A textual critic can hardly pass over the interference of annotations and corrections in the transmission of ancient texts. Building on her previous studies and critical editions of Greek and Latin prose texts, Giuseppina Magnaldi’s latest volume sheds light on ancient correction methods, specifically focussing on the so-called ‘signal-word’ (parola segnale). This method is best illustrated through an example drawn from Magnaldi’s selection (pp. 37-8: Apul. Socr. 2.3 = 120 Oudendorp; the text and selective apparatus are from Magnaldi’s OCT):
uaria quippe curriculi sui specie sed una semper aequabili pernicitate, tunc <uero> progressus [tunc uero], tum autem regressus mirabili uicissitudine assimulant pro situ et flexu et abstituto circulorum.
tunc uero transtuli ut supplementum ad tunc1 (del. Vulcanius, cf. Magnaldib 407-10) : tunc (tum z) … tunc (tum R) uero codd. (post uero vac. rel. M)
tunc uero is evidently misplaced and does not make sense in the context. Magnaldi suggests that a scribe, noticing the omission of uero, wrote it in the margin with an accompanying signal-word—in this case, tunc2—to indicate where it had to be incorporated, i.e. after tunc1. Failing to interpret this correction, a later scribe keyed in the whole annotation instead, leading to the awkward reduplication of tunc at such a short distance (tunc … tunc uero). Dittographies of this sort loom large in ancient texts, but their identification is not always straightforward because wrongly integrated corrections often appear next to a different word from the intended one and / or are modified to align with the context in ways that obfuscate their original purpose. By drawing on a wide-ranging selection of case-studies, this volume makes a strong case for readers to be alert to the possibility that dittographies, repetitions, and other comparable signs of interference might stem from misinterpretations of this correction method.
The first chapter provides a general introduction to this usus emendandi wherein a signal-word flags the correction of a miswritten term or the integration of an omitted pericope. As shown above, signal-words generally coincide with and duplicate either the word(s) before or the word(s) after the precise point in the text in which the correction / integration was supposed to be reinstated: selected examples neatly illustrate this practice in the second (marginal corrections) and third chapters (linear and interlinear corrections). The fourth chapter puts Magnaldi’s line of textual research in perspective, pointing to passages where early modern and modern editors appear to recognise the existence and function of signal-words, emending the text accordingly. In the fifth chapter, Magnaldi draws attention to signal-words—or traces thereof—that confirm previously overlooked textual proposals, thus demonstrating how understanding this mechanism may effectively help editors adjudicate on emendations and transpositions.
The sixth chapter considers examples of signal-words in the transmission of twelve prose writers (in order of appearance: Apuleius, Pseudo-Apuleius, Cicero, Gellius, Livy, Macrobius, Petronius, Seneca the Younger, Seneca the Elder, Suetonius, Tacitus, and Varro). Some examples are newly presented in this volume, others—those which Magnaldi has either dealt with in her publications or printed in her editions—are expanded upon with a step-by-step account of the genesis of the transmitted reading, a detailed exposition of the proposed solution, and a translation of the emended text. This chapter offers a significant contribution to the constitutio textus of each passage under discussion. Hence, scholars planning to edit or comment on the aforementioned authors (especially Apuleius, Cicero, and Seneca the Younger) are well-advised to keep this volume at hand. As the chapter takes up over three quarters of the book (158 out of 216 pp.), reviewing each proposal in detail is impractical. Instead, I will limit myself to a few comments on some dubious cases to stimulate further debate.
