Barnaby Marsh


Named on the documents as	”Barnaby Marsh” in the From and To lines of his own e-mail; “Barnaby Marsh, D.Phil.” at the head of a letter addressed to him; “BARNABY MARSH” in capitals in two trust instruments; “Barnaby” in his own sign-offs and “B.” and “-B.” at the foot of several of his messages; “Marsh” alone in two conference lists; “Dr. Barnaby Marsh” in his own signature block on a message of 18 March 2019¹²³⁴⁵⁶⁷
Titles the documents give him	”Executive Vice President, Strategic Initiatives / and Managing Director, Philanthropic Leadership Network / John Templeton Foundation”, on the inside address of an Enhanced Education letter of 2 December 2015; “Executive Vice President, John Templeton Foundation” on a LinkedIn connection notice of 21 August 2013; and “Founder, / Saint Partners Philanthropy =ervices” in his own signature block on a message of 18 March 2019, which also prints “www.saintpartners.com” (the `=` is a quoted-printable soft break on the face of that sheet). The sheets read at the image for this page give him no other title²⁸⁷
The institution he acts for in these documents	The John Templeton Foundation, of West Conshohocken, Pennsylvania²⁸⁹
Amounts he names in his own words	”$11M grant” and a “$6M” match, with “$6M of our grant” to go into “an external philanthropic vehicle” and “about $5M through Harvard” (12 May 2009); an “additional $2M. supplement” he says he will ask Templeton to approve (12 May 2009); “the $2 to $4MM level” for a proposal with Joi Ito (29 September 2013); “the range of $3-4 million for each” for Ito and Martin Nowak (12 November 2014)¹¹⁰¹¹
Instruments naming him	The Jeffrey E. Epstein 2018 Trust, dated 8 May 2018, Article VII §7.1(B): he is the first-named successor trustee. The 1953 Trust, made as of 8 August 2019, Article VI §6.1: he is the second-named successor trustee, after Boris Nikolic⁴³
Documents cited here run from	16 April 2009 (Epstein to him, “time to catch up”, cited from the search index and not rendered) to 8 August 2019 (the 1953 Trust). The last correspondence between the two that any search set out below reaches is of 19 March 2019¹²³¹³¹⁴

Barnaby Marsh is the John Templeton Foundation executive who corresponded with Jeffrey Epstein from April 2009 to March 2019. In May 2009 he wrote to Epstein about an “$11M grant” his foundation had made, which he was splitting between an external vehicle and Harvard for Martin Nowak’s centre, and asked Epstein whether he would “be in a position to make any additional contributions at all”. In September 2013 he proposed to Joi Ito, copying Epstein, a Templeton proposal “at the $2=to $4MM level”; in November 2014, asked by Epstein “do you need me to do a matching grant for jjoi and or martin”, he answered that his teams would “probably do something in the range of $3-4 million for each”. He is named in two of Epstein’s trust instruments as a successor trustee: first-named in the 2018 trust and second-named, after Boris Nikolic, in the 1953 Trust made two days before Epstein’s death.¹¹⁰¹¹¹⁵⁴³

This page rests on a small part of a very large return, and says so at the outset. The phrase "Barnaby Marsh" reaches 2,882 documents in the release, and the union of that phrase with eighteen damaged spellings of his name reaches 4,731. But the record is overwhelmingly one of arrangement rather than of action: classified mechanically, 2,432 of those 4,731 documents (51.4%) carry two or more scheduling terms and fewer than three distinct funding or science terms, and 209 (4.4%) carry three or more funding or science terms. Most of the substance set out below comes from that 4.4%, but not all of it does, and the exception matters: the fullest funding statement attributed to him in the 2009 material, Lawrence Krauss’s report of a meeting with him, sits outside the 4,731 entirely, because the extraction of that sheet misreads his forename and the sheet never prints his surname at all. The full decomposition, the searches behind it and their limits are set out below under What the search returned.¹⁴¹⁶

Nothing on this page is a finding of wrongdoing. He answers Epstein constantly in this record and many of his replies are quoted below, including two in 2019 about press coverage: one of 2 January about the press coverage of James Watson, and one of 5 March about the press coverage of Epstein himself. What no sheet read for this page records is any response by him to either trust instrument that names him: no acceptance, no refusal, and nothing showing that he was told. The searches run to test that are set out below and were built so that they could return one.¹⁴

2009: the Templeton side of the Harvard grant

The earliest message in the returned set is Epstein’s, of 16 April 2009, and it is four words long: “time to catch up”.¹²

Three weeks later the correspondence is about money and people. On Thursday 7 May 2009 Marsh wrote under the salutation “Dear Jeff:”, reporting that he had “finally heard from Chuck Harper, who is visiting anthropological sites in South Africa at the moment and has been a little hard to reach”, and that on Eastern Europe there had been “not too much luck- these are the names I got, and I don’t know any of them, so probably not too useful”. Four names follow as a bulleted list, each with a university and a barred contact value: an associate professor of philosophy of science in Estonia and three academics in Romania. He went on: “Chuck suggested Tegmark at MIT, and I think that we might be able to do something related to a new grants competition programs that look at creativity, innovation, and efficiency in the context of biological membranes, but the question is who are the people?” The message closes with a paragraph about a third party: “By the way I had a quick follow-up chat with Krauss in Arizona- what he is doing is important, and I have encouraged him to think about long-term directions/strategies (bigger than simply raising the endowment and convening famous people, which is only a start). I’ll reconnect with him soon.” It is signed “Barnaby”.¹

Epstein’s reply is one line: “come spend some time in florida”.¹

The unlabelled clock fields in this May 2009 run are Eastern Daylight Time, and the sheet fixes it: its own header reads Tue, 12 May 2009 20:35:37 +0000, and the message it answers is quoted as sent at 4:27 PM the same day. Only an offset of UTC−4 puts that message about eight minutes before the reply that quotes it; at UTC−5 it would fall after it. The times below are given on that basis and no zone is asserted for any sheet outside this run.

Five days later, at 10:52 AM Eastern on Tuesday 12 May 2009, Marsh set out the position of the Harvard programme in detail, and the money in it is the foundation’s:

A visit to WPB in a few weeks (2nd week in June) would work nicely. I’ve been working with Martin and Harvard. They are not doing much at all. I’m going to put $6M of our grant in an external philanthropic vehicle (either a DAF or simply an external account; with will save the project $600k in overhead) and put about $5M through Harvard, but nonetheless his center needs to be viable. As before, we can pick up the costs of administration but he needs a real solution for the rent, which we can’t fund because of charter restrictions. If we can’t do that he has to close PED. There is no other option. Perhaps we could get together some co-funders that could cover the rent part. I’m going to try Richard Watson of Cleveland gave Harvard $2M a few years back (and who is interested in game theory), and a few others that I know. Would you be in a position to make any additional contributions at all? Does anyone come to mind who Martin might approach (and who we might talk to first?). I want to make this work! Ideas welcome… –B.¹

The sentence beginning “I’m going to try Richard Watson of Cleveland gave Harvard $2M” is printed that way on the sheet and is not repaired here. “PED” is the Program for Evolutionary Dynamics at Harvard, which Martin Nowak directed.¹

Epstein answered at 12:47 PM Eastern: “how much is actually needed and on what time shedule? also , i look forward to your trip”. At 4:27 PM Eastern the same day Marsh answered. His message opens with a date, and that opening is what Epstein’s reply answers:

There is lots to discuss regarding other work, interesting people, etc…how about spending an afternoon together, the 15th or 16th of June? Does either of these dates work for you?

Then comes the arithmetic, and it is the passage in which the eleven-million-dollar figure appears:

About Martin, these are the details: his rent and the associated overhead that Harvard takes is about $400k per year, or $2m over 5 years (this is absolute minimum to keep him there and operating); his additional amounts for scholarship/scholars outside of the Templeton grant is about twice that amount; his current running budget is $1.2M per year (or $6M over 5 years). When we made the $11M grant, we though that this would scale up and expand his exceptional center and work, so if we could somehow raise $6M to add our $11M to that would be ideal. That is what I think we should aim for.

Also, just FYI, I’m going to be asking Templeton to approve an additional $2M. supplement for broad philosophical/applied questions after year 2 of the project, so this will further amplify the importance of Martin’s work, which can not be understated- he is a genius.¹

“we though” is as printed. Epstein’s reply, at the head of the same sheet, is “16th ,,,would be perfect.”¹

Five weeks after that, on Thursday 18 June 2009, Epstein wrote to Marsh with a list of subjects, unprompted by any message quoted on the sheet: “I thought about what we should fund, Just initial thinking.. my sugestion is Signal processing,,, algorithms , information theory ( pattern recognition)… Synthetic Intelligence. ( what is means to think ) ( not brain related at all, ) one step up. and Quantum computing.” He continued: “If succesful , a process could be developed where all the information , currently available to the world could be analyzed , in a few days, if conventional and arguably, with in minutes if done on a quantum level. - as you would be looking at all permuations of patterns all at once. similar to quantum key encryption.” The misspellings are as printed, and the Subject field on that sheet carries its label with nothing after it.¹⁷

A second document of the same day records what a third party said Marsh was proposing, and an earlier draft of this page reduced it to a date. On Thursday 18 June 2009 Lawrence Krauss wrote to Epstein under the subject “Re: templeton”:

met with Barnaby in NYC, and I think I can work with him.. I believe he is honorable. We decided would be best to do anonymous gift, which templeton could not then promote.. he wants to start with $2M and then go to $10M if we perform. Would be nice to talk to you.. Would you be willing to match?

