Epstein Data Wiki

Jacqui Safra

Properties2
tagsperson, edge-foundation, science-funding, property, napa, jerusalem
aliasesSafra, Jacqui, Jacqui Safra, Jaqui Safra, Jackie Safra, Jacque Safra, Jacgui Safra, J.E. Safra, J E Safra, j e safra, J.E. (Jacqui) Safra, jaqui

58 min read

Named on the documents as”Jacqui Safra” in a subject line and in the body of messages about him; “j e safra” in the sender field of his own e-mail, and again in the forwarded header blocks of the copies Epstein’s assistant passed on; “J.E. (Jacqui) Safra” in the circulation lists of Edge Foundation circulars, the form in which a sheet joins the two names; “J.E. Safra/Doumanian” as a single quoted display name in the copy lists of six further Edge circulars; “Jacqui” alone in staff correspondence about arranging a call; and “jaqui” in his own sign-off12345
What a document says he is”Swiss Investor, Chairman, Encyclopædia Britannica; Owner, Spring =ountain Vineyards” on the guest list of a New York dinner of February 2011, and “Investor, Art Collector; Encyclopædia Britannica, Spring Mountain Vineyards” on an Edge invitation list of June 2011. Both are John Brockman’s descriptions of a guest on his own list67
His own words in the releaseTwo messages, both of 24 March 2014, both about a telephone call with Jeffrey Epstein. The first runs three printed lines, the second of which reads “yes he can call me all morning,or in afternoon 2 30 pm to 5 30 pm about…” and the third “jaqui”; the second message reads “yes,i am in ny”12
The intermediary in every strand read hereJohn Brockman, president of the Edge Foundation, Inc., who supplied Epstein’s office with the contact route, wrote the property material, and reported the dinners389
Properties a document attributes to himThe Sherover Residence in the Talbiyeh neighbourhood of Jerusalem; 88 acres at Calistoga, California; Spring Mountain Vineyard at St. Helena, Napa County; and a Nevada ranch — all four named in one memorandum written by Brockman, not by him8
Documents cited here run from20 July 2009 (Peggy Siegal to Epstein, naming him in a party of three that had left Portofino that morning and was driving to London) to 13 November 2016 (Brockman to Epstein, writing that Safra “spring for” a Paris Edge event in 2009 — his word, and no sheet read here records a payment)1011
What no sheet read for this page showsHim writing to Epstein about any property; any signed instrument, payment, transfer or completed transaction; and any attendance by him at any event, as distinct from his name on a list8412

Jacqui Safra is, in John Brockman’s description on the only two sheets read for this page that describe him at all, a “Swiss Investor, Chairman, Encyclopædia Britannica; Owner, Spring =ountain Vineyards” (the = is a soft break on the sheet). That is Brockman’s label for a guest on his own Edge Foundation list, no sheet read here carries a title record behind it, and it is the nearest thing to an identification these documents supply. His name reaches 162 documents in the release under the spellings set out below, and his own words survive in two of its messages. Both are short messages of 24 March 2014 arranging a telephone call with Jeffrey Epstein, and both carry his own sender field. Everything else read for this page is other people writing about him. The second of the two descriptions names the same two institutions beside him, and Brockman writes the rest of it as well: offering Epstein his Jerusalem house, describing his properties in a four-page memorandum, reporting a dinner at which Safra was “scrambling because his loan at one of the big banks is not getting renewed”, holidaying “at Safra’s place” in Napa, and naming him on the circulation lists of Edge invitations to a conference held at his own vineyard.121389144

Nothing read for this page alleges wrongdoing by him. Offering to sell a house, giving a window for a telephone call and being named on a guest list are ordinary acts, and the page says so rather than letting proximity to the subject of this wiki do an allegation’s work.

One thing has to be said before any of it, because it is the reason this page exists in the shape it does. The desk’s own instrument for asking whether a record shows a person acting, act_rate.py, returns zero sender-field acts on the phrase "Jacqui Safra" when the name key given to it is Jacqui. That zero is a fact about the query, not about the man: his mailbox does not print “Jacqui Safra” in its sender field, it prints j e safra. Run over the same 101 documents with the key Safra the figure is 2; run over the union of all his name forms it is 8 documents in a sender field, 6 of them with the key at the head of the field. The arithmetic is set out under What the search returned, and the general lesson is the one the tool’s own documentation gives: being the top-level sender is evidence of acting, and not the only way to act.1215

How the record names him, and how the two halves were joined

The two name forms that matter, the one people use about him and the one his own mail prints, do not share a token, and no generator of damaged spellings reaches from either to the other. They are joined on the face of one class of document. The circulation list of the Edge Master Class Update of 12 May 2011, forwarded to Epstein the next day, runs:

TO: Paul Allen, Jeff Bezos, Stewart Brand, Sergey Brin, David Brooks, George Church, George Dyson, Ari Emanuel, Jeffrey Epstein, Tony Fadell, Bill Gates, W. Daniel Hillis, Bill Joy, Daniel Kahneman, Salar Kamangar, Dean Kamen, Elon Musk, Nathan Myhrvold, Pierre Omidyar, Tim O’Reilly, Larry Page, Sean Parker, Jean Pigozzi, Ricardo Salinas Pilego, Nicholas Pritzker, J.E. (Jacqui) Safra, Charles Simonyi, Jeff Skoll, Anne Treisman, Craig Venter, Jimmy Wales, Evan Williams, Mark Zuckerberg4

J.E. (Jacqui) Safra appears on 23 documents in the release, and it is the identification: one sheet, both forms, one person. Six further Edge circulars carry the two names written as one quoted display label, "J.E. Safra/Doumanian", in their copy lists. What that label is, on the sheet, is one entry in a mailing list. No document read for this page states a relationship between the two people it names, and this page supplies none.4512

What a document says he is

Two Edge Foundation participant lists give him a description, and they are the only sheets read for this page that do. Both were written by John Brockman in 2011.

The first is a dinner list. On Wednesday 9 February 2011 Brockman circulated the details of a dinner to be held at Le Cirque in New York on Wednesday 16 February, with a guest list running to several pages. The entry reads:

Jacqui Safra Swiss Investor, Chairman, Encyclopædia Britannica; Owner, Spring =ountain Vineyards

followed by a link to a biography page on the Edge Foundation’s own website. The entry immediately below it is headed ”& Jean Doumanian” and describes her as “Former producer, Saturday Night Live; Broadway & Movie Producer”. The = is a quoted-printable soft break on the face of the sheet; the aesc ligature in “Encyclopædia” is printed. He is not only described on that sheet, he is addressed by it: the To: line names him, as J.E. Safra, among twenty-one recipients.6

The second is the status list of the 2011 Master Class. On 2 June 2011 Brockman wrote to Epstein that “the usual billionaire crowd is not showing up for this one” and that “The list status is below.” The sheet that follows gives one line per invitee, and his reads:

J.E. (Jacqui) Safra, Investor, Art Collector; Encyclopædia Britannica, Spring Mountain Vineyards7

These two lines are the strongest documentary statement in the release of who he is, and they are Brockman’s description of a guest on his own list rather than anything Safra wrote. The first of the two is also where the word “Owner” is attached to the vineyard beside his name. That word is printed, and it is not an inference from an address; it is also Brockman’s label for a guest on his own list, and no sheet read for this page carries a title record behind it. (The sheet prints “Owner, Spring =ountain Vineyards”; the = is the soft break and is not repaired here or anywhere else on this page.)

The status list carries its own key, and that makes one mark worth reporting. Its heading reads “EDGE MASTER CLASS 2011 — INVITATION LIST (√ confirmed; x regrets; * maybe)”, and most lines carry one of the three marks in the left margin. His carries none of them. What that records is the state of his reply on 2 June 2011, six weeks before the class: not confirmed, not declined, not a maybe. It is not a record of whether he went, and this page makes no claim either way; a list of responses and a record of attendance are different documents, and only the first is here.7

24 March 2014: the call

Nine messages are narrated below, all of them on Monday 24 March 2014, with Brockman’s of the previous Thursday making ten. The strand of it that ran between Groff and Safra, every copy printing her opening line to him, survives in 13 documents, which are copies of one another and not separate events; the reminder to Epstein, his answer to it and Brockman’s message standing alone are three further documents outside that 13.12 It begins with Epstein’s office, not with him.

