observations: 35

Where each form was seen: corpus (and its reproducible record when there is one), date, observer, the verbatim note (original language), its English translation, and the exact probe that produced the figures when one was kept.

This data as json

id	form_id	form_name	corpus	corpus_id	document	date	observer	excerpt_en	excerpt_original	lang	probe
35	form-064	Period typographic conventions	PubMed abstracts on arginine, before 1990 (3,642 records, harvested 2026-07-29)	corpus-pubmed-arginine-pre1990-3642		2026-08-08	house observer (LLM agent), see report (Korela For Science, document 04 and annex 64)	184 titles in full capitals in a single corpus — "STUDIES ON THE NATURE OF RESISTANCE OF GRAM-NEGATIVE BACILLI TO PENICILLIN". A convention of the period, which breaks any case-sensitive entity recognition. It is not noise (23): it is a period norm, regular and systematic. Noise is corrected by tolerance; a convention is recognised. The time span is huge: 1917 to 1989 in a single corpus.	184 titres en majuscules intégrales dans un seul corpus — « STUDIES ON THE NATURE OF RESISTANCE OF GRAM-NEGATIVE BACILLI TO PENICILLIN ». Convention d'époque, qui casse toute reconnaissance d'entité sensible à la casse. Ce n'est pas du bruit (23) : c'est une norme d'époque, régulière et systématique. Un bruit se corrige par tolérance ; une convention se reconnaît. L'étendue temporelle est énorme : 1917 à 1989 dans un seul corpus.	fr	probes/2026-08/line-064-arginine-titles.py