Epstein’s reply, on the same sheet and of the same day, is “did i hear thank you,,”.¹⁶

Three things about that passage, and the page states all three rather than resting on it. It is Krauss’s account of what Marsh wanted, not Marsh’s own words, and nothing read for this page records Marsh saying any of it. The sheet names him only as “Barnaby”; the surname is supplied on Krauss’s page from a different document, and this page supplies none from this sheet. And it cuts against the shape of the rest of the record: an anonymous gift structured so that the foundation “could not then promote” it, a figure rising from $2M to $10M “if we perform”, and a request that Epstein match. It is a proposed arrangement, described in detail by a third party, on the same day as Epstein’s list of subjects to fund. Where the page says the record is overwhelmingly one of arrangement rather than of action, this sheet is the strongest counter-example it has found, and it was found only after the first pass had counted the date and not the document.¹⁶¹⁷

Three weeks later the subject was how to measure a grant. On 7 July 2009 Epstein wrote: “sorry i missed your call . metrics for success of the grants?” Marsh replied the same day: “No problem… about metrics, have you seen this?:”, gave a link to a document on the foundation’s own website headed cost_effectiveness.pdf, and added, “it is old, but contains the crux. I have additional ways… lets talk. Thursday and Friday look best- which is better for you?” He signed it “B.”⁵

The title the documents give him

Three documents in the release state what he was, and the third is set out further down, in his own signature block of March 2019. The first is a LinkedIn connection notice sent to Epstein’s address on 21 August 2013, whose message body opens “Jeffrey, Barnaby has accepted your invitation” and gives, under his name, “Executive Vice President, John Templeton Foundation” and, below that, “Greater Philadelphia Area | Nonprofit Organization Management Industry”. The invitation ran from Epstein to him, not the other way: the notice records the acceptance of Epstein’s invitation.⁸

The second is the inside address of a letter, and it is fuller. A letter on Enhanced Education letterhead dated 2 December 2015 is addressed to:

Barnaby Marsh, D.Phil. Executive Vice President, Strategic Initiatives and Managing Director, Philanthropic Leadership Network John Templeton Foundation

with the foundation’s street address in West Conshohocken, Pennsylvania beneath it. The letter is signed in ink by Darren Indyke over the typed line “Darren Indyke / Vice President”.² The instrument is described below.

A third document, not correspondence, records an institutional act of his outside the Epstein material. A book manuscript held in the release, Invisible Forces and Powerful Beliefs: Gravity, Gods, and Minds, “By the Chicago Social Brain Network”, carrying the imprint “FT Press, Upper Saddle River, NJ”, closes its preface: “We owe a debt of thanks to many for their contributions and support over the years, but we owe special thanks to Barnaby Marsh for approaching us with the idea of forming such a network and for his many contributions to the network, and to the John Templeton Foundation for their support and for their encouragement to pursue questions, ideas, and conclusions of our science regardless of where they led.”¹⁸

September 2013 to November 2014: the proposals, and the match

On Sunday 29 September 2013 Marsh wrote to Joi Ito, copying Epstein, under the subject “Re: some thoughts on designing around our “little mind"". The sheet carries quoted-printable soft breaks on its face and they are reproduced here as it prints them:

Thanks for sending me your travel schedule earlier. Mine looks similar but =e are not in the same cities, so lets have a call. I want us to focus on p=tting together a proposal for submission in Spring 2014, perhaps at the $2=to $4MM level, building on some of your earlier ideas. We need to narrow t=e focus, and identify the people on your side and on my side who will work=on the background documentation, if that timeframe works for you.

Maybe we can tie it into the projects evolving with Jeffrey, which would be=a lot of fun.¹⁰

He added that he would “ask one of my assistants to look for dates that might work for a cal= in the next week or two…”.¹⁰

Just under a year later the draft existed. On Wednesday 3 September 2014 Ito wrote to Marsh, copying Epstein, attaching “Templeton Media Lab proposal October 2014 Joi.docx”: “This is very very early draft that we haven’t even shared internally =et, but Jeffrey thought it would be good for you to have it in advance =f our call on Friday.” That message and that attachment are set out from the Media Lab end on Joi Ito’s page.¹⁹

A year later Epstein put the matching question directly. On Wednesday 12 November 2014 he wrote, in one line: “do you need me to do a matching grant for jjoi and or martin.” Marsh answered at 16:36:29 +0000 the same day:

Looking into the timing and logistics today…will get back to you about it in a few days after I hear from my teams working on this with Joi and Martin. We will probably do something in the range of $3-4 million for each… perhaps I will aim at covering the background administration part, and so you can concentrate on the fellows?! Also, the only parts I can’t do much on are building rent and more the 10% costs for travels; other than that, I have a lot of flexibility with the funding. More soon.¹¹

Epstein’s reply, timed 16:44:22 +0000 on the sheet that carries it, is “thanks”.²⁰

Eleven days after that, on Sunday 23 November 2014, Epstein wrote: “can we talk about your funding for joi and martin so that our mutual involvment gets the best result”. Marsh’s answer, timed 13:38:22 +0000 the same day, is the fullest description in the release of how he was working the two grants, and it treats the two institutions differently:

Yes I was going to talk to Martin today.. there are several parts.

Some of the things I am figuring out on my end:

For Joi- a first question is whether this is connected with MIT (overhead) or goes another route. The latter seems desirable, but I hear that MIT is not too happy with this. We need to makes sure there is not blockage. The bigger, more important issue is that the money enables something very unusual. I still am not convinced the Joi has all of the background details figured out. My team is working with his people who do the background stuff, but it is not clear where the loyalties are here - to the Lab, or to Joi.

For Martin- He has a current Templeton grant that runs through 2015 (already funded) and a second now in the works that will begin in late 2015 and extend through 2019. The second grant is not yet reviewed or approved. Any offer to match on the second grant will help secure it. Basically, all he needs is a short letter or statement (even from him) that says that additional donated money will augment and boost any of the ideas and work that he is working on in the grant. However, it can be used any way he wishes, including new directions, such as special fellowships. The work is broad, so, it should be easy (as far as you are concerned). On my end, the thing that is taking long with Martin is just the documentation- each time we need a proposal, and he, I, and others have to put together the justification ourselves. He does not have a grant-writing team, like many others do. Sometimes, he uses grad students or postdocs, but in these cases the result does not have the polish or completeness that is needed in a competitive process with adversarial peer review. There are reasons that I won’t go into for why I don’t want to cut corners on this sort of thing, and so, it just takes longer.¹⁵

“makes sure” and “the Joi” are as printed. The message ends “I agree that we should talk about how to get the best result… I can call anytime in the next few days, just let me know when. / -B.”¹⁵

A month earlier, on 17 October 2014, Epstein had written to him: “yes, gates is going to give 2 million if that helps”. The message it answers is not legible on the sheet read for this page beyond the opening words “can’t make the”.²¹

November and December 2015: the letter addressed to him

Marsh is the addressee of the matching-funds letter whose drafting Martin Nowak’s page sets out from the other end. That page records Nowak’s draft of 19 November 2015 and Richard Kahn’s report six days later that the letter had gone to Marsh, that Marsh “approved [l]anguage”, and Kahn’s question whether Darren should sign it.²²

Two things this page adds, from the sheets themselves. The first is the inside address quoted above, which is where the release states his office.² The second is the answer to Kahn’s question: the letter of 2 December 2015 is signed in ink by Darren Indyke over the typed line “Darren Indyke / Vice President”, and the body is written in the name of Enhanced Education rather than of Epstein personally: “Martin Nowak has been sharing with Enhanced Education his proposed new research applications of evolutionary game theory to the behavioral sciences. As you know, Enhanced Education is very interested in supporting Martin’s work.” The letter then states the sum it is written about: the foundation is “currently considering a project proposal in which Enhanced Education would have the opportunity to partner with the John Templeton Foundation through contributions which Enhanced Education has made to Harvard University’s Program for Evolutionary Dynamics and allocated to Harvard University Fund 347150, $1,500,328 [$1,304,633 direct costs + $195,695 indirect costs] of which contributions have been budgeted for this project.” That is the same figure, and the same fund, as the draft Nowak’s page quotes from the other end of the drafting.²

A second executed version of the same 2 December 2015 letter exists in the release, with the same body and its own ink signature, and it differs in its salutation: where the first reads “Dear Barnaby:”, the second reads “Dear Dr. Marsh:”.²³

The scheduling record

More than half the documents in this set are arrangements, and they run continuously from 2009 to January 2019. They take four shapes, and each is represented by one sheet read at the image for this page.

Epstein’s own messages to him are typically one line: “im in new york mon , tues”; “are you around fri sat”; “tues ok”; “skype?”¹⁴²⁴ His replies are of the same length: “be there in 10 mins…”, “Confirmed!”, “yes- tomorrow- sunday morning ca 7?”²⁴¹⁴

Epstein’s office arranged the meetings. A day sheet for February 2018, set in italic type with the times in bold, carries under “Thurs. Feb. 1, 2018 NY-PB” the entries “12:30pm LUNCH w/Dan Kahneman, Woody and Soon Yi”, “2:30pm Appt w/Arine de Rothschild”, “3:15pm Appt w/Barnaby Marsh”, “4:00pm Appt w/Leon Black” and “8:05pm Depart NY”. The bold is the sheet’s own and is not this page’s emphasis: every time label in that day’s block is bold italic, and in two of the five entries the word after the time is bold as well. A contact value after his name is barred, as are contact values after several other entries on the sheet; the entry at 6:00pm has its subject barred and the words “at the house” printing after it.²⁵

Calendar reminders generated by Epstein’s office form their own class. “Reminder: 5:15 Appt w/Barnaby Marsh”, “Reminder: when can Barnaby Marsh meet w/JE and Botstein?”, “Alert - reminder: Barnaby Marsh due back in NY after Dec. 7” are three of them. A forensic extraction of one of Epstein’s devices records Skype and message traffic with him, including an entry of 8 August 2016 whose whole content is “im here”.¹⁴²⁶

And his name recurs in the arrangements of other people’s meetings: the release records Epstein’s office working with the office of Leon Botstein through November and December 2012 to find a date on which Botstein and Marsh could both see Epstein, and Marsh answering, “I won’t be back in the city until after ther 7th of December… would love to see Jeffrey anytime in early December after that, if he is in town? Best, Barnaby”.¹⁴

Two lists

He appears twice in lists that are not schedules.

