Thanks Bernd. I did some digging on this with a deep research on Claude and GPT and then comparing them.
The standard treatment is J. J. G. Alexander, Medieval Illuminators and Their Methods of Work (Yale University Press, 1992), chapter 3, "Programmes and Instructions for Illuminators", pp. 52-71, plus a census of related preliminary marginal drawings in his Appendix 2. Publisher page: You are not allowed to view links.
Register or
Login to view. ; searchable preview: You are not allowed to view links.
Register or
Login to view.
Short version: Written instructions to the artists are attested from Late Antiquity. The fifth-century Quedlinburg Itala already carries directions to its painters, partly hidden under the paint (Alexander pp. 4-5, fig. 1). The surviving evidence then increases from the thirteenth century, when production moved to lay professional workshops and the work was split between scribe, draughtsman and colourist.
The functions Alexander documents: subject programmes telling the artist what to paint (the fullest is Jean Lebègue's written programme for a Sallust, Bodleian MS D'Orville 141); colour specifications; placement and layout fixes; checking and correction of finished work; and miniature counts tied to payment. These notes were working apparatus, expected to be erased or trimmed off at binding, so what survives skews toward unfinished books or, as in the Cgm 728 case, carelessly finished ones.
Examples by region:
France. The Bible moralisée (Paris, c. 1230, in volumes split between Oxford, Paris and London): a checker went through the finished roundels noting errors in French, including "Metez ihesu la robe blanche" (BL Harley 1526); the robe was painted purple anyway, so the correction was never made and the note never erased (Alexander p. 62). The Oxford volume is fully digitised: You are not allowed to view links.
Register or
Login to view. . A thirteenth-century Paris Psalter, BnF lat. 108, preserves a marginal letter code, 'a' = azur, next to strokes possibly recording payment (Alexander p. 64 and p. 166 n. 62, citing P. Stirnemann, "Nouvelles pratiques en matière d'enluminure au temps de Philippe Auguste", 1982). I could not locate a digitisation of lat. 108, so that one rests on the printed citation. BnF fr. 823 (Deguileville, dated 1393) has "Remiet ne faites rien çy...", an instruction naming the artist (Alexander p. 166 n. 50).
England. Cambridge, St John's College MS 262 (K.21), folio 50v: an Anglo-Norman instruction above a miniature, "Comment Barrabas li leres fu livere hors de prison" (Alexander p. 166 n. 49).
Iberia. The Roda Bible, BnF lat. 6 (Catalonia, eleventh century): a Latin corrector's note in volume 3, folio 97, complaining that text and miniature are misplaced (Alexander p. 62). Volume 3 on Gallica: You are not allowed to view links.
Register or
Login to view.
Germany. The fullest instructional witness is the Göttinger Musterbuch (c. 1450, SUB Göttingen, 8° Cod. Ms. Uffenb. 51 Cim.), a painter's manual giving colour by colour directions ("Wie du alle varbe temperieren vnd riben sullent"). Digitised: You are not allowed to view links.
Register or
Login to view. ; transcription and translation: You are not allowed to view links.
Register or
Login to view. . The Cgm 728 and the Strasbourg BNU 2152 copy sit at the very tail of this tradition, into the print era.
Italy. Notably thin: Alexander's census records only a single Italian instance.
One caveat: Alexander is essentially silent on herbals. His examples cluster in Bibles, Psalters and Books of Hours, and his index of manuscripts contains none of the classic herbal witnesses (no Egerton 747, no Carrara Herbal, no Sloane 4016). So the comparative base for colour annotations in herbal-type books is thin in the published literature..
The process of annotate first, paint badly later is a centuries-old workflow, so the presence of colour annotations in the VM barely constrains its date on its own.