TY - JOUR
AU - Rhatigan,, Emma
AB - Abstract This article explores the editing of early-modern sermons, with a particular focus on the challenge of recovering the sermon in performance. Taking as its starting point the fact that most sermons were not written out until after their delivery in the pulpit, it considers the ways in which sermons resist conventional editorial methods based on the identification of ‘error’ and the reconstruction of a holograph text. It argues for a new approach to editing and a new perspective on error which uses these moments of textual complexity in order to shed light on a sermons evolution from sermon notes and pulpit delivery to written text. It is a Paradox of Early Modern Editing that in the pursuit of an error-free text editors spend considerable time analysing error.1 But error is a slippery thing. An error exists only in relation to the ‘correct’ text an editor is seeking to establish, and yet that very text has been the subject of extended scholarly debate. For some it is a reconstructed holograph, for others a final revised text, for still others it can be a particular ‘version’ of a work.2 One editor's error is another's (legitimate) variant. As Spenser's narrator in The Faerie Queene sagely observes, ‘God helpe the man so wrapt in Errours endlesse traine’.3 Error is also refracted by a text's form and genre and nowhere is this more true than in the case of early modern sermons. Unlike other performance genres such as plays and masques, the vast majority of sermons were transcribed in full only after delivery in the pulpit.4 In other words, in contrast to a genre like poetry where it can make sense to think of composing and writing as synonymous, most sermons existed in a form prior to any paper witness. This challenges assumptions about editing which are often based on the synchronicity of the author's holograph and the composition of the work. If an editor of a sermon is to recover a performance, not a holograph text, then the editing process will need to involve a new approach to paper witnesses and a new perspective on error. A specific performance of an early modern sermon will always, to a greater or lesser extent, be lost in time. However, I want to suggest that traces of this event can be captured if we are willing to re-think the status of moments of textual complexity. Focusing on ‘error’, not through the lens of ‘corruption’, but in terms of a sermon's evolution from performance to text, offers a route, albeit imperfect, back to the sermon in the pulpit. The editor's heroic struggle with error is for the most part invisible to the reader, consigned to a marginal textual apparatus which few will study in detail. But in this article I want to place the margins of early modern texts centre page, peeling back the textual apparatus to consider again how editors identify and use error. Specifically, I will be asking how work editing the performance texts of Donne's sermons can shed new light on these challenges, questioning what error might look like in the complex processes by which a performance becomes a text, or a series of texts. I will argue that the sermon offers a unique perspective from which to re-examine the endeavour of editing early modern works, just as, in turn, the process of editing sheds new light on the complicated relationships between a pulpit performance and its textual afterlives. The Place of Error Attitudes to error need to be understood in the context of theories of editing. Crucial here is W. W. Greg's seminal essay ‘The Rationale of Copy-Text’.5 Greg's main focus is the principles which should guide an editor's interventions to their copy text. He argues that an editor should follow their copy text in relation to accidentals, which he defines as readings which mainly affect a text's ‘formal presentation’, but be prepared to exercise their own judgement in the case of substantive variants, which affect an author's meaning. These substantive variants can be legitimately ‘corrected’ using another witness if that establishes a ‘better’ text.6 By ‘better’, Greg means closest to the author's original. Indeed, Greg's entire argument is predicated on the assumption that an editor's duty is to restore a text to its original state as it was conceived and written by the author. Greg's work was developed by Fredson Bowers and G. Thomas Tanselle who make explicit the argument that the author's intention should be understood as their ‘final intention’, in other words their ‘final wishes about the version of [their] work to be presented to the public’.7 In the context of these editorial principles ‘error’ represents any way in which a text may have become corrupt from what the author originally wrote, or, in Bowers's formulation, intended to publish. In an early modern context such errors might refer to variants produced through scribal deviation from copy, non-authorial editing, or interventions made during the process of printing. Since, however, any changes introduced by the author in subsequent editions should be accepted as their ‘final intention’, an editor needs to remain alert in order to distinguish ‘error’ from authorial revision. This ideal of reconstructing an authorial ‘original’ text was articulated alongside the appropriation of stemmatics as a way of thinking about how to navigate multiple witnesses to a literary text. Stemmatics involves placing the different witnesses in a genealogical relationship to each other in order to establish the most authoritative text. First systematized by Karl Lachmann in the early nineteenth century, when it was applied to classical, biblical, and medieval texts, the method was adapted by William Ringler in the twentieth century to think about early modern texts.8 The key premise behind stemmatics is the concept of shared error. These errors, termed by Paul Maas, ‘indicative errors’, are used to construct relationships of difference and agreement.9 On the most basic level, if two or more witnesses share an error then an editor can hypothesize that they stem from a shared ancestor. By contrast, if a witness contains an error not present in another witness, then it can be assumed that they have either come down different lines of descent or that one witness needs to be placed further down the same line of descent, having introduced a further error. Stemmatics thus offers a means whereby an editor might seek to reconstruct a hypothetical original text. Error in this model paradoxically offers the key to its own erasure; an editor will use error in order to establish the original error-free text. These editorial theories have not been accepted without question. Most fundamental have been the attacks of critics such as Jerome McGann who has iconoclastically questioned the entire ‘stemmatic quest for the lost original’ and the attempt to reconstruct an authorial text.10 Criticizing the distinction between an idealised ‘work’ and a series of corrupt texts, McGann suggests that editors need to focus instead on the ‘dynamic social relations which always exist in literary production’. In other words, for McGann, focusing too narrowly on the author can blind us to the work of all the other people involved in the production of a text; work which cannot always be dismissed as ‘corruption’. Thus, rather than viewing a text's history as a story of ‘textual contaminations’, McGann argues we should analyse it as a reflection of ‘the systematic transformation which all literary works undergo in their production’.11 Consequently, what Greg and Bowers might identify as error, is, for McGann, important evidence of the complex processes of production which themselves constitute part of a text's meaning. In an early modern context, these debates have been refracted through certain distinctive complicating features of the period. As Michael Hunter has emphasized, this was a period when print and manuscript existed in an ongoing complex relation to one another. Thus an editor of an early modern text may well find themselves working with both print and manuscript witnesses, a practice which differs from the editing work of medievalists who are focused entirely on manuscripts, but also from the work of New Bibliographers, such as Greg and Bowers, who were focused on print.12 Recent editing work in the period has also turned its attention to genres such as letters, scientific writings, and parliamentary speeches, which are presenting new challenges.13 Sermons are an important example. It