John Templeton Foundation

Type	Grantmaking foundation. Its own application form calls it “the Foundation” throughout and describes it as “legally and ethically bound by Sir John Templeton’s charter to fund only those proposed projects that have the capacity to advance his philanthropic vision”1
Founded	1987, by the investor Sir John Templeton. Both the year and the founder are stated in the Foundation’s own letter of 5 November 20142
Addresses printed in the release	”Radnor, Pa” on the title sheet of its cost-effectiveness paper; “300 CONSHOHOCKEN STATE ROAD, STE 500 / CONSHOHOCKEN, PA 19428” on a grantee’s 2015 Form 990; “300 Conshohocken State Road Suite 500 / West Conshohocken, PA 19428 USA” on the inside address of a letter written to one of its officers345
Abbreviation it uses of itself	JTF, in its own letter of 5 November 2014 and in its own cost-effectiveness paper267
Officers the sheets name	John M. Templeton, Jr., M.D., “President and Chairman”, signing a full-proposal invitation; Charles L. Harper, Jr., D.Phil., “Executive Director/Senior Vice-President”, authoring its grantmaking paper; Barnaby Marsh, “Executive Vice President, Strategic Initiatives / and Managing Director, Philanthropic Leadership Network”; John Seel, given as “of Templeton”; Ayako Fukui, placed “at the foundation” with no title835910
Its own grant rules, as its form states them	Grants in “US dollars, euros, and British pounds”; not “for longer than three years except in exceptional circumstances, and in no case for more than five years”; a grant “only ‘closed’ when the final payment is made (typically 10% of the total budget)“111
Largest sum the release attaches to it	$1,125,000, entered as contributor no. 5 on the 2015 Form 990 Schedule B of NAUTILUS THINK, INC.4
How often it appears	"Templeton Foundation" returns 194 pages in 166 documents; "John Templeton Foundation" 82 pages in 74 documents; the bare token Templeton 757 pages in 651 documents before homographs are subtracted12

The John Templeton Foundation is the Pennsylvania grantmaking foundation whose application process, grant rules, officers and money run through Jeffrey Epstein’s files from about 2000 to 2018. The release holds the Foundation’s own voice three times: a letter of 5 November 2014 inviting the Harvard professor Elisa New to submit a full proposal, over the printed name of John M. Templeton, Jr., M.D., as “President and Chairman”; a completed Letter of Intent on the Foundation’s own form, whose printed instructions set out how long it will fund a project and when a grant is closed; and a thirty-five-sheet paper on grantmaking method written by Charles L. Harper, Jr., its Executive Director and Senior Vice-President.^28131113 It holds one of its officers approving the wording of a donor’s matching-funds letter in three words of his own.¹⁴ And it holds a figure: $1,125,000, entered against the Foundation’s name and Conshohocken address on the 2015 Form 990 Schedule B of the company that publishes Nautilus.⁴

This page is about the institution, and it exists because nobody searched for the institution. The Foundation is named on seven published pages of this wiki and every one of them reached it through a person: through Barnaby Marsh, Martin Nowak, Lawrence Krauss, Joi Ito, Robert Trivers, Michael Segal, or Harvard.¹⁵ Marsh’s own page reports fourteen search forms and every one of them is a variant of his name. 107 of the 166 documents that use the phrase Templeton Foundation contain none of 42 spellings of his.¹⁶ The strongest single item on this page, the Foundation writing in its own voice over its president’s name, sits in a two-sheet document that no person-scoped search for anyone already covered would return.

Nothing on this page is a finding of wrongdoing by the Foundation or by anyone named on it.

What counts as the Foundation acting, and how often it does

A measurement pass adopted by the desk defines an institutional act narrowly, and this page uses the definition unchanged: funding or support of an identified recipient or project · an invitation · a decision · an approval · or representation by a named Foundation officer. Not an act: a proposal, a hope, a criticism, a biographical title, a sponsor list, or a statement that an unnamed “head” was present.

That definition has one boundary it does not settle, and the boundary decides the figure: does a recipient’s own statement that the Foundation funded him count? A journal acknowledgment, a CV line, a solicitation letter’s self-description and a grant application’s account of existing support are all third parties saying the Foundation paid for something. The first version of this page counted three of them and excluded four materially identical ones. That is not a rule but an oversight, and a gate reading the sheets found it.

The rule this page now applies, stated so it can be re-run: a document records the Foundation acting if it states that the Foundation funded or supported an identified recipient or project, whoever states it (the Foundation, the recipient, a journalist, or any other third party reporting a completed award), and it does not if the funding is sought, hoped for, proposed, promised, refused, merely named as a publisher or sponsor, or attached to no identifiable recipient or project. On that rule an acknowledgment counts and an application does not, and the four journal acknowledgments that were excluded are now counted.

Two words in that rule were bought by documents that turn on them. Promised was added after a re-gate: two sheets report that a physicist “has a promise of several million dollars of funding from the Templeton foundation”, which is neither a proposal nor a completed award, and the earlier wording reached neither side of it.¹⁷ And a journalist is there because the rule as first written did not mention press at all, while the table quietly excluded a press report under a class label inherited from an older taxonomy. A reader re-running the rule got a different number from the one printed, which is the one failure a stated rule cannot survive. The document is a Wall Street Journal obituary reporting that “The foundation’s annual Templeton Prize, valued at $1.7 million, this year went to the founder of a global network of communities where people with intellectual disabilities, and their caregivers, live and work together.”¹⁸ That is a completed award, of a stated sum, to a recipient the sentence identifies by a unique description though it does not name him. It is counted. The alternative was to write a press exclusion after seeing which single document it would exclude, which is fitting the rule to the answer. A reader who thinks a newspaper report is weaker evidence than a grantee’s own acknowledgment can subtract it: without it the figure is 11 of 42, or 26.2%.

That test is the whole of the difficulty with this subject. The Foundation’s name is common in these files and the Foundation’s conduct is less so, and much of what looks at first like the Foundation is a title in somebody’s signature block or a line on somebody’s schedule.

The sample is systematic and reproducible: the 166-document "Templeton Foundation" population, sorted by identifier, every fourth document taken beginning at the first. It returns 42 documents and 54 hit pages, all 54 read.¹⁹

class	documents	share
the Foundation acting	12 of 42	28.6%
discussed by third parties	24	57.1%
grant or programme named in passing	1	2.4%
letterhead, title, sponsor or connection list	5	11.9%

The old table carried a fifth class, “press or biography”, holding one document; that class is now empty and is gone from the table.

A little over one document in four records the Foundation doing something; the rest are other people talking about it. Ten of the twelve were read at the image and none changed class there. The remaining two are House Oversight documents, and they are unreadable at the image by anyone: epstein-data.com/pdf/ returns HTTP 403 for both, measured 2026-09-11 against a control identifier that returns 200. On the image-verified subset alone the figure is 10 of 42, or 23.8%; the 28.6% counts the two 403 documents on the same rule as the rest, and says so rather than quietly dropping them.¹⁹

There is no cross-check on this figure, and an earlier version of this page implied there was. It offered the agreement between this sample at 16.7% and the desk’s disjoint sample at 19.0% as corroboration. It is not corroboration. Both samples were classified by the same reader-rule against the same text layer, so their agreement measures how reproducibly one rule can be applied, not whether the rule is right; and the rule was in fact wrong in both, in the same direction, for the same reason. Two runs of one inconsistent classifier agreeing is the thing that made the error hard to see. What moved the figure was reading the documents rather than counting them. The inconsistency was visible in the text layer, to anyone who read forty-two passages instead of tallying forty-two rows. What the second medium added was different and also necessary: the image confirmed that the numerator fabricated nothing, and it produced a grant number that no text-layer reading could have found, because that sheet’s extraction does not contain it.

Duplication is not spread evenly across the classes, and it moves the figure again. Of the 42 documents, 25 are second, third or fourth copies of eleven underlying items, so the sample is 28 distinct items. On that basis the Foundation acts in 11 of 28, or 39.3%. The document count understates the share because the heavily duplicated material is disproportionately the scheduling and introduction traffic, and none of the five documents added to the count since the first version, the four acknowledgments and the obituary, has a duplicate in the sample.¹⁹

And the phrase population is not free of homographs, which an earlier version of this page implied. One of the 42, EFTA00760692 p.8, uses “Templeton Foundation” only as “Templeton Foundation Press”, the Foundation’s publishing imprint, in a bibliography entry. At 1 of 42 sampled that is 2.4% of the phrase population, far below the bare token’s contamination, but it is not zero.²⁰

The Foundation writing in its own voice

The invitation of 5 November 2014

EFTA01805788 is a two-sheet Gmail export. At the top of the first sheet is a message from Lisa New to Epstein of Thursday 6 November 2014, and beneath it, reproduced in full across both sheets, is the Foundation’s letter of the day before.

On behalf of my father, Sir John Templeton, who establishe= the John Templeton Foundation, I want to thank you for taking the time to=submit an “Online Funding Inquiry” (OFI) to the John Templ=ton Foundation.

We have carefully reviewed the initial =roject you submitted entitled: Poetry in America: The Poetry of Character =evelopment (#58234). We now wish to invite you to submit a Full Prop=sal to us by 3/2/201=, or sooner.²

The = characters are on the face of the document, not introduced here. They are quoted-printable soft breaks carried through into the export and printed on the image at 400 dpi; each replaces one character at roughly fixed intervals, which is why the project title loses its D and the deadline loses its final digit.²

The letter goes on to describe what the Foundation will do with the proposal (“As the relevant JTF staff review your Fu=l Proposal, they will carefully study both the Big Questions you provide t= have greater clarity about the project; and the plausible (albeit not gua=anteed) Outcomes that you will be proposing”), and to say where the Foundation’s own vocabulary comes from: “Big Questions”- =94such as, “Have we begun to realize we are here for a purpose? =80 — were prominent and significant tools for orienting thinking=about projects proposed by applicants for review by Foundation staff and T=ustees.”² It sets out one further step of the process: “You will soon be receiving a se=ond email notification from the Foundation department head who will be ove=seeing the submission and review process for your proposal.”⁸

It closes over a printed name and title:

In Sir John’s gratitude,=br> John M. Templeton, Jr., M.D. President and Chairman John Temp=eton Foundation

ref: a07C000000mmPnv⁸

That closing is printed type and there is no ink on the sheet. It is an e-mail letter reproduced in an export, not an executed instrument: no By: rule, no manuscript signature, no empty rule where one would go. The name and the title are as much part of the printed text as the salutation is. Beneath the addressee block on the second sheet a solid rectangular bar covers what the producing party withheld.⁸

The action and the signature cross the page boundary, so both sheets were rendered and read.

