70 ( Bundle 31. )
S1. Source editor’s annotation There is also an undated leaf (bundle 22 ), in Joubert’s hand, and very enigmatic. I believe these are merely notes taken while reading—but reading what?... Here it is: “The eternal ones. — A path lined with suns. — Eloa, what do you see? — I see etc. — Future friend of cherubim and angels. — O creator. — A dying insect. — A cloud descended and covered one of the twelve seats of gold. — It has passed, it has passed, the first hour, etc. — He looked toward Thabord, But the judgment still endured. — By what name do you wish me to name you before your son? — You will name me by the name of father etc. — Are you all present? — yes, we are all present? — There is a voice missing. Are you all present? etc. — He imagines the days rising toward the heavens and the nights descending from them in the form of unfurled veils. — It seemed to him that his high priest’s robe was falling to the ground with a crash and suddenly crumbling to dust. — He took blood in his hand and cast it against the sky. — His blood flows! his blood flows! — He comes. —”
T170. All universal negatives, the paradoxical limit short of extreme wickedness, the long list, tiger comparison and qualified perhaps/doubtless remain. Consulted and betrayed follows the complete literal consulté et trahi , confirmed in the documented original-size color and native collation. The contextual difficulty remains; no conjectural insulté or other French repair is substituted. The finite relative preserves the reported acts of qui l’ait consulté et trahi rather than implying recruitment or purpose. Celle-ci clearly resumes colère ; English the latter retains that relation. Masculine plural ceux-là can resume the nearer caprices , while rois or vices also remain possible; English them leaves that plural antecedent open rather than fixing it to whims through former. Acrid blood preserves the bodily phrase. French retains complette , tirans , mocquent and plain paroitra/inouies . The English enclosing dashes around I mean clarify the source’s comma-marked expansion; the colon after every eye replaces a source comma. English uses conventional imperial names without changing the French. The judgments of emperors remain the source’s claims, not independently authenticated historical findings. The complete main paragraph is roman; only Bundle 31 is italic in its separate undated label. The final source call follows menti /lied before the terminal period.
T171. The source editor’s explicitly undated separate leaf, uncertain reading source, one annotation paragraph and every dash-separated fragment remain. Only liasse 22 and bundle 22 are italic, within roman parentheses; the quoted fragments are roman. The repeated questions, the reply’s question mark, lower-case yes/his, repeated naming and final He comes. — before the closing quotation remain. These printed fragments and quoi?... are not reproduction losses. Literal Eloa, Thabord, douzes with final s , plain O and Etes , and jetta with double t remain in French; the English does not silently identify or normalize the names. The source’s capital Mais after a comma corresponds to capital But, and first Son /later son to His/his. Seats of gold follows sièges d’or ; imagines gives a contextual literary reading of feint , with the two source finite image clauses recast as English rising and descending. This choice does not identify the unknown work or uniquely settle its literary force. No title, author, mythological identity or date is invented. The physical closing rule is not translated text.
T278. In the French source, printed item 70 is one complete roman main paragraph and one separate source annotation. Liasse 31. alone is italic in the heading; neither manuscript-group label supplies a date. The final raised call 1 follows menti before the main terminal period. Literal consulté is complete and supported by both original-size color and native 469. Its contextual difficulty does not authorize replacing it with insulté or another anticipated verb. The finite relative qui l’ait consulté et trahi remains; no lexical loss or conjectural French repair. Literal complette , tirans , mocquent and plain paroitra/inouies remain. Inouies has one ordinary i dot in color/native, not a certified diaeresis. The source’s universal negatives, en deçà paradox, maybe/doubtless qualifications, imperial names and repeated comparatives remain unaltered. The printed main uses commas around the je dis expansion, not enclosing dashes. The final historical accusations remain the source’s assertions. No outside historical authentication or reconstructed intention is claimed.
T279. In the French source, the entire annotation is one paragraph. Only liasse 22 is italic, within roman parentheses. Its explicitly undated separate leaf and unknown reading source remain unresolved; the closing rule is not text. Literal Eloa, O créateur. , douzes with final s , Thabord, capital Mais after the comma, plain Etes twice, jetta with double t and the first capital Son /later lowercase son remain. No title, author, date or mythological identity is supplied. All dash-separated fragments, the repeated present questions including the reply’s question mark, quoi?... and final Il vient. — before the closing guillemet remain. The printed ellipsis and sequence of fragments are not reproduction losses. Il feint remains a literary-image verb in the source; any English imagines interpretation is contextual and does not establish a historical source. These readings follow the source specialist’s documented color/native comparisons, with the stated qualifications retained.