The clocks first, because the order of events depends on them. Some copies carry an RFC header (Date: Mon, 24 Mar 2014 15:17:17 +0000) or an explicit EDT stamp; others carry a bare Sent: field with no zone. In this subject’s own documents the bare field is UTC, and that is measured rather than assumed: six messages survive in both shapes. Three of the six carry a bare field running exactly four hours ahead of the same message’s EDT stamp; the other three carry a bare field whose clock value is identical, to the second, to the same message’s explicit +0000 header, which is the shape no quoting client can explain away. One of the six settles it outright. The sheet carrying Brockman’s message of 20 March as Sent: Thursday, March 20, 2014 8:24 PM also prints, in an Apple property-list block below the signature, a machine date-received value of 1395347016, which is 20:23:36 UTC on 20 March 2014. All EDT times below are converted on that basis, and the footnotes give the printed field for each.16

At 7:36:09 AM EDT on Monday 24 March 2014, Lesley Groff sent Epstein a one-line reminder, “Reminder: did you want to set up a call with Jacqui?”, above a message Brockman had sent four days earlier, on Thursday 20 March at 4:24:20 PM EDT, under the subject “Jacqui’s Jerusalem residence”. Brockman’s message reads in full:

JE,

Jacqui wants to know if you would like to have a serious conversation about his Jerusalem residence that he discussed with you Monday night. If so:

Beneath it the sheet prints “Jacqui Safra” and then a barred block.17

Epstein answered thirty-eight minutes later: “yes, and Paul allen,”. What the “yes” answers is not in doubt, because Groff had asked one question, and the call was arranged within the next four hours. What the second clause means, and why that name is in it, no sheet read for this page settles.18

Groff wrote to Safra at 9:39 AM EDT: “Hello Jacqui. Jeffrey would like to set up a time to speak with you. Might you be available sometime today or tomorrow? Please let me know.”19 At 10:44:02 AM EDT he answered, and this is the first of the two messages in the release that are his:

hello… yes he can call me all morning,or in afternoon 2 30 pm to 5 30 pm about… jaqui

That copy is a top-level message whose header block prints From: and, beside it, j e safra.1

At 10:46 AM EDT Groff asked him to confirm the zone, “Terrific! Just double checking…is this EST?”, and put to him the number Epstein should use.20 At 11:16:49 AM EDT he answered “yes,i am in ny”, above a line carrying a single lower-case j. That copy, too, is a top-level message under his own sender field.2

Twenty-eight seconds later, at 11:17:17, Groff wrote to Epstein under the subject “Jacqui Safra”:

Please see below..I asked him to confirm this is EST or what time zone, but he has not responded as of yet…the number Brockman gave us for Jacqui is:

and the value after the colon is barred.3 Thirty-seven seconds after that, at 11:17:54, she wrote again under the subject “Jacqui is in NY-EST zone applies!”, with the body “just rec’d this …” above his reply.20 At 11:18:04 she wrote to him: “terrific..thank you!”21

Two things in that sequence are worth separating from the arrangements. The first is Groff’s sentence: the number Brockman gave us for Jacqui. Epstein’s office did not have a route to him; Brockman supplied one. The second is that the two men do not correspond. Every message here is between Safra and an assistant, and the call itself, whether it happened and what was said, is not in any document read for this page.312

The Jerusalem house, 2011 to 2014

On 27 October 2011 Brockman sent Epstein a four-page memorandum under the subject “Safra properties”. Its first line is “Below is the info on Jacqui Safra’s properties for sale”, and what follows is a prospectus for three of them, written by Brockman and attributed throughout to what Safra told him.

The first is the Jerusalem house. Brockman’s summary paragraph is damaged by quoted-printable soft breaks carried onto the face of the sheet, which are not repaired here:

One is the =herover REsidence, the alpha property in Jerusalem close by the =fficial residences of the President and Prime Minister. He’s been =ffering it to Russian oligarch for $100,000,000 but know it will go for =ess.

The pages that follow set out the house, in the Talbiyeh neighbourhood, between Dubnov and Pinsker Streets, built in 1958 by Miles Sherover, largely by quoting a press account, and add a development plan: “PLAN: Build the top luxurious hotel in Jerusalem, with some luxury =partments.”8

Sixteen months later the same house came back, and this time with no figure on it. On 26 February 2013, writing from a hotel room in Long Beach, California, Brockman sent Epstein a message headed “YOUR HOUSE IN JERUSALEM”:

Jacqui Safra is here with me. Wants to sell you his Jerusalem house in Talbieh neighborhood known as “The Sherover House” across st from Jerusalem Theatre (that the Sherovers built and donated to city), and a few doors down from President’s house. Two houses on acre and a half. When built in 50s, the most prestigious house in Jerusalem. He says he was in negotiations with Ronald Lauder but it didn’t happen. It’s now off the market. But for you, a TED special

That message survives in 10 documents; the sentence “Wants to sell you his Jerusalem house” is the phrase that reaches all ten.1312

A year after that it was still unresolved. Brockman’s message of 20 March 2014, the one Groff’s reminder carried, reports that Safra “wants to know if you would like to have a serious conversation about his Jerusalem residence that he discussed with you Monday night”, which places a prior conversation between the two men on a Monday evening and tells us nothing else about it.17

Nothing read for this page records a sale, an offer in writing, a price agreed, a deposit, or any further discussion after 24 March 2014. The searches run to look for one are set out below.12

Napa: the vineyard, the land at Calistoga, and what the documents actually say about whose they are

Spring Mountain Vineyard enters this release first as an address on an invitation, and it would be easy to leave it there. Two sentences written three years apart by the same man take it further.

On 20 August 2010 Brockman circulated a preview of that year’s Edge Master Class and explained why it was thin: “The result is that among the regulars, Stewart Brand, Danny and I spent some time at Jacqui Safra’s Spring Mountain Vineyard in St. Helena”.22 On 21 August 2013 he wrote to Epstein from holiday: “On vacation in Napa (at Safra’s place) with Kahneman and Thaler.” The postscript to the same message reads: “p.s. Got a hundred M for Jacqui’s property in Calistoga - a high-end resort. He spent 15 years trying to get approval. The town just had a referendum and voted approval.”14

The 2011 memorandum treats all of it as his and distinguishes the two holdings. Calistoga is “88 acres on a hill acquired in 1998”, acquired, Brockman writes, on the advice of Adrian Zecha, and “about 10 miles from Jacqui’s Spring Mountain Vineyard in =t. Helena”. The vineyard is not for sale: “His =ig investment is Spring Mountain (900 contiguous acres). The Vineyard =oses money every year but the and is about $300,000 an acre.” The = characters and the “but the and” are on the face of the sheet.8

Two cautions the page keeps rather than smoothing over. Calistoga and St. Helena are different places about ten miles apart, and the memorandum says so; a reader should not fold “Safra’s place”, “Jacqui’s property in Calistoga” and “Spring Mountain Vineyard” into one address. And every one of these sentences is Brockman’s. No sheet read for this page carries a deed, a title record, a company filing or any statement by Safra himself about any of these properties. What the documents establish is that the man who supplied Epstein’s office with the route to him described the vineyard as his in writing on four separate occasions across three years: the Master Class preview of 20 August 2010, the dinner list of 9 February 2011 (in the flat form “Owner, Spring =ountain Vineyards”), the invitation list of 2 June 2011 and the properties memorandum of 27 October 2011. In the last of those he invited Epstein to “stay at =he mansion at the Vineyard for a couple of nights.”8221467