New’s own reply, at the head of the first sheet the next morning, runs to five sentences, of which the first two are these: “Well I’ve cleared the first hurdle at Templeton. If you helpe= in any way, thank you so much.” The remaining three concern a separate proposal and a family wedding.²

The Letter of Intent form

EFTA01149129 is a seven-sheet completed application on the Foundation’s own form, headed “Templeton Foundation LOI / Project Information”. The project is a two-year, $150,000 study of the effect of New York Zen Center for Contemplative Care chaplains on cancer inpatients at Beth Israel Medical Center, with Benjamin Kligler and Roberta Lee as co-investigators and a proposed start date of 1 July 2012.¹³¹¹

The value of the document is not the application but the form. Interleaved with the applicant’s answers are the Foundation’s own printed instructions, and they are the most detailed statement in the release of how the Foundation says it works:

The Foundation will not fund a project for longer than three years except in exceptional circumstances, and in no case for more than five years. Ordinarily grants must be completely closed within three years. These constraints are required by the philanthropic principles laid down by Sir John Templeton in his Charter for the Foundation. Exceptions, including “no-cost extensions” beyond 36 months, will not generally be granted.

A grant is only “closed” when the final payment is made (typically 10% of the total budget). Because the final payment is contingent on the submission and approval of final narrative and financial reports, which can take many weeks to process, all contractually required final reports are due to the Foundation at least three months before the end of a 36-month grant.¹

That paragraph pair is printed twice on the same sheet, once under “Proposed Project Start Date” and once under “Proposed Project End Date”.¹ Elsewhere the form states that “The Foundation awards grants in US dollars, euros, and British pounds”, and asks the applicant to “describe how your vision for the proposed project aligns with Sir John’s Donor Intent”, the last three words set as a blue underlined hyperlink. The response box under that prompt is empty and the counter beneath it reads “1000 characters remaining”.¹¹¹

A separate document records that a Templeton Letter of Intent went to the Foundation through Epstein. On 12 October 2011 Woodson Merrell forwarded to Epstein a message from Benjamin Kligler under the subject “Templeton proposal”, with an attachment named “Templeton Foundation LOIfinal,doc”, and wrote: “Thank you for forwarding this to Templeton. This proposal seems directly on target for what they look to fund in spirituality—here using metrics to evaluate its value in healthcare and wellness.”²¹ That sentence records Epstein’s act and not the Foundation’s. Whether the completed form at EFTA01149129 is the attachment named in that message is not stated by either document. The two share the form name, the co-investigator and the subject matter, and the form’s proposed start date falls nine months after the message; no sheet read for this page joins them, and this page does not join them either.

The cost-effectiveness paper

EFTA00605502 is a thirty-five-sheet paper whose title sheet reads, in bold italic and then in regular type:

Optimizing Program Impact and Cost-Effectiveness Philanthropic Venture Partnership Opportunities for the Non-profit Community Charles L. Harper, Jr., D.Phil. Executive Director/Senior Vice-President John Templeton Foundation Radnor, Pa³

It speaks as the institution throughout (“It is important to recognize that the John Templeton Foundation certainly does not view itself this way. Our activities equate more to philanthropic venture capital investments”), and it sets out a method for scoring grants, including a “simple heuristic success equation” relating success outcomes to money, talent and motivation.²²²³ The paper carries no date on its title sheet; its worked examples cite a Speakers Bureau programme running 11/1/97 to 2/29/00 and the acquisition of Time-Warner by America Online, so it is not earlier than 2000.²⁴²⁵

Its second half applies six named metrics to the Foundation’s Freedom Project academic course competition, among them “Partnership cost leverage”, defined as the ratio R of outside partner funding to Templeton funding, with the stated target that “The program will be successful if (R) increases by 10% each year for the first three years, and 15% each following year until John Templeton funds are no longer necessary to maintain the program”, and “Long-Term Institutional Cost Leverage”, defined as the ratio of donor money institutionalised to “JTF venture seed capital”.⁶⁷ A worked scenario names a real grant: “The Freedom Project gave two professors at Boston University $48,000 to teach a year-long course to 25 students”, for a course that ran from September 1998 through May 1999.⁷

The desk’s evidence file cites pages 1 to 28 of 35, so seven sheets of this document sit outside the extent anyone has recorded reading, and they are not filler. Sheet 28, the last the evidence file reaches, introduces a symposium webcast; sheets 29 to 35 are where that section and the two after it are actually costed. They carry three webcast audience scenarios against a stated budget, the Speaker Bureau figures, the weighted-audience table and the Academic Lecture Series formulae: the paper’s arithmetic.²⁶²⁵²⁷ All seven are now read at the image at 400 dpi, one (p.31) for this page and six by the gate of 2026-09-11. Every sum on the other six is correct to the unit, and p.31 is the exception rather than the rule:

On a quantitative basis, this outreach resulted in a cost of $1,675 per lecture, and $17.92 per audience member (Total grant of $514,364 / 307 lectures = $1,675.45; Total grant of $514,364 / 28,700 total audience = $17.92.20).

With a significant reduction in costs in this renewal, if ABC CORP.’s speakers only reach the same total of 28,700 people during the next two years, the cost per audience member would be reduced to $14.56 (total grant of $417,817 / 28,700 people - $14.46 per person).²⁵

“$17.92.20” and the disagreement between $14.56 in the sentence and $14.46 in the parenthesis are both on the image at 400 dpi. Two sheets later the same paper prints “$17.92 per person” cleanly, which settles the reading: the figure the Foundation meant is $17.92 and p.31 mis-sets it, rather than the paper computing some other quantity. $417,817 divided by 28,700 is $14.56, so $14.46 is the error and $14.56 the right figure.²⁷ These two faults are confined to p.31. Every other sum across pp.29 to 35 multiplies and adds correctly: the nine weighted rows total exactly 70,202, $514,364 over 70,202 is $7.33, $417,817 over 70,202 is $5.95, and 21,000 plus 6,750 plus 27,600 is 55,350 at $7.33 a head. The paper is not generally unreliable; one sheet of it is.²⁶²⁷ The grantee in this section is anonymised by the paper itself as “ABC CORP.”; the Foundation’s paper does not name it and this page does not supply a name.²⁵

A second complete copy of the same paper is in the release at EFTA02444770.²⁸

The approval, in the officer’s own words

The matching-funds letter that Martin Nowak drafted in November 2015 for Epstein’s signature is set out on Nowak’s page from the applicant’s end and on Barnaby Marsh’s page from the addressee’s. Both record Richard Kahn’s report that Marsh “approved [l]anguage”. Neither carries what the Foundation officer actually wrote, and that is the difference between a third party’s characterisation and an institutional act.

EFTA02719709 carries both, one above the other. Kahn’s covering message of Wednesday 25 November 2015 reads: “i sent letter to Barnaby yesterday and he approved =anguage please advise if ok for darren to sign and =end back to barnaby thank you”. Immediately beneath it, in the forwarded block, is Marsh’s own reply, with a full RFC-style header line:

From: Barnaby Marsh Subject: Re: Matching Funds Letter Date: November 24, 2015 at 8:11:43 PM EST Cc: “Nowak, Martin”

looks good- thanks¹⁴

The zone is printed on the header, so no inference is needed: the approval is timed 8:11:43 p.m. Eastern Standard Time on 24 November 2015, and Kahn’s account of it went to Epstein the following afternoon. The attachment named on the covering message is “Enhanced Letter - John Templeton Foundation.doc”: the letter went out under the name of Enhanced Education rather than Epstein’s own.¹⁴

Three words are what the Foundation’s side of that transaction amounts to in this release. Everything else about the 2015 Harvard grant, from the draft and the $1,500,328 and the fund number to the executed letter and the Harvard review’s findings, belongs to Nowak’s page, Marsh’s and Harvard’s, and is not restated here.

Money the record attaches to identified recipients

Nautilus. The 2015 Form 990 of NAUTILUS THINK, INC. lists the Foundation on Schedule B as contributor no. 5: “JOHN TEMPLETON FOUNDATION / 300 CONSHOHOCKEN STATE ROAD, STE 500 / CONSHOHOCKEN, PA 19428”, total contributions “1,125,000.”, with the “Person” box marked. It is the second-largest of the six contributors listed on the sheet, and the largest of the five marked “Person” rather than “Noncash”.⁴ Separately, the magazine’s own promotional material states that “Nautilus was started in 2012 with a grant from the John Templeton Foundation”, and its prospectus that it was “Founded in June 2012 with a grant from the John Templeton Foundation”.²⁹³⁰ The magazine’s own story, its awards and its publisher’s finances are on Michael Segal’s page; the $1,125,000 line is not on any published page of this wiki, and it is the only figure on any sheet read for this page that is entered against the Foundation’s own name and address on a financial return.

Two cautions about that sum. A Schedule B entry is a recipient’s return of what it received in a tax year, not the Foundation’s own record, and the sheet gives no grant date, grant number or purpose. And the company is Nautilus Think, Inc., publisher of the science magazine. It is not Nautilus, Inc., the U.S. Virgin Islands corporation that held Little St. James, which is a different company sharing a name.