was not until Jeanne Shami's 1996 edition of John Donne's 1622 Gunpowder Plot sermon and Peter McCullough's 2005 edition of Lancelot Andrewes's sermons that editors started to address the issue of how to edit the early modern sermon, while remaining alert to its complex status as a performance text which may be extant in both manuscript and print witnesses.14 George Potter and Evelyn Simpson's ten-volume edition of Donne's sermons is monumental, but their decision to present the latest revised forms of the sermons as preserved in the three posthumous folios (1640, 1649, 1660) meant the issue of performance and the complications of navigating multiple manuscript and print witnesses were never their main focus.15 Foremost among the textual difficulties presented by sermons is the fact that the original did not exist as a written text. In the vast majority of cases sermons were performances delivered from memory, and they were usually only written out after delivery from the pulpit. This adds a new dimension to the heated debates on if and how far an editor should be attempting to recover an original. Put crudely, the editor of a sermon cannot seek to reconstruct an original authorial version because this may not have been written down until some time (possibly years) after the sermon's composition and performance. Theories of ‘final intention’ are equally problematic. In the case of a sermon the text you produce cannot be the author's final version because the sermon was meant for both performance (potentially on multiple occasions) and reading, and it would be altered in each case. What then are an editor's options? McGann's pursuit of a critical edition which does not privilege the authority of any one text or document has taken him to explore the possibilities of hypertext. Presenting every individual artefact in electronic form encourages, he argues, a ‘greater decentralization of design’, an approach which is essential in order to create a ‘social text’.16 However, if an editor is producing a single text in a print edition, albeit one with an apparatus that documents and explains the relationship of this text to other versions of the work, then there is no escaping the need to pick a particular version of the sermon to use as the copy text. As the editors of the Oxford Shakespeare observe when confronting the choice between a text closest to Shakespeare's foul papers or one closest to the theatrical promptbook, ‘a choice has to be made’, notwithstanding an appreciation that this is not the only ‘rational choice’.17 Broadly speaking, the editor of a sermon faces an analogous choice to that confronted by Shakespeare's editors; they can either present a text of the sermon as it was prepared, perhaps by the preacher, as a reading text (the option pursued by Potter and Simpson in their edition of Donne's sermons), or they can prefer the text which is closer to the sermon as it was first delivered in the pulpit. This second option of attempting to recover the sermon in performance is the policy of the Oxford Edition of the Sermons of John Donne, and is the editorial aim I will be exploring in this article. This option frees the editor from the impossible task of seeking to capture the preacher's ‘final intentions’, focusing instead on the occasionality of performance, and reflects recent attention to the sermon as ‘literary art inextricably engaged in the public sphere’.18 However, it is also an editorial decision which leaves the editor with the not inconsiderable challenge of devising a methodology to attempt to recover this performance. Crucially, such a methodology needs to focus not on a purported moment of authorial creation, but instead on a text's history in order to disentangle its transmission. This will involve a very specific perspective on error. Errors in this context are not readings which are erroneous with respect to the author's ‘original’ or ‘intended text’, but errors in relation to other copies, some extant, some lost. Investigation of these errors can provide an editor with evidence of the text's evolution since it is analysis of the differences between the textual witnesses (the ‘errors’ relative to other copies) which enables us to trace the history of the sermon through the various processes of copying and printing. It is then this understanding of the sermon's movement through different textual states which allows us to get closer to lost copies and perhaps even the performance. Composition, Circulation, Publication Donne's sermons are a good example of the editorial complexities presented by the early modern sermon. A comparatively large number survive, but most of these are extant only in the posthumously printed folios LXXX Sermons (1640), Fifty Sermons (1649), and XXVISermons (1661). There are, however, seven sermons which are also extant in one or more printed quartos, publications we know Donne was involved with, and twenty sermons which are included in one or more of ten manuscripts, although there are no holograph texts. In addition, there are sermon notes recorded by members of the congregation or early readers of paper witnesses. I will be focusing in particular on two Lincoln's Inn sermons which are extant in multiple manuscript and print witnesses: one preached on Psalm 38. 9, and one on Colossians i. 24.19 With the exception of the sermon notes, all the texts of Donne's sermons were produced after the first performance and relate to this performance in different ways. Thus before an editor starts work it is necessary to establish what place each of these witnesses represents in the complex processes of composition, circulation in manuscript, and evolution into print. Isaak Walton describes Donne's first steps when composing a sermon: … as he usually preached once a week, if not oftner, so after his Sermon he never gave his eyes rest, till he had chosen out a new Text, and that night cast his Sermon into a forme, and his Text into divisions, and next day betook himself to consult the Fathers, and so commit his meditations to his memory, which was excellent.20 Walton sets out the recognized process of constructing a sermon whereby a Biblical text would be ‘divided’, in other words broken down into steps in an argument, so the preacher could guide the congregation through his chosen verse or verses, explicating the doctrines they contain and applying them to the congregation. These headings could then be elaborated on through the use of further Biblical proof texts or authorities such as the Church Fathers. The resulting ‘text’ would be a series of notes under the original divisions which the preacher could commit to memory, as Walton tells us Donne did, or use in the pulpit as an aide-memoire.21 It is hard to be sure how often Donne then took the step of writing out these notes into continuous prose. Certainly this was not standard practice in the period, and what evidence we have does not suggest that it was something Donne did habitually. In response to a request for two of his sermons, Donne writes ‘I send you a Copy of that Sermon, but it is not my copy, which I thought my L. of South-hampton would have sent me backe … for the other, I will pretermit no time to write it; though in good faith, I have half forgot it’.22 Clearly one of the two sermons as yet exists only in note form. There would have been clear practical reasons for not writing out a sermon in full. Donne states that it would take him eight hours to make a fair copy of one sermon and, as Walton describes, he would usually be delivering at least one sermon a week.23 Thus although Donne doubtless gave fuller attention to particularly important sermons, he would quite simply not have had time routinely to write out his sermons.24 If Donne did write up his notes, then this holograph text could be copied and circulated more widely in manuscript. Such copies would deviate from each other insofar as Donne or a scribe would accidently deviate from their copy, but there is also evidence that on some occasions Donne subsequently revised his holograph copy. In a 1625 letter written to Sir Thomas Roe from Chelsea, where he was taking refuge from the plague, Donne stated that he was using his enforced absence from the city as an opportunity both to write out his notes and to ‘review’ his sermons: I have reviewd as many of my Sermons, as I had kept any notes of; and I have wrtten out, a great many, and hope to do more. I ame allready come to the number of 8o: of which my sonne