On 15 November 2011, in a chain headed “Some other Possible Invitees for NEXT conference” between Martin Nowak, Epstein and Cecile de Jongh, de Jongh’s message reads: “Inculdes:” followed by “Krauss / Church / Nowak / Marsh / Minsky”, then “Declines:” followed by “Parker / Sinofsky / Watson / Rosen”, then “I have asked Marvin about his contact with the others you mentioned last week but no firm answer. Will try again.” It is not the whole message: the sheet continues “With warm regards, / Cecile” and a disclaimer. A second message of the same planning sequence, of 27 October 2011, numbers the same group “1. Church 2. Nowak 3. Krauss 4. Marsh 5. Minsky 6. Rosen 7. Wolfman 8.Mayne or Milner 9. Corrina”.⁶²⁷

Neither sheet gives a forename and this page supplies none from either. It does not have to. Three further messages in the same de Jongh planning sequence of autumn 2011, all read at the image for this page, print “Barnaby Marsh” in full among the same group of names: her message of 23 September 2011 to Epstein, which says “So far we have 9 people including Minsky” over a numbered list ending “9. Barnaby Marsh”; her request of 14 October 2011 for contact information for eight named people, the last of them “Barnaby Marsh”, sent to a recipient the producing party has barred; and her “Other participants” list of 18 October 2011, forwarded to Epstein two days later, which ends “Barnaby Marsh”. So the release itself names the Marsh in this planning group, in full, three times, and the identification of the bare surname in the two lists above is not this desk’s inference. The company the bare surname keeps, Krauss and Church and Nowak and Minsky, all of whom appear beside him elsewhere in this record, corroborates it and is no longer what it rests on.²⁸

The second is an undated sheet headed “92nd St Y potential list”, which gives “Total – 63 people,” and “Blue eyes – 49” above a three-column table of name, field and eye colour. Marsh’s row reads “Barnaby Marsh”, an empty printed FIELD cell, and “blue”.²⁹ A message of 25 December 2014 in the release carries “92nd St Y potential list” as its subject line.³⁰

Science, and the last correspondence

Not all of it was scheduling. On Monday 28 March 2016 the two exchanged eight messages in a day, five from Marsh and three from Epstein, and the thread is quoted here as the sheets print it, soft breaks and all. Marsh: “in the context of risk with chance events, always the=e problems of dealing with various “freak” outliers and coincide=ces providing exceptions and wrecking models, but let’s turn it on its=head- how are contingencies structured to block, bridge, or amplify other =ontingencies, to shift the entire game space? Is this really what evolutio= is doing all the time??” Epstein, earlier the same day: “i have some ide=s. are you around tomorw or wed. I might hav= to cancel fri trip. . I think your chance luck idea, is=critical”. And Marsh’s last message on the run, the one at the head of the first sheet: “Of course it can be purely deterministic but interface with schrodinger mea=ure problems very strange”.

The message that starts the day is Marsh’s of 8:48 AM, and it runs over onto the second sheet, where it ends: “Lets discuss. FY=, I have been thinking about pioneering new sciences of causality, rare ev=nts causing cascades within systems, understanding distributions, and how =o represent these in fresh ways to allow new kinds of science to happen.” What he was answering is printed directly beneath it: Nowak’s message of the previous afternoon, “what about an institute in cambridge for / math of life, biological computing, the next step of evolution / beyond language”. Nowak’s line is in the chain because Marsh’s first message of the day was a reply to it.³¹

In January 2019 the subject was James Watson. Epstein wrote on 2 January 2019, “any news on watson. the press is awful to him.”; Marsh answered, “No. Gerry Ohrstrom is hosting a screening of a new movie about Jim W. at his home tonight (I can’t go); I imagine Jim will be there…” and then, later the same morning: “just saw the ny times piece. irrespective of views being “correct”, no room for any contrarian viewpoints. there are some top people quoted whom we know, and it is disappointing”.³²

Then the subject became Epstein himself. On Friday 1 March 2019 Marsh wrote, “i’m here - what is happening?”; Epstein answered, “Nothing special, want to help”; and Marsh replied:

:) all good, but keeping in mind that life is far too=short - there is fun to be had - and history to be made… so lets do what=we can when we can…

I’m heading out of town in a few h=urs but back sunday morning…are you stlll here sunday or next week?if no=, lets skype any morning next week…³³

On Monday 4 March 2019 he wrote under the subject line “Darren Indyke?”: “i’m getting messages from this guy asking to clic= to download files - i don’t open things like this but just typed his =ame in and saw he was once connected to you… any idea, other than him be=ng hacked? / call tomorrow morning?” Epstein’s one-word answer the same afternoon, on a second sheet, is “Hack”.³⁴³⁵

Later that day Epstein sent him a link to a set of letters published by the New York Times about himself. Marsh replied just after midnight UTC on 5 March 2019: “sad to see all of this- not a legal issue at all, just a media circus”.³⁶

On Saturday 16 March 2019 Epstein wrote “skype?”; Marsh answered at 13:18:37 +0000, “yes- tomorrow- sunday morning ca 7?”; and Epstein, at 13:50:26 +0000, “ok”. The quoted clock inside the first sheet carries no zone and no interval is asserted from it.²⁴

The correspondence runs three days further than that, and its subject is not scheduling. The last exchange between them that any search set out below reaches is of Monday 18 and Tuesday 19 March 2019, under the subject “Re: humans sense magnetic fields???”. The sheet that carries it is Epstein’s, dated on its face Tuesday, March 19, 2019 12:20 AM, and its whole body is “i saw it , thought of our conversation”; the message it answers is quoted on the same sheet as Marsh’s of “Mar 18, 2019 at 8:02 PM”, with a link to a Science news item on magnetic sense. Neither field carries a zone and no interval is asserted between them.¹³

A second sheet of the same date is not correspondence between the two but a third party’s forward to Epstein, and it is worth the page for two reasons. It carries a message Marsh wrote, dated on the forwarded header “March 18, 2019 at 7:59:47 PM =DT”, reading “Wish we had more time! Lots to work on, and possibilities are =ast! / Best, / Barnaby”, addressed to a recipient the producing party has barred, under a subject line the forward carries as “Fwd: Great to see you today”; the forwarder, whose name and address are also barred, adds ”:) I guess he pinged me after you !” And beneath Marsh’s sign-off the sheet prints a signature block giving a title that appears nowhere else in the sheets read for this page: “Dr. Barnaby Marsh / Founder, / Saint Partners Philanthropy =ervices / www.saintpartners.com”. This page does not say who Marsh met that day, or who the forwarder is: the recipient of his message and the sender of the forward are both barred, and Epstein is named on the sheet only as the person the forward was sent to.⁷

The two trust instruments

He is named in two of Epstein’s trust instruments, and his position in them is not the same.

The Jeffrey E. Epstein 2018 Trust, dated 8 May 2018, by and among Jeffrey E. Epstein as Grantor and Darren K. Indyke, Richard D. Kahn and Katheryn H. Ruemmler as Trustees, provides at Article VII, Section 7.1(B) that where a Trustee of the Trust resigns, is removed, becomes incapacitated or is unwilling or unable to serve, “such that the total number of Trustees acting on behalf of the Trust at all times shall be three. BARNABY MARSH shall first be appointed to fill any such vacancy, and thereafter the successor Trustee appointed to fill any such vacancy shall be appointed by the Trustees then acting, including, but not limited to, if there are fewer than three Trustees then acting”.⁴

The 1953 Trust, “made as of August 8, 2019”, by and among Jeffrey E. Epstein as Grantor and Darren K. Indyke and Richard D. Kahn as Trustees, provides at Article VI, Section 6.1 that the number of Trustees “at all times shall be two”, and sets out a chain:

BORIS NIKOLIC shall be appointed to fill any such vacancy, and if he is unwilling or unable to serve, BARNABY MARSH shall be appointed to fill any such vacancy. If BARNABY MARSH is unwilling or unable to serve, then ANTHONY BARRETT shall be appointed to fill such vacancy. If ANTHONY BARRETT is unwilling or unable to serve, then KATHY REUMMLER shall be appointed to fill such vacancy. If KATHY REUMMLER is unwilling or unable to serve, then DAVID MITCHELL shall be appointed to fill such vacancy. If DAVID MITCHELL is unwilling or unable to serve, then STEVE HANSON shall be appointed to fill such vacancy. If STEVE HANSON is unwilling or unable to serve, then EVA DUBIN shall be appointed to fill such vacancy and thereafter the successor Trustee appointed to fill any such vacancy shall be appointed by the Trustees then acting, or, if as a result of said vacancy, only one Trustee is then acting, by the sole remaining Trustee then acting.³

“KATHY REUMMLER” is as printed on that sheet; the 2018 trust’s own first page spells the same surname “KATHERYN H. RUEMMLER”. Neither sheet read for this page states that Marsh was asked, consented, or knew. Both instruments make the same requirement, in the same words: Section 6.2 of the 1953 Trust and Section 7.2 of the 2018 Trust each provide that “The appointment of a successor Trustee hereunder shall become effective when such successor Trustee signs an acceptance of the trust”, and no acceptance signed by him appears in any search set out below.³⁴¹⁴

The 1953 Trust is dated 8 August 2019, thirty-three days after Epstein’s arrest on 6 July 2019 and two days before his death on 10 August 2019. The correspondence between him and Marsh in the searches run for this page ends 142 days earlier, on 19 March 2019.³¹³¹⁴

What the search returned

The corpus of released files was searched with the FTS5 index on 2026-09-10. Counts are documents, with pages given beside them. FTS5 folds case, so these are upper bounds for exact-case matching and lower bounds wherever the extraction has damaged the name.¹⁴

form searched	pages	documents
`"Barnaby Marsh"`	2,985	2,882
`"Barnaby"` (forename alone)	4,062	3,882
`"Marsh"` (surname alone)	5,687	5,280
`Barnaby*` (prefix)	4,066	3,886
`Marsh*` (prefix)	11,375	8,435
`"Marsh, Barnaby"`	146	139
`"B. Marsh"`	5	5
`"Dr. Barnaby Marsh"`	1	1
`"bmarsh"`	7	7

The surname alone and the forename alone were both searched, and each reaches documents the full name does not. Every document returned by "Barnaby Marsh" is inside both the forename set and the surname set, so neither adds anything by itself; what they add is reached through damaged spellings, below, and through the bare surname, at the end of this section.¹⁴

Damaged spellings, and the largest single finding of this page. The extraction reads the letter pair rn as m throughout this corpus, so Barnaby prints as Bamaby in the text layer of a very large part of the record. Measured:

"Bamaby Marsh": 2,350 pages in 2,311 documents, of which 1,733 are outside the "Barnaby Marsh" set entirely
"Bamaby" (the forename alone): 2,999 pages in 2,913 documents. Searched, listed, and deliberately NOT in the union below; see the paragraph after next
"Barnab Marsh" (trailing y lost): 34 pages in 34 documents, 10 outside
"wlBarnaby" (the day sheets’ w/ read as wl): 36 pages in 36 documents, all 36 outside
"w Barnaby" (the day sheets’ w/ tokenised): 584 pages in 573 documents, 45 of them outside the "Barnaby Marsh" set
"withBarnaby" and "withBamaby": 4 documents, all outside
"Samaby Marsh" 6, "Banaby Marsh" 4, "Baamaby Marsh" 4, "Baarnaby Marsh" 4, "Barnabi Marsh" 2, "Barnaby Marsha" 2, "Bamaby Marsha" 2, "Baroaby Marsh" 1, "Banmby Marsh" 1, "Bamaby Marsb" 2, "Ramahv Marsh" 1, "arnaby Marsh" 11

Taking "Barnaby Marsh" and the eighteen spellings above together, the subject is reachable in 4,731 documents, and 1,849 of them, 39.1% of that union, are invisible to "Barnaby Marsh". The union was built as a set of identifiers and counted, not summed: "Bamaby Marsh" alone returns 2,311 documents, 578 of which are also in the "Barnaby Marsh" set, leaving 1,733; the remaining 116 come from the other seventeen forms. Five of the documents read at the image for this page are in the outside class: EFTA01000685, EFTA01001452, EFTA01009708, EFTA01031541 and EFTA01035784 all print “Barnaby Marsh” correctly on the page image while their extraction reads “Bamaby Marsh”. EFTA00749518 is not one of them, although its extraction carries the damaged form too: that sheet’s extraction prints both spellings, so the correctly spelled name reaches it and it sits in the overlap of 578. This is a fault in the text layer and not a second person: twenty-five of the 1,733 were read at their rows for this page and every one is From:/To:/Subject: traffic with Epstein’s address, with Lesley Groff, or an entry in an Epstein day sheet. Not one is a different Marsh.¹⁴

A nineteenth damaged form was searched and is deliberately not in that union. "Bamaby", the damaged forename standing alone, returns 2,913 documents and would add 523 that no other form above reaches. But a sample of twelve of those 523 found eleven that are this subject and one that is not, a different person whose forename is the surname Barnaby. The bare correct forename is excluded from the union for the same reason, so the bare damaged forename is excluded with it, and the 4,731 is the smaller and safer figure. The cost of that caution is visible on this page: EFTA01829404, the 18 June 2009 Krauss sheet quoted above, is in the 523 and not in the 4,731, because its extraction reads “met with Bamaby in NYC” and the sheet never prints the surname at all.¹⁴¹⁶

Why this subject loses 39% when the same fault costs most subjects nothing: it is a property of the scan batch, not of the name. The corpus records each document’s production batch. Of the 4,731, batch 9 holds 2,082, batch 10 holds 1,361 and batch 11 holds 1,165, with 123 unassigned or in small batches. The damaged-only class is not spread evenly across them:

production batch	documents in the union	of those, reachable only through a damaged spelling	share
9	2,082	1,249	60.0%
10	1,361	461	33.9%
11	1,165	103	8.8%

So two thirds of the 1,849 sit in one batch, whose scans are the poorest in the release for this correspondence. Measured against ordinary words elsewhere in this release the same rn→m fault costs 0 to 4%, with Osborne losing 4% and International 0%, and the desk’s own measurement of the same fault across fifteen further terms found it reaching 70% on only one of them, a foreign forename. The 39.1% here is an outlier with a documented cause, and it should not be read as a claim that the release loses two documents in five generally.¹⁴

These returned nothing: "Bamaby Marsli", "Barnaby Marsli", "Marsli", "Bernaby", "Barnady", "Bamaby Mardi", "Barnaby Marah", "Bsrnaby", "Bartlaby", "Barnaby Marslt", "Marhs", "Marrsh", "Masrh", "Barnaaby", "Bnrnaby", "Baraaby", "Bamady", "Barnaby Mash", "Bamaby Mash", "Bamnby", "Bansby", "Bomaby", "Barnaby Msrsh", "Bamaby Msrsh", "Barnaby Marsn", "Bamaby Marsn".¹⁴

Homographs, named and excluded. Marsh is a common surname, a common place-name element and a common word. The surname alone returns 5,280 documents, and 626 of them are reached by no spelling of the full name: not "Barnaby Marsh", not a damaged spelling of it, not "Marsh, Barnaby", "B. Marsh" or "Dr. Barnaby Marsh". Those 626 were classified by a first-match-wins keyword classifier, in the order printed here:

class	documents
Marsh Law Firm and James R. Marsh, counsel for Epstein claimants	98
Marsh & McLennan, Marsh USA and Mercer Marsh, the insurance broker	15
Marsh Harbour, Abaco	20
Russo Marsh, a political consultancy	14
Old Marsh Golf Club	10
”March” misread as “Marsh” by the extraction, in a date or statement-period line	6
case citations naming a party surnamed Marsh	21
other individuals surnamed Marsh, matched on a forename beside the surname	36
wetland or place-name sense	21
not classified (see below)	385

A homograph of the forename also sits in the "Barnaby" set and is not this subject: a Yuko Barnaby, who appears in her own right in the correspondence. Barnaby is her surname, and it is his forename that collides with it. The damage to her name is the extraction’s and not the sheet’s: a sheet of hers was read at the image at 400 dpi for this page, and it prints “Yuko Barnaby” correctly, with rn, at each of the four places her name appears on it, while the extraction of that same sheet reads “Yuko Bamaby” at all four. That is the same rn-read-as-m fault that damages his forename. Across the corpus "Yuko Barnaby" returns 51 documents and "Yuko Bamaby" 60. She was found in three documents of a forty-document sample of the 1,280 that the bare forename reaches and no full-name spelling does; that class was not classified exhaustively and no count on this page rests on it.¹⁴³⁷

The classifier is crude and the “not classified” row is large; it is not a claim that those 385 are other Marshes. Of the 626 residual documents, 61 carry the bare surname within 250 characters of Nowak, Krauss, Minsky, Templeton, Ito, Groff, Church or Botstein, and the two conference lists quoted above are in that group. The rule that produces that 61 is stated here because an earlier pass gave the figure as 64 with the rule unstated, and the figure moves with the rule: the window is 250 characters on each side of the match, Marsh and each of the eight names are matched on word boundaries, case is folded, and a document counts if any page of it has one such window. Dropping case-folding gives 51; taking the 250 characters before the match only gives 46; after only, 35; halving the window to 125 each side gives 33. And “excluded” in the heading above over-promises by 34: thirty-four of the 626 are inside the 4,731-document union already, reached by one of the four forename-only damaged forms, so they are residual to the surname search and excluded from nothing. So the bare surname does reach this subject, and no count on this page treats the residual as wholly excluded. Nothing in the substance above rests on a bare-surname document except the two lists, and those are now carried by three sheets that print the full name in the same planning sequence.¹⁴⁶²⁷²⁸

Act against mention. Every one of the 4,731 documents was classified mechanically on its strongest page shape, first match winning in the order printed:

class	documents	share
he is the author of a message on the sheet (a `From:` line carrying his name, or ”… Marsh … wrote:“)	2,070	43.8%
addressed to him on a `To:`, `Cc:` or `Bcc:` line, and not the author of anything on the sheet	1,041	22.0%
a day sheet, calendar entry or schedule (two or more of: a `Rec. Loc`, a `w/`-form appointment, a dated weekday heading, a `Reminder:`, a clock time)	972	20.5%
his name in a subject line only	250	5.3%
something else	398	8.4%

The five figures in that table are one pass’s classification and a second reader cannot re-derive them from the rule as stated. The rule does not fix how many characters may separate a header label from its value, what string counts as his name, or whether a document is classified from one page or from all of its pages, and those choices move the counts a long way: a re-gate of this page implemented the rule seven ways and reproduced none of the five, its closest attempt putting the authored class at 2,357 against the 2,070 here. What does hold is that the five classes are disjoint and exhaustive and sum to 4,731, that every share rounds correctly, and that on every one of those implementations which behaved sensibly the authored class was still the largest. On this classification, then, about two in five of the returned documents carry something he wrote, a higher proportion than this desk has measured on comparable pages, and still the wrong measure to stop at. The same 4,731 documents were classified a second time on vocabulary, after the standard confidentiality boilerplate and the Apple property-list block were stripped, against a fifteen-term funding list and a twenty-two-term science list:

class	documents	share
two or more scheduling terms, and fewer than three distinct funding or science terms	2,432	51.4%
neither vocabulary, overwhelmingly one- and two-line messages	2,090	44.2%
three or more distinct funding or science terms	209	4.4%

The authorship figure is high because he answered a great many one-line messages about breakfast. A hand-read sample of thirty documents from the “something else” class found device-extraction reports, LinkedIn digests, other people’s schedules that name him, contact lists, and messages between third parties about arranging to see him; the two trust instruments are in that class.

The vocabulary classifier is a crude proxy, and the size of its substance class is a property of the word list rather than of the record. The 209 (4.4%) in the table is the figure for the thirty-seven-term list printed in [^search], and that list is printed so that the figure can be re-derived; a re-gate of this page re-derived all three rows of the table exactly from the lists as printed. Two further runs were made, with a deliberately narrower list and a deliberately wider one, and their figures are not given here, because the word lists behind them were not recorded and neither figure can be re-derived by a second reader. What those runs showed is a property of the method rather than of the record, and it can be stated without them: the count is governed by the word list, so the 209 is a floor and not a census. The classifier carries false positives in both directions: some of the 209 are Krauss–Epstein exchanges that happen to name him, and at least one substantive message, the 28 March 2016 thread quoted above, is short enough to fall outside it, which is a demonstrated false negative rather than an estimated one. These two tables were re-derived for this page over the 4,731-document union; an earlier pass classified a working set of 4,743 documents whose composition could not be reproduced from the spellings listed above, and its counts are not carried here.¹⁴

Absence claims, and the searches built to falsify them. Two statements on this page are negative and both were tested with searches that could have returned a disconfirming document.