Edge: the lists, the venue, and one figure

The invitation gives the 2011 Edge Master Class a “Where:” of “Spring Mountain Vineyard, St. Helena, Napa, CA (Lodgings at Harvest Inn, 2.5 miles away)” and a “When:” of “July 15-17, 2011”, with Michael Gazzaniga, Martin Nowak, Steven Pinker and Vilayanur S. Ramachandran teaching. Safra is on the circulation list. He is on the list; the list does not say he came. The desk’s standing rule is that an invitation establishes inclusion and not attendance, and no sheet read for this page records his presence at that class or any other.412

The vineyard was the venue again four years later, and the document that says so is not a Brockman circular but Epstein’s own printed schedule. Four sheets of it in the House Oversight production carry the line, three of them under the day heading “Fri. July 31, 2015 PB-Zorro”:

Reminder: EDGE Master Class 2015 - Invitation “A SHORT COURSE IN SUPER-FORECASTING WITH PHILIP TETLOCK” Friday, July 31 - Sunday, August 2, 2015 Spring Mountain Vineyard, St. Helena, Napa (Hotel: Harvest Inn, St. Helena, Napa)

None of the four carries his name in any form, and none is reachable by any query on it; they were found by the venue and not by him. What they add is a second year and a different class of document for the same proposition: that the vineyard the lists call his is a working venue inside the Edge network. What they add about who attended, his own presence included, is nothing at all. They are reminder entries on a schedule printed for Epstein by a legal assistant, not a record of anyone being there.23

Two documents put money beside his name, and both are Brockman’s report rather than a record of payment:

  • 16 June 2010. Describing a plan to rebuild the Edge website: “Need backing. Jacqui is in for $25K. Interested?”9
  • 13 November 2016. Writing about a trip to Paris: “I haven’t been there since 2009 when Jacqui spring for a French EDGE event at the Ritz featuring a talk by Stanislas Deheane, Europe’s leading neuroscientist”. The verb is as printed.11

Neither is a receipt. They are two statements by the organiser of the Edge Foundation that Safra had put money into Edge projects, made six years apart; nothing read for this page confirms either payment. This page does not restate the science-funding material set out at Martin Nowak, Lawrence Krauss, Robert Trivers and Barnaby Marsh; the vineyard is a venue inside that network and those pages are where the funding runs.

16 June 2010: a loan, a request, and a reply whose referent is not settled

At 20:37:52 on 16 June 2010 Brockman wrote to Epstein under the subject “Jacqui Safra”. The message has four paragraphs and two asks in it:

Had dinner with Jacqui last night, whose is in town scrambling because his loan at one of the big banks is not getting renewed. Something for you??

He took us out to celebrate Max’s 30th birthday (today!!).

Saw Ronna Tanenbaum in SF last week. (Remember her?) She’s going to put together a team to re-engineer the EDGE website and create a new Internet platform. Need backing. Jacqui is in for $25K. Interested?

“whose” is as printed. Two minutes and forty seconds later, at 20:40:32, Epstein replied, and his whole message is:

im always in. whatever you need.

It has been reported that Epstein agreed to finance Safra. The documents read for this page do not establish that, and the reason is structural rather than a matter of degree. A five-word affirmative follows a message containing two distinct requests, a reported loan problem ending “Something for you??” and a website project ending “Need backing. Jacqui is in for $25K. Interested?”, and nothing in the reply says which it answers. If anything the wording points the other way, and in two respects. “im always in” echoes “Jacqui is in for $25K”, which belongs to the website paragraph and not to the loan paragraph. And the reply’s second person points the same way: the words “whatever you need” are addressed to Brockman, and the only need Brockman states anywhere in the message is his own, “Need backing”, in that same paragraph, whereas the loan paragraph reports Safra’s difficulty and offers it to Epstein as an opportunity, “Something for you??”, rather than as anything Brockman needs. Both the lexis and the grammar of the reply therefore point at the website and not at the loan. Neither settles it. The honest statement is that the referent is undetermined, and that no sheet read for this page records any money moving in either direction.92412

Elsewhere in the record

  • 20 July 2009. Peggy Siegal wrote to Epstein: “Jaqui Safra, Jean Doumanian and I left Portofino this morning- drove to Genoa and flew into Farnborough…now driving to London.” The chain survives in 7 documents, found by the phrase "left Portofino this morning". Four of the seven extract the name cleanly as Jaqui Safra and are inside the union below. The other three print the same name on the sheet and are reachable by no query on it at all, because their extractions read the surname “Safia”, “Sofro.” and “Salm”. All three were read at the image; all three print “Jaqui Safra”.10
  • 14 September 2015. Brockman to Epstein, subject “Safra in town - got me into Cornell Weill”: “Being met at ER by top cardiologist. All set.”25 The sheet gives no forename, and this page does not supply one; it is recorded because it is the latest dated item in the material read here that uses the surname in this correspondence.

What the search returned

The matching rule for every figure in this section, without exception: an FTS5 phrase query over the full-text index, counted in documents, not pages. FTS5 ignores punctuation and case, so "J.E. Safra", "J E Safra" and "j e safra" are one query returning one number, and "J.E. (Jacqui) Safra" and "J.E. Jacqui Safra" likewise. The consequences of that rule are not all convenient, and one of them cost the largest correction on this page: see the note on the second reply, below.

The name forms

formdocumentsnew to the unionnote
"Jacqui Safra"101101
"Jackie Safra"1313a human variant, not damage
"Jaqui Safra"1313the spelling of his own sign-off
"J.E. Safra"3229his own sender field; only 3 overlap the Jacqui forms
"Jacque Safra"33no overlap with anything
"Jacqui Satre"10not a spelling. See below.
"Safra Doumanian"71= "Safra/Doumanian", two names as one display label in a copy list
"Jacgui Safra"11q→g; found by name_forms.py, verified REAL
"Jacqui afra"11the sheet prints Jacqui =afra
union162

Two of the artefacts in this material turn his surname into “Satre”, and neither is a spelling. "J.E. Satre" returns one document, EFTA01799443, whose To: line prints J.E. Safra at the image.6 And "Jacqui Satre" is the same kind of thing. It returns one document, EFTA01930185, and that sheet, read at the image at 600 dpi, prints Subject: Jacqui Safra. The row is kept in the table because a reader re-running the query will get the hit; it is excluded from the union because the one sheet that produces the hit does not print it.26

ocr_variants.py --wide --count on "Jacqui Safra" generated 31 single-edit forms and returned two non-zero: Jaqui Safra (13) and Jacqui afra (1), both already above. name_forms.py on the surname returns no variant form at all, which its own output calls a real answer rather than a failure, and on the forename returns one, Jacgui (1 document), which --verify classes REAL. The forms that actually carry him and that no generator reaches are the human ones, Jackie and Jacque, and, in the other direction, the sheets whose extraction is too badly damaged for any form to match: the three Siegal copies print “Jaqui Safra” and are extracted as “Safia”, “Sofro.” and “Salm”.1210

Clean negatives, measured, each returning zero documents: "Sweetland Films" · "J.E. Beaucaire" · "Beaucaire" · "Jacob Safra" · "Moise Safra". The film and pseudonym strand of his outside biography is simply not in this release, and this page does not imply that it is. "Encyclopaedia Britannica" returns 5 documents and "Britannica" 45. Two of the 45 fall inside the union, and both are the Edge participant lists quoted above under What a document says he is ‑ they are how Britannica enters this page at all. Neither of the five "Encyclopaedia Britannica" documents is one of them, because the two sheets print the aesc ligature and the query does not. "Republic National Bank" returns 6 and "Edmond Safra" 2, with 0 and 1 inside the union respectively.12

Reading past the union

Seventeen further documents read for this page concern him and print no form in the table above. They were found by reading the surname residual and sampling the forename residual, not by any query on his name. But every one of them is reachable by a phrase printed here, so a reader can re-derive the set rather than take the reading on trust:

  • 8 carry Brockman’s line “On vacation in Napa (at Safra’s place)”. All eight carry his Calistoga sentence as well; five of them print it in the form quoted under Napa above, “Jacqui’s property in Calistoga”, and the other three carry the same sentence with his name damaged in extraction. Route: "at Safra's place" (8 documents, none of them in the union).
  • 3 are copies of the Siegal chain of 20 July 2009 whose sheets print “Jaqui Safra” and whose extractions do not. Route: "left Portofino this morning" (7 documents, 4 of them in the union).
  • 2 print a damaged subject form the union cannot reach: EFTA01750543 (“Jac ui Safra”, where the sheet itself prints “Jacqui Safra”) and EFTA01976399 (”.E. Safra”). Route: row 9 of the surname decomposition below.
  • 2 carry Brockman’s Paris sentence with the forename only. Route: "Jacqui spring for" (2 documents).
  • 1 is a guest list whose extraction welds his surname to the next name. Route: "Jacqui Safrallean" (1 document).
  • 1 is the “Safra in town” message of 2015. Route: "Safra in town" (1 document).