The Santa Fe Institute workshop. A Santa Fe Institute letter of 16 July 2012 invites a recipient whose name is barred to “The Principles of Complexity: Life, Scale and Civilization” workshop of 6–8 August 2012 and states: “This event is generously supported through a grant from the John Templeton Foundation (JTF).” The letter is subscribed “David Krakauer Ginger Richardson, Jerry Sabloff, Geoffrey West, / JTF Project Team Leaders”, in printed type over no rule and with no ink.³¹ A second copy is in the release at EFTA01179154.³²

The philosophy of cosmology consortium. A notice forwarded to Epstein on 24 March 2012 advertises two postdoctoral fellowships at Cambridge and Oxford, running from 1 January 2013 to 31 August 2014, and states that the appointments “are part of the initiative ‘establishing the philosophy of cosmology’, involving a consortium of universities in the UK and USA, funded by the John Templeton Foundation.”³³ The covering message calls them “some Templeton Foundation-funded post-docs in ‘philosophy of cosmology’ at Cambridge and Oxford”.³³

Two named grants, with numbers. The published version of a Nowak co-authored paper, “Cooperating Without Looking”, acknowledges: “This research was funded, in part, by John Templeton Foundation Grant RFP-12-11 from the Foundational Questions in Evolutionary Biology Fund, National Science Foundation Grant 0905645, and Army Research Office Grant W911NF-11-1-0363.”³⁴ A second number belongs to the same fund. A PNAS paper by Bozic and Nowak records: “We are grateful for the support from Foundational Questions in Evolutionary Biology Grant RFP-12-17 and the John Templeton Foundation.”³⁵ These two are the only Foundation grant numbers the searches run for this page reach, and the bound is the searches, not the release.³⁵

One of the two is invisible to every search that found it. A Nature research letter in this page’s own 42-document sample prints, at the image, “I.B. was supported by Foundational Questions in Evolutionary Biology Grant RFP-12-17”, and the string RFP does not occur in that sheet’s extraction at all. The text layer drops the grant number and the half-sentence around it and runs the next sentence straight on. No phrase search can return a document for a string its extraction does not contain, so the count of grant numbers in the release is a floor and cannot be anything else.³⁵

A current grant, reported by its holder. On 8 January 2014 the physicist Antony Valentini wrote to Epstein seeking funding for work on hidden variables in quantum mechanics and said of his existing support: “I am currently focusing on finding evidence for hidden variables in the early universe, in the form of anomalies in the cosmic microwave background. This research is currently funded by the Templeton foundation. / My funding will soon run out.”³⁶

The acknowledgments. The commonest way the Foundation appears as a funder in this release is the least dramatic: a line at the foot of a scientific paper. Four sit in the 42-document sample alone. A Nature research letter records “Support from The John Templeton Foundation is gratefully acknowledged.”³⁵ A Proceedings of the Royal Society A paper records, under the heading “Funding”, that a named author whose initials the producing party has barred “gratefully acknowledges support from the John Templeton Foundation.”³⁷ A Journal of Theoretical Biology paper records that “This publication was made possible through the support of a grant from the John Templeton Foundation”, and adds a disclaimer naming the Foundation twice over: “The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the John Templeton Foundation.”³⁸ And a Science paper acknowledges support from the Templeton Foundation to Martin Nowak’s Harvard Program for Evolutionary Dynamics. That last one is not quoted here, and the reason is worth stating: its sheet cannot be rendered, and its extraction interleaves the paper’s two columns, so no continuous sentence can be lifted from it.³⁹ As a class these are the most checkable records of Foundation funding in the release, being printed, published and attached to an identified project, and they are the four this page originally mis-classified. Three of the four are read at the image here; the fourth cannot be.

The prize, and the only total. The one document in this sample that gives the Foundation’s grantmaking as a whole is a newspaper obituary of its president, forwarded to Epstein by John Brockman the day after it ran. It reports the Foundation’s flagship award as a completed transaction: “The foundation’s annual Templeton Prize, valued at $1.7 million, this year went to the founder of a global network of communities where people with intellectual disabilities, and their caregivers, live and work together.” The recipient is identified by description and is not named on the sheet, and no name is supplied here. The next paragraph gives the Foundation’s own account of its purpose, “the big questions of human purpose and ultimate reality”, and the only cumulative grantmaking figures these searches reach: “Since its creation, it has awarded a total of $966 million, including $103 million in 2013, the latest year in which figures are available.”¹⁸ Those totals are a journalist’s, not the Foundation’s, and the sheet attributes them to nobody; they are the scale against which every other figure on this page should be read, and they are the reason the $1,125,000 to Nautilus is a large grant and not a large share.

Where the release attaches money only to an institution or a project and names no personal recipient, as with the Santa Fe workshop, the cosmology consortium and the Boston University course, this page leaves it there and does not invent one.

The officers the sheets name, and the ones they withhold

Six Foundation people are named in the sheets read for this page. The counts below are corpus-wide phrase searches over the full index, not counts of Foundation acts, and the two are very different numbers.⁴⁰

as the sheets print the name	what a sheet attaches to it	documents, full-index phrase search
John M. Templeton, Jr., M.D.	”President and Chairman”, subscribing the full-proposal invitation	2 for "Templeton Jr"
Charles L. Harper, Jr.	”Executive Director/Senior Vice-President”, authoring the grantmaking paper	140 for "Charles L. Harper"
Barnaby Marsh	”Executive Vice President, Strategic Initiatives / and Managing Director, Philanthropic Leadership Network”	2,985 for the canonical spelling alone
John Seel	”John Seel of Templeton”, meeting an applicant about her project	5
Gary Rosen	described by a third party as “the former was external communications guy at JTF”	25
Ayako Fukui	placed “at the foundation” beside Marsh in an applicant’s account of his conversations	2

Two of those counts are worth stating plainly because they cut against the method this page is built on. Charles L. Harper, Jr. appears in 140 documents under that phrase, and a sample of ten of them shows direct correspondence between him and Epstein through 2010, letterhead reading “FROM THE DESK OF CHARLES L. HARPER JR.”, and meeting arrangements, of which the Templeton Foundation search reaches only a small fraction. One of the ten sampled is a false hit on an unrelated page. An entity-scoped search under-reaches its own officers exactly as a person-scoped search under-reaches the entity, and the 140 documents have not been classified; this page makes no claim about what is in them beyond the ten it read.⁴¹

And no unnamed person is identified here. The sheets repeatedly place a Foundation figure in a room without naming him: “the head of templeton foundation is there”; “Templeton foundation described me in a meeting yesterday”; “I met some Templeton people this weekend”; “the Foundation department head who will be ove=seeing the submission and review process”.⁴²⁴³⁹⁸ The producing party withheld those names or the writer never wrote them, and this page supplies none of them. Two further Foundation-side people appear under a forename only: “Ashley”, who is described as having sent proposal reviews to the trustees, and “Ivette”, to whom a third party attributes a promise of Foundation money. Both are reproduced only as the documents print them, with no surname, title or role beyond what the sheet itself states.⁴⁴¹⁷

The Foundation as other people describe it

Fifty-seven per cent of this page’s sample is somebody else talking about the Foundation, and the substance of it is worth setting out because it is what the Foundation looked like from outside.

It was a prospective partner. On 26 July 2010 Tom Pritzker asked Epstein: “Would Templeton Foundation be a logical partner to work on Pritzker Consortium … The question is whether this would tickle Templeton’s fancy. I can go on as the sole funder and would only want a partner if they were like minded and easy to work with.”⁴⁵

It was a source of news awaited. On 20 August 2011 Ellis Rubinstein of the New York Academy of Sciences wrote: “I owe you a conversation re the Executive Committee reaction (they met on Tuesday) to your generous offer and you owe me news from the Templeton Foundation.”⁴⁶

It was an introduction Epstein made repeatedly, in nine words: “barnaby marsh, head of templeton, foundation, is someone you must meet.” One recipient answered “love to”; another answered “Don’t they do religion----and I am on my way to rehearse the St. John Passion”, and, told that the man “hes has a giant founation , loves music, very bright,”, replied “Great… Have fun with Gates”.⁴⁷⁴⁸

And it was criticised, from inside the network. On 9 February 2013 Boris Nikolic wrote to Epstein: “I would stay away of Barnaby — I know that you are liking him but Templeton Foundation has a bad reputation of religious mambo jumbo wrapped up with pseudo science.” Epstein answered: “your call, but bamaby is not like that at all. he went to oxford, only Hikes real science. the princpals are wackos”. Nikolic came back: “Bill is one of the least religious people you will ever meet. We are working with a top scientists world wide on a daily basis. It might be risky to get involved/have any connection with Tempelton at this stage”.⁴⁹ That is an adverse characterisation by Boris Nikolic and it records no act by the Foundation. It is set out here with Epstein’s answer beside it because the record holds both.

One application, followed end to end

The Elisa New sequence is the only Foundation transaction in the release that can be followed from first approach to outcome, and it is assembled here from seven documents.

date	what the record holds	document
1 Oct 2014	New sends Epstein a draft budget from Harvard’s EdLabs: “I met some Templeton people this weekend, and they are being very kind and receptive. I’m meeting with John Seel of Templeton on the 21st to talk about this project.” Epstein forwards it to Larry Summers with the line “Wgbh 300k -graphics, music 100k 999999HELP!!!!!!”	EFTA008693189
5 Nov 2014	The Foundation invites a full proposal by 3/2/201=, over the printed name of its president and chairman, giving the project number #58234 and the reference a07C000000mmPnv	EFTA0180578828
6 Nov 2014	New to Epstein: “Well I’ve cleared the first hurdle at Templeton. If you helpe= in any way, thank you so much.”	EFTA018057882
17 Feb 2015	New on the final drafts: “there’s some disagreement on the t=am about just how much to stress the fact that I have alre=dy attracted substantial investment for my projects with other funde=s. Their rule of thumb is that you should always tell funders that they ar= your only hope.” Epstein replies “yes and i will get a private read”, then asks “is the greant ( sic ) in your name?”; she answers “the grant is in my name. Yes. I’m the project lead”	EFTA02399540, EFTA00864303, EFTA00864305, EFTA00696295505152
3 Oct 2015	New to Epstein, listing her needs: “Had I gotten the Templeton grant I’d have hired 2-2/12 a half more people to produce, edit and to take this sample content into schools to test … It’s a shame to miss the windo[w]“	EFTA0063069653

Between the invitation of 5 November 2014 and the applicant’s statement of 3 October 2015 that the grant did not come, the release holds her side of the correspondence and nothing at all from the Foundation. The outcome is recorded only by the person it happened to. The searches run to test whether a Foundation decision document exists are set out below.

