RIGHT…” ⚠⚠ CORRECTION: an earlier version of this page said the memorandum’s addressee was blacked out. It is not. There is no addressee line on the sheet at all: the quoted-message format carries none, and the bar on that page covers his own address. An absence had been mislabelled as a redaction, which is a worse error than either, because it invents a decision by the producer. The memorandum’s text distinguishes its reader from “Alistair” and attributes a Pre-Budget Report decision to that reader; that is material a reader may weigh, and this page names nobody.¹¹

In early 2010 the traffic turns to banking policy. In January, Jes Staley of JPMorgan asks Epstein “Were you able to arrange a meeting with Mandelson or Darling?”, and Epstein answers the same day that he has passed the contact on. On 30 March 2010 Staley wrote to Mandelson directly, sending “some brief speaking points that we would use in discussing the Volcker plan with Summers”, and that message was forwarded to Epstein. The next morning at 08:15 Mandelson wrote to Epstein: “Have spoken to Chancellor re Larry. In good position. Seeing him in hour.” ⚠ That message is a reply on a different thread, whose subject is an unrelated forward Epstein had sent four hours earlier; it is not a reply to the Volcker note.¹²

⚠ “Larry” is not identified on that sheet, and this page does not complete it. Larry Summers is named elsewhere in the same correspondence and in Staley’s note of the previous day, which is a reason to ask the question, not a basis for answering it.¹²

On polling day, 6 May 2010, he wrote: “we are praying for a hung parliament.”⁸

Out of office

After May 2010 the correspondence becomes, in large part, a job search conducted in front of Epstein.

⚠ And the anchor act is repeated. On 13 May 2012 he forwarded Epstein a draft newspaper article he had co-written with a serving shadow chancellor and asked what he thought. Epstein criticised it and told him “I would [l]ike you to be bold”, and in the same thread asked “I think you should consider introducing me to anyone you think could be a huge business participant.” The 2009 memorandum is the clearest instance of this shape, not the only one.¹³

Two further forwards of his own work belong in the same section. On 18 June 2010 he sent Epstein a speech he was to deliver, under cover of “please review - good for business world ?”, forwarding beneath it his own instruction to a speechwriter; every page of the sixteen-page speech is headed EMBARGO UNTIL DELIVERED. On 16 July 2010 he forwarded his own letter, written that morning, to the chief executive of Glencore, seeking a board seat before the company’s flotation. Epstein’s reply, in full: “Number to call?”¹⁴

• June 2010: he sends Epstein successive drafts of the introduction and the banking chapter of his memoir, then asks “so if I re- do the beginning, you like the rest ?”
• June 2010, reporting a three-hour dinner with Lee Kuan Yew: “He is prepared to help me in China. Thinks that I need a ‘perch’, a bank.”
• August 2010, sending a “refined version of biz prospectus”, and then, answering Epstein: “Of course I want to follow the money trail (if someone helps put me on it) but I also need a parallel business platform.” ⚠⚠ The phrase is EPSTEIN’S, not his. Epstein had written it six minutes earlier, and it stands in the quoted block beneath his reply: “I have always suggested that the key is for you to follow the money trail”. He is repeating it back. An earlier version of this page said the phrase was his, which reversed the finding the correction was supposed to record.
• 25 December 2010: “Reflecting on our phone call, I hope I can do something with JPM.” The message runs to four paragraphs on how such work could be routed through his own firm, and closes “I do not want to live by salary alone.”¹⁵

⚠ In July 2014 Epstein wrote that a third party “has asked me to encourage you to join his board”, and he replied asking about the risk of “over leveraged parent company” before saying he would discuss it. That exchange belongs as much to DP World as here.¹⁶

March 2011, and afterwards

Three days after the drafted statement, on 16 March 2011, Epstein forwarded him a question from a newspaper. His reply was: “Categorically untrue. I will say that but not sure you sd comment.” Epstein answered: “I am saying nothing,, where are they getting this?”, the doubled comma as printed, and a bar follows on the same line and forwarded the same question to two other people, one of them before his reply rather than after it, across seventeen minutes.¹⁷

⚠ The content of what the reporter put is the newspaper’s, not this wiki’s, and it is reported here only as a question a journalist asked and an answer he gave. The page does not restate the allegation as though the release established it.¹⁷