First, the date on which the correspondence ends. An earlier pass of this page gave that date as 16 March 2019, and it was wrong by three days. The census now run for it takes every D Mon YYYY, Mon D, YYYY and M/D/YYYY date string on every page of the 4,731-document union, takes the maximum per document, and sorts: above the 16 March cluster stand EFTA02638818 and EFTA02639014, both of 19 March 2019, and above those only the two trust instruments of 8 August 2019. Nothing in the union carries a later date than those. What that establishes is that no correspondence between them after 19 March 2019 is reachable by these searches, not that none exists.¹⁴¹³⁷

How the earlier date survived is worth recording, because the defect was in the probe and not in the reading. The 19 March sheets were inside the 2,882 documents the correct spelling already returned, and one of them is the single document the "Dr. Barnaby Marsh" row of the table above reports: tabulated, counted, and never opened. The probe that was supposed to catch them ran a month list reading Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec … 2019: it begins at April, so a document dated 19 March 2019 could not have been returned by it whatever it contained. An absence claim is only as good as the probe that tested it, and this probe could not have found a disconfirming document in the one month where a disconfirming document sat. Say what a census covers. This one covers three date forms, on every page of a stated 4,731-document set, with no month list and no keyword; it does not cover date forms it does not parse, documents outside that union, of which the Krauss sheet above is one, or anything the extraction has damaged past recognition.¹⁴

Second, that no acceptance of either trusteeship appears. "acceptance of the trust" and "signs an acceptance" return the two instruments’ own boilerplate and no executed acceptance naming him; "BARNABY MARSH" AND "Trustees" returns nine pages in four documents, which are the two instruments in two copies each. Verifying the negative further would require the probate record, which is not in this corpus.¹⁴³⁴

What an outside report claimed, and what these documents do and do not establish. A working report held by this desk states that he was Executive Vice President of the John Templeton Foundation and managed an $11 million grant, and that he appears across a decade to within days of the arrest. Of that: the title is established, by the inside address of the December 2015 letter and by the 2013 LinkedIn notice.²⁸ An $11M grant is established as his own words, in the message of 12 May 2009, where he calls it “our grant” and describes splitting it $6M / about $5M. The sheet does not say which entity received it beyond naming Harvard and Nowak’s centre, and it does not say he “managed” it.¹ The decade is established, from April 2009 to March 2019.¹²¹³ “To within days of the arrest” is not established by anything read for this page: the correspondence in these searches stops in March 2019, and the document that falls closest to the arrest is not correspondence at all but the trust instrument of 8 August 2019.¹⁴³

Note on the documents

Thirty-six sheets across thirty-two documents were read at the image at 400 dpi for this page, and two of those sheets were re-read at 600 dpi to settle one character each; every quotation above is taken from a page image. Thirty of the thirty-seven footnotes below assert an image read; seven carry an UNVERIFIED mark on at least one sheet they cite, each in those words and each saying what verifying it would cost. One footnote is in both lists, asserting an image read on the sheet it quotes and an UNVERIFIED mark on two further sheets of the same thread; [^search] claims neither, because it describes searches rather than a sheet. 30 plus 7, less the one counted twice, plus [^search], is the thirty-seven.

Every block-quoted line on this page was verified by hand against its sheet. Automated quotation checking on this wiki binds only spans carried inside quotation marks on a single line, so none of the nineteen block-quoted lines here is covered by it. Each was matched, line by line, against its own document, and four of the nineteen diverge from the extraction of their own sheet: the two settled at 600 dpi below, the $3-4 million line, and the Krauss line of 18 June 2009, whose extraction reads “met with Bamaby in NYC” where the sheet prints “Barnaby”.

Six faults in the text layer matter to anyone re-deriving this page.

One. The extraction reads rn as m, so the subject’s forename is Bamaby across a large part of the corpus, and 1,849 documents in the union described above are beyond the reach of the correctly spelled name. Five of the sheets read at the image for this page are in that class and every one prints “Barnaby Marsh” correctly. EFTA00749518 was listed as a sixth by an earlier pass and is not one: its extraction carries both spellings, so the correct spelling reaches it.¹⁴

Two, and it is the more dangerous. EFTA01000685 and EFTA01000686 are two sheets carrying the same message of 12 November 2014. At 400 dpi both print “$3-4 million”; the extraction of EFTA01000685 drops the hyphen and reads “$34 million”, while the extraction of EFTA01000686 reads it correctly. A page working from the extraction alone would find two copies of one sentence disagreeing by a factor of roughly ten, and would have a finding about a foundation’s grant-making that does not exist. Both sheets were opened before this page said anything about the figure.¹¹²⁰

Three. The extraction corrupts URLs the sheets print correctly: the Templeton document link on EFTA02439712 extracts as templeton.oro/pdfs/submittino a proposal/cost effectiveness.odf where the sheet prints http://www.templeton.org/pdfs/submitting_a_proposal/cost_effectiveness.pdf, and the New York Times link on EFTA01031541 extracts as hups://www.nytimes.corn/… where the sheet prints https://www.nytimes.com/….⁵³⁶

Four. On the same May 2009 sheet the extraction reads “I want to make this world Ideas welcome… —S.” where the image prints “I want to make this work! Ideas welcome… –B.”¹

Five, and both were settled at 600 dpi because a single character decided each. On EFTA00749518 p.1 the extraction reads “an extemal account” where the sheet prints “an external account”. That is the same rn-read-as-m fault that hides 1,849 documents from the correctly spelled name, appearing here inside an ordinary word on a sheet this page quotes at length. And on EFTA02633099 p.1 the extraction reads “are you still here sunday” where the sheet prints “are you stlll here sunday”, with three undotted l glyphs against the dotted i of “in” and “if” on the same line; “is” is on that sheet but in the preceding paragraph, not on that line. The extraction silently repairs the writer’s own typing slip. Each was noticed only because every blockquote was matched against its document by hand, and in each case the image was opened before the page settled on a reading.¹³³

Six. Several sheets in this correspondence carry quoted-printable soft breaks and HTML fragments on their face, among them =e are not in the same cities, p=tting together, $2=to $4MM, clic= to download, be=ng hacked, =/div>, i&=39;m here, too=short and =blockquote class="gmail_quote". These are what the documents print and they are reproduced unrepaired.¹⁰³⁴³³³¹

Field states, since a rule is not a signature and a blank is not a bar. The Enhanced Education letter at [^enhanced2015] is signed in ink, in blue, over the typed name, and no value on that sheet is barred. The second executed version at [^enhanced2015b] is also signed in ink, and two values on that sheet are barred by the producing party: the organisation’s own business telephone line on the letterhead and the academic post-nominal in the inside address. Neither class is protected and no value of either is taken from either copy. On [^signal] and on the lower block of [^metrics] the Subject label prints with nothing after it, which is an empty printed field and not a redaction. On [^sfi2010] the From: field is barred and this page supplies no sender. On [^mar18fwd] the forwarder’s name and address are both barred and this page supplies neither, and on [^minskylists] one of the three sheets has its whole To: line barred and this page supplies no recipient. On [^daysheet2018] a contact value after his name is barred, as are values after several other entries on that sheet. On [^may2009], [^nov12a], [^nov12b], [^nov23], [^ito2013], [^itoproposal], [^watson], [^circus], [^last2019], [^indyke], [^march1], [^schrodinger], [^krauss2009], [^mar19], [^dejongh2] and [^signal] at least one address value inside the angle brackets of a header or a quoted header is barred; the counts differ from sheet to sheet and are given in the footnotes, and no value of that class is taken from any of them. On [^dejongh1] four address values are barred. On the two trust sheets at [^trust1953] and [^trust2018] nothing is barred at all. On [^ninetysecond] Marsh’s FIELD cell is an empty printed table cell, not a bar; one other row on each of that document’s two sheets carries a green highlight behind the name.

Clock times. Where a clock time is given above it is either taken from an RFC Date: header carrying +0000 and printed in that form, or it is labelled Eastern, which is done only for the May 2009 run whose own header fixes the offset, as set out where those times are given. Unzoned Sent: fields are otherwise not converted into clock times on this page, no interval is asserted between two unzoned clocks, and no zone is asserted for any sheet outside a run that fixes one.

Protected values. Identity numbers, account numbers, telephone and fax numbers, e-mail addresses, dates of birth and residential addresses printed on these sheets are not reproduced on this page. Each of those seven categories was searched for by hand across this page’s text as well as by the desk sweep, including the non-US telephone shapes and the spaced sixteen-digit account shape the sweep does not see, and the dash-date forms of a date of birth that it does not see either.

People: Jeffrey Epstein; Martin Nowak; Joi Ito; Lawrence Krauss; Darren Indyke; Richard Kahn; Lesley Groff; Leon Black; Robert Trivers
Entities: Enhanced Education; Epstein and Harvard University; MIT Media Lab

