George Church

Named on the documents as	george church in lower case in the sender field of his own messages, and George Church in the same field on others; Church, George where an address book supplies the display name; George M. Church in formal contexts; Geo Church twice; Dr. Church in third-party scheduling traffic; and --Geo as his own sign-off12
Range of documents cited here	November 2013 to March 2017; the wider population’s parsed dates run 7 October 2010 to 25 March 20173
Size of the file	2,196 documents on the exact phrase "George Church"; 2,240 over the union of seven forms established below. The surname alone is not a key: church returns 4,780 and is a common English noun41
The act this page is built on	On 11 July 2014, the day after Epstein proposed a call, he sent him a provisional allocation of $10,000,000 across nine of his own ventures and a held reserve, itemised line by line, two days after Epstein proposed a call56
What no document read here shows	That any of the $10,000,000 was paid, in whole or in part; any completed transaction between them in either direction; that he set foot on Little Saint James; or anything about his conduct beyond what the documents cited here record
Covered elsewhere on this wiki	The science-philanthropy network around Epstein: Nathan Myhrvold, Roger Schank

George Church, the Harvard geneticist, appears in the released Epstein files as the author of an itemised $10,000,000 investment proposal he sent to Jeffrey Epstein in July 2014, and as a laboratory head who twice pressed his own staff on Epstein’s behalf. Nothing on this page is a finding of wrongdoing. He is not accused of anything, and no document read for this page suggests he knew of any offence. Nor does any document read here show that the $10,000,000 was ever paid, his own message says the exercise had cost nothing so far, and the population contains no completion record.

The $10,000,000 allocation

On 10 July 2014 Epstein wrote to him proposing a call the same day. The next day, 11 July at 17:04, Church replied with an allocation he called provisional, against a stated sum of $10,000,000, together with a paragraph thanking Epstein for what he had said the previous morning.⁵

The list is itemised and it sums exactly. Nine of his own ventures and research efforts carry a figure each, totalling $7,500,000, and the remaining $2,500,000 is held back, to go later to whichever effort progressed furthest. Nine items plus the reserve reach the printed total of $10,000,000, and the desk’s own addition of the line items reproduces that total to the dollar.⁶

The nine efforts span the whole of his laboratory’s programme, a genome-collection study, an organ-transplantation venture, industrial biochemical sensing, protein and virus-resistance engineering, aging reversal, invasive-species control, a sequencing method meant to take share from clinical diagnostic imaging, human genome engineering for spaceflight, and cold-resistance engineering in elephants. Each is attributed on the sheet to named colleagues, twelve individuals in all, three of the efforts carrying two names each. The sheet’s own wording varies: seven of the nine say the effort is led by the person named, one says the work was collected by them, one naming them as founding partners in the venture. Two separate merge notes appear, touching four of the nine efforts, not one pair.⁶

Three things about the document are worth stating precisely. He wrote that the exercise was already interesting to him even though nothing had been spent, so the sheet records a proposal, not a transaction. He offered it unprompted and at length, noting that it might be more than Epstein wanted at that stage. And he framed it partly as a departure from ordinary venture-capital practice, recalling that in 2003 investors in a sequencing company had taken another scientist’s advice over his own, so that his technology ended up inside a company already trading rather than a start-up; that company was sold for $117,000,000 in 2012 while the other went bankrupt. The sheet gives that other scientist by forename only and this page does not supply a surname.⁵⁷

Two days later, on 13 July, he offered a Skype address and a mobile number and agreed a time to speak. Neither value is reproduced here, and the reason is recorded in the note.⁸

The genome request, and a laboratory pressed more than once

In November 2013 Church wrote to Epstein’s assistant, Lesley Groff, about a request then sitting in his laboratory, and she forwarded his reply to Epstein the same evening under a subject line naming him and the word genome.⁹

What he told them is a sequence of three facts. The matter had passed out of the hands of the colleague who had been handling it. The obstacle was how the laboratory tracked its samples, and he had asked his colleagues to improve how samples were tracked. The team had just received a grant of $500,000 which did not cover fixing it. He added that his colleagues objected to him taking up single cases himself, and that he meant to carry on regardless.⁹

That last clause is the act. A laboratory head records that his own staff object to him taking up single cases himself, and states that he will take up this one anyway. The request itself is not described in the documents read for this page, and this page does not infer what it was.