What the release does not hold

Each of the following search forms was written down before it was run, as a description of what a disconfirming document would say in its own words rather than in this page’s, and every form is reported with what it returned. All searches are FTS5 phrase matches over the full-text index, run 2026-09-11, and cover every production in the index rather than EFTA alone.⁵⁴

form searched	conjoined with	returned
"grant agreement"	Templeton	0 pages, 0 documents
"grant number"	Templeton	0, 0
"payment schedule"	Templeton	0, 0
"final report"	Templeton	0, 0
"counter-signed"	Templeton	0, 0
"regret to inform"	Templeton	0, 0
"your proposal has been"	Templeton	0, 0
"not be funded"	Templeton	0, 0
"we are pleased to inform"	Templeton	0, 0. The phrase alone returns 97 pages in 97 documents, none of them conjoined with Templeton
"award letter"	Templeton	2 pages in 2 documents, both read: both are the MIT Media Lab “Finding Genius” proposal, where the phrase names an award letter from a different donor
"unable to fund"	(phrase alone)	5 pages in 5 documents, all 5 read: none concerns the Foundation
"declined"	Templeton	7 pages in 7 documents, all 7 read: three copies of a privilege-log argument, a credit-card processing memo, a labour-force analysis, a lexicography paper and a passage on eugenic sterilisation. None concerns the Foundation
"trustees"	Templeton	2 pages in 2 documents, both read: a bank trust record and a magazine article about a university trustee. Neither concerns the Foundation

The one document in the release that does record a Foundation trustee question uses JTF and never the word Templeton, so no Templeton-scoped search reaches it. That is why the JTF form was run separately, and it is the clearest illustration on this page of a query-scoped search answering only the query.⁴⁴

So: no sheet read for this page, and no sheet returned by any of the searches above, is a document of the Foundation’s recording a decision: no grant agreement, no award notice, no decline notice, no payment record and no final report. The one grant number the searches reach, RFP-12-11, is printed in a grantee’s acknowledgments and not in any Foundation record.³⁴ The Foundation’s own documents in this release are an invitation to apply, an application form, and a paper about method. What the Foundation decided, in every case where the release shows something being asked of it, is recorded, where it is recorded at all, by the applicant.

Two limits on that. These are searches of a text layer, and a phrase search counts the copies whose extraction survived; a decision letter whose scan misread one character in “Templeton” would not be returned by any form above. And the release is Epstein’s files: it would hold a Foundation decision letter only if a correspondent sent him one.

What the extraction gets wrong on these sheets

Every quotation above was read at the image. Five faults found in the process are recorded because each would have produced a false statement, and the last of them produced one that survived into a draft:

sheet	the extraction reads	the sheet prints
EFTA01805788 p.2	”John M. Templeton, Jr., M.O.”	M.D.8
EFTA01805788 p.1	the sender as “Lisa Nev”	Lisa New, with her e-mail address barred immediately after it2
EFTA01149129 p.6	”The Foundation will not find a project for longer than three years” in the first of two identical paragraphs, and “not fund” in the second, which reads as the Foundation’s own boilerplate contradicting itself on one sheet	”fund” in both, at 400 dpi. The disagreement exists only in the text layer1
EFTA01876643 p.1	”August 6’h”, “August 8h", "July 20"" and “ymw.santafe.edu”	superscript ordinals, a bold “July 20th”, and “www.santafe.edu”31
EFTA00603737 p.4	…1.8. was supported by FoundationalQuestions in Evolutionary Bio Grant S.M.acknowledge support from The Virginia and M. Ludwig Fund… A RFP[- ]?\d+ search over the whole row returns no match	…I.B. was supported by Foundational Questions in Evolutionary Biology Grant RFP-12-17. ███, R.H.H. and B.V. acknowledge support from The Virginia and ███. Ludwig Fund… The extraction deletes the grant number, the word “logy” and a redaction bar, and manufactures the initial-pair S.M., which is on no part of the sheet35

The = characters throughout the Gmail-export quotations run the other way: they are on the image and are not repaired here.

The last row is the one that cost this page a false sentence, and it is two faults in one passage. The first four corrupt a string a reader can see is corrupted: August 6'h and ymw.santafe.edu announce themselves. The fifth removes content and invents content in the same sentence, and leaves no visible corruption marker behind: nothing in Grant S.M.acknowledge support from tells a reader that a grant number and a redaction bar used to sit between those words, even though the fragment is not grammatical. An earlier version of this page consequently asserted that the release holds one Foundation grant number when it holds two. A phrase search cannot return a document for a string its extraction does not contain, so every count of a rare string on this page is a floor, and this is the sheet that proves it.

Related

The Foundation’s own acts are this page’s; a person’s dealings with it are that person’s, and are not restated here.

• The four person pages that already carry Foundation material. Barnaby Marsh, the Foundation executive who corresponded with Epstein from 2009 to 2019: his titles, the 2009 Harvard grant arithmetic, the Ito and Nowak proposals, the Chicago Social Brain Network acknowledgment and the two trust instruments naming him. Martin Nowak, the 2015 matching-funds draft and Harvard’s findings about it. Lawrence Krauss, his 2009 meeting with Marsh and the anonymous-gift structure he reported. Robert Trivers, his statement that “even large foundations like the templeton which solicited a proposal, turn me down”.
• Three more published pages that name the Foundation. Joi Ito, the MIT Media Lab proposals including “Finding Genius in an Antidisciplinary World”. Michael Segal, Nautilus, its publisher and its 2015 Form 990. Epstein and Harvard University, the 2020 review, which recommended that Harvard consider further action over “the misstatements made to the John Templeton Foundation about the source of matching funds”.
• Entities: Enhanced Education, the vehicle in whose name the 2015 matching-funds letter went to the Foundation.
• People on the Epstein side of this correspondence: Jeffrey Epstein; Richard Kahn; Darren Indyke; Lesley Groff; Tyler Shears.

Note on the documents

Every page quoted or described above was rendered from the release PDF and read at the image, except where the footnote says otherwise in terms; the dpi and the reader are given in each footnote. Where a sheet was read at the image by the gate of 2026-09-11 rather than by this page, the footnote attributes it to that reader and does not claim it. Two documents cited here, HOUSE_OVERSIGHT_017001 and HOUSE_OVERSIGHT_026521, were until 2026-09-14 described on this page as impossible for anyone to read at the image. That was wrong, and it is corrected here and in the two footnotes. epstein-data.com/pdf/ does return HTTP 403 for both, measured 2026-09-11 against a control that returns 200. But that is a fact about one route, not about the sheets. Both were read from the production’s page-image store on 2026-09-14; both are ordinary page images; and the Templeton acknowledgment on 017001 reads continuously at the image, which is the reading given at ³⁹. Both remain inside this page’s act count.

The counts in the infobox and in the sample table were re-derived for this page against pages_fts on 2026-09-11 and are not inherited. Population is stated with every number, and no count anywhere on this page is scoped by documents.dataset, which drops 8,717 of 30,507 House Oversight documents that carry that column NULL; every split here is assigned from the identifier prefix.¹²

Sources

Footnotes

1. Same application, EFTA01149129 p.6. Read at the image at 400 dpi on 2026-09-11. The sheet is printed type and carries no redaction. The two paragraphs quoted are printed twice on this sheet, once under “Proposed Project Start Date: * 7/1/2012” and once under “Proposed Project End Date: 6/30/2014”, and at the image both instances read “will not fund”. The extraction renders the first instance “will not find a project”, which manufactures a disagreement between two copies of one paragraph that does not exist on the sheet. The words “Sir John’s Donor Intent” are set as a blue underlined hyperlink. Empty rectangular response boxes are on the sheet — they are printed form fields with white interiors, not redaction bars — and beneath the Donor Intent prompt the counter prints “1000 characters remaining”; a paragraph of the applicant’s text continues below the last box. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷

2. John M. Templeton, Jr., to Elisa New, November 5, 2014, reproduced beneath a message from Lisa New to Jeffrey Epstein of Thursday, November 6, 2014 2:21 AM. EFTA01805788 p.1. Read at the image at 400 dpi on 2026-09-11. The sheet is printed type. Two solid rectangular bars are on it: one covering the sender’s e-mail address on the From: line immediately after “Lisa New”, and one after the words “gram Officer”. The extraction renders the sender as “Lisa Nev”; the sheet prints “Lisa New”. The = characters throughout are quoted-printable soft breaks carried onto the face of the sheet, each replacing one character; they are reproduced here as printed and are not repaired. The Sent: field carries no time zone and none is asserted here; the letter’s own date line, “November 5, 2014”, carries no time. The run is 2 sheets. Production stamps on this sheet: EFTA_R1_00152683 and EFTA01805788. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰

3. Charles L. Harper, Jr., Optimizing Program Impact and Cost-Effectiveness: Philanthropic Venture Partnership Opportunities for the Non-profit Community, John Templeton Foundation, Radnor, Pa, undated. EFTA00605502 p.1. Read at the image at 400 dpi on 2026-09-11. The sheet is printed type and carries no redaction. The title is bold italic; the author block, the affiliation and “Radnor, Pa” are regular type; “Contents” is bold. No date appears on the sheet. The run is 35 sheets, confirmed against documents.total_pages and against 35 pages rows. ↩ ↩² ↩³ ↩⁴

4. NAUTILUS THINK, INC., Schedule B (Form 990, 990-EZ, or 990-PF) (2015), Part I, Contributors, page 2 of the schedule. EFTA00805777 p.22. Read at the image at 400 dpi on 2026-09-11. The sheet is printed type on the IRS form and carries no redaction. The Foundation is entry no. 5 of six on the sheet, printed as “JOHN TEMPLETON FOUNDATION / 300 CONSHOHOCKEN STATE ROAD, STE 500 / CONSHOHOCKEN, PA 19428” with total contributions “1,125,000.” and an X in the “Person” box; the other five entries are Thomas Campbell Jackson $335,000, Gerry Ohrstrom $85,000, Howard Hughes Medical Institute $150,000, Simons Foundation $700,000 and Nautilus Ventures LLC $1,300,000, the last marked “Noncash”. The sheet gives no grant date, grant number or purpose, and a Schedule B is the recipient’s return of what it received in a tax year rather than the Foundation’s own record. The form’s first sheet gives the filer as NAUTILUS THINK, INC., principal officer John Steele, exempt under 501(c)(3), gross receipts $3,822,786. The run is 25 sheets. This document is cited three times on Michael Segal; that page carries the gross-receipts figure and does not carry the Schedule B contributor list. ↩ ↩² ↩³ ↩⁴

