My appreciation of textual criticism – a nowadays somewhat marginalized subdiscipline that continues nevertheless to provide the foundation of our subject – has been vastly enhanced by Richard Tarrant's new book on the subject. I read it from cover to cover with great pleasure and satisfaction (several times laughing out loud, which doesn't happen often with works of scholarship), with great interest, and with dismay at my own ignorance, and I came away determined to be a better Classicist. This little volume is the fourteenth ‘suggestive essay’ published in CUP's Roman Literature and its Contexts series (established in 1990 by Denis Feeney and Stephen Hinds), but it does not – sadly – mark a revival of this excellent series, but rather a late addition. (There cannot be many Latinists of my generation who did not, as young scholars, aspire one day to be the author of one of these elegantly concise yet ground-breaking volumes.) On the face of it this volume is rather different from its predecessors, which usually engaged with cutting-edge theory from a Classical perspective; instead, Texts, Editors and Readers opens up to non-initiates such as myself a whole world of existing scholarship into which many literary scholars seldom venture, inhabited not only by the towering ‘heroic editors’ of the past (Chapter 1) but also by colourful characters such as ‘interpolation hunters’ (86), freewheeling neo-sceptics (77), elegant minimalists, and unrestrained maximalists. With a combination of vivid characterization, lucid explanation, and delicious detail, Tarrant outlines the challenges of establishing a decent text, and the techniques involved; in Chapters 3 to 5 we learn about recension, conjecture, interpolation, collaboration, and intertextuality. He also makes exceptionally clear the issues that are at stake in editing a text, and the tensions with which the discipline is charged. At every stage of the process, from the selection of manuscripts for scrutiny to the display of information in the final edition, choices need to be made that are bound to provoke dissent. The twin aims of providing a legible text and legible apparatus are often in conflict with one another. Eventually, to establish a readable text, an editor needs to choose a single solution and put all alternatives in the apparatus, which must then record the evidence and the decision process as far as possible. Done well, it allows us to understand the process by which the text of the edition has been established, and the contributions made by scholars over the years. But within Classics there is no agreement about precisely how this should be achieved, as Tarrant points out. As he makes clear with his comparison of two reviews of the same edition, one reviewer's ‘accuracy’ and ‘methodological rigor’ is another's ‘frivolous superfluities’ (25–6). Tarrant comments that one would hardly believe these evaluations pertained to the same edition of Lucan, but in fact the picture is consistent and the divergence of opinion is telling; what comes across strongly is that these two reviewers want something very different from their editions. The disagreement here is between a scholar who wants progress towards a better text, amending scribal errors and providing confident, robust conjectures, and another who is glad to find a text relatively untouched, but in the apparatus all the material that enables a reader to come to their own decisions about the variants to be preferred. The merits of both are clear; the tensions are between the aspiration for a readable, usable text and the desire to be transparent about the difficulties involved in establishing that text. A decisive reading may obscure ambiguities; excessive hedging muddies the reading. Every choice involves compromise: minimalists may omit important information that might allow the reader to draw different conclusions; maximalists risk cluttering up the page and seeming undiscriminating. Tarrant (a self-confessed minimalist) alarms us on pages 130–1 with the sight of the monstrous apparatus produced by an unrestrained maximalist. Meanwhile, while conservative critics are averse to new conjectures and stick as close to the manuscript reading as possible, conjecture emerges as a creative art form, where natural talent is enhanced by intimate appreciation of Latin literature and style (73); it can attract great admiration. I now aspire to be able someday to compile, as Tarrant does, my own list of favourite conjectures – a bit like a montage of favourite sporting moments, as one revels in the pleasure of seeing the execution of skilful manoeuvres. Chapter 6 brings our attention to a representative case where textual tradition and literary interpretation cannot be disentangled: is Propertius a ‘difficult’ poet, prone to elliptical writing, or is he an elegant writer whose text has been unfortunately mangled in transmission? In other words, where the text is hard to understand, do we spend our energies reading his poetry as if he were a modernist poet, teasing out cryptic meaning, or do we channel our energies into amending the text to