3.1The primary texts
Two primary texts are read as evidence, and one as frame. All three are in the public domain and are held as files in the parent study's sources.
Greenspan's testimony of 1 October 1998 (Greenspan 1998) is the text served by the Federal Reserve Board's website, fetched in October 2026. It is born-digital, not a machine reading of a scan; the site's navigation and footer were removed and blank lines closed up, and nothing else was changed. It keeps its own spelling slips ("liqudiation", "arrray", "decidely"), which are reproduced where quoted. The testimony refers more than once to the account of "President McDonough", given just before it; the file holds only the heading "Testimony of William J. McDonough", not the statement, which is not read.
The Board's press release of 14 September 2008 (Board of Governors 2008) is likewise the text served by the Board's website, born-digital, with navigation removed. It is short: a heading, a paragraph of announcement, two paragraphs of statements by the Chairman, and five paragraphs on the facilities and on section 23A.
Bagehot's Lombard Street (Bagehot 1873) is read in five excerpts from chapters II and VII, from Project Gutenberg's volunteer transcription; the file records that the transcription was not checked against a page image and should be treated as uncorrected. It is not a machine reading of a scan, so the article quotes it, sparingly, exactly as the file gives it, including the file's double hyphens for dashes. It is used only to frame the question in Section 4.
Every quotation of these texts is copied character for character from the files. Omissions are marked with an ellipsis. The texts are quoted inside the article's sentences; the modern records, below, are set apart as block quotations.
3.2The records
Each episode also has a record on Skeptron, a studio of the RISCON family that publishes questions about the past drawn from public sources. A Skeptron history case gives a dated event, a short scenario, and five questions with Skeptron's answers: the date of the record, a concrete fact, the boundary of what can be judged, a calendar check, and the condition of a later comparison, with the public source the studio checked. The records are written in Korean; the article quotes the English translation made once for the parent study, in whole sentences.
The parent study calls the two records The LTCM Recapitalisation (Skeptron's title: LTCM rescue negotiations: sources of funds and changes in control) and Lehman's Bankruptcy (Skeptron's title: Lehman headquarters bankruptcy and dealer subsidiary liquidity). The first is dated as of 14 September 2026 and names one source, Greenspan's testimony itself on the Board's website. The second is dated as of 13 September 2026 and names one source, an essay titled Support for Specific Institutions on the Federal Reserve History website. The second record does not cite the Board's release.
These records are modern summaries written in 2026 for a public platform. They are not archival documents, and they are not contemporary with the events. Nor are they synthetic scenes: each is a compressed reading of a public source. Each also carries sentences that the studio repeats across many cases with only names and dates changed; the parent study calls these the frame and does not treat them as evidence about any case. One of them, in Lehman's record, reads: "Announcements, decisions, implementation, and outcomes must not be combined into a single event." The fourth question of each case, a weekday computed by calendar arithmetic, says nothing about the event and is not used. The closing answer of each fifth question, on what would keep or withdraw a later comparison, is the same in substance in both records and is set aside.
For the Lehman record the parent study states a rule that holds wherever the subject is used, and the article adopts it in the study's own words: Skeptron's record of Lehman holds two entities, keeping the holding company's bankruptcy of 15 September 2008 apart from its dealer subsidiary's use of the Federal Reserve's existing credit facilities, and this study does not say that Lehman as a whole received or was refused a loan.
3.3The ledger
The two records belong to a set of 84 Skeptron cases that the parent study, Meridian's study of emergency acts in money and credit, read on a common set of questions. Each record was read twice, independently, by one model run twice. The readings answered fixed questions about the record's own words, with the frame set aside: whether the record shows an act by an authority; what kind; whether the act is presented as temporary or for a term; whether the record names a condition, date or event on which it would end; what became of it; whether two different dated moments are given; whether a figure is set apart from what was available or paid later; and whether the record names who bore the cost. These questions and five hypotheses about their answers were registered on 6 October 2026, before any case was coded for them. Where the two readings agree, the answer is treated as settled; where they differ, it is left open.
The ledger reads Skeptron's summaries, not the primary texts. What it reports is what a reader of the summary is told. The case sections state its results in plain words, and the figures are gathered in the Appendix.