who, I hope, will take the same profession, or some other in the world of middle understandinge, may hereafter make some use.25 Preachers would routinely revise their initial manuscript copy before releasing it for circulation, adding marginal references, providing translations, and using different scripts to designate quotations. A preacher comfortable in a multilingual milieu and confident that he could recognize quotations would not need such additional textual aids, but other readers might. There would also, of course, be an opportunity for more substantial revision of the text. Mary Morrissey helpfully terms these revised manuscript copies ‘manuscript sermon-books’ in order to distinguish them from less ‘finished’ manuscript copies.26 We cannot be sure of the fate of these sermons which Donne ‘reviewd’ in 1625, but the first folio was, of course, LXXX Sermons and there is certainly a possibility that these sermons were the eighty which Donne ‘wrote out’ in 1625. This would imply that the ‘reviewing’ process involved some of the finishing work required in creating a manuscript book which would be necessary prior to print publication.27 By contrast, we know that the sermons which made their way into Fifty Sermons were less finished, as the printer needed to employ an Oxford graduate, Nicholas Langford, to ‘poynt & Emphasis’ the sermons in preparation for print.28 This suggests that the printer was dealing with manuscript sermons, rather than sermon books. The implication is that while the printed texts of Donne's sermons all seem to stem from notes which he had written out, the state of the copy could vary considerably, ranging from those sermons written out and revised (to a greater or lesser degree) as sermon books, to those sermons which existed as less finished manuscript copies. This has certainly been the finding of editors for the Oxford edition of Donne's sermons working on the printed texts, who have identified significant variation in the state of sermons, even within the same folio.29 The journey by which a Donne sermon evolved from pulpit performance to a text written out in continuous prose could, then, be one which took considerable time, involving a number of interventions by Donne as well as scribes and editors. But this is not to say that the sermon performance is entirely lost to us. Even when revising a sermon, a preacher would rarely recast their initial division. As Morrissey emphasises, ‘When preachers say that they have revised the sermon, they are at pains to say that [the] essential structure has been unaltered … The structure of the sermon and the doctrines it expounded … formed a kind of “core” that changed little between media’.30 Donne is no exception. Editors confronting multiple copies of a Donne sermon are always dealing with the same recognisable work, even though these copies were made at different stages of its textual transmission. And, crucially, it is fair to assume that these copies relate in a fundamental way to what Donne delivered in the pulpit. There is, then, a performance to recover. It just remains for the editor to ascertain how the surviving copies might be used to capture it and which will then act as the copy text. Copy Text and the Evidence of Error The starting point for an editor is the choice of copy text, and logic dictates that this should be the copy closest to the text the edition is seeking to present. But how can an editor confronting multiple manuscript and print copies of a Donne sermon determine which copy is closest to the sermon he delivered in the pulpit? Greg's ‘Rationale of Copy Text’ is little help here. Because Greg was thinking about print witnesses his method focuses on texts which form an ‘ancestral series’, in other words copies which move from a holograph, to a printed text, to successive editions. It is in this scenario that he argues an editor should select an early text and follow it as regards accidentals. But texts circulating in manuscript offer no clear linear evolution. Thus while both the manuscript and the printed texts of Donne's sermon derive from Donne's post-performance holograph copy, it is impossible to say which of these would be closest to that document in terms of accidentals. The manuscript copies are earlier and so could be said to conform more to the conventions of the period in which Donne was writing, but, as Greg acknowledges, while scribes often exhibited great care in the copying of words, this care did not extend to accidentals where they tended to follow ‘their own habits or inclination’.31 An editor seeking to capture what a preacher said, not what he wrote, also needs to question how far accidentals might reflect performance. Some, for example, capitalisation, are obviously not spoken and consequently of far less significance than they might be for an editor seeking to capture a written text. Others, such as punctuation and spelling, could reflect oral delivery, but analysis of such variants has only rarely provided consistent evidence linking a particular text with performance.32 If accidentals offer no clear route back to the sermon's first delivery, then an alternative approach might be to consider the provenance of the witnesses. To give one striking example, the witnesses to Donne's 1622 Gunpowder Plot sermon at Paul's Cross include the manuscript Royal MS 17B. xx in which the text is corrected in Donne's own hand. As Shami shows, this is, most likely, a copy of the sermon prepared by Donne for James I, and for an edition seeking the text closest to the sermon in performance it is an obvious choice for copy text.33 Sadly, however, this sermon is the exception not the rule, and it has not been possible to identify such a clear provenance for any of the other surviving manuscripts which contain Donne's sermons.34 Consequently, if, as is the case with the vast majority of those Donne sermons extant in multiple copies, neither accidentals nor provenance history indicate a text close to performance, then the copy text needs to be determined, contrary to Greg's ‘Rationale’, on the basis of substantive variants. The contingencies of transmission mean there is no guarantee that an earlier witness will offer a more accurate text, thus all copies, print and manuscript, need to be collated. And, because the sermons were copied as separate texts, this needs to be completed on a sermon-by-sermon basis.35 Donne's Lincoln's Inn sermon on Psalms 38. 9 offers an example of the sort of data this collation produces. This sermon is a useful starting point as it is unique among Donne's sermons in being extant only in manuscript witnesses; it is to be found in Harvard University, Houghton Library, MS Eng D 66. 4 (Dobell), Bodleian Library, MS Eng. th. e. 102 (Dowden), National Library of Scotland, MS 5767 (Lothian), and Bodleian Library, MS Eng. th. c. 71 (Merton). It is likely that all these manuscripts derive from Donne's manuscript copy of the sermon, written up after delivery. Collation of these witnesses reveals over 800 places where the wording in one or more manuscript varies from another. All these variants seem to be the result of scribal deviation from copy, rather than the result of revision. Having collated the texts, it becomes possible to compare readings across the manuscripts and identify places where a witness is most likely in error. Working with what Maas terms ‘indicative’ errors, in other words omissions or mistranscriptions which result in a distortion of sense, the texts in Lothian and Merton manuscripts emerge as least reliable and consequently unlikely choices for copy text.36 This leaves a choice of Dobell or Dowden, but since both these manuscripts present a similar degree of error, the collation on its own provides no clear evidence on which to base a decision. Fig. 1 Open in new tabDownload slide Stemma for Sermon on Psalm 38.9 Fig. 1 Open in new tabDownload slide Stemma for Sermon on Psalm 38.9 This is a situation in which it makes sense to draw on the method of stem- matics since, although the aim here is not to reconstruct an original text, the methodology can help in understanding how the different copies relate to each other. Applying the methodology of stemmatics to the collation of the Psalms 38. 