The correspondence runs on for another five and a half years, and it is heaviest in the three years this page says least about: on a count of documents naming him, 2011, 2012 and 2013 carry more than two thousand between them, continuously month by month. ⚠ The later gaps are real and are stated here rather than left to be discovered: no document naming him is dated between August 2014 and February 2015, or between August 2015 and September 2016. In March 2015, on his friendship with Andrew, Epstein wrote that he supposed Mandelson had been right; the reply eleven days later was “All well thanks. Time heals.” On the evening of 6 November 2016 Epstein wrote to him “63 years old. . you made it”, and twenty-six minutes later he answered “Just. I have decided to extend my life by spending more of it in the US”; and, two days before the American election, “By way, as political practitioner, donald is phenomenal. The craft and tenacity are amazing”, which is unpunctuated at its end on the sheet and is printed here as it stands. Three days later he wrote again with nothing in the body and only a subject line: “Well? What now?”⁷

That the exchange continued is a documented fact on its dates. What it means is not something this page decides.

What this page does not say

⚠⚠ The most quotable lines in this correspondence are not his, and cannot be shown to be. Several of the sharpest messages in the 2009 banking sequence sit in threads he is part of but carry a sender field blacked out in every copy in the release, in one case all seventeen. Attributing those to him would be the allegation-by-adjacency error in its purest form, and this page does not do it. There is ample material where the sender prints.

⚠⚠ A body of material in the release names him and is deliberately excluded. No disposition of it could be found in the release. This page records that it exists and does nothing else with it: no allegation is stated, nothing is quoted, no identity is resolved, no identifier is listed, and nothing is said about who wrote it, because a characterisation of the author is itself information about the author. ⚠ An earlier version of this notice both described the author and listed “no denial” among the things not found. A denial is not the kind of thing this corpus could show, so its absence measures nothing. Both are removed.

⚠ No telephone number, e-mail address or other contact value is reproduced from any document cited here, though several print them in message bodies.

⚠ Where a copy 403s on the public PDF route, that is a fact about the route and not about the release, and an equivalent is cited instead. ⚠⚠ An earlier version of this line claimed that NOTHING on the page rests on an unopened document. That was never tested and is withdrawn; much of the page is read at the text layer and says so footnote by footnote.

Related

• DP World: carries the 2009 approaches to him as Business Secretary, framed there and not restated here
• Jeffrey Epstein
• Ghislaine Maxwell
• Bill Gates and Boris Nikolic: the February 2011 dinner
• Mortimer Zuckerman: the April 2013 dinner
• Eduardo Teodorani-Fabbri
• Epstein and global influence