5. Enhanced Education to Barnaby Marsh, December 2, 2015, on Enhanced Education letterhead. EFTA00306058 p.1, and a second copy at EFTA00584873 p.1. UNVERIFIED, from search hits: neither sheet was rendered for this page. It is cited here only for the inside address, which prints the Foundation’s street address and the fullest statement in the release of Marsh’s Foundation title: “Barnaby Marsh, D.Phil. / Executive Vice President, Strategic Initiatives / and Managing Director, Philanthropic Leadership Network / John Templeton Foundation / 300 Conshohocken State Road Suite 500 / West Conshohocken, PA 19428 USA”. The letter itself, its body, its figure of $1,500,328, its fund number and its ink signature are set out on Barnaby Marsh’s page, which read both copies at the image, and are not restated here. The first copy’s extraction truncates the sender’s own address to “St. T” where the second gives “St. Thomas. USVI 00802”; that is scan variation between two copies and not a difference between the documents. The letter states that the Foundation “is currently considering a project proposal” and records no Foundation decision. ↩ ↩²

6. Same paper, EFTA00605502 pp.24 and 25. UNVERIFIED, from a search hit: these sheets were not rendered. P.24 carries the six named Freedom Project metrics and the phrase “JTF venture capital ‘seed’ investment”; p.25 carries the quality-score and buy-in formulae. ↩ ↩²

7. Same paper, EFTA00605502 p.26. UNVERIFIED, from a search hit: this sheet was not rendered. It carries the partner-funding ratio R and its targets, the “Long-Term Institutional Cost Leverage” definition, and the Boston University scenario with the $48,000 figure and the September 1998 to May 1999 course. ↩ ↩² ↩³

8. Same letter, EFTA01805788 p.2. Read at the image at 400 dpi on 2026-09-11. The closing, the name “John M. Templeton, Jr., M.D.”, the titles “President and Chairman” and “John Temp=eton Foundation”, the reference line and the addressee block are all printed type: there is no manuscript signature, no By: rule and no empty rule anywhere on the sheet, so this is an e-mail letter in an export and not an executed instrument. The extraction gives the postnominal as “M.O.”; the sheet prints “M.D.” Beneath the addressee block a solid rectangular bar covers a block of several lines; its content is not reproduced here and its class is not stated, because the sheet does not state it. The reference a07C000000mmPnv returns this sheet and no other when searched as a whole token, and returns nothing at all when the leading fragment "a07C000000" is searched as a quoted phrase — the tokeniser keeps the reference whole, so a quoted sub-token matches nothing. Production stamps: EFTA_R1_00152684 and EFTA01805789. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸

9. Elisa New to Epstein, Wednesday, October 1, 2014, “a budget”, forwarded by Epstein to Larry Summers the same day at 13:47:27 +0000 with the attachment “Draft_Project_Budget_—_Poetry_of Character Development_v1J3).xlsx”. EFTA00869318 p.1. UNVERIFIED, from a search hit: this sheet was not rendered. Two further copies of the same forward are at EFTA00869314 p.1 and EFTA01146592 p.1, whose extractions give “Pm meeting” where this one gives “I’m meeting”. The phrase “some Templeton people” is the writer’s and names nobody; no identity is supplied here. The run is 4 sheets. ↩ ↩² ↩³

10. MIT Media Lab, “Finding Genius in an Antidisciplinary World”, draft Templeton Foundation proposal. EFTA01190591 p.5, item 17: “This proposal and plan resulted, in part, from conversations that Joi Ito has had with Barnaby Marsh and Ayako Fukui at the foundation.” UNVERIFIED, from a search hit: this sheet was not rendered. The sheet places Ayako Fukui at the Foundation and gives her no title; none is supplied here. The same sheet carries the requested amount and the supplemental-funding section; those and the rest of the Media Lab proposals belong to Joi Ito’s page and are not set out here. A second copy of the proposal is at EFTA01190700 p.5, a six-sheet run, also not rendered; its extraction gives the forename as “Bamaby Marsh” where this copy gives “Barnaby Marsh”, which is the rn read as m fault and not a difference between the documents. ↩

11. Same application, EFTA01149129 p.5. UNVERIFIED, from a search hit: this sheet was not rendered. It carries the currency instruction, “Request Amount: $150,000 over two years” and “Total Project Amount: $150,000”. ↩ ↩² ↩³ ↩⁴

12. Full-text index searches of the release, run 2026-09-11 against pages_fts in full_text_corpus.db, FTS5 phrase matches on text_content, no LIMIT, rowid selected first and identifiers resolved in batches. "John Templeton Foundation" returns 82 pages in 74 documents (77 pages / 69 documents with an EFTA prefix, 5 / 5 HOUSE_OVERSIGHT_); "Templeton Foundation" 194 pages in 166 documents (185 / 157 EFTA, 9 / 9 House Oversight); the bare token Templeton 757 pages in 651 documents (710 / 611 EFTA, 47 / 40 House Oversight); JTF 62 pages in 46 documents, all EFTA. The three phrase forms nest exactly: full name ⊂ short name ⊂ bare token, so the counts overlap and must not be summed. Prefix splits are assigned from pages.efta_number; no documents.dataset condition was used anywhere. The desk’s evidence file gives the bare-token split as 719 / 38 pages and 618 / 33 documents. Those figures sum to the correct totals of 757 and 651 and are wrong by nine pages and seven documents; the re-derivation above was run twice. The bare-token population is a high-recall superset containing the measured homograph families — Franklin Templeton and named Templeton funds, Templeton World Charity Foundation and Templeton Religion Trust as separate entities, Templeton & Company LLP, unrelated people surnamed Templeton, three case captions, and donor-list surnames — and this page rests on the phrase forms rather than on the bare token. ↩ ↩²

13. “Templeton Foundation LOI / Project Information”, completed application for “Impact of the Zen Chaplaincy Program on Quality of Life. Symptom Control, and Costs of Care in Patients Hospitalized for Cancer”. EFTA01149129 p.1. UNVERIFIED, from a search hit: this sheet was not rendered. It carries the form heading, the project title, the executive summary naming Benjamin Kligler and Roberta Lee as co-investigators and the New York Zen Center for Contemplative Care, and the figure “$150,000 study”. The run is 7 sheets. ↩ ↩²

14. Richard Kahn to Jeffrey Epstein, Wednesday, November 25, 2015 4:00 PM, subject “Fwd: Matching Funds Letter”, forwarding Barnaby Marsh’s reply of November 24, 2015 and Kahn’s own message of the same day. EFTA02719709 p.1. Read at the image at 400 dpi on 2026-09-11. The sheet is printed type. Solid bars cover the e-mail addresses on the outer From: line, on Marsh’s From: line, on the inner To: lines and on the Cc: lines, and cover the values in the three labelled contact lines of Kahn’s signature block; none of those values is reproduced here. The attachment line prints “Enhanced Letter - John Templeton Foundation.doc; Untitled attachment 00091.htm”. Marsh’s reply carries a full date line with an explicit zone — “Date: November 24, 2015 at 8:11:43 PM EST” — so no zone inference is needed for it; Kahn’s own top-of-message Sent: field carries no zone and none is asserted. The = characters are quoted-printable soft breaks on the face of the sheet. The second sheet of the run carries only Kahn’s original one-line message and a page number. This document is cited once each on Barnaby Marsh and Martin Nowak, in both cases for Kahn’s report; neither page quotes Marsh’s own reply or its timestamp. ↩ ↩² ↩³

15. Case-sensitive text.count over every .md file in the wiki tree outside _desk, run 2026-09-11, counting Templeton, templeton and TEMPLETON: people/barnaby-marsh 32, people/martin-nowak 8, people/lawrence-krauss 5, people/joi-ito 4, people/michael-segal-nautilus 2, entities/epstein-and-harvard-university 1, people/robert-trivers 1. All seven are on _desk/PUBLISHED.txt; no unpublished draft in the tree names the Foundation. overlap_check.py --phrase "Templeton" returns the same seven pages with the same counts, which is not always true of that tool, since it case-folds and counts footnote keys as mentions. ↩

16. The 42-form union of "Barnaby Marsh" with the 40 damaged spellings generated by scripts/ocr_variants.py "Barnaby Marsh" --wide and with the doubly damaged "Bamaby Marshnalaillik", intersected against the 166-document "Templeton Foundation" population. Re-derived for this page 2026-09-11: the union reaches 4,680 documents corpus-wide and intersects 59 of the 166, leaving 107 documents (99 EFTA, 8 House Oversight) that use the phrase and contain none of the 42 forms. A union of known damaged forms is not proof against every possible corruption, and a non-zero damaged form is not by itself an identification. ↩

17. James Tagg to Epstein, Monday 20 February 2017: “I should also say that Ivette, the Physics lead has a promise of several million dollars of funding from the Templeton foundation for the Bose-Einstein-Condensate sensor and currently has $200,000 in feasibility money”. EFTA00664252 p.1, and a second copy at EFTA02366785 p.1. UNVERIFIED, from search hits: neither sheet was rendered. This is a third party’s report of a promise and records no Foundation act. The forename is reproduced as printed and no surname is supplied. The figure is settled at the image: EFTA00664252 p.1 prints “$200,000”, read at 400 dpi by the re-gate of 2026-09-11. The second copy’s extraction gives it as “5200,000”, which is an extraction fault and not a difference between the documents. This pair is the reason the rule’s exclusion list names “promised”: the sheet’s own word is “a promise”, which is neither a proposal nor a completed award, and two documents of the sample turn on it. ↩ ↩²