So the population reachable by every route read for this page is 179 documents, and the name-form union of 162 is a floor, not a total. Eleven of the seventeen fall inside the surname population decomposed below; the other six do not contain the token Safra anywhere in their extraction. The bound on all of this is the forename residual, and it is stated rather than implied: "Jacqui" returns 172 documents, 110 of them inside the union and 62 carrying no name form in it at all. Nine of those 62 are among the seventeen above. The remaining 53 were not read in full for this page and nothing here is a claim about them.12

The surname is a different subject, and mostly a different man

"Safra" returns 366 documents. They decompose exactly, each document assigned to the first rule below that matches it, over whitespace-normalised full text:

assigned todocuments
the subject, by a name form in the union161
Steven Safra / Steven R. Safra / Steven Russell Safra — a Florida attorney of Cole, Scott & Kissane, counsel for Alan Dershowitz146
the same attorney, under forms outside those three queries (MR. SAFRA, Steve Safra, auen R. Safra)8
Safra Catz of Oracle — where Safra is a forename8
Andrew Saffir, rendered Andrew … Safra by the extraction11
the Safra banking family and institutions named for it (Edmond, Lily, Joseph, Deborah; Safra Bank; the Edmond and Lily Safra Center for Brain Sciences)11
the subject, by possessive only — “at Safra’s place”8
a membership-list marketing proposal in unrelated real-estate correspondence8
the subject, by a damaged name form2
algebraic notation, sAfrA, not a name at all1
the subject — Brockman, “Safra in town”, September 20151
unresolved: a handwritten sheet whose extraction reads SAfra-41421041
total366

161 + 146 + 59 = 366. The attorney’s population is very nearly as large as the subject’s and has zero overlap with it, and one of the 162 subject documents, the one printing Jacqui =afra, does not contain the token Safra at all, which is why the first row reads 161 and not 162.12

A message in this release refusing business with “the Safras” is not about him unless something on its own sheet makes it so, and no sheet read for this page does. The Safra banking family is a separate and much larger subject; this page supplies no relationship between him and it, because the documents read here state none.

The act-versus-mention measurement, re-derived

act_rate.py --selftest was run first and printed PASS. Over the union of all nine name forms, 162 documents, the tool returns:

documentsof 162
sender-field act (loose: key anywhere in the field)85%
sender-field act (strict: key at the head of the field)64%
addressed only42%
header-less, authored32%
header-less, press00%
neither14791%

The eight are all the March 2014 exchange. Six print j e safra at the head of the sender field; two are forwarded chains carrying Groff’s header first and his second. The same tool, on "Jacqui Safra" alone with the name key Jacqui, returns 0, and that zero, not the 8, is the number that was on record before this page was written. The gap is entirely the key: his mail does not print the word “Jacqui” in a sender field, and the form that does print there, J.E. Safra, only entered the query set when someone looked up what the initials stood for.1512

Read as a proportion the record is still overwhelmingly one of mention: 147 of 162 documents are neither authored by him nor addressed to him. Most of them are Edge circulation lists that carry his name beside thirty others. But the proportion is the wrong test for this subject, and the page says so plainly: a man whose only correspondence in a private production is a two-hour exchange with someone’s assistant will always score low, and what he is doing in the other 147, being offered to Epstein as a seller, a host, a funder and a fellow guest, is visible only by reading them.

What this page does not claim

  • That Epstein agreed to finance him. Set out above; the referent of “im always in. whatever you need.” is undetermined on the face of the exchange.924
  • That any property or lending transaction completed. Searches for a completion would have to return a document that says one occurred. "Sherover" returns 11 documents and "Wants to sell you his Jerusalem house" 10; all eleven were opened, and every one is either a copy of Brockman’s February 2013 message or the October 2011 memorandum. "Safra properties" returns 1 (the memorandum itself, and no copy of it), "Jacqui's Jerusalem residence" 3 and "serious conversation about his Jerusalem" 2, all copies of material already described here. No deed, contract, closing statement, escrow instruction, wire record or title document naming him was returned by any of those searches, and the disconfirming document would be exactly that: a sheet reading, in its own words, that a sale or a loan closed, with parties and a date. The claim is bounded to those searches and is not a claim about the whole release.12
  • That he attended anything. An invitation list establishes inclusion. The disconfirming document would be a sheet placing him somewhere on a date: a manifest, a car list, a photograph caption, a message saying he was there. Brockman’s two messages placing himself at the vineyard and at Safra’s place in Napa are the nearest thing read here, and neither says Safra was present.42214
  • Anything about his biography, family, wealth or holdings beyond what a document states and attributes. The figures in the 2011 memorandum ($3 billion, 130,000 acres, $300,000 an acre, $100,000,000) are Brockman’s report of what Safra told him, and are carried here in that grammar and no other.8
  • That any class of material about him is absent from the release. No search run for this page audited the release by document class, and a zero from a name-scoped pass is not a finding about what the release contains.

No link below asserts a relationship beyond what the cited sheet shows.

  • People named on sheets cited here: Jeffrey Epstein (the correspondent in every strand); Lesley Groff (arranged the call of 24 March 2014, and wrote to him directly); Peggy Siegal (names him, 20 July 2009); Martin Nowak (named on the 2011 invitation as teaching the Master Class at the vineyard); Robert Trivers (co-recipient of the February 2011 dinner invitation).
  • Where the Edge science-funding material this page does not restate is set out: Lawrence Krauss; Barnaby Marsh.

Footnotes

  1. E-mail, sender field j e safra, Sent: Mon 3/24/2014 2:44:02 PM (bare, UTC — see [^zones]; the same message’s EDT stamp is at [^groff1117]), Subject Re: Jeffrey Epstein, quoting Lesley Groff’s message of 9:39 AM. EFTA02110904 p.1, the whole run. Read at the image at 400 dpi. The sheet prints the label From:; the extraction of this document reads it Fran:, which is why the page states the label from the image and not from the text layer. The To: value is barred, and the sender value in the quoted 9:39 AM attribution line is barred, as is a block below “Thank you,”. The body is quoted here in full and nothing on the sheet follows “jaqui” except those quoted lines. The forms "Jacqui Satre" and <EM= that appear in the extractions of other copies are not on this sheet. ↩ ↩2 ↩3 ↩4

  2. E-mail, sender field j e safra, Sent: Mon 3/24/2014 3:16:49 PM (bare, UTC), Subject Re: Jeffrey Epstein, body “yes,i am in ny” above a line carrying a single lower-case j, and quoting Groff’s message of 10:46 AM and the earlier exchange. EFTA02109846 p.1, the whole run. Read at the image at 400 dpi. The To: value is barred. The sign-off is a lower-case j on the image, on this sheet and on EFTA00372330 p.1; the extractions of both read it as a capital J. Groff’s 10:46 AM line is quoted here only as far as “is this EST?”; the line continues “Is the number Jeffrey should call you on:” and the value after the colon is barred on this sheet. ↩ ↩2 ↩3 ↩4