Footnotes

Footnotes

1. The measurement, and why the page states it. "Peter Mandelson" returns 5,126 documents, the surname alone 5,409, "Lord Mandelson" 279. ⚠⚠ A COUNT THAT SIZE IS USUALLY AN ARTEFACT AND THIS ONE WAS TESTED. On this wiki a photographed book’s verso running head once inflated a subject to several hundred pages, and a firm name inflated another. Tests run before this page was written, with two figures corrected by the verification pass: 95.1% of the DOCUMENTS are e-mail header pages, and 7,134 of the 7,753 occurrences of the full name, 92%, fall in those documents (⚠⚠ an earlier version called the 95.1% an occurrence share, and a later one gave 7,134 as a count of distinct contexts with the share at 88%: the two had been crossed, and no definition of the share produces 88%); distinct textual contexts number 5,429 on a forty-character window across 5,371 pages, where a running head yields few contexts repeated many times; ⚠ documents.doc_type and document_classification.doc_type are different taxonomies and only the former gives 95.1%; 89% of messages to him are from Epstein, and about 87% of his own, among those with a parsed recipient, go to Epstein (⚠ an earlier version said 100%, which the page’s own 2002-03 section refuted three lines later). His own messages: about 880 DOCUMENTS by a loose sender match, 809 by a strict one, and about 704 after deduplication. ⚠⚠ An earlier version reported “880 and 889 agreeing to within 1%” as two independent methods. The 889 does not reproduce: twelve implementations of the stated recipe span 704 to 1,043, and no script for the original was kept. A convergence that cannot be re-derived is not a convergence. ⚠ The hinge is which sender field is read: the extraction’s sender column gives about 1,200 over this population, the From: lines on the sheets about 500, the first From: line alone about 400. An implementer must be told which, or the recipe is not a recipe. ⚠ No total of distinct items is stated on this page. ⚠ An earlier version gave 3,556 and 3,994; neither re-derives, and a figure whose recipe cannot be restated is not a figure. Containment over-merges, chaining a message with the message it quotes; a structural key merges records whose timestamps will not parse (⚠ an earlier version said it split them, which is backwards), and 2,198 of these 5,126 documents, 43%, have no parseable date header at all. Two oppositely biased estimators are not an estimate, and neither is reported as a number here. ⚠⚠ The identity space is not clean, and every headline count on this page is keyed to ONE spelling. Mendelson is partly an OCR variant of this man and partly other people and a law firm of that name, 266 documents outside the surname population, and an earlier version said it was only the former. Mandelsohn is the reverse: 154 pages, and on 150 of them a “Peter” or “Lord” stands within forty characters. ⚠ It is not OCR damage — on most it is a typed subject line — and none of it falls inside the 5,126, the 5,409, the 5,371 or the 11,720. A count keyed to one spelling bounds nothing, and neither do the absences measured over it. ⚠ No figure is given for Mandleson: the great majority of its 36 pages sit inside the withheld block this page’s withholding notice covers, so a count of that variant would be a count of the withheld body. ⚠ Four documents of the population are printed book pages from a manuscript, not two; that is still four documents and not a block, which is the point the figure was making. ↩ ↩² ↩³ ↩⁴ ↩⁵

2. The gap, and where it ends. The 2002-03 messages run on to 6 April 2003 (EFTA00581017 p.1, twin at EFTA00578839, “I will phone you in Paris.”), and no e-mail resumes until 3 October 2008: a gap of five years and five months. ⚠⚠ AN EARLIER VERSION SET THE GAP’S START AT JANUARY 2003 AND WAS WRONG BY TEN WEEKS. It had measured where his 2002-03 correspondence stops using his correspondence with Epstein, when the 2002-03 correspondence is with Ghislaine Maxwell, the distinction this page’s own [^origin] draws. Eight messages of March and April 2003 sit on the run [^origin] already cites. ⚠ Several of those eight print a telephone number in a signature block and none is reproduced. An earlier version also concealed the gap entirely by describing a continuous “2002 to 2016” exchange. The resumption is EFTA00748410 p.1: From: Jeffery Edwards / To: Peter Mandelson / Subject: / Date: Fri, 03 Oct 2008 16:52:31 +0000, opening “To start , this is one of the greatest poliical revival opportunitis of all time” (spelling as extracted). ⚠ “Jeffery Edwards” is Epstein, writing from his own address; the name appears in the sender field of the early resumed messages. Further messages of 4 and 7 October are at EFTA00748412 and EFTA00748413. ⚠ Read at the text layer. ⚠⚠ THE APPOINTMENT DATE IS NOT IN THE RELEASE AND THIS PAGE NO LONGER STATES IT. An earlier version set the resumption beside his entry to the Cabinet and marked the coincidence “stated-not-inferred”. On a living person’s page, a date with no source in the corpus is not something this desk may state at all, however carefully hedged; the body’s “the week he entered the Cabinet” went with it. ↩ ↩²

3. What is in the gap. ⚠⚠ An earlier version of this page said “nothing until October 2008”. That is false, and the error is instructive: the absence had been measured across one document type only. Three entries, on two different Epstein accounts, thirteen months apart.

   1. JPMorgan Private Bank, Business Checking, account styled “EPSTEIN INTERESTS”, 1-30 May 2003. EFTA01487811 p.1, read at the image at 700 dpi by the desk; twins at EFTA01487808 and EFTA01532275 p.2. “May 14 · CHIPS Debit · $25,000.00 · VIA: BARCLAYS BANK PLC … BEN: PETER MANDELSON”, the balance running to $25,153.64, which is the statement’s own Ending Balance. It posted. ⚠⚠ THE BENEFICIARY ACCOUNT IS IN ANOTHER PRIVATE INDIVIDUAL’S NAME, AT A LONDON BRANCH. That person is not named here, no relationship is stated or implied, and the account details, sort code and address on the sheet are not reproduced. ⚠ Two other debits precede this one on the same sheet. The page draws no connection between them and this entry, and a reader should not either.

