ABSTRACT
I begin with a hypothetical question. Suppose that the ninety-seven letters which John of Salisbury wrote for his employer, Archbishop Theobald of Canterbury, had survived, not in the collection which John sent to his friend Abbot Peter of Celle but in an early Canterbury Register or letter-book, how would we know that they had been written by John, rather than by another member of the household, or even by Theobald himself? That question should send shivers down the spines of all editors, especially editors of medieval letters, for it draws attention to the fallibility of the sources upon which scholars rely to establish authorship. We know - or think we know - that John wrote many of Theobalds letters only because they survive in Johns first letter collection 1 ; and, even there, there are some uncertainties about whether a particular letter was written in the archbishops name or in Johns 2 . There is little differentiation of style between the two categories. It seems that John kept drafts of letters which he wrote on official business in Theobald’s curia, whether that business was the archbishops or his own: but he was interested in the literary composition, not the historical content, and the absence of fully explicit headings and, generally, of protocols - and sometimes even of the arenga - creates serious difficulties in the identification of recipients. The headings were of the briefest - Conventui Cant, ecclesie, Domino pape Adriano/Ad. 3 , Domino pape 4 , Adriano pape 5 , R. cancellario 6 , Bosoni carnerario 7 , episcopo Sar. 8 , Cellensi abbatil abbati Cellensildomino Cellensi/Cellensi 9 , cancellario Regis/cancellario 10 — or absent altogether 11 , and sometimes wrong 12 ; and the problems of identification so created were exacerbated by the omission of addresses. In Johns early letters, only eight protocols were preserved 13 . It might be thought that such omissions reflected the decision of a copyist, but they were characteristic of the best manuscript of John’s second collection also (MS Q: Paris, Bibliothéque Nationale, MS lat. 8562), which Millor and Brooke determined to be the closest to „an excellent, authentic text". In that collection, too, protocols are conspicuous by their absence and some headings are obscure or inaccurate 14 .