9 sermon, the first question is whether there are any instances of three manuscripts agreeing in error, and there are, in fact, six errors shared by Dowden, Lothian, and Merton that provide convincing evidence for (one or more) shared ancestors for these three witnesses.37 There are then forty- six errors shared by Dowden and Lothian, suggesting they descended via a further ancestor.38 It is worth noting at this point that this set of relationships is not one which is obvious from study of the manuscripts alone. For example, the Merton manuscript seems likely to be an early witness; the sermons in the manuscript date from December 1617 to Easter 1622 and one of the stocks of paper is Heawood 840, which is Dutch paper of 1619. This might suggest that copying was not that much later than 1622. It turns out, however, that even if it was copied earlier than the other witnesses it is, in fact, the most erroneous. Equally, there is on the face of it little to connect the Lothian and Dowden manuscripts, and yet it turns out their texts for this sermon clearly stem from a shared ancestor. Stemmatic analysis thus forces us to revise any assumptions we have made about the manuscripts’ texts on the basis of what we might know or be able to guess about their provenance or the date of transcription. On the basis of this stemma, Dobell emerges with the best claim for copy-text. It has the least number of indicative errors and offers the most direct line of transmission back to the first text of the sermon, presumably Donne's manuscript copy. Stemma Dilemmas Having established the copy text, the editor then needs to decide how far they will intervene in the text which it supplies. Clearly it makes sense to correct all the indicative errors in the copy text; these are readings which do not make sense and cannot possibly be what Donne spoke in the pulpit. However, since the stemma suggests that the shared ancestor of Dowden, Lothian, and Merton would have been a much better text than Dobell (it would have contained just the six errors connecting these witnesses), there is also a case for correcting Dobell where it is possible to reconstruct this ancestor through agreement between the three other witnesses. This would, however, involve extending the definition of error to encompass not just indicative error, but all readings in which the stemma shows Dobell to be in error in relation to the hypothetical original holograph text. This is a more difficult decision and, not surprisingly, it is at this point that some editors have been reluctant to allow their stemma to govern their editorial decisions. The Variorum edition of Donne's poetry is a case in point. This edition is committed to collation and stemmatic analysis of all witnesses, but in the case of the majority of the poems the editors recognize that ‘the possibility of missing copies and the intractability of the surviving evidence … make it impossible to construct a complete genealogy’. Consequently, while the editors draw on stemmatic analysis to select a copy text, they then only correct ‘obvious errors’, as opposed to errors which might be suggested on the basis of the stemmata.39 This concern over the ‘intractability’ of the evidence is important, because it constitutes a clear example of the difficulties scholars have experienced when seeking to apply the methodology of stemmatics. One of the most fundamental of these difficulties is the fact that the search for shared errors rarely produces unambiguous results. To return to the stemma for the Psalms 38. 9 sermon, the six errors shared by Dowden, Lothian, and Merton, constitute strong evidence for a shared hypothetical ancestor, but what this stemma cannot account for are four errors shared by Merton and Dobell; ‘panam’ for ‘poenam’, ‘sole’ for ‘solis’, ‘convert’ for ‘correct’ and the omission of ‘is’.40 Since Merton cannot be related to both Dobell and to Dowden and Lothian an editor needs to decide which set of errors are most convincing. The remaining errors must then be assumed to be the result of ‘independent variation’. As Ringler explains when faced with comparable conflicting evidence in the witnesses to the Arcadia poems, ‘Since most of these agreements in error indicate contradictory and therefore impossible patterns of relationship, they cannot derive from errors in the originals that were being copied but must result from mere chance. Human minds and hands are prone to similar lapses, and two or more scribes copying the same original will from time to time make identical mistakes’.41 In the case of the errors shared between Merton and Dobell the mistakes in the Latin are certainly errors which could be made by non-Latinate scribes working independently, as is the small omission of ‘is’. The misreading of ‘convert’ for ‘correct’ is also understandable in terms of the similarity of the letter forms, although an editor still has to balance the likelihood of two scribes independently misreading this word. Overall these errors seem to present less compelling evidence than those connecting Merton with Dowden and Lothian, but they also demonstrate very clearly how the construction of a stemma is always the result of individual judgement.42 For an editor working with multiple manuscript (and potentially also print) witnesses stemmatics is clearly a useful method to explain the relationship of witnesses to each other. This enables a choice of copy text and also facilitates the correction of indicative errors; an omission or nonsensical reading can be corrected using the other witnesses. The stemma also provides the editor with a framework within which to analyse other substantive variants between different witnesses. This may not necessarily be with a view to ‘correcting’ the copy text since, as discussed above, the stemma may not provide sufficiently definitive evidence to allow an editor to justify such intervention. Nor is it ever going to be the case that the stemma will take us back to a lost ‘original’, because the ‘original’ we are dealing with is a text derived from a performance. But what we do now have is a text and a set of variant readings from which we can start to chart the sermon's evolution. As we begin to think in more detail about performance, however, stemmatics, a method predicated on recovering a holograph text, will become increasingly inadequate to capture some of the complexities we will encounter. In particular, we will need to re-orientate ourselves in relation to that fulcrum of the stemma, error. Margins of Error For a vivid illustration of the ways in which a performance text demands a new perspective on error, we need only move our eyes beyond the main body of the text to consider the space of the margins. Margins present a particular challenge for the editor of a sermon.43 After all, if the editor's aim is to get back to the sermon as it was delivered, then we may well ask what place the unspoken marginalia have in this text at all. Marginalia function as a continual reminder that the ‘original’ text of a sermon may never have existed as a document written in continuous prose at the time of composition. Moreover, when comparing multiple witnesses to the same sermon it soon becomes evident that the margins are one of the least stable parts of a text, peculiarly resistant to textual analysis. Marginal references are frequently missing, routinely erroneous, and quite often to be found wandering errantly around the page, either higher or lower in the margin or within the text as opposed to in the margin. Take for example, the Biblical references in Donne's Lincoln's Inn sermon on Colossians i. 24, a sermon extant in the folio Fifty Sermons and four manuscript witnesses. Of twenty- six references, ten are in error and nine are omitted in one or more witness. This is probably not surprising; after all nothing is easier than mistranscribing or mis-placing a reference, and arguably numbers are particularly easy to confuse, which suggests why the biblical references are particularly susceptible to error. But where the margins differ from the main body of text is that this propensity to err is probably as true for the author as it is for a scribe or a compositor. For example, in one case all the witnesses to the Colossians i. 24 sermon are in error, citing Romans ii. 12 instead of Romans ii. 13. This sort of mistake could just as easily be an authorial error, which has then been faithfully copied, as it could be a scribal error, introduced by a shared ancestor to all the witnesses. Indeed, editors working on printed witnesses have in some cases argued that erroneous marginalia are more likely to be authorial, as this is the sort of error which editing work before printing would seek to