something more easily comprehensible? One's prejudice about the nature of Propertius’ poetry inevitably shapes one's approach to editing the text. The question is insoluble, but the debates thereby evoked are illuminating. As Chapter 2 makes clear, this is a discipline that relies on persuasion and is characterized by strong rhetoric; the contempt and disgust that are directed at fellow scholars and inferior manuscripts are remarkable. Language is often emotive and moralizing; the bracketing of problematic lines described as ‘a coward's remedy’ (86, n. 2). Tarrant himself, who takes a light and genial tone throughout, doesn't shy away from describing a certain practice of citing scholars in the apparatus criticus as ‘an abomination’ (161). One of many evocative details is the idea of Housman storing up denunciations of editorial vices without a particular target yet in mind (68). Traditionally, self-belief and decisive authority have been the hallmarks of the ‘heroic’ style of editing, and these qualities are especially unfashionable in our own era, which prizes the acknowledgement of ambiguity and hermeneutic openness. Tarrant encourages us to accept that the notions of the ‘recoverable original’ or the ‘definitive edition’ are myths, but at the same time to acknowledge that they are necessary myths (40) for this ‘doomed yet noble’ endeavour (156). A critical edition is no more nor less than a provisional ‘working hypothesis’ which invites continued and continual engagement. As Tarrant puts it: ‘any edition, to the degree that it stimulates thinking about the text, begins the process that will lead to its being succeeded by another edition’ (147). Textual criticism should be, therefore, a collaborative endeavour to be marked by humility and an acceptance of the open-endedness of interpretation, of the hermeneutic work that an editor needs to undertake, and also of the overlap between the roles of editor and reader. It is easy to perceive textual criticism – with its heyday in the nineteenth century – as constituting the dry and dusty past of Classics, and indeed Tarrant treats us to a most entertaining account of its Heroic Age, when Housman et al. lashed one another with cruel wit and erudite put-downs. However, Tarrant also makes an irrefutable case for the continued relevance, and indeed the exciting future, of textual criticism – despite the fact that it has lost its position at the centre of our discipline, and so many of us are untrained and unable to appreciate its value. Tarrant's depiction of the discipline brings home the lesson – which we already knew, but now really get – that all classical scholars ought accordingly to be aware of these general issues and to have some grasp of the specific routes by which the text they are reading has been reached, the problematic aspects of that text, and the issues involved in attempting to resolve its problems. Such is the information that an apparatus criticus attempts to convey, and it may therefore be judged on how effectively and efficiently it does so. Having made all of this so clear and in such an engaging fashion, Tarrant concludes by providing as an appendix a helpful guide for the inexperienced to reading a critical apparatus. The final chapters explore two questions in particular: what can technological advances contribute (for instance in access to and presentation of manuscripts), and is the current model of the apparatus criticus fit for purpose? On the latter issue, Tarrant would like to see, at the very least, more scope for providing in the notes nuanced indication of the editor's feelings about the choices he or she has made. He proposes the wider use of phrases that allude to the internal struggles behind a rejected variant, for instance (such as utinam recte or aegre reieco) or the introduction of new symbols for the apparatus that would signal degrees of suspicion – although he doesn't go quite so far as to second Donaldson's suggestion for a pictorial symbol of ‘a small ostrich, with head in the sand’ to denote occasions where an editor follows a manuscript out of despair of making actual sense of the text (58, n. 25). Early in his essay, Tarrant expresses regret that new editions are less likely to be reviewed than other forms of scholarship, and, with the decline in the requisite editorial knowhow, it easy to see why: reviewing a new edition of a text is not a job that can be undertaken with confidence by most scholars of Latin literature. How can one pass judgement on an editor's decisions without a very sound knowledge not only of the work but also of the manuscripts available, of the relationships between them, and of the subsequent critical tradition? How can one comment on individual amendments or conjectures without an understanding of the entire interpretative framework which the critic has brought to bear? One of the many valuable things I have learned from Tarrant's book is that it not always necessary to comment on individual cruces; equally useful can be an evaluation of the general approach and principles upon which an edition is both established and communicated.