In general, observations on the usus scribendi are not always given due weight in assessing a reading. At pp. 31-2, Magnaldi examines Apul. met. 1.4.2. Her argument that et ante should be regarded as a signal-word (et) with the integration (ante) to be incorporated before the first et (et <ante> tamen Athenis proximo [et ante] Poecilen porticum isto gemino obtutu circulatorem aspexi…) runs against the lack of attestations of proximo with accusative (noted in footnote 4) and the fact that Apuleius rarely employs ante as an adverb without a further modifier (paulo ante, multo ante, and the like). Moreover, there are no parallels in his works for a sentence beginning et ante tamen, whereas parallels for et tamen are unsurprisingly widespread. While discussing Cic. Phil. 7.5 at p. 93, Magnaldi asserts that taking nomen as a signal-word and assuming that tanti honoris was meant to be inserted before nomine (quo <tanti honoris> nomine dignus est nemo, nisi qui [tanti honoris nomen] potest sustinere) gives ‘un testo nitido e asciutto’. The repetition nomine … nomen may indeed sound overly emphatic or jarring (good parallels for it are however provided in G. Manuwald, Cicero, Philippics 3-9, Berlin-New York: De Gruyter 2007, pp. 844-5), yet comparanda for modifiers wedged between a connecting relative and the noun it refers to as in quo tanti honoris nomine would have put Magnaldi’s argument on more solid grounds than a generic judgement on style, not to mention that in her reconstruction sustinere remains without an explicit object. Another spurious insertion—sim positus after promptus in Sen. dial. 9.1.10 (p. 135)—is read as suggesting that positus should be placed after the previous sim (sed ut … omnibus deinde mortalibus paratior utiliorque sim <positus>. promptus [sim positus] sequor Zenona…). The single parallel cited in support of this construction (Sen. epist. 109.16 securior et extra metum positus) involves the basic meaning of the verb along with a spatial determination (if metaphorical) and does not support the use of positus with predicatives which Magnaldi’s proposal entails. Magnaldi’s readiness to endorse such readings raises the question of whether detecting traces of a signal-word is sufficient to accept the implied correction / integration at face value, or if these traces are in fact just one element amongst others that editors should factor in when evaluating the transmitted text.
The majority of Magnaldi’s solutions are convincing or plausible enough to deserve mention in an apparatus or textual companion. Not all her arguments, however, are equally compelling. At pp. 165-6, she claims that per omnes figuras per casus handed down in the mss. of Suet. gram. 25.4 harbours a signal-word (per2) plus the term to be added (casus) after the signal-word’s referent (per1). If casus precedes omnes, then figuras may be a corruption of the infinitive needed to complement exponere and explicare. This prompts Magnaldi’s suggestion that figuras is a corruption of figurare, a verb which some later grammarians use in the sense of ‘inflect’ (nam et dicta praeclare per <casus> omnes figurare [per casus] et apologos … exponere et narrationes … explicare consuerant). Magnaldi dismisses Kaster’s idea that per casus may in fact be a gloss displacing the lost infinitive with the counterargument that casus is the expected term in a passage describing chriae; it is not difficult to object that precisely because of this a scribe may have used casus to gloss the less common figuras. Similarly, postulating that an archaism like quoi, never otherwise attested in Apuleius’ philosophical works, lies behind qui at Apul. mund. 25.5 = 345 Oudendorp (pp. 50-1) merely because of the adjacent ei—as if this was the relic of an interlinear correction –oi later turned into the pronoun (i.e. quioi > qui … ei)—strikes me as a somewhat strained explanation of the relatively common intrusion of a short word like ei where it does not belong (a corruption of uel as hypothesised by Goldbacher and printed in Beaujeu’s text is just as likely and would be much easier to explain palaeographically: for uel paululum see e.g. Apul. flor. 15.23 nihil ab hac secta uel paululum deuius).
Even if one may hold different opinions on some of her reconstructions, Magnaldi’s proposals are consistently thought-provoking and will no doubt encourage fresh insights into the passages she focusses on. Furthermore, adding new cases to the list of dittographies generated by this correction method is no small merit of her work. In addition to these specific contributions, the volume presents an excellent introduction to Magnaldi’s approach to textual criticism. Her attention to ancient practices of copying and emending a text is brought to bear on an effective strategy to detect scribal interventions and heal badly corrupted passages. A brief concluding appendix showcases twenty-one recently published textual proposals relying on the identification of instances where lectio falsa and lectio emendata, the latter sometimes accompanied by a signal-word, coexist in the paradosis. That textual critics are both sensitive to this ancient usus emendandi and better equipped to emend the marks it left on texts is a testament to the quality of Magnaldi’s research over the past three decades.
 Her book La forza dei segni. Parole-spia nella tradizione manoscritta dei prosatori latini (Amsterdam: Hakkert 2000) and her editions of Galen’s Περὶ παθῶν καὶ Περὶ ἁμαρτημάτων (Romae: Typis officinae polygraphicae 1999), Cicero’s Philippics (Alessandria: Edizioni dell’Orso 2008), and Apuleius’ Opera philosophica (OCT: Oxford 2020) are particularly notable amongst her numerous contributions, a more inclusive list of which can be found in the bibliography (pp. 208-9).
 Each text is introduced by a prospectus of the mss. and the main critical editions.
 An index locorum at pp. 211-3 directs readers to the pages where each passage is discussed. The bibliography at pp. 191-210 is divided into three sections respectively dedicated to mss., editions / translations / commentaries, and critical contributions.
 The volume is well printed and I did not notice any typos except for the unusual blank space underneath the heading of the sixth chapter (p. 29).