correct.44 This poses a fundamental challenge to stemmatic analysis as it effectively reverses the correlation between error and corruption introduced through the processes of copying or printing. In the space of the margin error can be as likely to signify an authorial ‘original’ as it can a corrupt copy. Recovering Performance As the margins demonstrate, when it comes to recovering the performance text of a sermon the methodology of stemmatics can only take the editor so far. Capturing performance frequently demands a different perspective on moments of textual complexity, one that focuses less on corruption and more on the sermon's evolution. In order to disentangle the sermon's transmission an editor needs to pursue a form of ‘textual archaeology’ whereby variants are analysed not only to deduce whether they constitute indicative error, but also to ascertain whether they may be traces of the sermon's movement through various forms (sermon notes, oral delivery, manuscript copy, and print). Donne's sermon on Colossians i. 24 offers a useful case study of how this analysis might work in practise. Extant in four manuscript witnesses, the folio Fifty Sermons, and a set of sermon notes, this sermon is one of the more textually complex of Donne's sermons. Analysing the results of the collation, what is immediately striking is the extent to which the folio text is distinct from the four manuscripts, for it shares no indicative errors with any of the manuscripts and also has eighty- four independent readings. From the point of view of stemmatics, these readings mean the folio can be placed on an independent line of descent. This is useful insofar as it goes; however, with further analysis, these variants can also reveal much more about both the textual history of this specific sermon and the nature of a ‘performance text’ of a sermon more generally. Of the folio's independent variants, at least forty constitute additional text which seems likely to be the result of authorial revision, rather than scribal deviation from copy. For example, a number of the additions function to provide further explanation or to clarify the sense. Thus when Donne references Gennadius, telling his congregation, ‘for wee knowe from Gennadius, that the Bushops of Girra did vse to con St Cyrills sermons’, the folio adds in a clause telling us that Gennadius is ‘an Ecclesiasticall author’.45 These sorts of clarifications would seem to be precisely the sort of activity Donne (or conceivably someone editing the sermon) might undertake.46 In other cases the insertions seem to have been made for rhetorical reasons. For example, in a celebration of joy in suffering, the manuscripts read ‘it is a fundamentall ioye, a radicall ioye, a viscerall ioye, that arises out of the wombe and bowells of the tribulac[i]on it self’ which the folio extends to ‘it is a fundamental! joy, a radicall joy, a viscerall, a gremiall joy, that arises out of the bosome and wombe and bowels of the tribulation it selfe’.47 Here the additional text functions as amplificatio. Thus ‘gremiall’ adds to the homoioteleuton of ‘fundamentall … radicall … viscerall’ and ‘bosome and’ extends the polysyndeton of ‘and wombe and bowels’. This evidence of revision has important implications for an editor thinking about the sermon's evolution. First of all, it is worth stating that it seems highly likely that the variants are the result of authorial intervention in the text. The use of the word ‘gremiall’ is particularly telling. The Oxford English Dictionary identifies this word as a Donne coinage, meaning ‘of or pertaining to the bosom or lap’, and cites a 1625 court sermon as the earliest recorded usage. There, Donne tells his congregation, ‘there is the land of Gold, centricall “Gold, viscerall Gold”, gremiall Gold’.48 In fact, a search on Early English Books Online produces two other earlier examples, dating back to 1591.49 Nonetheless, the word is clearly sufficiently rare that it is most likely to have been inserted by Donne, rather than a later scribe, compositor, or editor. As we do not know when the revisions to the Colossians sermon were made it is impossible to say whether the insertion predates Donne's other use of the word in his court sermon. However, the 1625 date is intriguing as it coincides with the period when Donne stated he was reviewing his sermons and the usage in both sermons is strikingly similar; on each occasion the adjective is deployed in a display of amplificatio, playing on the homoioteleuton of ‘viscerall … gremiall’. Could Donne have deployed this unusual word in his court sermon and then in the same period inserted it into the Lincoln's Inn sermon he was revising? It is a tempting hypothesis, which offers us a tantalizing glimpse of the sort of work Donne might have been involved in as he reviewed the copies of those sermons he had written out after delivery. Thus far, then, it seems that these independent variants in the folio are pointing towards Donne's revisions to the version of the sermon which he wrote out after its first delivery in the pulpit at Lincoln's Inn. Significantly, though, this revision seems not to have been an attempt to make the sermon less performative, because this sort of rhetorical amplification is precisely the kind of stylistic trait which would be particularly effective in oral delivery. Indeed, the cumulative effect of the inserted text means that, paradoxically, the printed folio text, a text which would have engaged readers, not a congregation, looks more performative than the manuscript text, which is presumably the text closest to delivery. Although counterintuitive, this is not as surprising as it might seem. Recent research has repeatedly challenged any easy distinction between a ‘performance text’ and a ‘reading text’ of early modern sermons. Many high-profile sermons were printed with a view to creating a record of the event.50 In other words, although it could never do this exactly, the printed text was looking to capture the performance, rather than transform the performance into a reading text. Equally, Arnold Hunt has stressed that not only were printed sermons intended for reading aloud, for example in a household setting, as well as for reading privately, but preachers revising their texts for print might deliberately seek to heighten the style of particular passages in order to compensate for the loss of the immediacy of performance in the pulpit.51 We could have, in Donne's sermon on Colossians, a perfect example of just this sort of editing, a reminder that variants which appear to point to a performative text may not be signalling the text in performance. There may, however, be another motivation behind Donne's performative revisions to this sermon. In his 1625 letter to Sir Thomas Roe, Donne states that he is ‘reviewing’ his sermons not in order to produce a reading text but so his son ‘or some other in the world of middle under- standinge’ might ‘make some use’ of them. Presumably Donne envisaged his son or another priest using his sermons as the basis for their own or perhaps even re-preaching them.52 Re-preaching sermons was standard practice in the period and there is evidence to suggest that this might have played a part in the revisions to the Colossians sermon. The Colossians sermon is one of two Donne sermons included in the commonplace book of John Burley (Trinity College Dublin, MS 419).53 The notes for the other sermon, one on Psalms 6. 4-5, are dated 16 October 1625, precisely the period when Donne was in Chelsea, and it seems likely this sermon was delivered either in Chelsea college (which Burley entered in April 1624) or in Chelsea parish church.54 The fact that these notes are followed by notes from the sermon on Colossians, which must have originally been preached at least three years earlier (before Donne left Lincoln's Inn in 1622), is puzzling. The physical evidence of the manuscript suggests that the notes for both sermons were written at the same time; there is no space between the two sets of notes and no change in the hand or ink. Burley could, of course, have been copying Donne's or someone else's notes or taking notes from a manuscript copy of the sermon (again, either belonging to Donne or someone else).55 However, there is also a distinct possibility that Donne re-preached the Colossians sermon in Chelsea in 1625. Indeed, the fact that the notes on the sermon on Psalm 6. 