   2 and 3. JPMorgan Private Bank, Premier Checking, Jeffrey E Epstein, 29 May to 30 June 2004, two entries, read at the image at 700 dpi by the desk and re-read at 700 dpi by the verification pass: EFTA01482503 p.1 (page 7 of 10), “Jun 24 · CHIPS Debit · $25,000.00 · VIA: HSBC BANK USA · A/C: PETER MANDELSON”; and EFTA01482505 p.1 (page 9 of 10), “Jun 30 · CHIPS Debit · $25,000.00 · VIA: HSBC BANK USA · A/C: ACHSBC PRIVATE BANK, 78 ST. JAMES STREET, LONDON · BEN: PETER MANDELSON”. ⚠ The copies of the 2004 sheets are in the same dataset; an earlier version called them a second production.

   ⚠⚠ BOTH 2004 DEBITS POSTED AND STAYED POSTED, settled by the balance column and not by the description text, every step checked by the desk: $1,162,075.07 − $25,000 = $1,137,075.07 on 24 June, carried down through the rest of that day; and $1,150,776.00 − $25,000 = $1,125,776.00 on 30 June, which is the statement’s own Ending Balance.

   ⚠⚠ AN EARLIER VERSION OF THIS FOOTNOTE SAID THE OPPOSITE IN BOLD: that the two 2004 entries were one payment rejected and re-sent, and “must not be added”. That reading came from the description text of a $25,000 Book Transfer Credit on the same sheet marked “REVERSAL OF ENTRY DD06/24/04 … BECAUSE NEED BBK OR VALID 9 DIGIT ACCT NUMBER TO CREDIT”. The balance column refutes it: that credit posts BEFORE the 24 June debit ($1,167,075.07 precedes $1,137,075.07), and a reversal cannot precede the entry it reverses. It is the only reversal in the statement period, and no credit anywhere in the period reverses either payment. The narrative said “reversal”; the numbers said otherwise. Read the ledger, not the description.

   ⚠⚠ AND THAT SCAN WAS OF THE WRONG SPACE. After the 2004 pair the desk swept the identifier range EFTA01482496 to EFTA01482511 and reported the name absent from the rest of it. The 2003 payment is on a different account in a different production, and the identifiers holding it were printed in the pre-verification’s own hazard section and listed in the desk’s seed queue. Four passes had them named in front of them and left them closed. An identifier neighbourhood is not a document class, and the sweep that would have found it is by class.

   ⚠⚠ THE IDENTITY IS A QUESTION AND THE PAGE SAYS SO IN THE BODY, NOT ONLY HERE. A bare beneficiary name on a bank statement does not establish which person of that name. The 30 June entry routes to a private bank in St James’s, London, which is consistent and is not proof. The page treats the identification as [Q] throughout. ⚠ Account numbers and the banks’ internal sequence numbers are on these sheets and none is reproduced; the sequence field is labelled in a way that resembles a personal identity number and is not one. ⚠ A 2005 telephone-message log also lists him, EFTA01308309 p.6, text layer. ⚠⚠ THE SET IS NOT KNOWN TO BE COMPLETE AND THE PAGE SAYS SO. 2005 to 2008 has not been swept, nor have Epstein’s other entity accounts, and the search depends on a surname surviving OCR on a statement page in a corpus that mangles surnames routinely. ⚠ A second, independent sweep of every date header in all 11,720 pages of the population returned no further in-gap material. ⚠⚠ But that sweep cannot bound this. It read Date:/Sent: header lines, and a bank statement is a ledger table with no such header, so the sweep was blind to exactly the shape of evidence that produced this finding. It is corroboration, not a bound, and the page claims no measured extent for the gap. ↩ ↩² ↩³

