This book, beautifully produced by Brepols, well illustrates the very high standards that are regularly achieved in the editing of the Church Fathers, especially in the Francophone countries. All the certainly genuine works of Cyprian have now taken their place in the Corpus Christianorum, and these new editions of the spurious De laude martyrii, De Iudaica incredulitate ad Vigilium and De rebaptismate replace editions in Hartel's third CSEL volume. The two editors state that they take joint responsibility for the volume, but pp. 1–505 seem to be almost entirely the work of Laetitia Ciccolini, who has edited the first three of the texts listed above, and pp. 507–96 of Paul Mattei.

Ciccolini's monumental discussion of the manuscripts is in a different league of excellence from that of earlier scholars. For example, she shows better than her predecessors how contamination occurring between late antiquity and the early Carolingian period precludes the use of the stemmatic method either to produce consistent groupings of the earliest witnesses or mechanically to eliminate readings. However, the most profitable use of my space may not be to summarise Ciccolini's general excellence but rather to point to the one significant weakness in her discussions: most of her many stemmata have at least one flaw, and manuscripts are far too often given a weighting equipollent to each other when in reality one is the source of (an)other(s). Although this does not affect constitution of the text, it does affect the classification of scores of recentiores.

Full justification of these observations would need a monograph and will be found when I publish my own researches into manuscripts of the Cyprianic corpus. I illustrate the problem with four examples where my collation (of other texts in the corpus, but the manuscipts are most unlikely to behave differently) has led me to different conclusions: (i) At pp. 134–9: Cicciolini makes 255 (Bodl. 210), 253 (Lambeth 106) and 257 (Cambridge, Pembroke 154) equipollent with each other, but 253 and 257 (also 651 [York XVI.I.1], which she confesses not to have seen) derive from 255, corrections to which they incorporate; (ii) At pp. 148, 358 and 464: 565 (Avignon 244) is not a younger sibling of 370 (Escorial S.I.11) but derives from it; (iii) At pp. 188–91 Cicciolini makes a host of Italian manuscripts equipollent to 216 (Δ, Turin Naz. D IV 37, s. xii), but Diercks's discovery that Δ was the source of this family of Italian manuscripts was his most important contribution to the study of the transmission of Cyprian; (iv) At pp. 347–8 Cicciolini makes 539 (Siena F V 13) and 566 (Reims 370) derive from a lost ancestor, which in turn shared another lost ancestor with 515 (Vat. lat. 9943); but the three shared errors that she claims to be found in 539 and 566 but not in 515 (De laude 8 formido] fortitudo, 75 quam] quod, and 82 ac] et are all found in 515. On p. 188 she rightly has 539 and 566 as descendants of 515, eliminating the two imaginary lost manuscripts.

Ciccolini edits her texts with care and discrimination. De habitu must have been the easiest to edit: the text is well preserved and there is rarely doubt about what Cyprian, who had a logical mind and knew how to express himself clearly, wrote. The other two texts are less well preserved, and editing is made difficult by the style of the authors, who did not always manage to express themselves with ease; many passages provoke doubt, but only very rarely did I conclude that another reading is clearly preferable to that chosen by Ciccolini. Her own conjectures include: De laude 21 fide deuotionis, 56 dominus for domus, 131 diuturnis factis for a desperate set of manuscript readings, Ad Vigilium 172 (deletion of qui), 213 promiscam for proximam or the like, 221 te for et. At De habitu 91–2 her repunctuation restores proper sense. I offer two objections. De laude 24–6 Nam et [et om. eAns] ‘quis est quem non ista res terreat, quis quem non admirationis suae pauore subuertat?’ The sequence ‘Nam et quis’ seems to be unparalleled in ancient Latin (as is ‘Nam ecquis est’), and since et is omitted by almost half the tradition, I should follow Hartel in deleting it. De laude 87–8 ‘Christus … cuius aguntur exempla quae petimus, cuius uirtus est qua repugnamus’. Only one manuscript, which on its own carries no authority, has est and omission of it seems better to suit the style of this author. He likes parallel clauses such as is given by cuius … cuius (often as here introduced by anaphora) and regularly has a part of the verb esse in the first member that is then understood in the second; cf. lines 24–5 (quoted above), 114–16, 119–20, 120–1. Even though aguntur is not quite sunt, the same principle seems to apply.

The transmission of the De rebaptismate is entirely different from that of the treatises edited by Cicciolini. A manuscript at Reims, destroyed by fire in 1774, served as the basis for the editions of Rigault (1648, and our earliest extant witness) and Baluze (1726) and for mss Vat. Reg. lat. 324 and Barb. lat. 653 (both s. xvii). Mattei reports these witnesses with great thoroughness. But the text must have been very difficult to edit. The author's thought is very often unclear and illogical, and there abound modes of expression that would have been intolerable to a traditionally educated writer like Cyprian. Very often it is impossible to know whether what is ungrammatical is due to the author or copyists. In a series of helpful notes Mattei shows that many of these ‘lapses’ can be paralleled in other writers of late substandard Latin; in principle, therefore, he must be right to adopt a conservative text. I rebel, however, at the appalling ‘et dixit dicens’ at 102, which could be removed by, for example, Schueler's ‘diuisit’ for ‘dixit’.