4–5 were taken from performance perhaps makes it more likely that the Colossians sermon was also being noted from delivery. Together the two sermons could have formed a mini-series preached by Donne during his stay in Chelsea. Both sermons include a focus on suffering which make them particularly suited to the circumstances of the plague and it would make perfect sense if, when ‘reviewing’ his sermons for appropriate material, Donne selected the Colossians sermon and revised it for re-delivery. One substantive difference between the notes and the text of the sermon in the other witnesses is particularly suggestive of a 1625 performance. The sermon as extant in the manuscripts and Fifty Sermons opens with a discussion of the controversial Roman Catholic doctrine of supererogation and includes a couple of allusions to sermons Donne has either preached, or intends to preach, at Lincoln's Inn. However, there is no reference to any of this opening material in Burley's notes. Instead, the version of the sermon he encountered seems to have included an exordium which connected the Biblical text with the suffering produced by the plague. His first note reads ‘death is entred into or windows. that is into or eyes we have seene if not tasted mortalitye’, alluding presumably to Jeremiah 9. 21 ‘For death is come up into our windows’.56 This arresting image does not feature anywhere in the sermon on Colossians 1. 24, but it does appear in a St Dunstan's sermon which Donne preached on 15 January 1626 ‘after our Dispersion by the Sickness’. There, elaborating on the desolation wrought by the plague, Donne does not only once again allude to Jeremiah, but he also extends the metaphor to the eyes in the manner suggested in Burley's notes, stating ‘The windows of this House are but our eyes’.57 Faced with a Chelsea congregation confronting the desolation of the plague, Donne could have cut his original exordium, suited to the particular location of Lincoln's Inn, and replaced it with an elaboration on Jeremiah 6. 12, a text which, as the St Dunstan sermon shows, he clearly viewed as speaking closely to the experience of London during the plague.58 Analysis of the variants in the folio text and a comparison with the 1625 sermon notes, the 1625 court sermon, and the 1626 St Dunstan's sermon allow us to posit a tentative timeline for Donne's Colossians sermon. First performed at Lincoln's Inn some time between 1616 and 1622, it was then written out by Donne and circulated in manuscript. Donne returned to the sermon in 1625, re-preaching it as part of a new series of sermons on suffering and revising his original manuscript copy. This revised text was then printed posthumously as part of Fifty Sermons in 1649. This textual detective work provides vital evidence for an editor looking to recover a performance except that, confusingly, in this case we have not one, but two performances. The editor now needs to establish which performance they want to recover. The theatre offers a useful analogy. For example, the Oxford Shakespeare's editorial policy is to present the text of a play which is closest to the theatre's prompt-book, as opposed to the playwright's foul papers. But these performance texts can include traces of multiple performances, such as act breaks which were retrospectively imposed on plays originally intended for continuous performance. Because they are committed to reflecting not just the text in performance, but the text in its first performance, the editors remove such breaks.59 Similarly, an editor seeking the first performance of Donne's Colossians sermon would need to reject the revisions to the folio text even though this material may well relate to performance. Finally, it is worth stressing that the reason this task of unpicking various performance (and, indeed, reading) texts is so difficult is because just as each dramatic performance, or ‘production’, of a play is still essentially the same play, so too editors confronting multiple copies of a Donne sermon are always dealing with the same recognisable work, all of which relate in an essential way to the sermon's initial delivery in the pulpit. For an important illustration of this, and what is arguably the final key piece of evidence for an editor seeking to capture a performance, we need to return to the margins. Performing the Margin As I have already discussed, the margins of early modern sermons can be one of the most textually unstable parts of the text. But an editor needs to remain open to the fact that not all marginalia are the same either in origin or function. The marginalia which delineate the preacher's careful structuring of the sermon are a crucial case in point. To stick with the example of the sermon on Colossians i. 24, this sermon includes ten structural markers in the margins. The first marker, ‘Divisio’, signals the moment when Donne divides his text and then each new branch is indicated with a marginal heading. Compared to the references to biblical texts or other authorities these entries are strikingly complete, consistent, and error free across the different witnesses.60 And, in contrast to the references, we can be relatively confident that this marginalium is authorial. Indeed, these markers are probably not only authorial, but actually take us back to the very early stages of the sermon's composition when, as Walton describes, Donne cast his Text ‘into divisions’. These marginal entries are what, if anything, Donne had in the pulpit and they thus have the potential to take us back to the sermon in delivery in a way the main body of the sermon cannot. If we are seeking to establish a text of what Donne spoke in the pulpit, this division is in some ways the closest we are going to get to it. The irony, of course, is that in one sense these crucial indicators of performance were never actually spoken; marginalia sit outside the main body of the spoken text. However, in another sense they are spoken since Donne weaves the headings into his prose as he guides his congregation through the sermon. For example, when he tells his auditory ‘To beginne then with the first branch of the first part’ he is essentially voicing his division.61 Moreover, the congregation, trained in what Hunt calls ‘the art of hearing’, would have been listening out for the division and then identifying and sometimes recording the structural markers. In the case of our Colossians i. 24 sermon we have evidence of this process in the Burley manuscript notes. As discussed earlier, there are some suggestive differences between these notes and the sermon as it is extant in the other witnesses, some of which might be the result of re-preaching or revision. However, the one aspect which remains constant is this recording of the structure. Burley carefully recreates Donne's division in his notes, including nearly all the headings which we then see in the margins of the complete sermon.62 These notes thus constitute vivid testimony of how marginalia can take us back to the sermon in performance. Margins have a peculiar habit of becoming central in the process of editing an early modern sermon. By foregrounding a sermon's division, they offer us a tantalizing glimpse of Donne in the pulpit, but their textual complexity also draws attention to the ways in which the genre of a sermon resists conventional editorial methods. Establishing the text closest to the sermon as delivered in the pulpit is rarely a straightforward process of identifying markers of ‘performance’, not least because, as the textual history of Donne's sermon on Colossians i. 24 illustrates, the performance text is not necessarily that which appears most performative. Nor can an editor rely on stemmatics. Such a methodology cannot capture the textual serendipity generated by a sermon's evolution through various states. Instead, the very particular textual history of a sermon, in which the performance precedes the text, demands a new approach to editing and a new perspective on that backbone of editing, ‘error’. The only way to get close to the sermon in performance is to use the surviving copies to unravel the sermon's evolution. This involves turning to those cruxes which editors have usually designated as ‘errors’, moments which unsettle a single authorial text, and re-envisaging them as clues to the sermon's history. In other words, it is by understanding ‘error’ not as erroneous deviation, but as a road map of how the text has travelled, that an editor can attempt to recover the sermon event. Footnotes 1 This is particularly true for editors drawing on Lachmannian stemmatics to mediate between multiple copies of the same text. As I discuss below, this methodology is based on the identification of errors shared between different witnesses. 