18. The word “grantmaking” is doing work in the sentence this note supports, and the bound is searched rather than assumed. Five phrase searches over the full index — "since its creation" (17 documents), "awarded a total of" (3), "total grants" (4), "has awarded a total" (1), "cumulative grants" and "in grants since" (0 each) — return four documents that both match a phrase and name Templeton, and exactly one that carries a cumulative Foundation figure: this one. The other three — EFTA00754357, EFTA00893585 and EFTA02410375 — are three copies of one item whose "total grants" hit is a prospective budget for a proposed grants competition, “a total grants budget of, say, $75K to $100K”, and whose Templeton mention falls on a different sheet and concerns Templeton support to a third party rather than any total. All three read. The claim is bounded to those searches and is not a claim about the whole release. It is narrowed to GRANTMAKING because this same document contradicts the wider form: its first sheet gives a cumulative ENDOWMENT figure for the same period, which is a different quantity. Kris Maher, “Philanthropist John Templeton Jr. Dies”, The Wall Street Journal, updated May 19, 2015 12:59 p.m. ET, forwarded to Epstein by John Brockman of the Edge Foundation on Wednesday, 20 May 2015 13:35:34 +0000 under the subject “Philanthropist John Templeton Jr. Dies”. EFTA01142598 pp.1 and 2. p.2 read at the image at 400 dpi on 2026-09-11 for this page — it is the sheet the classification turns on, and it had not been rendered by anyone before; p.1 was read at the image at 400 dpi by the re-gate of 2026-09-11. Production stamp on p.2: EFTA01142599. The prize sentence, the $966 million sentence and the quoted self-description are all on p.2 and are printed type; the article’s own subheading on p.1 reads “Surgeon and prominent donor’s family foundation awards the annual Templeton Prize”. The prize recipient is not named on either sheet, and no name is supplied here; the sentence identifies him by a unique description and that is what the document gives. Contact values in Brockman’s signature block at the foot of p.2 are barred, and no value from anywhere on the sheet is reproduced here. The same sheet reports a personal political donation by Dr. Templeton and his wife, attributed to the Philadelphia Inquirer: that is the man and not the Foundation, and this page does not carry it. The run is 2 sheets. ↩ ↩²

19. Systematic sample of the 166-document "Templeton Foundation" population: distinct pages.efta_number sorted ascending, then every fourth document taken beginning at the first, giving 42 documents (40 EFTA, 2 House Oversight) and 54 returned hit pages, all 54 read in the text layer 2026-09-11. Classification uses the rule stated in the body. Acting, 12: EFTA00603737, EFTA00626051, EFTA00805777, EFTA00981180, EFTA01142598, EFTA01165132, EFTA01202576, EFTA02011471, EFTA02444770, EFTA02702876, HOUSE_OVERSIGHT_017001, HOUSE_OVERSIGHT_026521. Third-party discussion 24; grant or programme named in passing 1 (EFTA00760692, see [^bibliography]); letterhead, title or sponsor list 5. 12/42 = 28.6%, 24/42 = 57.1%, 1/42 = 2.4%, 5/42 = 11.9%; the four shares sum to 100.0% and the four counts to 42. The class ‘press or biography’ held one document, EFTA01142598; that document is now in the acting count and the class is empty and removed. The count has been corrected twice and both corrections raised it. The first version put four of the twelve — EFTA00603737, EFTA01202576, EFTA02702876, HOUSE_OVERSIGHT_017001 — in ‘named in passing’ while counting three materially identical recipient statements as acting, giving 7 of 42. The second version fixed that but left the obituary out under an inherited class label the stated rule does not support, giving 11 of 42. Neither error was found by a checker; both were found by a reader re-running the rule against the documents. The duplicate families are: EFTA00306058/EFTA00584873; EFTA00626051/EFTA01165132; EFTA00650354/EFTA01794962; EFTA00664252/EFTA02366785; EFTA00744374/EFTA01822846; EFTA00782650/EFTA01211158; EFTA00679171/EFTA00844414/EFTA02485274; EFTA00895046/EFTA02408617; EFTA00917358/EFTA01860103; EFTA01763263/EFTA01902943/EFTA02566758/EFTA02679688; EFTA01904862/EFTA02724344 : 25 documents collapsing to 11 items, leaving 28 distinct items, of which 11 record the Foundation acting, 11/28 = 39.3%. None of the five documents added to the count since the first version belongs to a duplicate family. **Provenance of the twelve, stated document by document. Read at the image at 400 dpi for this page: EFTA00603737, EFTA00626051, EFTA00805777, EFTA01142598, EFTA01165132, EFTA01202576, EFTA02702876 — seven (of these, EFTA01142598 p.2 is the sheet the classification turns on and its p.1 was read by the re-gate). Read at the image at 400 dpi by the gate of 2026-09-11 and not re-rendered here: EFTA00981180, EFTA02011471, EFTA02444770 — three, attributed to that reader rather than claimed by this page. Read at the image on 2026-09-14 by the fix pass, from the production’s page-image store: HOUSE_OVERSIGHT_017001, HOUSE_OVERSIGHT_026521 — two. An earlier version of this footnote listed these two as “unreadable at the image by anyone”; they are not, and the label is withdrawn. What remains true of them is that epstein-data.com/pdf/ returns HTTP 403 for both, measured 2026-09-11 against a control that returns 200. So all twelve are image-verified and none changed class at the image. The other 30 documents of the 42 were classified from the text layer alone; all 42 have now been read against the current rule, 31 of them by a second reader. A sample of 42 from 166 carries sampling error this page does not quantify, and no independent cross-check of this figure exists — see the body on why the desk’s disjoint sample is not one. ↩ ↩² ↩³

20. Bibliography of a Program for Evolutionary Dynamics publication list. EFTA00760692 p.8. UNVERIFIED, from a search hit: this sheet was not rendered for this page. Its only occurrence of the phrase is the publisher line of a cited chapter, “In Spiritual Information, ed. CL Harper. Templeton Foundation Press”, reached through a hyperlink field the extraction renders with stray control characters. Templeton Foundation Press is the Foundation’s publishing imprint and a citation to it is not a record of the Foundation funding anything, so this document is classed as a mention. It is the only homograph found in the 42-document sample and the only one this page has looked for inside the phrase population; the homograph families subtracted from the bare token are a separate and much larger set. ↩

21. Woodson Merrell to Jeffrey Epstein, copied to his own office, Wednesday 12 October 2011 9:54:23 PM, subject “FW: Templeton proposal”, forwarding a message of the same day from Benjamin Kligler. EFTA01853133 p.1. UNVERIFIED, from a search hit: this sheet was not rendered. The attachment line prints “Templeton Foundation LOIfinal,doc” as the extraction has it. The sheet also prints a professional telephone and fax line for Kligler at Beth Israel, not reproduced here because nothing on this page turns on them. This document records Epstein’s act of forwarding and no act of the Foundation’s; it is the desk’s first supplied anchor, correctly demoted. The run is 2 sheets. ↩

22. Same paper, EFTA00605502 p.2. UNVERIFIED, from a search hit: this sheet was not rendered. It carries the sentence about how the Foundation views itself and the “leverage” definition. ↩

23. Same paper, EFTA00605502 pp.22 and 23. UNVERIFIED, from a search hit: these sheets were not rendered. They carry the lecture-host scenario and the heuristic success equation, whose exponent the extraction renders inconsistently across the two copies of the paper; no form of the equation is quoted on this page for that reason. ↩

24. Same paper, EFTA00605502 p.28. UNVERIFIED, from a search hit: this sheet was not rendered. It carries the “Extended Life, Eternal Life” symposium webcast section and the America Online / Time-Warner reference. ↩

25. Same paper, EFTA00605502 p.31, the “Speaker Bureau” section. Read at the image at 400 dpi on 2026-09-11. The sheet is printed type and carries no redaction; the heading “Speaker Bureau” and the sub-heading “Quantitative Impact” are bold, the latter underlined. Both arithmetic faults quoted in the body are on the image and are not extraction artefacts: the sheet prints ”= $17.92.20” and prints “$14.56” in the sentence against “$14.46” in the parenthesis that is meant to derive it, with a hyphen where an equals sign belongs. $417,817 divided by 28,700 is $14.56. The grantee is anonymised on the face of the paper as “ABC CORP.”; the Foundation’s own document does not name it, and no name is supplied here. ↩ ↩² ↩³ ↩⁴

26. Same paper, EFTA00605502 pp.29 and 30. Read at the image at 400 dpi by the gate of 2026-09-11 and not re-rendered for this page; production stamps EFTA00605530 and EFTA00605531. They are the first two of seven sheets beyond the extent the desk’s evidence file cites. They carry three webcast cost scenarios at a stated cost of $12,800, with per-person figures of $3.20, $0.64 and $0.13. The symposium is named on p.28 and on no sheet after it: the string Extended Life, Eternal Life is on p.28 of the run and on none of pp.29 to 35, so the name is cited to p.28 at [^metrics28] and not to the sheets that cost it. ↩ ↩²

27. Same paper, EFTA00605502 pp.32 to 35. Read at the image at 400 dpi by the gate of 2026-09-11 and not re-rendered for this page; production stamps EFTA00605533 to EFTA00605536. They carry the media-attention paragraph, the three-point audience rating scale, the weighted-audience table (beginning on p.32 and totalling on p.33), the Academic Lecture Series formulae on p.34 and the closing “Increased Productivity” section on p.35. p.33 is where the paper prints “$17.92 per person” cleanly, two sheets after p.31 mis-sets the same figure as “$17.92.20”. All seven sheets beyond the evidence file’s cited extent are now image-read: p.31 for this page, pp.29, 30 and 32 to 35 by the gate. One count on these sheets is deliberately not reported as a fault: the audience column totals 28,775 while the paper says and divides by “over 28,700”, and 28,775 is over 28,700, so the paper’s own rounding is not an error. ↩ ↩² ↩³

28. A second complete copy of the same paper is in the release at EFTA02444770, also 35 sheets, carrying EFTA_R1_ production stamps that the first copy lacks. Its p.1 was read at the image at 400 dpi by the gate of 2026-09-11 and not re-rendered for this page, which found it identical to EFTA00605502 p.1 and carrying the stamp EFTA_R1_01521700; no other sheet of this copy has been rendered by anyone. Its text layer differs from the first copy’s at several points: successfit against successful.., prima facie against prima fade, and Mother measure against Another measure. Those are extraction differences between two scans and are not evidence that the documents differ. ↩

29. John Steele, Publisher and Editorial Director of Nautilus, to Epstein, Thursday, August 21, 2014, forwarded to Epstein and Tyler Shears on Saturday, 30 August 2014 21:57:23 +0000 with the attachment “Nautilus_One-Sheet.docx”. EFTA00626051 p.1. Read at the image at 400 dpi on 2026-09-11. The sheet is printed type. The forwarding message’s Date: header carries an explicit +0000 offset. Several contact fields on the sheet are barred and no value from any of them is reproduced here. A second copy of the same forward is at EFTA01165132 p.1, also read at the image at 400 dpi; contact fields on that sheet are barred too, and no value from it is reproduced here either. The two extractions differ at “Beautiful”/“Beautifid” where both sheets print “Beautiful” at the image: that difference is scan quality and not a difference between the documents. Each run is 2 sheets. ↩