Sources

E-mail chain, Jeffrey Epstein and Barnaby Marsh, Thu, 07 May 2009 to Tue, 12 May 2009 20:35:37 +0000, three of Marsh’s messages nested under Epstein’s replies. https://epstein-data.com/EFTA00749518 p.1 and p.2. Both sheets read at the image at 400 dpi. The quoted messages are set in blue type on p.1 and partly in blue and partly in black serif type on p.2; the closing paragraphs of the 7 May message and the signature “Barnaby” are in black. No claim of emphasis is made from either. Three address values on p.1 and one on p.2 are barred, and a contact value after each of the four names in the bulleted list on p.2 is barred. The four names themselves print on the sheet with their universities, an associate professor in Estonia and three academics in Romania; it is the contact value after each that is barred. The 4:27 PM message is quoted here from its opening, not from its second paragraph: it begins “There is lots to discuss regarding other work, interesting people, etc…how about spending an afternoon together, the 15th or 16th of June? Does either of these dates work for you?”, with the ordinals set as superscripts on the sheet, and that is the sentence Epstein’s “16th ,,,would be perfect.” answers. The sheet prints “he has to close PED”, “we though that this would scale up”, and “I want to make this work! Ideas welcome… –B.”; the extraction reads the last of these “I want to make this world Ideas welcome… —S.” A second copy of the 12 May 12:47 PM message is at https://epstein-data.com/EFTA02524563 p.1, not rendered. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹²
Letter on Enhanced Education letterhead, St. Thomas, USVI, dated December 2, 2015, addressed to “Barnaby Marsh, D.Phil. / Executive Vice President, Strategic Initiatives / and Managing Director, Philanthropic Leadership Network / John Templeton Foundation,” salutation “Dear Barnaby:”, subscribed “Very truly yours,” over an ink signature and the typed line “Darren Indyke / Vice President.” https://epstein-data.com/EFTA00314149 p.1. Read at the image at 400 dpi. The instrument is signed in ink: the stroke width varies, it is in blue, it sits off the baseline and it runs over the typed name. No value on this sheet is barred. The letterhead carries the organisation’s own business telephone line. The body quoted in the text above continues to the sum the letter is written about — “Harvard University Fund 347150, $1,500,328 [$1,304,633 direct costs + $195,695 indirect costs]” — and then to a second paragraph naming a three-year period, “[January 1, 2016 - December 31, 2018]”. Further copies of the same letter are at https://epstein-data.com/EFTA00306058, https://epstein-data.com/EFTA00584873 and https://epstein-data.com/EFTA00584874, none rendered. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷
The 1953 Trust, trust agreement “made as of August 8, 2019,” by and among Jeffrey E. Epstein as Grantor and Darren K. Indyke and Richard D. Kahn as Trustees; Article VI, Sections 6.1 and 6.2. https://epstein-data.com/EFTA00099303 p.25 (of 32), bearing the page number 25 and the stamp GJ_00025. Read at the image at 400 dpi. Nothing on this sheet is barred. The successor chain is one continuous paragraph in regular weight, the names in full capitals. “KATHY REUMMLER” is as printed on this sheet. A second copy of the same instrument is at https://epstein-data.com/EFTA01266204 p.25, not rendered. The instrument’s first page and its execution pages were read in the extraction only and are cited here only for the parties and the date; no claim is made here about whether the instrument was signed, by whom, or what it disposed of. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹
The Jeffrey E. Epstein 2018 Trust, trust agreement dated May 8, 2018, by and among Jeffrey E. Epstein as Grantor and Darren K. Indyke, Richard D. Kahn and Katheryn H. Ruemmler as Trustees; Article VII, Section 7.1(B). https://epstein-data.com/EFTA00099449 p.23 (of 29), bearing the page number 23 and the stamp GJ_000451. Read at the image at 400 dpi. Nothing on this sheet is barred. The name prints in full capitals, “BARNABY MARSH,” in the same weight as the surrounding text. A second copy of the same instrument is at https://epstein-data.com/EFTA01266457 p.23, not rendered. The instrument’s first page and its execution pages were read in the extraction only and are cited here only for the parties, the date and the fact of execution. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶
E-mail, Barnaby Marsh to Jeffrey Epstein, Sent: Tue 7/7/2009 12:01:23 PM, “RE:”, above Epstein’s message of Tuesday, July 07, 2009 7:52 AM. https://epstein-data.com/EFTA02439712 p.1. Read at the image at 400 dpi. Neither Sent field carries a zone, so no clock time is stated in the body of this page for this exchange. The Subject label on the lower block prints with nothing after it. The sheet prints the link as http://www.templeton.org/pdfs/submitting_a_proposal/cost_effectiveness.pdf; the extraction reads templeton.oro/pdfs/submittino a proposal/cost effectiveness.odf, which is why no search on the correctly spelled host reaches this sheet. A second copy is at https://epstein-data.com/EFTA02439695 p.1, not rendered. ↩ ↩² ↩³
E-mail chain, “Nowak, Martin” to Jeffrey Epstein, CC Cecile de Jongh, Tue, 15 Nov 2011 23:03:13 +0000, Subject “Re: Some other Possible Invitees for NEXT conference,” quoting de Jongh’s message of Nov 15, 2011 at 10:22 PM. https://epstein-data.com/EFTA00661788 p.1. Read at the image at 400 dpi. Four address values on the sheet are barred, counting the inline attribution as a header, which is how this page counts them elsewhere. The list prints as two labelled blocks, “Inculdes:” (as spelled) and “Declines:”, each name on its own line in regular weight. It is not the whole of de Jongh’s message: it is followed on the sheet by “With warm regards, / Cecile” and a disclaimer paragraph. The sheet gives no forename for “Marsh” and this page supplies none. ↩ ↩² ↩³
E-mail forwarded to Jeffrey Epstein, Sent: Tuesday, March 19, 2019 2:05 AM, Subject “Fwd: Great to see you today”, whose forwarding line reads ”:) I guess he pinged me after you !” https://epstein-data.com/EFTA02638818 p.1. Read at the image at 400 dpi. The forwarder’s name and address are both barred, and this page supplies no name, role, employer or attribute for that person. The forwarded message is headed “From: =/b>“Dr. Barnaby Marsh"", “Subject: =/b>Great to see you =oday” and “Date: =/b>March 18, 2019 at 7:59:47 PM =DT”, above a To: line whose name and address are barred and a salutation which is barred; its body reads “Wish we had more time! Lots to work on, and possibilities are =ast! / Best, / Barnaby”, and beneath that a signature block reading “Dr. Barnaby Marsh / Founder, / Saint Partners Philanthropy =ervices / www.saintpartners.com”. Six name or address values on the sheet are barred. The =/b>, =oday, =ast, =ervices and =DT are quoted-printable soft breaks on the face of the sheet and are not repaired here; =DT stands where a time-zone abbreviation belongs and no zone is asserted from it. Nothing on this sheet states whom Marsh met on 18 March 2019: the recipient of his message is barred, and Epstein appears on the sheet only as the person to whom the forward was addressed. This is the single document returned by the "Dr. Barnaby Marsh" row of the search table above — one page, one document, counted there by an earlier pass and not opened. ↩ ↩² ↩³ ↩⁴
LinkedIn connection notice to Jeffrey Epstein, Sent: Wed 8/21/2013 1:52:13 PM, Subject “See Barnaby’s connections, experience, and more…”. https://epstein-data.com/EFTA01959904 p.1. Read at the image at 400 dpi. The sheet reads “Jeffrey, Barnaby has accepted your invitation” in bold, then “Barnaby Marsh” as a bold heading and “Executive Vice President, John Templeton Foundation” in regular weight below it. The profile photograph and two logo positions render as crossed placeholder boxes, which are missing images in the mail render and not redactions. Nothing on the sheet is barred. A second LinkedIn sheet at https://epstein-data.com/EFTA01960823 p.1, not rendered, shows the same title in a third party’s connections list. ↩ ↩² ↩³ ↩⁴
E-mail to Jeffrey Epstein, Sent: Tue 11/30/2010 3:14:26 PM, Subject “reminder”, whose whole body reads “Joint John Templeton Foundation & Santa Fe Institute Meeting on Frontiers of Complexity Science next week, Tues Dec 7 and Wed Dec 8 (Barnaby Marsh)”. https://epstein-data.com/EFTA02415750 p.1. Read at the image at 400 dpi. The From: field is barred by the producing party and this page supplies no sender, role or attribute for it. The Sent field carries no zone. A second copy, carrying the same line inside a longer thread, is at https://epstein-data.com/EFTA02415724 p.1, not rendered. ↩
E-mail, Barnaby Marsh to Joi Ito, copied to Epstein Jeffrey, Sent: Sunday, September 29, 2013 9:23 PM, Subject “Re: some thoughts on designing around our “little mind"". https://epstein-data.com/EFTA02575268 p.1. Read at the image at 400 dpi. The sender’s address value is barred. The = characters quoted in the body are quoted-printable soft breaks carried onto the face of the sheet and are not repaired here. The Sent field carries no zone. The sheet also prints an Apple property-list block beneath the signature. ↩ ↩² ↩³ ↩⁴ ↩⁵
E-mail, Barnaby Marsh to “jeffrey E.”, Wed, 12 Nov 2014 16:36:29 +0000, quoting Epstein’s message of Wed, Nov 12, 2014 at 5:24 AM. https://epstein-data.com/EFTA01000685 p.1. Read at the image at 400 dpi. The sender’s address value is barred. The sheet prints “$3-4 million”; the extraction of this document reads “$34 million”, the hyphen lost. “jjoi” is as printed. ↩ ↩² ↩³ ↩⁴
E-mail, Jeffrey Epstein to Barnaby Marsh, Thu, 16 Apr 2009 22:08:30 +0000, whose whole body reads “time to catch up”. https://epstein-data.com/EFTA00881265 p.1. UNVERIFIED, from a search hit: this sheet was not rendered. It is cited only as the earliest dated message in the returned set; verifying it would cost one page render. A second copy of the same message is at https://epstein-data.com/EFTA01828751 p.1, also not rendered. ↩ ↩² ↩³
E-mail, Jeffrey Epstein to Barnaby Marsh, Sent: Tuesday, March 19, 2019 12:20 AM, Subject “Re: humans sense magnetic fields???”, whose whole body reads “i saw it , thought of our conversation”, above a quoted attribution reading “On =on, Mar 18, 2019 at 8:02 PM Barnaby Marsh” and an indented line carrying message identifiers and a link to a sciencemag.org news item of March 2019. https://epstein-data.com/EFTA02639014 p.1. Read at the image at 400 dpi. Two address values on the sheet are barred. The spacing of the body, and the =on for “On” in the attribution — a quoted-printable soft break on the face of the sheet — are as printed and are not repaired here. Neither the Sent: field nor the quoted clock carries a zone, so no interval is asserted between them and no local date is asserted for either. This document is inside the 2,882-document "Barnaby Marsh" set and is returned at the top of the date census described under What the search returned. ↩ ↩² ↩³ ↩⁴ ↩⁵