  3. E-mail, Lesley Groff to Jeffrey Epstein, Date: Mon, 24 Mar 2014 15:17:17 +0000, Subject “Jacqui Safra”, forwarding his message of 10:44:02 AM EDT. EFTA00372329 p.1, the whole run. Read at the image at 400 dpi. The forwarded header block prints From: j e safra, Subject: Re: Jeffrey Epstein, Date: March 24, 2014 10:44:02 AM EDT, To: Lesley Groff, with the two address values barred; the extraction reads the date “March 24. 2014” with a full stop, and the sheet has a comma. The value after “the number Brockman gave us for Jacqui is:” is barred on this sheet. Groff’s signature block prints “Lesley / Assistant to Jeffrey Epstein” above a barred line. ↩ ↩2 ↩3 ↩4

  4. E-mail forwarding John Brockman’s “Edge Master Class Update” of Thu, May 12, 2011 at 6:47 PM to Jeffrey Epstein; the covering header is Date: Fri, 13 May 2011 15:54:56 +0000, Subject “Fwd: Edge Master Class Update”, Attachments “Brooks._Intro.docx; Jennie_Max.jpg”. EFTA00433872 p.1 of 2. Read at the image at 400 dpi. The covering From: and To: display names and address values are barred, as are Brockman’s address, Epstein’s address in the forwarded header, a Cc pair and two lines below the class link. The circulation list is quoted here verbatim and in full, with J.E. (Jacqui) Safra among the names. The “Who / Where / When” block is printed as quoted. p.2 carries the remainder of Brockman’s signature block. "J.E. (Jacqui) Safra" returns 23 documents and "Spring Mountain Vineyard" returns 65 across the whole index, of which 23 carry a name form in the union. 61 of the 65 carry the EFTA prefix and 4 are House Oversight sheets; the page’s stated matching rule counts every production, so 65 is the figure and 61 would be a scope this page does not use. The four are at [^mc2015]. ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7

  5. E-mail, John Brockman to Nathan Myhrvold, copied to a list of thirty-one, Date: Thu, 01 Dec 2011 00:42:28 +0000, Subject “Dyson on Kahneman - New York Review of Books”, Attachments “How_to_Dispel_Your_Illusions.docx”. EFTA00719772 p.1 of 2. Read at the image at 400 dpi. The Cc list prints "J.E. Safra/Doumanian" as a single quoted display name between Jean Pigozzi and Jeff Skoll, with the address value beside it barred, as are almost all the other address values in the list. "Safra/Doumanian" returns 7 documents; 6 of them print the J.E. prefix. ↩ ↩2

  6. E-mail, John Brockman, Sent: Wednesday, February 9, 2011 8:37 PM (bare), Subject “NY dinner details and update”, addressed to twenty-one named recipients including “Jeffrey Epstein”, “Gmax Maxwell”, “J.E. Safra”, “Jean Doumanian” and “Robert Trivers”. EFTA01799443 p.1 of 5 for the header and the dinner particulars, read at the image at 400 dpi, and p.2 for the guest entry, read at the image at 600 dpi because the ligature decides how the line is transcribed. The sheet prints J.E. Safra in the To: line; the extraction of this document reads it “J.E. Satre” — the second sheet on this page whose extraction turns his surname into “Satre”, the other being at [^satre]. On p.2 nothing is barred; on p.1 Brockman’s sender value, a mobile value and one line of the restaurant block are barred. The guest entry is quoted here in full and the =ountain is as printed. The aesc ligature in “Encyclopædia” is on the sheet at 600 dpi; the extraction of this document reads it “Encyclopmdia”, so no phrase query on the correct spelling reaches this line. The run is 5 sheets: p.3 continues the guest entries, p.4 carries a further list of names without descriptions, and p.5 an Apple property-list block. None of the three is cited here and his name is on none of them. ↩ ↩2 ↩3 ↩4

  7. E-mail, John Brockman to Jeffrey Epstein, Date: Thu, 02 Jun 2011 12:08:06 +0000, Subject “update”. EFTA00649356 p.1 for the covering message and p.2 for the list; both read at the image at 600 dpi, p.1 because the key turns on one glyph. p.1 carries the heading “EDGE MASTER CLASS 2011 — INVITATION LIST (√ confirmed; x regrets; * maybe)” at its foot, and p.2 the list itself, one line per invitee. At 600 dpi the first mark in that key is a tick; the extraction of this sheet renders it as the numeral 1, breaks the heading across two lines and turns the em dash into a hyphen, so the heading is transcribed here from the image and not from the text layer. The same substitution runs through p.2, where the extraction renders each tick as -4. His line is the twenty-eighth of thirty-five and carries no mark; thirteen of the thirty-five carry none. Brockman’s sender value, the recipient’s address value, a mobile value and the telephone, facsimile and address lines of his signature block are barred on p.1; nothing on p.2 is barred. The aesc ligature in “Encyclopædia” is on the sheet; the extraction of this document reads it “Encyclopwdia”. The covering message is also where Brockman asks Epstein for “some support for the event”, which this page does not treat as an act by Safra and mentions only because it is the sentence the list is attached to. ↩ ↩2 ↩3 ↩4

  8. E-mail, John Brockman to Jeffrey Epstein, Sent: Thursday, October 27, 2011 1:43 PM (bare), Subject “Safra properties”. EFTA01799889 p.1 of 4, and p.2 for the Calistoga and Jerusalem detail. Both sheets read at the image at 400 dpi. The = characters inside the quotations are quoted-printable soft breaks carried onto the face of the sheet, one replacing one character at a roughly fixed interval; they are not repaired here and they are not extraction damage. “but the and is about $300,000 an acre” is as printed. The salutation “JE,” prints twice. Brockman’s sender value is barred; nothing else on p.1 or p.2 is barred. The sheets print street addresses for the two Jerusalem houses and links to mapping services; this page gives the neighbourhood and the two street names as Brockman’s own summary line gives them and reproduces no house number. pp.3–4 carry the Nevada entry, a signature block and an Apple property-list block. The four-page run is one document and nothing quoted here continues onto a sheet not cited. "Safra properties" returns 1 document in the index — this one. No copy of the memorandum survives in the release under that phrase, which is why the searches for a completed transaction below reach nothing further through it. ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8

  9. E-mail, John Brockman to Jeffrey Epstein, Date: Wed, 16 Jun 2010 20:37:52 +0000, Subject “Jacqui Safra”. EFTA00749227 p.1, the whole run. Read at the image at 400 dpi. The three paragraphs quoted are the first, second and third of the message; a fourth, about a Silicon Valley dinner, follows and is not quoted. “whose is in town scrambling” is as printed. Brockman’s sender value and one line of his signature block are barred. A second copy, whose bare Sent: Wed 6/16/2010 8:37:52 PM field carries the identical clock value to this sheet’s +0000 header, is at EFTA02413047 p.1, read at the image at 400 dpi — that pair is one of the six instances of the bare-field-is-UTC control at [^zones], and one of the three in which the two shapes print the same number rather than differing by four hours, because the zoned member of the pair is an explicit +0000 header and not an Eastern stamp. ↩ ↩2 ↩3 ↩4 ↩5