The shape of the relationship

The correspondence is scheduled by third parties more often than by either man. Across 400 page-hits sampled from the population, the commonest sender field is not Epstein’s or Church’s but Lesley Groff’s, at 54, against 20 for Church in lower case and 19 for Epstein; a calendar service and two further assistants account for most of the rest.² The commonest subject line is Re: Jeffrey Epstein at 45, that is, other people corresponding about Epstein, with the July 2014 thread second at 22, its own subject line not reproduced here.¹⁰

The population’s density is concentrated in four years. Parsed date tokens run from 7 October 2010 to 25 March 2017, and the count by year peaks sharply across 2013 (283), 2014 (225) and 2015 (188), with 2011 and 2012 together at 158 and the years either side in single or low double figures.³ In January 2015 Groff was still circulating his availability windows to Epstein for a call the same day.¹¹ In August 2016 his name appears in a thread with the literary agent John Brockman under the subject RE: Jeffrey Epstein.¹²

One document in the population is not correspondence at all, and it is worth describing accurately because this page previously understated it. It is an unsigned and undated narrative in praise of Epstein’s science funding, held in his files, which names his conviction and the eighteen-month sentence in order to argue past them. Its actual subject is not Church. It records that a $30 million grant established Harvard’s Program for Evolutionary Dynamics in May 2003, with the cooperation of Lawrence Summers, then the university’s president, under the direction of Martin Nowak. Church appears in it once, in a list of researchers said to have been supported by 2003 beside three Nobel laureates in physics and, among others, Marvin Minsky and Stephen Hawking. It is a characterisation written by someone unnamed, not a record of any payment, and no payment to Church is established anywhere on this page.¹³

Identification: a common noun, and a middle initial that is not a subset

The surname alone is not a key. church returns 4,780 documents and is an ordinary English word: "the church" accounts for 480, "Church of" 219 and "Church Street" 249. No figure on this page is keyed to the bare surname.⁴

The exact phrase "George Church" returns 2,196 documents on the authoritative route. Six further forms add 44 between them, for a union of 2,240:¹

form	documents	outside the base set
"George Church"	2,196	,
"Church, George"	103	4
"church@harvard"	93	1
"George Church, Harvard"	89	0, a strict subset
"Church lab"	73	9
"George M. Church"	34	29
"Dr. Church"	32	4
"Geo Church"	2	1

The middle-initial form is the instructive one: 29 of its 34 documents lie outside the full-name set. The desk records this as a general mechanism, a middle-initial form and a bare form are not in a subset relation in either direction, so a page written against one misses part of its own record. Here it is nearly disjoint.¹

Forms tried that return nothing are given because a form that fails is evidence about the corpus: gchurch 0, churchg 0, "George Churchill" 0, "Prof. Church" 0, "Professor Church" 0.¹

The floor is 2,240 documents and it is a floor, not a total. Its stated bounds: it is the union of eight forms only; the bare surname was deliberately excluded as unsafe and its 4,780 documents were not individually classified; no Damerau–Levenshtein enumeration of the index was run for this page; and the church@harvard handle was measured as a string, not resolved against a full address. No claim of convergence is made.⁴

Why this page quotes nothing from either thread

Both exchanges exist in the release in more than one copy, and the copies are not identical in what they withhold. Quoting from either would hand a reader a string far narrower than the subject’s own name, which returns 2,196 documents. So this page states the content of both exchanges and quotes neither: the finding is reproduced, the citation is given, and a reader who wants the wording has the sheet. The desk’s working is in the pass report, not on this page.

Every figure that does appear was measured before it was printed, and none is a usable handle.¹⁴

---

Footnotes

1. Counts on the authoritative pages_fts route, scoped text_content, each form run as an exact phrase and the returned identifier sets compared by set difference against the "George Church" base. The union of the eight forms is 2,240 distinct documents. Zero-returning forms are listed in the body. ↩ ↩² ↩³ ↩⁴ ↩⁵

2. Sender fields extracted from the first From: line of each of 400 page-hits sampled from the "George Church" population. Lesley Groff 54, george church (lower case) 20, Jeffrey Epstein 19, Jeffrey E. 11, a calendar service 7, Joi Ito 6, Meghan Radden 6, George Church 4. This is a sample of page-hits, not a census of the population, and is not an act rate. ↩ ↩²

3. 233 date tokens of the form M/D/YYYY parsed from the same sample, earliest 7 October 2010 (EFTA01811639), latest 25 March 2017 (EFTA02208947). Year counts from a separate pass over the leading 3,000 characters of each hit: 2000 (2), 2003 (14), 2006 (7), 2011 (92), 2012 (66), 2013 (283), 2014 (225), 2015 (188), 2016 (36), 2017 (4). These are token counts, not document counts, and a document may contribute several. ↩ ↩²