30. Nautilus prospectus, “Concept to Completion”. EFTA00797032 p.4. Read at the image at 400 dpi on 2026-09-11. The sheet is justified printed type in a serif face and carries no redaction; “Nautilus” is italic throughout and the section headings “Concept to Completion” and “Critical Reception” are bold. The sentence prints “took nine months to defined its editorial approach” — “defined” is on the image and is the document’s own error, not an extraction fault. This sheet carries one production stamp, EFTA00797035, and no EFTA_R1_ stamp. The run is 13 sheets. This document is cited twice on Michael Segal, which carries the founding-grant claim. ↩

31. Santa Fe Institute to a recipient whose name is barred, July 16, 2012, “The Principles of Complexity: Life, Scale and Civilization” workshop. EFTA01876643 p.1. Read at the image at 400 dpi on 2026-09-11. The run is a single sheet, so nothing of this letter continues past it. The sheet is printed type on Santa Fe Institute letterhead with the Institute’s logo. Four solid bars are on it: the salutation after “Dear”, a parenthetical after “Hilary Skolnik at SFI”, and the values after “Tel:” and “Fax:” in the footer; none is reproduced here. The date and the ordinals print as superscripts — “July 16th, 2012”, “August 6th”, “August 8th” — where the extraction gives “July 16’”, “August 6’h” and “August 8h"; "**July 20th**" is bold on the image and the extraction gives "July 20""; the footer prints “www.santafe.edu” where the extraction gives “ymw.santafe.edu”. The sheet prints “no later then July 20th”, with “then” as the document has it. The subscription “David Krakauer Ginger Richardson, Jerry Sabloff, Geoffrey West, / JTF Project Team Leaders” is printed type over no rule, with no ink anywhere on the sheet. This is the desk’s fourth supplied anchor and the only one of the four that survives the operational test. ↩ ↩²

32. A second copy of the same letter is at EFTA01179154 p.1. Read at the image at 400 dpi by the re-gate of 2026-09-11, which also corroborated on this copy five things this page states about the first: the superscript ordinals, the bold “July 20th”, the document’s own “no later then”, www.santafe.edu, and the subscription in printed type over no rule. The funding sentence is verbatim on this copy too. The two copies are redacted to different extents in the footer, and what differs is an institution’s own contact block — a published street address, switchboard, fax line and a staff coordinator’s institutional e-mail. Under the editor’s ruling of 2026-09-11 none of that is a protected class; no value from either copy is reproduced here, and nothing on this page rests on the comparison. An earlier version of this footnote asserted this differential having opened only one of the two sheets, which is a claim about two documents made from one; the render settles it and widens it by one item. ↩

33. David Grosof to Epstein, Saturday 24 March 2012 2:39:17 AM, subject “Oxbridge post-does in philosophy of cosmology”, forwarding a notice circulated by A. Zee on 18 March 2012 6:53:18 PM EDT. EFTA02011471 pp.1 and 2. Both sheets read at the image at 400 dpi by the gate of 2026-09-11 and not re-rendered for this page, which confirmed the quoted sentence verbatim and confirmed that p.2 opens on “cosmology’,”. The quotation about the consortium runs over the page break: p.1 ends “These appointments are part of the initiative ‘establishing the philosophy / of” — the word “of” is the last on that sheet — and p.2 opens “cosmology’, involving a consortium of universities in the UK and USA, funded by the John Templeton Foundation”, which is why both sheets are cited. The inner notice carries an explicit EDT zone; the outer message’s Date: header carries +0000. The run is 2 sheets. ↩ ↩²

34. Moshe Hoffman, Erez Yoeli and Martin Nowak, “Cooperating Without Looking”, ACKNOWLEDGMENTS. EFTA00632631 p.5, the PNAS Early Edition version, footed “Hoffman et al.” and “5 of 6”. Read at the image at 400 dpi on 2026-09-11; the sheet is printed type, carries no redaction, and carries a diagonal “PNAS Embargoed” watermark. Production stamp EFTA00632635. The run is 6 sheets. The quotation above is the image text. The extraction of the same sheet hyphenates “Evolution-ary” across a line break and reads the Army Research Office number as “W91 INF-11-1-0363” where the sheet prints “W911NF-11-1-0363”, so a literal search for the sentence as printed does not return it. This pin replaces an earlier citation of the same acknowledgment to HOUSE_OVERSIGHT_026521, whose corpus row carries no usable page pin. That identifier’s row is a container: measured 2026-09-11, 34,367 characters in a single pages row with total_pages recorded as 1, against neighbouring House Oversight rows at 997, 8,588, 1,885 and 2,127 characters, while the paper’s own printed pagination runs past twenty; and epstein-data.com/pdf/HOUSE_OVERSIGHT_026521.pdf returns HTTP 403, the measured behaviour for that prefix. An earlier version of this footnote added that the identifier “cannot be rendered by anyone” and that “no pass can ever read it at the image”. Both are withdrawn. The sheet was read on 2026-09-14: it is printed page 1, folioed 1, of the Cooperating Without Looking manuscript — title, the three authors Moshe Hoffman, Erez Yoeli and Martin Nowak, their affiliations, and the opening of the abstract — so a pin to 026521 p.1 binds precisely, even though the row behind it merges the run. The distinction to hold is between the row, which merges, and the sheet, which does not. It holds the manuscript version, whose wording differs slightly from the published one quoted above: “This research was funded in part by the John Templeton Foundation, Grant RFP-12-11 from the Foundational Questions in Evolutionary Biology Fund”. Corpus-wide, "RFP-12-11" returns 3 documents — EFTA00632631, HOUSE_OVERSIGHT_026521 and HOUSE_OVERSIGHT_026533. ↩ ↩²

35. Ivana Bozic and Martin Nowak, PNAS, ACKNOWLEDGMENTS. EFTA01201684 p.5, footed “Bozic and Nowak” and “PNAS Early Edition | 5 of 5”. Read at the image at 400 dpi on 2026-09-11; printed type, no redaction on the sheet. Production stamp EFTA01201688. The run is 7 sheets. The second sheet carrying this number is EFTA00603737 p.4, a Nature research letter inside this page’s 42-document sample, footed “4 | NATURE | VOL 000 | 00 MONTH 2015” and so an author proof rather than the published version. Read at the image at 400 dpi on 2026-09-11: its Acknowledgements read “Support from The John Templeton Foundation is gratefully acknowledged” and, two sentences later, “I.B. was supported by Foundational Questions in Evolutionary Biology Grant RFP-12-17”. Several author initials and two reference entries on that sheet are barred and nothing from them is reproduced here; the recipients are given by initials on the sheet and no identity is supplied. Production stamp EFTA00603740. The run is 17 sheets. The extraction of that sheet contains no RFP string at all: a RFP[- ]?\d+ search over its text_content returns nothing, and the row reads “by FoundationalQuestions in Evolutionary Bio / Grant /” and then continues with an unrelated sentence, the grant number and the words around it having been dropped. "RFP-12-17" therefore returns 1 document in the text layer (EFTA01201684) and reaches a second only at the image. Both figures in this footnote are counts of what survived extraction, and the second document proves that such a count is a floor. ↩ ↩² ↩³ ↩⁴ ↩⁵

36. Antony Valentini to Epstein, forwarded to Epstein Wednesday, 08 Jan 2014 21:44:03 +0000, subject “Fwd: Funding for hidden variables in quantum mechanics?”, attachment “Science_Magazine.pdf”. EFTA00981180 p.1. Read at the image at 400 dpi by the gate of 2026-09-11 and not re-rendered for this page, which confirmed the sentence verbatim including the lower-case f in “Templeton foundation” as quoted above. The statement about current funding is the writer’s own account of his own grant, not a Foundation document. The run is 1 sheet. ↩

37. B. Adlam, K. Chatterjee and a third author whose initials the producing party has barred, Proceedings of the Royal Society A 471: 20150114, section headed “Funding”. EFTA01202576 p.10. Read at the image at 400 dpi on 2026-09-11. The sheet is printed type and carries a downloaded-from header dated September 3, 2015. The extraction reads that line’s heading as “Audio’s contributions” where the sheet prints “Authors’ contributions”. Two solid bars are on it, one in the “Authors’ contributions” line and one immediately before “gratefully acknowledges support from the John Templeton Foundation”; the barred initials are not supplied here, so the sheet identifies the supported project and not the supported person. The same line acknowledges an ERC Start grant, an Austrian Science Fund grant and an FWF NFN grant. The run is 12 sheets. ↩

38. J. W. Olejarz, B. Allen, C. Veller and M. A. Nowak, Journal of Theoretical Biology 399 (2016) 103–116, Acknowledgments. EFTA02702876 p.13, the article’s printed page 115. Read at the image at 400 dpi on 2026-09-11. The sheet is printed type in two columns and carries no redaction. Production stamps EFTA_R1_02076892 and EFTA02702888. The disclaimer sentence quoted above follows the funding sentence immediately. The run is 14 sheets. ↩

39. “Quantitative Analysis of Culture Using Millions of Digitized Books”, Science, published online 16 December 2010, acknowledgments. HOUSE_OVERSIGHT_017001 p.1, read at the image at 5× native on 2026-09-14. An earlier version of this footnote called the sheet UNVERIFIED and “unverifiable” and said that “no pass can read this sheet at the image”. That was false and is withdrawn. What is true is narrower and is kept: epstein-data.com/pdf/HOUSE_OVERSIGHT_017001.pdf returns HTTP 403, measured 2026-09-11 against a control identifier that returns 200 — a fact about that route — and the extraction still cannot be quoted, because it interleaves the two printed columns, reaching the row as fragments separated by figure-caption text: The Program for Evolutionary Dynamics acknowledges / support from the Templ / eton Foundation, NIH grant / RO1GM078986, and the Bill and Melinda Ga, with a sentence about dictionary headwords running between them. At the image the interleaving diagnosis is confirmed and the acknowledgment reads continuously. The sheet is footed Sciencexpress / www.sciencexpress.org / 16 December 2010 / Page 6 / 10.1126/science.1199644, and note 36 in its left column prints: The Program for Evolutionary Dynamics acknowledges support from the Templeton Foundation, NIH grant R01GM078986, and the Bill and Melinda Gates Foundation. The grant number begins R01, R-zero-one; the extraction’s RO1 with a capital letter O is extraction damage. The same note prints J-B.M. was supported by the Foundational Questions in Evolutionary Biology Prize Fellowship and the Systems Biology Program (Harvard Medical School), which is where the fragment Evolutionary Biology Prize Fellowship comes from. So this document no longer rests on the text layer, and the earlier warning that it is the weaker of the two in the act count is superseded. ↩ ↩²