  10. E-mail chain, Peggy Siegal and Jeffrey Epstein, 20 July 2009; the sentence quoted is Siegal’s, attributed on the sheet to “On Mon, Jul 20, 2009 at 8:55 AM”. EFTA01815746 p.1, the whole run; read at the image at 400 dpi. Message blocks are counted here as the top message plus each quoted or forwarded block beneath it, counted at the image; on that convention this sheet carries four. A second copy of the same chain, carrying two blocks, is EFTA01817895 p.1, read at the image at 600 dpi; a third, carrying five, is EFTA02440181 p.1, read at the image at 600 dpi. The third is the longest because it alone carries Siegal’s own 1:44:39 PM reply above the message that stands at the head of the first sheet. All three sheets print “Jaqui Safra, Jean Doumanian and I left Portofino this morning- drove to Genoa and flew into Farnborough…now driving to London.” The extraction of the first reads “Jaqui Safia” and “Famborough”; the extraction of the second reads “Jaqui Sofro.”; the extraction of the third reads “Jaqui Salm”. None of the three extractions is reachable by any form in the union, and on each of the three sheets the printed name is “Jaqui Safra” and the corruption is the extraction’s alone. The chain survives in 7 documents in all, found by "left Portofino this morning"; the other four extract the name as Jaqui Safra and are inside the union. Siegal’s address values are barred in the quoted attribution lines on all three sheets. The message is Siegal’s own first-person account of her own movements and names Safra as one of a party of three. It fixes a date and a route; it establishes nothing else about him. Epstein’s replies in this chain concern a different subject and are not quoted, on any of the three sheets. ↩ ↩2 ↩3

  11. E-mail, John Brockman to Jeffrey Epstein, Date: Sun, 13 Nov 2016 17:58:49 +0000, Subject “in Paris Fri Dec 9- Wed Dec 14”. EFTA00631232 p.1, the whole run. Read at the image at 400 dpi. “Jacqui spring for a French EDGE event” is as printed; the speaker’s surname is printed “Deheane” in the sentence quoted and “Dehaene” in the next paragraph, and neither is repaired here. Brockman’s sender value and a line below “JB” are barred. The sheet gives no surname; it is assigned to this subject because Brockman’s other correspondence with Epstein uses the same forename for the same man, and that is an inference from the surrounding record and not a statement on this sheet. EFTA01061915 p.1 is not a second copy of this message: it is Epstein’s reply to it, From: "jeffrey E." with the forename in lower case, Subject Re: in Paris Fri Dec 9- Wed Dec 14, Date: Sun, 13 Nov 2016 18:02:53 +0000, quoting Brockman’s message in full beneath it. Read at the image at 400 dpi. Its whole body is “check win closer to the time . i ll think about it” — the spacing is as printed, and the sheet prints “i ll” in lower case where the extraction of this document reads “i II”. The interval is 4 minutes 4 seconds, measured between two +0000 headers and therefore zone-independent. The reply answers the question about meeting people in Paris; it does not mention Safra, and nothing in it bears on the 2009 event or on who paid for it. ↩ ↩2

  12. Every figure in this footnote and in What the search returned was derived on 2026-09-11 by FTS5 phrase query over the full-text index, counted in DOCUMENTS. The index covers every production in the release, not only the EFTA- prefix. FTS5 folds case and ignores punctuation, so a quoted phrase matches on token adjacency alone: "J.E. Safra", "J E Safra" and "j e safra" are one query returning 32, and "JE Safra" — one token — returns 0. That property is not harmless, and it cost this page its largest single correction. The phrase "yes,i am in ny" returns 16 documents, and an earlier account of this subject reported 16 copies of his second reply. Six is the right number. Ten of the sixteen are other people writing “Yes I am in NY” in other years — Edward Jay Epstein on 8 March 2012 (EFTA00418383) and, on 21 August 2013, the sender the field prints as “Jean Luc Brunel” (EFTA00967577) among them — and they enter the count because the tokeniser cannot see the missing space or the comma. The six that carry the literal comma form are EFTA00372330, EFTA00372331, EFTA00986215, EFTA01930213, EFTA02109846 and EFTA02109869, identified by testing each of the sixteen returns against the regular expression yes\s*,\s*i\s*am\s*in\s*ny over its whole text. The 13-document figure for the exchange is the union of three of its printed lines: "Hello Jacqui. Jeffrey would like to set up a time to speak with you" (13 documents), "yes he can call me all" (11 — the stem is used because the copies split 8 to 3 between “morning” and the extraction’s “moming”, with no document carrying both), and the six above. The union is 13 and the first phrase alone reaches all 13. The 13 are the Groff–Safra strand and not the whole day: EFTA01750543 (the reminder to Epstein), EFTA01933564 (his answer to it) and EFTA02582179 (Brockman’s message of 20 March standing alone) carry none of the three phrases and are in none of the three populations, which is why the nine messages narrated in the body are more than the 13 documents cover. Three of the 13 print no form of his surname and are outside the name union: EFTA00372300, EFTA02110531 and EFTA02110942. All 13 were opened; 7 were read at the image: EFTA00372300, EFTA00372329, EFTA00372330, EFTA00372331, EFTA01930185, EFTA02109846 and EFTA02110904. The surname decomposition in the body assigns each of the 366 "Safra" documents to the first rule it matches in the order the table prints them, tested over whitespace-normalised full text; that is one specification for all twelve rows and every figure in the table obeys it. A first-matching rule makes the order of the rows part of the specification, so the cost of that order was measured rather than left as a caveat: all nine residual patterns were run against all 59 residual documents, and no document matches two of them. The assignment of rows 3 to 12 is therefore order-independent in fact, and re-ordering the nine rules would move no figure in the table. What the row-1-first order is actually load-bearing for is twenty-one documents, and nineteen of the twenty-one are the union form restating itself rather than a competing membership. Row 9’s second arm is \.E\. Safra, which is a substring of the union’s own form J.E. Safra, so every row-1 document whose extraction prints the initials with their points satisfies row 9 by construction: EFTA00636847, EFTA00656696, EFTA00719772, EFTA00738932, EFTA00740586, EFTA00740603, EFTA00841697, EFTA00871595, EFTA00924572, EFTA00947324, EFTA01752264, EFTA01764323, EFTA01977641, EFTA02396208, EFTA02548313, EFTA02552821, EFTA02563769, EFTA02694698 and EFTA02694835 — nineteen, each carrying the label inside a copy list in the same normalised text the rules are run over, checked document by document. (Of the 32 documents the punctuation-blind "J.E. Safra" query returns, those 19 are the ones whose extraction prints the initials unspaced, J.E. Safra, and they are the only ones row 9 reaches; the other 13 print j e safra, or in three cases the spaced J. E. Safra, and the pattern misses them all.) The remaining two are row-1 documents that also satisfy the banking-family pattern of row 6, and in each the matching passage was located and read rather than counted: EFTA01799889 is the properties memorandum, and what matches row 6 is a reference to a banker of the Safra family inside the Jerusalem press account it quotes, not a reference to him; EFTA02421758 is a 39-page list whose extraction carries his name in one entry and an Edmond Safra entry elsewhere in the same list. All twenty-one belong in row 1 because the subject is named in each, and the order of the rules put none of them in the wrong place — the order-independence that bears on the table’s figures is the one measured inside the residual, where it is exact. In the other direction, none of the 146 attorney documents matches any non-attorney pattern. Rows 1 and 2 are the two query sets already given (the nine-form union, and the three attorney queries, whose intersection is empty); rows 3 to 12 classify the 59 documents left over, by these case-insensitive patterns in this order: MR\. SAFRA|Ste(?:ven|ve) \S{0,4} ?Safra|R\. Safra, Florida|auen R\. Safra · safra[- ]?A?\.? ?catz · andrew (?:\S+ )?safra · Edmond (?:and Lily )?Safra|Edmond J\. Safra|Lily Safra|Joseph Safra|Deborah Safra|Safra Bank|Safra family|Safra Center|account \(Safra · at Safra's place · obtaining [=s]afra · Jac ui Safra|\.E\. Safra · sAfrA \+ · Safra in town, with one document matching none of them. The three attorney queries are "Steven Safra" (70), "Steven R. Safra" (79) and "Steven Russell Safra" (23), whose union is 146 and whose intersection with the subject union is 0. Damaged-spelling generation: python3 scripts/ocr_variants.py "Jacqui Safra" --wide --count (31 forms, 2 non-zero), python3 scripts/name_forms.py "Safra" (no variant returns a document), python3 scripts/name_forms.py "Jacqui" (one form, Jacgui, 1 document), and python3 scripts/name_forms.py --verify "Jacgui" (REAL, exit 0); each tool’s --selftest was run first and both printed PASS. The limit of all of that generation is measured on this page rather than asserted: three documents whose sheets print Jaqui Safra are unreachable by every generated form, because the damage is in the extraction and is too heavy for a single-edit generator ([^siegal]). Absence searches: "Sherover" 11 documents, all opened; "Wants to sell you his Jerusalem house" 10, all opened; "Safra properties" 1; "Jacqui's Jerusalem residence" 3; "serious conversation about his Jerusalem" 2; "did you want to set up a call with Jacqui" 2. Clean zeros: "Sweetland Films", "J.E. Beaucaire", "Beaucaire", "Jacob Safra", "Moise Safra". Context populations: "Spring Mountain Vineyard" 65 documents (23 in the union), of which 61 carry the EFTA prefix and 4 are the House Oversight schedule sheets at [^mc2015] — the rule at the head of this footnote counts every production, so 65 is the figure; 61 is the EFTA-only count and is not the scope this page uses anywhere; "Jean Doumanian" 44 (34), "Doumanian" 52 (41), "Britannica" 45 (2), "Encyclopaedia Britannica" 5 (0), "Republic National Bank" 6 (0), "Edmond Safra" 2 (1), "Edge Foundation" 547 (39), "Andrew Saffir" 18 (0), "Safra Catz" 7 (0). A reader re-running any query in this page should get the number printed beside it; where a number here is not a query result but a count of documents opened and read, the sentence says so. ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10 ↩11 ↩12 ↩13 ↩14 ↩15

