This study had its inception in a project to prepare the first modern edition of a particular medieval medical text, a Latin translation made in Montpellier in 1299 of a regimen of health attributed to the famous Spanish physician Ibn Zuhr (Latin, “Avenzoar”): a Jewish philosopher had translated the Arabic into a Romance vernacular and read it out loud to a Christian surgeon, who wrote it down in Latin. When the surviving manuscripts of the result were examined and collated, however, it began to appear as though there were not one but two similar yet distinct versions of the text scattered among them. A comparison with the original Arabic text might have helped establish the relation between the two Latin versions, but a search of the literature indicated that no manuscript of the original had survived. However, a Hebrew translation seemingly prepared by the Jewish member of the team was available in two copies, so the study had to expand: the Hebrew text had to be established in order to see what light it could throw on the two Latin versions and the lost Arabic that underpinned all three—for, as it proved, the Hebrew translator had left in transliteration the Arabic passages that he had been unable to understand.

Our comparative study of the three texts confirmed that our first perception had been correct, that the Latin texts were indeed different and, not only that, embodied a first draft and a subsequent revision of the translation. We also became convinced that our close comparison of those two texts turned up many signs of the actual translation procedure: e.g., a deliberate omission in the first draft of passages that the translators could not immediately understand and planned to come back to on a second pass; a misunderstanding of words spoken out loud as they passed from one member of the team to the other; stylistic and rhetorical “improvements” made to the revision, with the aim of matching its language to the needs and expectations of the academic community for which it was being composed. The result that we arrived at was a picture in unparalleled detail of how such a “four-hands” translation was produced, of how the different backgrounds and preparations of the two members of the team played into—negotiated, perhaps—the form and language of the eventual product.

To substantiate these conclusions, we offer here 1) an extended historical introduction incorporating the rationale for our argument; 2) an edition of what we believe to be the initial draft hastily prepared by our translators, followed by an edition of the final translation as they eventually revised it; 3) an edition of the Hebrew text, including all its Arabisms; and 4) an annotated English translation of that Hebrew text, which comes as close as is presently possible to communicating the content of the original Arabic regimen of health that launched this long tradition. We hope that providing all this evidence will allow readers to evaluate our conclusions and to decide whether we have proved our case. At the very least, however, we have achieved (in #2 and #4 above) the limited goal that originally motivated us, an edition and English translation of this Arabic regimen of health in the form in which it was read and studied in later medieval Europe.

We are grateful to the history departments of the University of North Carolina and of Duke University for their willingness to provide logistical support for two of their emeritus members over the past several years. We are also particularly grateful to a number of scholars for advice and substantive contributions: to Cristina Alvarez Millán for advice pertaining to the lost Arabic original; to Joël Chandelier for confirming manuscript readings; to Seymour Mauskopf for bibliographical researches; and to Julia Zwink and Guido Mensching for help in identifying Romance (Occitan) terminology.

One of the greatest pleasures of this research program, carried on over the last half-dozen years, has been for us to appreciate how much can be gained from the close interaction of three individuals trained in three rather different scholarly traditions; it has been a wonderful experience. Readers will find a mark of that in our individual practice in such things as transliteration—e.g., that Arabic ج is sometimes rendered as j, sometimes as ǧ, and once (in a quotation) as ŷ. We thought of course of imposing a standard system, but in the end we decided to let the inconsistency remain, to bring home to readers the active character of this collaboration that has meant so much to us.

The Regimen sanitatis of “Avenzoar”

Stages in the Production of a Medieval Translation