4. Epstein’s line-edits, 20 December 2009. From Jeffrey Epstein to “PETER MANDELSON”, subject EDITS, timed 13:15:11 +0000. EFTA00743941 p.1, read at the image at 700 dpi. ⚠⚠ THE FORWARDED ORIGINAL IS A SEPARATE DOCUMENT AND AN EARLIER VERSION OF THIS PAGE CITED NEITHER IT NOR ITS TIMESTAMP. It is EFTA00766661 p.1, twin at EFTA02433709 p.1, read at the image at 700 dpi: From: “PETER MANDELSON” / To: Epstein’s personal address / Subject: “Fw:” / Date: Sun, 20 Dec 2009 12:24:24 +0000, above the line ”--- On Sun, 20/12/09, PETER MANDELSON wrote:”. Every pre-edit wording quoted on this page comes from that sheet; EFTA00743941 carries only the returned text. 12:24:24 to 13:15:11 is fifty minutes and forty-seven seconds, and both are produced headers on the same clock. The edits are highlighted in yellow on the sheet and number sixteen, not the “roughly twenty” an earlier version gave; the four quoted here are examples, not the whole. THIS IS RIGHT becomes I AGREE, and further insertions appear at “INSTEAD THEY WILL RIDE EASILY ON BACK OF THE MOOD FOR CHANGE”, at “AGAINST THIS, WE ARE LEFT TO HOPE FOR BRIGHTENING ECONOMIC PROSPECTS AND POSITIVE GROWTH WITH THE CHANCE OF A ‘BETTER THE DEVIL YOU KNOW’ CONCLUSION”, and at a further point in the closing paragraphs. ⚠ The recipient address is barred and is not supplied, and no other copy is consulted for it; Epstein’s own address prints and is not reproduced. ⚠ The page does not say the edits were adopted, and nothing cited here shows that they were. ⚠ This document was cited by no page on this wiki before 2026-09-19, and it was the verification pass that found it, after the page had been built on a weaker act. ↩ ↩²

5. The drafted statement, 13 March 2011. EFTA00638085 p.1 and EFTA00638092 p.1, both read at the image at 700 dpi. The first is from Peter Mandelson to Epstein, timed 17:46:17 +0000, subject “Fw:”, quoting beneath it his own message of 16:54 that day; the second is from Benjamin Wegg-Prosser at a Global Counsel address, timed 18:11:58 +0000, and the statement text is quoted in full, including its opening quotation marks as printed. ⚠ No address, and no other contact value, is reproduced from either sheet. ⚠ The recipient name on the second sheet is barred and is not supplied. ⚠ This page reports the routing and says nothing about the truth of the statement’s content. ↩ ↩²

6. 2002-2003. Messages to Ghislaine Maxwell: EFTA00580646 p.1 (22 September 2002); EFTA00578749 p.1, twin at EFTA02332468 (22 January 2003); The April 2002 message is EFTA00578608 p.1, twin at EFTA02332381 p.1: From: “Peter Mandelson” / To: “G Max” / Date: Mon, 01 Apr 2002 21:40:56 +0000. ⚠ The businessman is named on the sheet and is not named here, because nothing else on this page turns on him. Courier records on Epstein’s shipping account: the November 2002 entry addressed to him at the House of Commons is at EFTA01313812 p.4, and the August 2002 entry running from him to Epstein’s New York address is at EFTA01317758 p.4. ⚠ An earlier version of this footnote gave the second identifier for both. ⚠ Read at the text layer except where a footnote says otherwise. ⚠ The quoted fragments are his own words in messages whose sender field prints. ↩ ↩² ↩³

7. After 2011. EFTA00653241 p.1 (March 2015); EFTA01062531 p.1 and EFTA01062564 p.1 (6 November 2016); EFTA02670723 p.1 (9 November 2016). ⚠ Read at the text layer. ⚠ “Well? What now?” is the SUBJECT LINE of EFTA02670723, and the body is empty but for a device footer and a firm disclaimer. The double space is as printed. ⚠ This sheet uses Sent:-style headers rather than Date:. ⚠ The continuation of the correspondence is a documented fact on its dates; what it signifies is not stated here. ↩ ↩²

8. In office, 2008-2010. June 2009 messages: EFTA00775771 p.1 and EFTA00775779 p.1. On the news alert, 8 November 2009: EFTA00768390 p.1. On write-downs, 1 November 2009: EFTA00768023 p.1. Polling day, 6 May 2010: EFTA02424951 p.1. ⚠ Read at the text layer. ⚠ All quoted lines are from messages whose sender field prints as his. ⚠ A longer line in the polling-day message is not reproduced. ↩ ↩² ↩³