40. Full-index FTS5 phrase searches over text_content, run 2026-09-11, counting distinct pages.efta_number: "Templeton Jr" 2 documents; "Charles L. Harper" 140; "Barnaby Marsh" canonical spelling 2,985 documents (the 42-form union reaches 4,680); "John Seel" 5; "Gary Rosen" 25; "Ayako Fukui" 2. These are corpus-wide name counts and not counts of Foundation acts, and they are a different population from the desk’s evidence file’s figures (Marsh 147, Harper 16, Rosen 12, Seel 5, Templeton Jr 1, Sarah Clement 1), which count only pages carrying both the name and the token Templeton. Neither set of figures is wrong; they answer different questions, and the gap between 140 and 16 for Harper is the point made in the body. "Sarah Clement" returns 2 documents on the corpus-wide form. ↩

41. The 140-document "Charles L. Harper" population was sampled at ten documents drawn with a fixed seed on 2026-09-11 and each hit passage read in the text layer: nine are the same man — direct correspondence with Epstein in 2010, a letterhead reading “FROM THE DESK OF CHARLES L. HARPER JR.”, a research-meeting proposal, and travel and meeting arrangements — and one, EFTA01206755 p.2, is an unrelated social-media page returned by extraction noise. The remaining 130 documents were not read and are not classified here, and this page asserts nothing about them beyond the sampled ten. No sheet in the sample was rendered. ↩

42. Epstein to a correspondent, Tuesday 12 January 2010, quoted in a thread of the same day: “tricky, i can see you , but the head of templeton foundation is there. you can sit in, but it will be difficult to have private time”. EFTA00744374 p.1, and a second copy at EFTA01822846 p.1. UNVERIFIED, from search hits: neither sheet was rendered. Neither sheet names the head, and no identity is supplied here. This is the desk’s second supplied anchor, correctly demoted: the named-officer branch of the test fails on it. The first run is 3 sheets. ↩

43. iMessage of 2018-12-08 17:03:07 (UTC), sender self: “Templeton foundation described me in a meeting yesterday. As the misunderstood playful genius, girls were enraged”. EFTA00782650 pp.15 and 16, and a second extraction of the same thread at EFTA01211158 pp.2 and 3. UNVERIFIED, from search hits: no sheet was rendered. Both extractions print each message twice under different thread indices, which is a property of the forensic report format and not two sendings. The other party’s identifier is barred throughout and no identity is supplied. The timestamps carry an explicit (UTC) label on the face of the report. ↩

44. Avi Loeb to “BHC colleagues”, February 16, 2016, summarising a meeting with Harvard’s FAS Dean and Science Dean, forwarded on through Shing Tung Yau and Lesley Groff to Epstein on 19 February 2016. EFTA00328899 pp.1 and 2. UNVERIFIED, from search hits: neither sheet was rendered. The passage reads: “Nina will send the official invitation as soon as we hear back from the trustees of JTF on Friday (per Ashley’s latest e-mail) on their decision regarding an accelerated review for our proposal. My understanding is that Ashley sent the reviews that she received last Friday to them for their perusal before reaching their final decision on Friday.” On the operational test this is a third party’s account of a Foundation decision that had not yet been made, not a Foundation act; the desk’s evidence file classed it under JTF as a trustee decision, and this page does not. “Ashley” and “Nina” are reproduced as the sheet prints them, with no surname, employer or title supplied for either, and the sheet states of “Ashley” only that she sent reviews to the trustees. This document uses JTF four times and never the word Templeton, so no Templeton-scoped search returns it. The run is 2 sheets. ↩ ↩²

45. Tom Pritzker to Epstein, Mon 7/26/2010 9:31:17 PM, and Epstein’s reply of Mon, 26 Jul 2010 21:42:34 +0000. EFTA00895046 p.1, and a second copy at EFTA02408617 p.1. UNVERIFIED, from search hits: neither sheet was rendered. ↩

46. Ellis Rubinstein to Epstein, Sat, 20 Aug 2011 22:30:42 +0000, quoting his own message of the same day. EFTA00917358 p.1, and a second copy at EFTA01860103 p.1. UNVERIFIED, from search hits: neither sheet was rendered. The second copy bars a telephone number that the first prints; no value from either is reproduced here. ↩

47. Epstein to a correspondent whose address block prints only as “President”, Saturday, February 23, 2013 9:04:32 AM: “barnaby marsh, head of templeton, foundation, is someone you must meet.” Four copies are in this page’s sample: EFTA01763263 p.1, EFTA01902943 p.1, EFTA02566758 p.1 and EFTA02679688 p.1. UNVERIFIED, from search hits: no sheet was rendered. Two of the four carry a trailing Apple property-list block from the export. The correspondent is identified on these sheets only by the word “President” in the address field and no identity is supplied here. ↩

48. The same thread, continued. EFTA01763263 p.1 carries the replies “Don’t they do religion----and I am on my way to rehearse the St. John Passion”, Epstein’s “hes has a giant founation , loves music, very bright,” and the answer “Great… Have fun with Gates” of Saturday, February 23, 2013 5:51 PM. EFTA01902943 p.1 carries the reply “love to” of Sat 2/23/2013 2:32:42 PM. UNVERIFIED, from search hits: neither sheet was rendered. The misspelling “founation” is as the extraction has it and has not been checked at the image. None of the Sent: fields on these sheets carries a zone and none is asserted. ↩

49. Boris Nikolic to Epstein and Epstein’s reply, 9 February 2013, with Nikolic’s further message quoted above them. The two copies do not agree on the clock and no time is asserted here: one prints Epstein’s reply under a top-of-message field reading “Sun 2/10/2013 1:12:45 AM” and the other under “Saturday, February 09, 2013 5:13 PM”, while both quote Nikolic’s message as “On Sat, Feb 9, 2013 at 9:10 PM” and Epstein’s earlier forward as “Saturday, February 09, 2013 5:05 PM”. None of those fields carries a zone. EFTA01904862 p.1 and EFTA02724344 pp.2 and 3. UNVERIFIED, from search hits: no sheet was rendered. “Hikes” for “likes”, “princpals”, “alien”/“allen” for “Allen” and “Tempelton” are as the extractions have them; the two copies differ at “scn”/“sen” and “alien”/“allen”, which is scan variation. This is the desk’s third supplied anchor, correctly demoted: it is adverse commentary by Boris Nikolic and records no Foundation act, and it is usable only with that attribution. ↩

50. Elisa New to Epstein, Tuesday, February 17, 2015 11:50 AM, “advice re Templeton”. EFTA02399540 p.1. UNVERIFIED, from a search hit: this sheet was not rendered. The = characters are quoted-printable soft breaks in the extraction of this Gmail export and have not been checked at the image on this sheet; they are reproduced as the extraction has them. The run is 1 sheet. ↩

51. Epstein to Lisa New, Tue, 17 Feb 2015 11:59:30 +0000: “yes and i will get a private read”. EFTA00864303 p.1. UNVERIFIED, from a search hit: this sheet was not rendered. ↩

52. Epstein to Lisa New, Tue, 17 Feb 2015 12:00:40 +0000: “is the greant ( sic ) in your name?” — the ”( sic )” is Epstein’s own, on the face of his message. EFTA00864305 p.1. Her reply of 12:12:43 +0000, “the grant is in my name. Yes. I’m the project lead / Elisa New—Poetry of Character Development”, is at EFTA00696295 p.1, and a further reply of 12:12:06 +0000 about the budget draft is at EFTA00632026 p.1. UNVERIFIED, from search hits: no sheet was rendered. All three Date: headers carry an explicit +0000 offset. ↩

53. Lisa New to Epstein, Sat, 03 Oct 2015 12:47:03 +0000, “poetry”. EFTA00630696 p.1. UNVERIFIED, from a search hit: this sheet was not rendered. Epstein’s reply of 14:10:20 +0000 the same day, quoting the whole message back, is at EFTA00844477 p.1. The two extractions differ at “ISO per episode” against “150 per episode” and at “I)” against “1)”, which is scan variation; the sentence quoted on this page is identical in both. This is the applicant’s own statement that the grant did not come. It is not a Foundation document and no Foundation document recording the outcome was located — see the search list in the body. Both runs are 2 sheets. ↩

54. Absence searches, run 2026-09-11 against pages_fts over the whole index, FTS5 phrase and conjunction matches on text_content, with no LIMIT. The forms were written down first as descriptions of what a Foundation decision document would say in its own words, and none of them is drawn from wording already on this page. Returns: "grant agreement" + Templeton 0/0; "grant number" + Templeton 0/0; "payment schedule" + Templeton 0/0; "final report" + Templeton 0/0; "counter-signed" + Templeton 0/0; "regret to inform" + Templeton 0/0; "your proposal has been" + Templeton 0/0; "not be funded" + Templeton 0/0; "we are pleased to inform" 97 pages / 97 documents alone and 0/0 conjoined with Templeton; "award letter" + Templeton 2/2, both the Media Lab proposal, read; "unable to fund" 5/5, all read, none concerning the Foundation; "declined" + Templeton 7/7, all read and listed in the body; "trustees" + Templeton 2/2, both read; "Online Funding Inquiry" 1/1, being the invitation itself; "Full Proposal" + Templeton 2/2, being the invitation and the Letter of Intent form. A conjunction is scoped to a single page, so a decision letter whose body and letterhead fall on different sheets could evade the conjunctive forms; the phrase-only forms above are not so scoped. These are searches of a text layer and a phrase search reaches only the copies whose extraction survived: a decision letter misread in one character of “Templeton” would be returned by none of them. And this release is Epstein’s files, so it would hold such a letter only if a correspondent sent him one. The claim made in the body is bounded to the sheets read for this page and to these searches, and this page does not assert that no such document exists. ↩