2 See the discussion of W. W. Greg's and Fredson Bowers's theories of editing below. The term ‘version’ is used by Jerome J. McGann in A Critique of Modern Textual Criticism (Chicago and London: University of Chicago Press, 1983), but is derived from Hans Zeller, ‘A New Approach to the Critical Constitution of Literary Texts’, Studies in Bibliography, 28 (1975), 231–64. 3 Edmund Spenser, The Faerie Queene, ed. by A. C. Hamilton, Hiroshi Yamashita, and Toshiyuki Suzuki (Harlow: Pearson Longman, 2001), 1.1.18.9. On the troubling persistence of error in scholarly editing, see W. Speed Hill, ‘The Calculus of Error, or Confessions of a General Editor’, Modern Philology, 75 (1978), 247–60, and Cathy Shrank, ‘On Error’, in Claire Loffman and Harriet Phillips, eds, A Handbook of Editing Early Modern Texts (Abingdon: Routledge, 2018), pp. 209–17. 4 A few preachers wrote their sermons out in full and brought these scripts into the pulpit. Lancelot Andrewes is a famous example. This was not, however, standard practice. See Mary Morrissey, Politics and the Paul's Cross Sermons, 1558–1642 (Oxford: Oxford University Press, 2011), pp. 36–7, and Peter McCullough, ed., Lancelot Andrewes: Selected Sermons & Lectures (Oxford: Oxford University Press, 2005), pp. li–lii. 5 W. W. Greg, ‘The Rationale of Copy-Text’, Studies in Bibliography, 3 (1950–51), 19–36. 6 Greg's distinction between accidentals and substantives is not always easy to sustain. Accidentals are usually taken to include punctuation, capitalization, and spelling, but arguably these also affect meaning. 7 Fredson Bowers, ‘Some Principles for Scholarly Editions of Nineteenth-Century American Authors’, Studies in Bibliography, 17 (1964), 223–28 (p. 227), and G. Thomas Tanselle, ‘The Editorial Problem of Final Authorial Intention’, Studies in Bibliography, 29 (1976), 167–211 (p. 168). See also Fredson Bowers, Bibliography and Textual Criticism (Oxford: Clarendon Press, 1964), Textual and Literary Criticism (Cambridge: Cambridge University Press, 1959), and Essays in Bibliography, Text, and Editing (Charlottesville: Bibliographical Society of the University of Virginia, 1975). 8 See Ringler's edition of Sidney's poetry, which is based on stemmatic analysis of the witnesses; William A. Ringler, Jr., ed., The Poems of Sir Philip Sidney (Oxford: Clarendon Press, 1962). 9 See Paul Maas, Textual Criticism, trans. by Barbara Flower (Oxford: Clarendon Press, 1958). Maas offers a brief, but extremely useful, overview of the principles behind stemmatics. 10 McGann, Critique, p. 66. For a comparable critique, see D. F. McKenzie, Bibliography and the Sociology of Texts (Cambridge: Cambridge University Press, 1999). 11 McGann, Critique, pp. 8i, 62. 12 Michael Hunter, Editing Early Modern Texts: An Introduction to Principles and Practice (Basingstoke: Palgrave Macmillan, 2009), pp. 7–8. 13 See Hunter, Editing Early Modern Texts, pp. 5–6. 14 Jeanne Shami, ed., John Donne's 1622 Gunpowder Plot Sermon, A Parallel-Text Edition (Pittsburgh: Duquesne University Press, 1996) and McCullough, ed., Lancelot Andrewes. 15 George Potter and Evelyn Simpson, eds, The Sermons of John Donne, 10 vols (Berkeley, CA: University of California Press, 1953–62). 16 Jerome McGann, Radiant Textualities: Literature after the World Wide Web (New York: Palgrave Macmillan US, 2001), pp. 24–25, 70–71. 17 Stanley Wells and Gary Taylor, with John Jowett and William Montgomery, William Shakespeare: A Textual Companion (Oxford: Clarendon Press, 1987), p. 15. 18 Lori Anne Ferrell and Peter McCullough, ‘Introduction: Revising the study of the English sermon’, in The English Sermon Revised: Religion, Literature and History, 1600–1750 (Manchester: Manchester University Press, 2000), p. 2. 19 These sermons are number 6 in vol. 2 and number i6 in vol. 3 of Potter and Simpson, eds, Sermons. 20 Izaak Walton, The Life of John Donne (London: Richard Marriot, 1658), pp. 86–87. 21 For a useful overview of this process, see Morrissey, Politics, pp. 50–61, and ‘Sermon-Notes and Seventeenth-Century Manuscript Communities’, Huntington Library Quarterly, 80 (2017), 293–307 (pp. 295–98). See also Arnold Hunt, The Art of Hearing: English Preachers and Their Audiences, 1590– 1640 (Cambridge: Cambridge University Press, 2010), pp. 131–35. Hunt provides examples of both preachers who would write out their sermons in full and then memorize them and preachers who relied purely on the headings in their notes. 22 John Donne, Letters to Severall Persons of Honour (London: Richard Marriot, 1651), p. 200. This letter is addressed to Sir Thomas Lucy, but I. A. Shapiro argues persuasively that the recipient must have been Henry Goodyer. See ‘The Text of Donne's Letters to Severall Persons’, The Review of English Studies, 7 (1931), 291–301. 23 Donne, Letters, p. 154. 24 For example, when in 1627 Laud takes offence at his court sermon and requests a copy, Donne writes in a letter to Robert Ker that he has ‘exscribed’ it, a verb which, as the Oxford English Dictionary states, could mean either ‘to write out’ or ‘to copy’ (Letters, p. 308). This suggests he had a full draft, if not a fair copy, of that sermon intended for a prestigious pulpit. On this episode, see R. C. Bald, John Donne: A Life, reprinted with corrections (Oxford: Clarendon Press, 1986), pp. 491–94, and David Colclough, ed., The Oxford Edition of the Sermons of John Donne. Vol. 3, Sermons Preached at the Court of Charles I (Oxford: Oxford University Press, 2013), pp. 319–23. 25 The letter is cited in Bald, John Donne, pp. 478–80. Donne appears to have undertaken similar ‘reviewing’ work during his stay in Aldborough Hatch in 1630. See Bald, John Donne, pp. 522–23. 26 Morrissey, Politics, pp. 57–8, and ‘Sermon-notes’, pp. 299–301. On the ways in which preachers would revise their sermons for print, see also Hunt, Art of Hearing, pp. 147–63. Sebastiaan Verweij's work on Francis Russell's notes on Donne's sermons provides important new evidence of the possible gaps between the sermons as delivered and then as printed. See ‘Sermon Notes from John Donne in the Manuscripts of Francis Russell, Fourth Earl of Bedford’, English Literary Renaissance, 46 (2016), 278– 3…3. 27 This fits with Henry King's claim, in a letter to Izaak Walton, that ‘three days before his death [Donne] delivered into my hands those excellent Sermons of his now made publick: professing … that he had prepared them for the Press’, cited in Robert Kreuger, ‘The Publication of John Donne's Sermons’, Review of English Studies, n.s. 15 (1964), 151–60 (p. 160). King also claims Donne gave him his ‘Sermon-Notes’. 28 Oxford University Archives, Chancellor's Court Papers, 1640/121. 29 The folio XXVI Sermons presents the most divergence. For an example of a sermon which appears to have been set from a manuscript copy, rather than from a sermon book, see Donne's 1615 sermon on Isaiah 52. 3, which, as Peter McCullough shows, exhibits many signs of being an unrevised text (Peter McCullough, ed., The Oxford Edition of the Sermons of John Donne. Vol. I, Sermons Preached at the Jacobean Courts, 1615–1619 (Oxford: Oxford University Press, 2015), p. 145). 30 Morrissey, Politics, p. 49. 31 Greg, ‘Rationale’, pp. 21–2. Ted-Larry Pebworth addresses the limitations of Greg's ‘Rationale’ when editing late sixteenth- and early seventeenth-century manuscript poetry, including that by Donne, in ‘Manuscript Transmission and the Selection of Copy-Text in Renaissance Coterie Poetry’, Text, 7 (1994), 243–61. 32 Punctuation and spelling have thus far had a bearing on the choice of copy text for just one sermon in the edition, see McCullough, ed., Sermons, p. 182–85, and ‘Sermon 4’, ll. 126 and 257 and comments. 33 Shami, ed., Gunpowder Plot Sermon, pp. 11–14. 34 Henry Woudhuysen discusses the challenge of marrying work on manuscript circulation and the process of editing in ‘Sidney's Manuscripts (again)’, Sidney Journal, 30 (2012), 117–25. 35 Woudhuysen draws attention to Ringler's edition of Sidney's poetry in which, despite rigorous analysis of the manuscript witnesses, he concluded that the later printed texts of the Old Arcadia poems, the Certain Sonnets and most of Astrophil and Stella offered the best copy text (‘Sidney's Manuscripts (again)’, p. 120). 