This book makes a somewhat mixed impression: on one hand, it does present a wealth of information. On the other hand, structured arguments and overall theory are conspicuously missing.
First, let us state a rather important detail: the book subtitle is largely misgiving. The "relique de Roncevaux" is certainly not the focus of the study, which deals, rather, with various literary texts and the many occurrences of a horn in fictional works. Although "real" horns that figure in church inventories as relics or precious heirlooms are present in the book, along with a few very nice black-and-white pictures at the end of the volume, they are not all that important. And as far as Roland's horn, or "olifant", is concerned, the author does not really deal with it until the third chapter of her third part--actually, the last chapter of her book. She suggests it should make this special horn the climax of her study. It does not work, however; if anything, these pages are rather disappointing compared to the rest.
Another caveat may be necessary: although Magnusdottir acknowledges at the beginning of her study the complexity of the horn as a multi-purpose object in medieval texts, the body of her book fails to implement the distinction between its different functions. One is led from one occurrence of a drinking horn to another about a blowing horn without any attempt at classifying the corpus or establishing working definitions. A number of the most interesting passages, in fact, have no relation whatsoever, at least none I have been able to discover, with the main topic of the chapter, or of the book.
Magnusdottir scans a large corpus, from Chretien de Troyes' Erec and Enidei, written about 1165, to the late fourteenth-century "chanson de geste" Lion de Bourges-- to mention only two French texts. Indeed, she demonstrates an impressive knowledge of various traditions and languages, and can quote as easily Norse sagas as Irish tales or Latin chronicles. However, one does not always see the connection between these texts. Too often, the only argument used to put them into perspective is the dangerous all-encompassing sentence "it is not only chance that makes this motive reappear in another text", far removed in time and space from the first one considered. Magnusdottir's method is to juxtapose examples, leaving the reader to draw conclusions from them. She hardly establishes any distinctions between time periods, literary genres, or thought frames. All documents, whether written (and among these, whether literary or "historical") or archeological, are considered equal in her approach.
Since she herself is probably conscious of the problems this may create, Magnusdottir often chooses not to conclude-- to leave the matter open, as it were, to further questioning. An astonishing number of chapters end up with a series of questions. While this may at first seem thought-provoking, it rapidly becomes a little frustrating. After all, the reader would like to get some more definite opinions about the issues the book is supposed to deal with! Instead of what, he/she is left to his/her own devices to determine not only what is at stake, but what one may reasonably deduce from the material presented. The theoretical tools used by the author have mainly to do with the anthropological and structural perspective exemplified in G. Durand's seminal book Les Structures anthropologiques de l'imaginaire. This approach has been successfully applied to specific medieval texts and problems by Ph. Walter, under whose direction this book was first written as Magnusdottir's dissertation. It requires a great deal of prudence, however, and a judicious amount of distance critique: there does not seem to be enough of that in La voix du cor.
That is especially apparent in the second part, "Les enjeux calendaires du cor": Walter's meticulous study of the calendar elements in several texts has allowed him to draw some qualified conclusions, that have significantly improved our understanding of these works. Here, on the whole, one is not convinced of the relevance of the perspective, and there are much too many assumptions that nothing really confirms. Since solid arguments are not available, there are many preposterous affirmations, based on very doubtful notions, and simply introduced by sweeping formulas of the kind "cela va sans dire" ("one does not need to say..."). The wealth of scholarly references that supposedly sustains these claims tends in fact to undermine them, since they mix indiscriminately classical or Medieval authors, and modern researchers whose conclusions are usually much more prudent (see for instance p. 310).
On the other hand, as a source of information, this book is a wonder; pleasant to read, vividly written, it takes the reader on a fascinating journey from the twelfth century to the fifteenth, from ancient Rome to medieval Iceland, and from Isidore of Seville to Claude Gaignebet. Small facts, textual details, innovative insights are woven together into a tapestry that provides deep enjoyment--as long as one's critical sense is not too much awake. Magnusdottir really knows much about drinking and blowing horns, and she is very good at sharing her knowledge. Even texts that one would have thought already well researched are coaxed to suggest new interpretations. In not a few cases, the juxtaposition of previously unrelated passages does open original perspectives.
Technically, the book's division into three parts, according to the old academic tradition in France, does not work terribly well. Since each of the three parts address the issue from a different point of view, there are many redundancies when texts or even passages are studied a second or a third time under another angle. There are also two appendices, the first one a translation of Robert Biket's Lai du cor analysed earlier in the book, and the second a translation of a significant passage of the Karlamagnussaga. Since other translations of these texts are available, one does not quite see the point of the exercise. The bibliography is rich and well researched; one wonders, however, why there are no italics in it for any titles. The same phenomenon occurs in the notes with a somewhat disconcerting effect, since it is hard to determine whether the item mentioned is an article or a book, or even a paraphrasis of a real title. A very complete index (that does not take into account names present only in the footnotes) shows at first glance how eclectic the book really is: simultaneously its greatest quality, and maybe the reason for its ultimate failure.