9. The travel invoices, and the ledger that is not reproduced. Two invoices dated 4 and 11 April 2003 are billed in his name by Epstein’s travel agent, in a “Find Report / All Transactions” ledger. EFTA01265994 p.4, read at the image at 700 dpi. A DOJ-OGR-* copy of the same ledger 403s on the public /pdf/ route. ⚠ That is a fact about the route and not about the release, and it is not the reason an EFTA identifier is cited: the EFTA copy is cited because this page does not reproduce the ledger from any copy. ⚠⚠ THE LEDGER IS NOT REPRODUCED AND MUST NOT BE. On the image roughly half the Name column is covered by bars, and the text extraction does not preserve the row structure: names and amounts come out of the extraction silently mis-aligned, so a reader working from searchable text can attach the wrong figure to the wrong traveller and will see nothing to warn them. Only the two rows bearing this subject’s name were read, and only they are reported. ⚠ Nothing is inferred about who paid or why. ↩

10. 17 December 2009, three days before the memorandum. EFTA00743792 p.1, From: Jeffrey Epstein / To: PETER MANDELSON / Date: Thu, 17 Dec 2009 05:33:54 +0000, subject blank. Copies at EFTA00886687 p.1, EFTA01821158 p.1, and EFTA02027662 p.1, the last an FBI Electronic Communication. ⚠ The two first names on the sheet are not expanded there and are not completed here. ⚠⚠ A message sent at 09:14:41 the same morning, replying in the same sequence, CARRIES A BLACKED-OUT SENDER FIELD AND IS NOT ATTRIBUTED TO ANYONE ON THIS PAGE, OR QUOTED. ↩

11. The election-strategy memorandum, 20 December 2009, sent by him to Epstein on the day he wrote it. ⚠ An earlier version of this footnote carried no document identifier; the memorandum’s text survives in two sheets and the page quotes both: the forwarded original he sent, EFTA00766661 p.1 (twin EFTA02433709 p.1), which carries the pre-edit wording, and Epstein’s returned copy at EFTA00743941 p.1, which carries the edits. Both read at the image at 700 dpi. ⚠ An earlier version cited only the returned copy, and so sourced the before-text to a sheet that does not contain it. ⚠⚠ AND AN EARLIER VERSION SAID ITS ADDRESSEE WAS BLACKED OUT. IT IS NOT. The quoted-message format carries no addressee line at all, and the bar on that sheet covers his own address. An absence had been mislabelled as a redaction, which invents a decision by the producer that the producer did not make. The text distinguishes its reader from “Alistair” and places its reader, with “Alistair”, inside the Pre-Budget Report handling; that is evidence a reader may weigh, and this page names nobody. ⚠ Only the lines bearing Epstein’s edits are quoted. The rest is party strategy of no evidentiary bearing here. ↩

12. Banking policy, January to March 2010. Staley’s question and Epstein’s reply, 27 January 2010: EFTA01300443 p.1 and EFTA01824159 p.1. Staley to Mandelson with the Volcker speaking points, Tue, 30 Mar 2010 14:26:24 -0400, forwarded to Epstein the same evening at 20:08:15 +0000: EFTA02428040 p.1, copies at EFTA01826780 and EFTA00889933, p.1 each. ⚠⚠ The 08:15 message is not on this thread. EFTA02427999 p.1 carries the subject “Re: Fwd: Attachement: Tom Ford’s House” and quotes beneath it Epstein’s forward of 03:40:37 -0400 that morning on that subject. Its proximity to Staley’s note is a proximity in time only, which is a further reason the page leaves “Larry” open. Mandelson to Epstein, 31 March 2010 at 08:15: EFTA02427999 p.1. ⚠ Read at the text layer. ⚠⚠ “Larry” is not identified on the sheet and this page does not complete it. Larry Summers is named in the same correspondence eight weeks earlier and in Staley’s note of the previous day; that makes the identification a reasonable question and not a documented fact, and the page keeps it in the weaker form. ⚠ “Chancellor” is likewise not expanded to a name on the sheet. ↩ ↩²