36 I have identified fifty-four errors in Dobell, seventy-six in Dowden, one hundred and twenty-two in Lothian, and one hundred and fifty-eight in Merton. 37 These errors are ‘perseuers’ instead of ‘perseuere’, ‘too’ instead of ‘soe’, ‘profitts’ instead of ‘perfitts’, ‘profunditur’ instead of ‘perfunditor’, and two omissions of text. They suggest the scribe of the shared ancestor to these witnesses had particular difficulty distinguishing the contraction for ‘per’ from that for ‘pro’. 38 Both Dowden and Lothian also misplace an extended passage of the sermon, providing further evidence that they descend from a shared ancestor. 39 Gary Stringer, et al. eds, The Variorum Edition of the Poetry of John Donne, Volume 2: The Elegies (Bloomington and Indianapolis: Indiana University Press, 2000), pp. lii, liv. In the case of some poems it has been possible to construct a complete stemma, and here the editors have pursued what they term a more ‘aggressive’ editorial principle. As they explain in the textual note to ‘His parting from her’, ‘That the evidence has lent itself to postulation of a complete stemma, however, has enabled a principled attempt to recover lost authorial readings; and we have correspondingly suspended our usual conservative policy of emending only where necessary to prevent misreading in favor of a more thorough going effort to reconstruct the [lost original holograph]’ (p. 349, p. xciii). For an overview of the editorial principles of the edition, see Richard Todd, ‘Editing Donne's Poetry: The Donne Variorum and Beyond’ in The Oxford Handbook of John Donne, ed. by Jeanne Shami, Dennis Flynn, and M. Thomas Hester (Oxford: Oxford University Press, 2011), pp. 56–64. 40 There are no errors shared by Dobell and either Dowden or Lothian. 41 Ringler, ed., Poems, p. 367, fn. 1. 42 An alternative solution would be to imagine a scenario whereby the scribe of Merton, or an ancestor of Merton, was drawing on both a descendent of the hypothetical ancestor of Dowden and Lothian and an ancestor shared with Dobell. This situation is termed ‘contamination’. See Maas, Textual Criticism, pp. 7–8. For further specific discussion of the potential problems when applying the Lachmann method, see Ralph Hanna, ‘The Application of Thought to Textual Criticism in All Modes– with Apologies to A. E. Housman’, Studies in Bibliography, 53 (2000), 163–72, and Paul Oskar Kristeller, ‘The Lachman Method: Merits and Limitations’, Text, i (1984), 11–20. 43 Other genres do, of course, contain marginalia. Ringler discusses the treatment of marginalia in the Arcadia in Poems, pp. 367–68. The nature of marginalia and the textual challenges it poses are, however, often genre dependent. 44 See McCullough, ed., Sermons, p. 145. 45 Cambridge University Library, MS Add. 8469, fol. 5r; John Donne, Fifty Sermons (London: James Flesher, 1649), p. 131. 46 McCullough finds similar evidence of clarifying work in the folio text of Donne's sermon to Queen Anne in 1615 (McCullough, ed., Sermons, Sermon 4, ll. 106, 257–58, 319, and comments), as does Shami in the 1622 Gunpowder Plot Sermon, where she argues the clarifications often involve careful changes to the sermon's political commentary (Gunpowder Plot Sermon, pp. 24–35). 47 Cambridge University Library, MS Add. 8469, fol. 9r; Donne, Fift'y Sermons, p. 133. 48 Potter and Simpson, eds, Sermons, vi, 231. 49 Thomas Ridley, A View of the Civile and Ecclestaical Law (London: Company of Stationers, 1607), p. 179, and Adrien Saravia, Of the Diverse Degrees of the Ministers of the Gospell (London: John Perin, 1591), sig. A1v. 50 See Morrissey, Politics, pp. 64–65, and ‘Sermon-Notes’, pp. 303–4. Similarly, Tiffany Stern argues that plays printed from notes might be seeking to draw attention to the oral nature of the text in order to celebrate its ‘performance heritage’. See ‘Sermons, Plays and Note-Takers: Hamlet Q1 as a ‘Noted’ Text’, Shakespeare Survey, 66 (2013), 1–23 (p. 21). 51 Hunt, Art of Hearing, pp. 170–3, 159–63. See also Morrissey, Politics, p. 58. 52 See Bald, John Donne, pp. 479–80. On the practice of reading printed sermons in the pulpit, see Hunt, Art of Hearing, pp. 179–82. There is some evidence that the Donne sermons in the Dowden manuscript were annotated with a view to re-delivery. See Sebastiaan Verweij and Peter McCullough, Textual Companion, Oxford Edition of the Sermons of John Donne (forthcoming). 53 For a discussion of these notes, see P. G. Stanwood, ‘John Donne's Sermon Notes’, Review of English Studies, n. s. 29 (1978), 313–20. For a comparable example of the important evidence that can be deduced from sermon notes, see Verweij, ‘Sermon Notes’. Stern explores the phenomenon of note taking in relation to sermons, plays, and parliamentary speeches in ‘Sermons, Plays and Note-Takers’. 54 There is evidence that this Psalms 6. 4–5 sermon had been composed and delivered on a previous occasion and was being re-preached at Chelsea. The sermon seems to have been constructed from two sermons taken from a series on Psalm 6 (numbers 50–55 in LXXX Sermons). It seems likely that a 1628 court sermon on Psalms 6. 6–7 (Colclough, ed., Sermons, Sermon 7) also originally derived from this series. See I. A. Shapiro, ‘Donne's Sermon Dates’, Review of English Studies, n.s. 31 (1980), 54–6, P. G. Stanwood, ‘Donne's Earliest Sermons and the Penitential Tradition’, in John Donne's Religious Imagination, ed. by Raymond-Jean Frontain and Frances M. Malpezzi (Conway, AR: UCA Press, 1995), pp. 366–79 (p. 369), and Philip Michael George, ‘The Sacramental Art of John Donne's Sermons on the Penitential Psalms’ (Ph. D. diss., University of British Columbia, 1996), pp. 96–97. 55 I. A. Shapiro argues some of the notes show evidence of mishearing and are therefore most likely to have been taken down from delivery, see ‘Correspondence’, Review of English Studies, n.s. 30 (1979). 194. 56 Trinity College Dublin, MS 419, fol. 75r. 57 Potter and Simpson, eds, Sermons, vi, 356. 58 Donne appears to have undertaken a similar re-shaping of the exordium when revising the sermon on Psalms 6. 6–7 for redelivery at court in 1628. See Colclough, ed., Sermons, p. 353, George, ‘Sacramental Art’, pp. 96–7, Shapiro, ‘Donne's Sermon Dates’, p. 55, and Stanwood, ‘Donne's Earliest Sermons’, p. 369. 59 Where later imposed act breaks co-exist with other revisions the editors retain them because such integral changes are too difficult to undo. See Wells and Taylor, Shakespeare, p. i6. For discussion of the Oxford Shakespeare, including this particular editorial decision, see Gabriel Egan, The Struggle for Shakespeare's Text: Twentieth-Century Editorial Theory and Practice (Cambridge: Cambridge University Press, 2010), esp. pp. 178–79. 60 There is just one omission in the Lothian manuscript and then two small errors, both in Fifty Sermons (‘Tristitia’ mistakenly given as ‘Tristitiae’ and ‘Gandium’ for ‘Gaudium’). The only variant of note is the placing of ‘Divisio’. The folio places this at the start of the paragraph while the manuscripts place it part way through the paragraph where Donne actually begins the division with the words ‘The parts then will be but two’ (Potter and Simpson, eds, Sermons, iii, 333). This change might be the result of authorial revision, but is more likely an adjustment introduced by a compositor who felt the marginal entry looked more regular at the start of the paragraph. 61 Potter and Simpson, eds, Sermons, iii, 334. 62 The one exception is the very end of the sermon where the headings ‘suffering’ and ‘my suffering’ are combined and the final heading ‘pro vobis’ is omitted. Burley's notes become less detailed as the sermon continues. This article and all my editorial work on Donne's sermons is deeply indebted to Steve May. I am extremely grateful to him for his kindness, generosity, and good humour in sharing his expertise and knowledge. I am also grateful to Catherine Evans, Marcus Nevitt, Tom Rutter, Cathy Shrank, Sebastiaan Verweij, the two anonymous readers for The Library, and, in particular, Mary Morrissey, who all read and commented on earlier versions of this article. © The Author 2020; all rights reserved This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Margins of Error: Performance, Text, and the Editing of Early Modern Sermons
JF - The Library
DO - 10.1093/library/21.4.423
DA - 2020-12-18
UR - https://www.deepdyve.com/lp/oxford-university-press/margins-of-error-performance-text-and-the-editing-of-early-modern-GQmvdfURcf
SP - 423
EP - 444
VL - 21
IS - 4
DP - DeepDyve
ER -