  13. E-mail, John Brockman to Jeffrey Epstein, copied to Lesley Groff, Date: Tue, 26 Feb 2013 18:04:11 +0000, Subject “today”. EFTA00394883 p.1, the whole run. Read at the image at 400 dpi. The passage quoted is printed under the heading “YOUR HOUSE IN JERUSALEM” and is followed on the sheet by a line carrying only an ellipsis; the page’s quotation ends where the sheet’s sentence ends. Brockman’s sender value, Groff’s copy value, a mobile value, two lines below the Jerusalem paragraph, the hotel and room values in the itinerary block and the telephone and facsimile values in the signature block are all barred. The top line prints “Westin”, then a barred value, then a room label and its number; no hotel or room value is taken from this sheet. A forwarded copy of the same message, with a second sheet, is at EFTA00394884 pp.1–2, not rendered. The message survives in 10 documents, found by the phrase "Wants to sell you his Jerusalem house"; three of the ten are reachable only as Jacque Safra. ↩ ↩2

  14. E-mail, John Brockman to Jeffrey Epstein, copied to Lesley Groff, Date: Wed, 21 Aug 2013 18:26:51 +0000, Subject “Re:”. EFTA00385244 p.1 of 2. Read at the image at 400 dpi. Both quoted sentences are on p.1; the postscript follows a barred block beneath “JB”, and the rest of the sheet is a draft invitation to a weekend at Brockman’s own farm in Connecticut. Brockman’s sender value and Groff’s are barred. Brockman’s line places himself, Kahneman and Thaler at “Safra’s place”; it does not say Safra was there, and this page does not say so. The same message survives in 8 documents, found by the phrase "at Safra's place", none of which carries any form of his name that the union below can reach; five of the eight also print the phrase "Jacqui's property in Calistoga", and the remaining three carry the same postscript sentence with the name damaged in extraction, so all eight carry the Calistoga line and five are reachable by the phrase. ↩ ↩2 ↩3 ↩4

  15. scripts/act_rate.py, run 2026-09-11. --selftest was run first and printed PASS (population 100, Epstein 46, Harper 38, overlap 0, and the stacked-header fixture returning __UNDETERMINED__). The figures in the body are: act_rate.py '"Jacqui Safra"' Jacqui → 0 sender-field acts of 101; act_rate.py '"Jacqui Safra"' Safra → 2 of 101; act_rate.py '"J.E. Safra"' Safra → 8 of 32; and the nine-form disjunction with the key Safra → 8 of 162, whose eight identifiers the tool prints under --show: EFTA02110904, EFTA02109846, EFTA01930213, EFTA01930185, EFTA00986215, EFTA00986212, EFTA00372330, EFTA00372329. The strict figure of 6 is derived by reading the sender field the tool prints for each: the first six carry j e safra at the head; the last two are forwarded chains carrying Lesley Groff first and j e safra second. The tool does not print a strict figure of its own and none is inherited here. The tool’s own documentation states the limit that matters on this subject — “Being the top-level sender is evidence of acting; it is not the only way to act” — and records that a public-relations firm and a modelling agency both score zero and both became published pages. ↩ ↩2

  16. The zone finding, and its control. Six messages in this subject’s material survive both as a bare Sent: field and as an explicitly zoned stamp of the same message. Three of the six pair a bare field against an EDT stamp and run +4h. They are: Brockman’s of 20 March 2014, Sent: Thursday, March 20, 2014 8:24 PM on EFTA02582179 p.1 against Date: March 20, 2014, 4:24:20 PM EDT on EFTA01750543 p.1; his own first reply, Sent: Mon 3/24/2014 2:44:02 PM on EFTA02110904 p.1 against March 24, 2014 10:44:02 AM EDT on EFTA00372329 p.1; and his second, Sent: Mon 3/24/2014 3:16:49 PM on EFTA02109846 p.1 against March 24, 2014 11:16:49 AM EDT on EFTA00372330 p.1. All three run +4h, which is UTC against US Eastern Daylight Time. All six sheets read at the image at 400 dpi. The independent check is a machine value, not a human-typed one: EFTA02582179 p.1 prints an Apple property-list block below the signature containing <key>date-received</key> <integer>1395347016</integer>, which is 20:23:36 UTC on 20 March 2014 — the same clock as the bare 8:24 PM, and four hours from the EDT stamp. The other three pairs settle it from a direction no quoting client can explain away, because in each of them the bare field stands against the same message’s own +0000 header rather than against someone else’s attribution — and all three show no offset at all. EFTA02413047 p.1 prints Sent: Wed 6/16/2010 8:37:52 PM for the message whose header on EFTA00749227 p.1 reads Date: Wed, 16 Jun 2010 20:37:52 +0000. EFTA01812980 p.1 prints Sent: Wed 6/16/2010 8:40:32 PM for the reply whose header on EFTA00735612 p.1 reads Date: Wed, 16 Jun 2010 20:40:32 +0000. And on the day of the call itself, EFTA01930185 p.1 prints Sent: Mon 3/24/2014 3:17:17 PM for the message whose header on EFTA00372329 p.1 reads Date: Mon, 24 Mar 2014 15:17:17 +0000. Identical to the second in all three cases. All six of those sheets read at the image at 400 dpi. A second, structural check points the same way: on the Eastern reading of the bare fields, Epstein’s go-ahead at 12:14:25 would fall almost an hour after Groff had finished arranging the call — her last message of the sequence carries the header Date: Mon, 24 Mar 2014 15:18:04 +0000 on EFTA00372331 p.1, read at the image at 400 dpi — which the sequence cannot bear. The desk’s general measurement — that a bare Sent: field in these exports is UTC and a zoned one is Eastern — is recorded in _desk/TOOLS.md; this page re-derives it from its own documents rather than inheriting it. ↩

  17. E-mail, Lesley Groff to Jeffrey Epstein, Sent: Monday, March 24, 2014 11:36:09 AM (bare, UTC — 7:36:09 AM EDT; see [^zones]), Subject “Reminder: JOhn Brockman/Jacqui” with the capital O as printed, forwarding John Brockman’s message of March 20, 2014, 4:24:20 PM EDT, Subject “Jacqui’s Jerusalem residence”. EFTA01750543 p.1, the whole run. Read at the image at 400 dpi. The sheet prints “Jacqui Safra” beneath “If so:”; the extraction of this document reads “Jac ui Safra”, which is why no query on his name reaches it. The block immediately below that name is barred, as are Groff’s own sender value, Brockman’s sender value, the Cc value and a block below his sign-off. A second copy of Brockman’s message, standing alone, is at EFTA02582179 p.1, read at the image at 400 dpi; a third, inside Epstein’s reply, is at [^epsteinyes]. ↩ ↩2