4. Counts on the authoritative pages_fts route, scoped text_content. "George Church" 2,196; union of eight forms 2,240. The bare surname church returns 4,780 and was deliberately excluded as unsafe: "the church" 480, "Church of" 219, "Church Street" 249, "Church St" 18. The 4,780 were not individually classified and no figure here is keyed to the surname. Triage figures taken on the faster document-level index were confirmed against this route before use. ↩ ↩² ↩³

5. https://epstein-data.com/EFTA01916477 pp. 1-2. Read at the image at 400 dpi by a verification pass, which established that Epstein’s message of 10 July sits on printed p. 2, not p. 1. Church to Epstein, Friday 11 July 2014 17:04, on a thread whose produced header is stamped Sunday 13 July 2014 15:38:09; Epstein’s message of Thursday 10 July 2014 08:07 is quoted within it. The sheet is one of several copies; see the leak note. ↩ ↩² ↩³

6. https://epstein-data.com/EFTA01916477 pp. 1–2. The nine line items and the reserve are printed as a column of figures beside a column of labels. The desk added the ten printed figures independently: $7,500,000 across the nine efforts, with $2,500,000 held back, reaching the printed total of $10,000,000 exactly. Nine named individuals are given as leads. Read at the image at 400 dpi by a verification pass. ↩ ↩² ↩³

7. Same sheet, p. 1. The 2003 episode, the two companies and the $117,000,000 sale figure are Church’s own account in his own message. The other scientist is given by forename only on the sheet and this page supplies no surname, no corpus count and nothing from which one could be inferred. Neither company is named here: the sheet names both, and a reader holding this note holds the sheet. ↩

8. https://epstein-data.com/EFTA01917808 p. 1 and https://epstein-data.com/EFTA01916477 p. 1. He offers a Skype address and a mobile number and later agrees a time. Both values are withheld here. They are contact values, which this wiki does not reproduce for anyone, and one sibling copy of this family prints the Skype handle where eleven bar it, see the leak note. ↩

9. https://epstein-data.com/EFTA01754222 p. 1. Lesley Groff to Epstein on the evening of Monday 25 November 2013, forwarding Church’s message to her earlier that afternoon. Read at the image at 400 dpi by a verification pass. One of several copies; see the leak note. ↩ ↩²

10. Subject lines from the same 400 page-hits: Re: Jeffrey Epstein 45, the July 2014 thread’s own line 22 (not transcribed: it reaches every copy of a family whose copies differ), Jeffrey Epstein 19, George Church 12, and eight further lines in single figures. ↩

11. https://epstein-data.com/EFTA01747101 p. 1. Lesley Groff to Epstein, Monday 12 January 2015, circulating Church’s availability for a call that day. The produced Sent: stamp on this production is UTC and in-body quoted stamps are local Eastern, established independently by a verification pass from a four-hour offset between the same message’s produced and quoted stamps on two copies and corroborated by two forward latencies; the message was therefore sent in the early afternoon Eastern, not the evening, and this page states no clock time for it. ↩

12. https://epstein-data.com/EFTA02044506 p. 1. John Brockman, Friday 19 August 2016 21:34:42, subject RE: Jeffrey Epstein, confirming dates. Church is named in the thread; the exchange is between other parties. ↩

13. https://epstein-data.com/EFTA01966271 pp. 1-3. Read at the image at 400 dpi by a verification pass. Unsigned and undated. A narrative in praise of Epstein’s science funding, whose subject is the Harvard programme and its director rather than Church; it names the conviction and the sentence and argues past them. Church is named once, in a list. The $30 million figure, the May 2003 date, the programme and the two named principals are the document’s own; the desk records them here as the document’s content and claims nothing from them about Church. The wiki covers the programme’s principals on their own pages. ↩

14. Measurements taken by the desk on the authoritative pages_fts route before any figure was printed. Each dollar figure here was checked against the document families it belongs to and none is selective: $10,000,000 returns 5,204 documents, $2,500,000 returns 1,972, $500,000 returns 1,143, and the sale figure as $117M returns 26 - every one broader than, or of the same order as, the 2,196 a reader already has from the subject’s own name. One itemised figure from the July 2014 sheet was measured, found selective, and is not printed here. The desk swept the whole rendered page as one continuous stream of words at a one-word floor and adjudicated every span occurring in any cited document; the working and the residual are recorded in the pass report rather than on this page. ↩