13. 13 May 2012, the anchor act repeated. He forwards a draft newspaper article co-written with a serving shadow chancellor and asks what Epstein thinks; Epstein criticises it, writes “I would [l]ike you to be bold”, and in the same thread asks “I think you should consider introducing me to anyone you think could be a huge business participant.” Seven documents, all with produced +0000 headers: EFTA00647710 p.1, EFTA00719354 p.1, EFTA00719359 p.1, EFTA00934936 p.1, EFTA00719364 pp.1-2. ⚠ An earlier version of this page presented the December 2009 memorandum as singular. ⚠ Whether the article was published in the form Epstein saw is not recorded here. ↩

14. Two further forwards of his own work, June and July 2010. The speech: EFTA02411628 p.1, From: PETER MANDELSON, Subject: “Fw: Re: Your speech in Lujiazui Forum”, “please review - good for business world ?”, forwarding beneath it his own instruction to a speechwriter; the speech itself is EFTA02411630, sixteen pages, every one headed EMBARGO UNTIL DELIVERED. The Glencore letter, 16 July 2010: EFTA00739722 p.1, Subject: “Fw: Future”, forwarding his own letter of that morning to the company’s chief executive seeking a board seat ahead of the flotation; Epstein’s reply is EFTA00628319 p.1, in full “Number to call?”. ⚠ Both read from the text layer, not the image, and nothing is quoted from either enclosure beyond the embargo legend. ⚠ Whether Epstein did anything in response to either is not recorded. ↩

15. The job search, 2010. Memoir drafts, June 2010: EFTA00747689, EFTA00747697, EFTA00747715, all p.1. The dinner note, 24 June 2010: EFTA02411106 p.1. The prospectus, 3 August 2010: EFTA00735268 pp.1-2; the “money trail” line, 5 August 2010: EFTA00735628 p.1. On JPMorgan, 25 December 2010: EFTA01832618 p.1; Epstein’s reply the same evening at EFTA00901039 p.1. ⚠ A third copy has a corrupt text layer in which the quoted-printable soft breaks survive as literal characters and “JPM” reads as “WM”; it is not cited. ⚠ Only two sentences of a four-paragraph message are quoted; the rest concerns a named retainer and is not reproduced. ⚠ Read at the text layer. ⚠ Third parties named in these messages are named only where the sheet names them, and no inference is drawn about whether any approach succeeded. ↩

16. July 2014. EFTA00993821 p.1 and EFTA00993827 p.1, rendered and read. ⚠ The person Epstein names as making the request is named on the sheet and is not reproduced here, because the approach is already framed on DP World and this page does not loosen that page’s handling. This exchange is new to the wiki and belongs on both. ↩

17. 16 March 2011. His reply, “Categorically untrue. I will say that but not sure you sd comment.”: EFTA00907114 p.1 ONLY: From: Peter Mandelson … Subject: “Re: Fwd: FW: Question from the Daily Mail RE Jeffrey Epstein” … Wed, 16 Mar 2011 15:46:11 +0000. Epstein’s “I am saying nothing, where are they getting this?”: EFTA00907119 p.1. ⚠⚠ CITE p.1 AND NOT THE RUN. Later pages of this run carry contact values belonging to a journalist, and this page neither reproduces them nor says which values sit on which sheet. ⚠⚠ A CORRECTION OF A CORRECTION, AND THE WORST ERROR THIS PAGE HAS HAD. An earlier version of this footnote carried no identifier at all; the version that fixed that supplied three identifiers which were all wrong. They are pages of a press-clipping compilation about a different man and a survivor’s allegations, and they matched only because the phrase “categorically untrue” appears there as a palace denial of allegations concerning minors. A correction had pinned this man’s quotation to survivor-allegation material about someone else: allegation by adjacency, manufactured inside a citation. Caught by the verification pass, never served. ⚠ Read at the text layer. ⚠ One of the two onward forwards preceded his reply rather than following it; the page no longer implies a single ordered sequence. ⚠⚠ What the reporter put is the newspaper’s assertion, sourced in its own words to having “been told”. It is reported here only as a question asked and an answer given, which is the tier the document supports. This page does not restate the allegation as though the release established it, and does not reproduce its content. ↩ ↩²