  18. E-mail, Jeffrey Epstein, Sent: Mon 3/24/2014 12:14:25 PM (bare, UTC — 8:14:25 AM EDT), Subject “Re: Reminder: JOhn Brockman/Jacqui”, whose whole body is “yes, and Paul allen,”, above Groff’s reminder rendered by his own client as “On Mon, Mar 24, 2014 at 5:36 AM” and Brockman’s forwarded message. EFTA01933564 p.1 of 2. Read at the image at 400 dpi. The sheet prints “allen”; the extraction of this document reads “alien”, and the body is quoted here from the image. The To: value, the quoted sender value, Brockman’s sender value, the Cc value and the blocks below “Jacqui Safra” and below “John Brockman” are all barred. p.2 carries the continuation of a boilerplate confidentiality notice and nothing else. The quoted attribution “at 5:36 AM” is the rendering of Epstein’s own client and is not used here to fix any zone; the 38-minute interval between the two messages is measured within one stamp kind and is zone-independent. ↩

  19. E-mail, Lesley Groff, Date: Mon, 24 Mar 2014 13:39:18 +0000, Subject “Jeffrey Epstein”, whose whole body is the two sentences quoted, above “Thank you,”. EFTA00372300 p.1, the whole run. Read at the image at 400 dpi. The sender’s address value and both the recipient’s display name and address value are barred, and so is the sign-off name below “Thank you,”; the From: line prints “Lesley Groff” and the role line prints “Assistant to Jeffrey Epstein”. This sheet carries no form of the subject’s surname and is therefore outside the 162-document union: it is reached only through the body sentence. The same sentence is on all 13 documents of the exchange. ↩

  20. E-mail, Lesley Groff to Jeffrey Epstein, Date: Mon, 24 Mar 2014 15:17:54 +0000, Subject “Jacqui is in NY-EST zone applies!”, body “just rec’d this …” above a forwarded block printing From: j e safra and Date: March 24, 2014 11:16:49 AM EDT. EFTA00372330 p.1, the whole run. Read at the image at 400 dpi. The address values in the top header, in the forwarded header block and in all three quoted attribution lines are barred, as is the value at the end of Groff’s quoted 10:46 AM line and a block below her signature. The forwarded header block on this sheet names him as sender in terms, and the same reply also survives under a clean top-level header of his own in two further copies ([^reply1], [^reply2]), so the attribution of these words to him does not rest on a thread boundary. ↩ ↩2

  21. E-mail, Lesley Groff to j e safra, Date: Mon, 24 Mar 2014 15:18:04 +0000, Subject “Re: Jeffrey Epstein”, whose whole body is “terrific..thank you!”. EFTA00372331 p.1, the whole run. Read at the image at 400 dpi. The To: field carries his name, To: j e safra, with the address value barred. It is one of two sheets read for this page that address him by name in a top-level header; the other is the dinner invitation at [^lecirque], whose To: line names him as J.E. Safra. The sheet’s other address values, the value at the end of the quoted 10:46 AM line, and a block below Groff’s signature are all barred. ↩

  22. E-mail, John Brockman to Jeffrey Epstein, Date: Fri, 20 Aug 2010 18:19:36 +0000, Subject Preview - Danny HIllis on “Cancering” (Edge Master Class 2010) video/audio/text. EFTA00757793 p.1 of 2. Read at the image at 400 dpi. The sentence quoted is the opening of the fourth paragraph. Brockman’s sender value is barred and a line below “Eastover Farm” is barred; a distribution list runs on at the foot of the sheet and onto p.2. This is the earliest sheet read for this page that states whose the vineyard is, and it states it in passing, in the course of explaining why that year’s class was under-attended. Nothing on either sheet says Safra was present. ↩ ↩2 ↩3

  23. Four one-page sheets of Jeffrey Epstein’s weekly schedule, House Oversight production: HOUSE_OVERSIGHT_001817, HOUSE_OVERSIGHT_001820, HOUSE_OVERSIGHT_001827 and HOUSE_OVERSIGHT_001831, each the whole run. All four read at the image. This production is released as page images rather than as PDFs, so each was fetched as its released page image and read at the native raster, 2550×3369 pixels, which is 300 dpi on a letter sheet; the line is transcribed from that image and not from the text layer. Three of the four print it under the day heading “Fri. July 31, 2015 PB-Zorro”; the fourth, HOUSE_OVERSIGHT_001817, is a continuation sheet carrying the printed page number 4 and no day heading. The hyphens, the curly quotation marks around the course title and the two spaces after “Reminder:” are as printed. The extraction of HOUSE_OVERSIGHT_001817 reads “St Helena” without the point where the sheet prints “St. Helena”, which is why the line is transcribed from the image on every sheet. Two of the four, HOUSE_OVERSIGHT_001827 and HOUSE_OVERSIGHT_001831, carry below the entry the signature block of a legal assistant at Darren K. Indyke, PLLC, with the name and the telephone value barred; HOUSE_OVERSIGHT_001827 also quotes, beneath that block, a message of 16 July 2015 from Epstein’s own mailbox reading in full “print out a scheudle for the week, thanks”, the spelling and the double space as printed. That is what fixes the document class: these are sheets of a week’s schedule printed for him on request, not a circular sent to him. These four are the whole of the difference between "Spring Mountain Vineyard" at 61 documents and at 65: they are the only documents returned by that phrase that do not carry the EFTA prefix. None of the four carries any form of his name, so the 23-in-the-union figure at [^mc2011] does not move, and nothing here places him, or anyone else, at the 2015 class. ↩

  24. E-mail, Jeffrey Epstein to John Brockman, Date: Wed, 16 Jun 2010 20:40:32 +0000, Subject “Re: Jacqui Safra”, whose whole body is “im always in. whatever you need.” above the quoted message at [^jun2010a]. EFTA00735612 p.1 of 2. Read at the image at 400 dpi. The double space after the full stop is on the sheet. The recipient’s address value and two lines of Brockman’s quoted signature block are barred. p.2 carries the continuation of a boilerplate confidentiality notice. The interval to the message it answers is 2 minutes 40 seconds, measured between two +0000 headers and therefore zone-independent. Two further copies survive. EFTA01812980 p.1, read at the image at 400 dpi, carries the same reply under a bare field, Sent: Wed 6/16/2010 8:40:32 PM, against this sheet’s +0000 header: a sixth pair for the control at [^zones], and a third at zero offset. EFTA00893568 p.1 is a third copy, read at the image at 400 dpi; nothing is taken from it. ↩ ↩2

  25. E-mail, John Brockman to Jeffrey Epstein, Sent: Monday, September 14, 2015 2:56 PM (bare), Subject “Safra in town - got me into Cornell Weill”, whose whole body is “Being met at ER by top cardiologist. All set.” above an emoticon glyph. EFTA02357201 p.1, the whole run. Read at the image at 400 dpi. Brockman’s sender value and a line below “JB” are barred. The sheet gives no forename and this page supplies none. It is the latest dated item read for this page that uses the surname in this correspondence, and it is recorded as that and nothing more. ↩

  26. EFTA01930185 p.1, the whole run — Lesley Groff to Jeffrey Epstein, Sent: Mon 3/24/2014 3:17:17 PM (bare), the same message as [^groff1117] in a different export shape. Read at the image at 600 dpi, because one character decides the point. The subject line prints Jacqui Safra. The extraction of this document reads it “Jacqui Satre”, and that single row is the whole of the "Jacqui Satre" population in the release. The same extraction also inserts <EM= after “j e safra” in the forwarded header block; no such string is on the sheet. The value after “the number Brockman gave us for Jacqui is:” is barred here. The address values in the forwarded header block and in the quoted 9:39 AM attribution line are barred, as is a block below the signature. ↩

Backlinks