Sources of the Elara S. Novikova archive 422 source records across 71 dossiers of four novels, one per distinct address rather than per work. 423 records: those 422 and one for the sources page itself. https://elaranovikova.com/sources.html Research date 29 July 2026. Links re-checked 30 July 2026. File last revised 2026-08-07. Encoding UTF-8. Every record carries the full citation as it was recorded, in its note field. Author, title, year, journal, volume, issue, pages, DOI and ISBN are filled only where they could be taken from that citation word for word; where they could not, the citation stands in the title field and the record is complete all the same. The note of every record also carries the verification state of that source: whether it was retrieved in full, seen as a catalogue record or found only as a search result. A record without an access date was never opened on the research date. The research files hold 480 entries, which come to 423 distinct works. 1 of them is not in this file: an internal working note that says of itself that it is not evidence. This is the research base of works of fiction. The sources are real publications; the archive built on them is a novelist's. FILES novikova-archive-sources.bib BibTeX novikova-archive-sources.ris RIS novikova-archive-sources.xml EndNote novikova-archive-sources.json CSL-JSON novikova-archive-sources.csv CSV WHAT IS IN A RECORD 423 records. 314 of them carry a title taken out of the citation, 163 an author, 188 a year, 403 an address. The rest carry the citation itself in the title field. No record is short of information because of it: the full citation is in the note field of every single one, parsed or not. A field was filled only where its value stands in the recorded citation word for word. Nothing was inferred, completed or looked up elsewhere. That is why a nineteenth-century Russian encyclopedia entry has no author here: the record does not name one. Records are deduplicated by address, not by work. A work cited in twelve dossiers is one record whose note names all twelve. But a work reached through three different addresses stays three records, one per address, and eleven titles in this file appear more than once for that reason. Which of the two you want depends on whether you are counting works or places a work was read; this file counts the latter, because that is what the research recorded. NOTES ON THE FORMATS .bib BibLaTeX with Biber. UTF-8, Cyrillic left as it is, no transliteration. Entry type @online is biblatex, not bibtex. .ris CRLF line endings, one ER per record, UTF-8 without BOM. Imports into EndNote, Zotero, Mendeley, RefWorks, Citavi. .xml EndNote XML. Reference types are restricted to the ones that carry the same number in every EndNote version; where the exact type would have been a guess, the record says Web Page or Generic rather than something more specific and wrong. .json CSL-JSON for Zotero and Pandoc. Author names are given as literal, because the records do not separate given and family name and a machine split would invent one. .csv RFC 4180, CRLF, UTF-8 without BOM, every field quoted. CITING THE ARCHIVE The first record of every file is the sources page itself.