Full-text search of the release (full_text_corpus.db, pages_fts, FTS5, rowid-first) on 2026-09-10, in the forms and with the counts set out under “What the search returned”. Every count in that section is a count of documents unless a page figure is given beside it. FTS5 folds case and folds combining accents, so proper-noun counts are upper bounds for exact-case matching and lower bounds wherever the extraction has damaged the name; searching the non-folding letters separately was not required here, the name carrying none. Statements about what the search returned describe those searches and are not claims about the contents of the release as a whole. The union of 4,731 documents is "Barnaby Marsh" together with the eighteen damaged forms listed under “Damaged spellings”; the nineteenth form there, the bare damaged forename "Bamaby", is listed but is NOT in the union, for the reason given at that point. It was built as a set of identifiers and its members were counted, not summed. Every figure in this section was re-derived from the index on 2026-09-11; an earlier pass of this page published a union of 4,743, an outside class of 1,861, a "w Barnaby" outside figure of 41 and a residual of 591, none of which is reproducible from any stated composition, and none of those figures is carried here. The residual homograph table was produced by a first-match-wins classifier, in the printed order, over the 626 documents that the surname search returns and that no spelling of the full name reaches; its “not classified” row is 385 documents and is not a claim about them, and 61 of the 626 carry the bare surname within 250 characters of Nowak, Krauss, Minsky, Templeton, Ito, Groff, Church or Botstein, under the window and matching rule stated where that figure is given. Thirty-four of the 626 are inside the 4,731-document union, reached by one of the four forename-only damaged forms. The act-against-mention table was produced by classifying every page of every document in the union mechanically, first match winning, on From:-line, To:/Cc:/Bcc:-line, day-sheet and subject-line shapes, allowing for the extraction’s habit of putting a header label and its value on separate lines. That description does not fix the character window, the name-matching rule or the page-versus-document unit, so the five counts it produced are not independently re-derivable and are reported as one pass’s classification; the limitation is stated in the text beside the table. The vocabulary table was produced over the same documents after stripping the standard confidentiality boilerplate and the Apple property-list block, counting distinct stems from a fifteen-term funding list (grant, funding, donation, donor, gift, endowment, philanthrop, foundation, matching, contribution, proposal, budget, overhead, million, templeton) and a twenty-two-term science list (science, scientific, research, evolution, physics, quantum, mathemat, biolog, genome, genetic, neuroscience, cognitive, algorithm, institute, professor, harvard, laborator, fellowship, postdoc, experiment, cosmolog, theory); the scheduling class requires two or more of breakfast, lunch, dinner, meeting, appt, appointment, skype, schedule, calendar, tomorrow, the weekday names, flight and confirm. A sample of thirty documents from the “something else” class was hand-read in the extraction only, and twenty-five of the 1,733 damaged-only documents and twelve of the 523 bare-forename documents were read at their rows. The date census that establishes 19 March 2019 as the latest correspondence date was taken over every D Mon YYYY, Mon D, YYYY and M/D/YYYY date string on every page of the union, taking each document’s maximum, and sorting; it was not a keyword search and carried no month list. It cannot see a date form it does not parse, a document outside the union, or a date the extraction has damaged. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹ ↩²⁰ ↩²¹ ↩²² ↩²³
E-mail, Barnaby Marsh to “jeffrey E.”, Sun, 23 Nov 2014 13:38:22 +0000, quoting Epstein’s message of Sun, Nov 23, 2014 at 7:07 AM. https://epstein-data.com/EFTA01001452 p.1. Read at the image at 400 dpi. The sender’s address value is barred. The whole message is in regular weight; nothing on the sheet is emphasised. “makes sure”, “I still am not convinced the Joi” and “involvment” are as printed. Two further sheets in the same thread, https://epstein-data.com/EFTA01001490 p.1 and https://epstein-data.com/EFTA01001492 p.1, carry Marsh’s follow-up of the same evening — “Did speak with Martin- all should be good. Will call in the morning, if that works…” — and Epstein’s reply “Ok”; neither was rendered. UNVERIFIED to that extent; verifying would cost two page renders. ↩ ↩² ↩³
E-mail, Lawrence Krauss to Jeffrey Epstein, quoted on Epstein’s reply of Sent: Thur 6/18/2009 1:52:10 PM, Subject “Re: templeton”, whose body is “did i hear thank you,,”; Krauss’s message is attributed on the sheet to “Thu, Jun 18, 2009 at 9:28 AM”. https://epstein-data.com/EFTA01829404 p.1. Read at the image at 400 dpi. Neither clock carries a zone and no interval is asserted between them. The sheet names him only as “Barnaby.” The surname is supplied on Lawrence Krauss’s page from a different document; nothing on this sheet supplies it and this page supplies none from this sheet. The quoted passage is Krauss’s account of his own meeting and of what he says Marsh wanted: it is not Marsh’s words, and no sheet read for this page records Marsh saying any of it. Krauss’s message continues past the quotation with “I have been thinking a lot about your last email, btw.”, “LMK” and a postscript, above a signature block giving his university titles. Five values in the header and that signature block are barred. The extraction of this sheet reads “met with Bamaby in NYC” where the sheet prints “Barnaby”, which is why the document falls outside the 4,731-document union described under What the search returned. ↩ ↩² ↩³ ↩⁴
E-mail, Jeffrey Epstein to Barnaby Marsh, Thu, 18 Jun 2009 14:11:19 +0000. https://epstein-data.com/EFTA00750209 p.1. Read at the image at 400 dpi. The Subject label prints with nothing after it — an empty printed field, not a redaction. The recipient’s address value is barred. The misspellings “sugestion”, “succesful”, “permuations” and the tripled commas are on the face of the sheet. ↩ ↩²
Invisible Forces and Powerful Beliefs: Gravity, Gods, and Minds, “By the Chicago Social Brain Network,” FT Press, Upper Saddle River, NJ — a manuscript held in the release; the acknowledgement is the closing paragraph of the preface. https://epstein-data.com/EFTA02444849 p.6. Read at the image at 400 dpi. The paragraph is in regular weight throughout and nothing on the sheet is barred; the sheet’s own Bates stamp is EFTA02444854, the run’s numbering advancing per page. The title page is p.1 of the same document. A second copy of the same passage is at https://epstein-data.com/HOUSE_OVERSIGHT_021247, not rendered. ↩
E-mail, Joi Ito to Barnaby Marsh, copied to Jeffrey Epstein, Sent: Wednesday, September 3, 2014 10:02 AM, Subject “draft of proposal”, Attachments “Templeton Media Lab proposal October 2014 Joi.docx; Untitled attachment 00004.txt; signature.asc”. https://epstein-data.com/EFTA02708809 p.1. Read at the image at 400 dpi. The sender’s address value is barred. The =et and =f in the quotation are quoted-printable soft breaks on the face of the sheet and are not repaired here. The Sent: field carries no zone. The same message and the same attachment are set out from the Media Lab end on Joi Ito’s page. ↩
E-mail, “jeffrey E.” to Barnaby Marsh, Wed, 12 Nov 2014 16:44:22 +0000, whose body is the single word “thanks” and which quotes the message at [^nov12a] and Epstein’s own earlier message. https://epstein-data.com/EFTA01000686 p.1. Read at the image at 400 dpi. The recipient’s address value is barred. This sheet also prints “$3-4 million”, and its extraction reads it correctly — so the two copies’ extractions disagree while the two sheets agree. ↩ ↩²
E-mail, “jeffrey E.” to Barnaby Marsh, Fri, 17 Oct 2014 23:05:09 +0000: “yes, gates is going to give 2 million if that helps”, above a message from Marsh of the same evening whose legible opening is “can’t make the”. https://epstein-data.com/EFTA00868955 p.1. UNVERIFIED, from a search hit: this sheet was not rendered; verifying it would cost one page render. A second copy is at https://epstein-data.com/EFTA02515924 p.1, also not rendered. The sheet is cited only for Epstein’s sentence; this page makes no claim about what the quoted message said or about any gift. ↩
The drafting of the November 2015 matching-funds letter, Nowak’s draft text of Nov 19, 2015, and Richard Kahn’s report of Nov 25, 2015 that the letter had been sent to Marsh and that Marsh “approved [l]anguage,” are set out on Martin Nowak’s page from https://epstein-data.com/EFTA02068001 and https://epstein-data.com/EFTA02719709. Neither sheet was rendered for this page and neither is quoted here beyond what that page already carries; this page cites them only to place the letter it does quote. UNVERIFIED to that extent; verifying would cost two page renders. ↩
A second executed version of the same December 2, 2015 Enhanced Education letter, identical in body, salutation “Dear Dr. Marsh:”. https://epstein-data.com/EFTA00602286 p.1. Read at the image at 400 dpi. Also signed in ink, in blue, over the typed line “Darren Indyke / Vice President.” Two values on this sheet are barred by the producing party where the first executed copy bars nothing, and the class of each is stated here because neither is protected: one is the organisation’s own business telephone line on its letterhead, and the other is the academic post-nominal after his name in the inside address. No value of either is taken from either copy. A desk ruling of 2026-09-10 forbade writing this comparison at all; it was amended on 2026-09-11 on the ground that the two differing values are a business contact line and a degree, and the ruling now records both its original wording and the reason it was changed. ↩
E-mail, Barnaby Marsh to “J”, Sat, 16 Mar 2019 13:18:37 +0000 — “yes- tomorrow- sunday morning ca 7?” — quoting Epstein’s message of Sat, Mar 16, 2019 at 8:47 AM, “skype?”. https://epstein-data.com/EFTA01035784 p.1. Read at the image at 400 dpi. The sender’s address value is barred. Epstein’s reply “ok”, timed Sat, 16 Mar 2019 13:50:26 +0000, is on a separate sheet at https://epstein-data.com/EFTA01035782 p.1, not rendered; the third sheet of the run, carrying Epstein’s “skype?” as sent, is at https://epstein-data.com/EFTA01035788 p.1, also not rendered. ↩ ↩² ↩³
Epstein day sheet headed “2018 February,” the entry under “Thurs. Feb. 1, 2018 NY-PB” reading “3:15pm Appt w/Barnaby Marsh” followed by a barred contact value. https://epstein-data.com/EFTA00285523 p.1. Read at the image at 400 dpi. The sheet is set in italic type throughout and the date headings are bold italic and underlined. Every time label in the Thurs. Feb. 1 block is bold italic, which is why all five times quoted in the text above are bolded there; in two of those five the word after the time is bold as well, “12:30pm LUNCH” and “4:00pm Appt”. The bolding is not universal on the sheet: two travel entries under Fri. Feb. 2 print their time labels in regular italic. No emphasis is claimed for his name, which is regular italic like every other name on the sheet. Contact values after several entries, one entry’s whole subject, four record locators and several traveller names are barred. ↩
Forensic extraction report of one of Epstein’s devices, the Skype log entry timed 2016-08-08 10:17:05 (UTC) whose content is “im here” and whose participants are given as “Bamaby Marsh” and “Self ( jeevacation )”. https://epstein-data.com/EFTA01209205 p.9. UNVERIFIED, from a search hit: this sheet was not rendered; verifying it would cost one page render. The report is cited only for the existence of the log class. A second such report at https://epstein-data.com/EFTA01617279 p.4 carries a line dated 2019/02/05 reading “Barnaby Marsh are you in the city?”; also not rendered. ↩
E-mail, Cecile de Jongh to Jeffrey Epstein, Sent: Thur 10/27/2011 7:55:46 PM, Subject “Re:”: “We already asked for 12 rooms. That should be ok, right?” above the numbered list “1. Church 2. Nowak 3. Krauss 4. Marsh 5. Minsky 6. Rosen 7. Wolfman 8.Mayne or Milner 9. Corrina”, then “With warm regards, / Cecile”. https://epstein-data.com/EFTA01854708 p.1. Read at the image at 400 dpi. “8.Mayne” is set without a space after the point, as printed. An earlier draft of this page quoted the eighth entry as “8.Mayne or Milne…”, truncating inside the surname; the sheet prints Milner, and a ninth entry follows. One address value on the sheet is barred. The sheet also carries Epstein’s own message of the same afternoon, “you can book the hotel.. marving will call steven wolfram, , mayne ur milner, , or another harvard mathtmentician corrina, or one of martins friends, so book for the confirmed plus 3 others.” A second copy is at https://epstein-data.com/EFTA02028175 p.1, not rendered. ↩ ↩²
Three messages from Cecile de Jongh in the 2011 conference-planning sequence, each printing “Barnaby Marsh” in full. (1) To Jeffrey Epstein, Sent: Fri 9/23/2011 2:25:39 PM, Subject “Re: Minsky Conference”: “So far we have 9 people including Minsky (plus spouses so I about 14 - 16 with them included) Only 9 will be in on the meetings plus you (and Gloria???).” above a numbered list running “1. Marvin Minsky 2. Lawrence Krauss 3. George Church 4. Martin Novak 5. Sean Parker 6. Lee Smolin 7. Steve Sinofsky 8. Jim Watson 9. Barnaby Marsh”. https://epstein-data.com/EFTA01855058 p.1. (2) Sent: Fri 10/14/2011 5:01:09 PM, Subject “Minsky Conference”: “Can you get me the contact information for the following persons please? Thanks.” above eight names ending “Barnaby Marsh”. The whole To: line is barred, as is the name in the salutation, and this page supplies no recipient, role or attribute for either. https://epstein-data.com/EFTA02182019 p.1. (3) To Epstein, Sent: Thur 10/20/2011 9:17:55 AM, Subject “Fwd: Time Sensitive - Jeffrey Epstein Conference”, forwarding her own message of “Tue, Oct 18, 2011 6:43 pm” whose “Other participants:” list runs “Marvin Minsky / George Church / Martin Nowak / Lee Smolin / Sean Parker / Lawrence Krauss / Jim Watson / Barnaby Marsh”. https://epstein-data.com/EFTA02022150 p.1. All three sheets read at the image at 400 dpi. “Martin Novak” is as printed on (1) and (2); “Martin Nowak” is as printed on (3). Two values are barred on (1) and five on (3); on (2) the whole To: line and the salutation name are the barred values. EFTA01855058 and EFTA02022150 each run to two sheets and only the first of each was read; nothing is claimed here about their second sheets, and each list is complete on the sheet read. These three are cited for the identification of the bare surname in the two lists quoted above and for nothing else. ↩ ↩²
Undated sheet headed “92nd St Y potential list,” giving “Total – 63 people,” “Blue eyes – 49” and “Unclear (might be blue, but not 100% sure) - 2” above a three-column table of NAME, FiELD and an unlabelled third column carrying an eye colour. https://epstein-data.com/EFTA00583600 p.1 and p.2. Both sheets read at the image at 400 dpi. The heading “FiELD” is as printed. Marsh’s row is on p.2 and reads “Barnaby Marsh”, an empty printed table cell in the FIELD column, and “blue”. Most names print in blue underlined link type; two names on the two sheets carry a green highlight behind them. Nothing on either sheet is barred. The sheets carry no date. ↩
E-mail to “jeffrey E.,” Sent: Thursday, December 25, 2014 5:07 PM, Subject “Re: 92nd St Y potential list,” quoting Epstein’s message of the same day: “we will =ave to go through each on of these names and people videos etc”. https://epstein-data.com/EFTA02343458 p.1. UNVERIFIED, from a search hit: this sheet was not rendered; verifying it would cost one page render. It is cited only for the subject line, and no claim is made that it transmitted the sheet at [^ninetysecond]. ↩
E-mail chain, Barnaby Marsh and “jeffrey E.”, Sent: Monday, March 28, 2016 8:15 PM. https://epstein-data.com/EFTA02466937 pp.1, 2 and 3. All three sheets read at the image at 400 dpi. Eight messages of 28 March 2016 are nested in the run — Marsh at 8:48 AM, Epstein at 8:59 AM, Marsh (“Yes, might be around both days.. Morning mee=ing either day???”, quoted without a clock time), Epstein at 11:07 AM, Marsh at 12:35 PM, Epstein at 2:01 PM, Marsh at 4:10 PM and Marsh at 8:15 PM — five of them his and three Epstein’s; beneath them is “Nowak, Martin” of Sun, Mar 27, 2016 at 4:49 PM. An earlier draft of this page and a desk ruling both said five, which is the number Marsh sent, and both said this footnote had read only p.1; both have been corrected. p.1 carries the 8:15 PM message and everything down to the opening of the 8:48 AM one. p.2 carries the rest of the 8:48 AM message — the “Lets discuss. FY=…” paragraph quoted in the text above — then Nowak’s quoted message in full, then two blocks of confidentiality boilerplate. p.3 carries the remainder of that boilerplate and nothing else. The sheets print quoted-printable soft breaks and raw =blockquote class="gmail_quote" HTML fragments on their face and they are not repaired here. Several address values on this document are barred. ↩ ↩²
E-mail, Barnaby Marsh to “J”, Wed, 02 Jan 2019 14:27:23 +0000, quoting his own message of Jan 2, 2019 at 9:06 AM and Epstein’s of Jan 2, 2019 at 8:21 AM. https://epstein-data.com/EFTA01009708 p.1. Read at the image at 400 dpi. The sender’s address values are barred in both headers. “Gerry Ohrstrom” and “Jim W.” are as printed. Two further sheets in the same January 2019 Watson sequence, https://epstein-data.com/EFTA02625518 p.1 and https://epstein-data.com/EFTA02605773 p.1, were not rendered. ↩
E-mail, Barnaby Marsh, Sent: Friday, March 1, 2019 5:17 PM, quoting Epstein’s message of Fri, Mar 1, 2019 at 11:5= AM and Marsh’s own of 11:29 AM. https://epstein-data.com/EFTA02633099 p.1. Read at the image at 400 dpi. Address values in the headers are barred. The sheet prints “too=short”, “what=we”, “h=urs”, “stlll”, “if no=” and “i&=39;m here”; none is repaired here. The Sent field carries no zone and the quoted clocks do not reconcile with it on one zone, so no interval is asserted. Two further sheets of the same exchange, https://epstein-data.com/EFTA02633109 p.1 and https://epstein-data.com/EFTA02633074 p.1, were not rendered. ↩ ↩² ↩³
E-mail, Barnaby Marsh to Jeffrey Epstein, Sent: Monday, March 4, 2019 4:44 PM, Subject “Darren Indyke?”. https://epstein-data.com/EFTA02631725 p.1. Read at the image at 400 dpi. The sender’s address value is barred. The sheet prints “clic=”, “=ame”, “be=ng” and a trailing ”=/div>”; none is repaired here. The Sent field carries no zone. ↩ ↩²
E-mail, Epstein to Barnaby Marsh, Sent: Monday, March 4, 2019 4:45 PM, Subject “Re: Darren Indyke?”, whose body is the single word “Hack”. https://epstein-data.com/EFTA02631671 p.1. UNVERIFIED, from a search hit: this sheet was not rendered; verifying it would cost one page render. ↩
E-mail, Barnaby Marsh to “J”, Tue, 05 Mar 2019 00:18:34 +0000, quoting Epstein’s message of Mon, Mar 4, 2019 at 6:29 PM, whose body is a link to a set of letters published by the New York Times on 4 March 2019. https://epstein-data.com/EFTA01031541 p.1. Read at the image at 400 dpi. The sender’s address value is barred. The sheet prints the link correctly as https://www.nytimes.com/2019/03/04/opinion/letters/jeffrey-epstein.html; the extraction reads hups://www.nytimes.corn/…, which is why no search on the correct host reaches this sheet. A second copy is at https://epstein-data.com/EFTA02632463 p.1, not rendered. ↩ ↩²
E-mail, Yuko Barnaby, Sent: Thur 12/1/2016 4:31:34 PM, Subject “Re: Dinner”. https://epstein-data.com/EFTA02196336 p.1. Read at the image at 400 dpi. The sheet prints “Yuko Barnaby”, with rn, at four places on its face: the From: line, the attribution of her own earlier message, and each of her two signature blocks. The extraction of the same sheet reads “Yuko Bamaby” at all four. Several values on this sheet are barred and none is taken. This sheet is cited for the spelling of her name and for nothing else: this page states no role, employer or other attribute for her, and asserts no connection between her and the subject beyond the collision of his forename with her surname. ↩

tags	person, science-funding, john-templeton-foundation, enhanced-education, mit-media-lab, epstein-and-harvard-university, trusts
aliases	Marsh, Barnaby, Barnaby Marsh, D.Phil., Marsh, Barnaby, Bamaby Marsh, Bamaby, Barnab Marsh, Banaby Marsh, Baamaby Marsh, Baarnaby Marsh, Samaby Marsh, Baroaby Marsh, Banmby Marsh, Barnabi Marsh, Ramahv Marsh

Epstein Data Wiki

Barnaby Marsh

2009: the Templeton side of the Harvard grant

The title the documents give him

September 2013 to November 2014: the proposals, and the match

November and December 2015: the letter addressed to him

The scheduling record

Two lists

Science, and the last correspondence

The two trust instruments

What the search returned

Note on the documents

Sources

Table of Contents

Backlinks

Epstein Data Wiki

Barnaby Marsh

2009: the Templeton side of the Harvard grant

The title the documents give him

September 2013 to November 2014: the proposals, and the match

November and December 2015: the letter addressed to him

The scheduling record

Two lists

Science, and the last correspondence

The two trust instruments

What the search returned

Note on the documents

Related

Sources

Footnotes

Table of Contents

Backlinks