University of Virginia Library

The "Setting Foorth" of Harington's Ariosto
by
Simon Cauchi

Just as book design has only recently become established as a specialist craft in which a skilled workman may make a living and be credited for his work, so the historical study of book design is a relatively new field in inquiry, not yet thoroughly incorporated into the conventional wisdom of historical bibliography and of textual and literary criticism. Adrian Wilson is a pioneer in both fields, and his book The Making of the Nuremberg Chronicle (1976) is a remarkable demonstration that book design was practised in a recognizable way in the earliest years of printing—and indeed before—and that evidence of its practice survives. Somewhat similar evidence survives in relation to Harington's Ariosto,[1] which Philip Gaskell has described as "an unusually well-documented example of the progress from writer to reader of a work of literature of the late Elizabethan period". Yet Gaskell in his study of Robert McNulty's critical edition pays hardly any attention to questions of book design, preferring to concentrate his discussion on the transmission of the text and on McNulty's treatment of it.[2] The "text" is understood narrowly as being constituted by the "words" and their "details" (spelling, punctuation, etc.) but such matters as the layout of the words on the page are regarded as "relatively minor" (p. 15) and the

fact that, for example, much of Harington's apparatus, like the text of the Nuremberg Chronicle, was "written to fit" the space available for it is not even mentioned. Tom Davis in an article on modern textual editing[3] adopted a similar approach, acknowledging that Harington "specified the typeface and ornaments" (p. 64) of his book, but arguing that his authorial intentions in such matters could be ignored by a modern editor, "since they do not affect what readers would declare to constitute the work" (p. 65).

In this article I shall be arguing, on the contrary, that the design of Harington's book (the 1591 edition) did indeed affect the text of his apparatus and does indeed affect—in ways that Harington intended—a reader's understanding and enjoyment of the work as a whole. Harington chose to publish his verse-translation in the form of a fine, elaborately-presented printed book, in which he addressed his readers not only in the verse of his translation and in the verse and prose of his apparatus but also in the ancillary languages of picture and typography. I believe that, if these ancillary languages are interpreted aright, one is led to a new respect for his work as a translator; or, in other words, that his work as a book-designer is an essential part of his achievement in "setting foorth" Ariosto's poem for the reading public of Elizabethan England. Like the poem itself and much of the apparatus, the book's design was largely "filcht from forren lands"[4] —that is to say, the pictures and typography are modelled on Italian originals, notably the Franceschi edition of 1584—but Harington imposed his own treatment on all three elements of the book (poem, apparatus, and design) and did his best to reconcile and integrate the diverse materials he energetically brought together, imitated, adapted, and improvised. The verse-translation was written first, and the length of each of the forty-six cantos seems to have been determined by purely literary constraints (Harington's time, patience, and skill); but there is ample evidence to suggest that the apparatus was shaped to fit the design that Harington had in mind, and that the design itself was shaped to express the form and indeed some of the significances of the work. No doubt it is going a little far to call Harington a "book-designer" in his own right—he was a "print-customer", really—but the term does serve to emphasize two well-known facts: that the printer, Richard Field, took his instructions from Harington in these matters, and that Harington "gaue direction" for the making of the book's engraved title-page and forty-six engraved illustrations. It seems clear that Harington intended the book's design to serve as an expressive means of communication with his readers, modifying as well as reinforcing the verbal meanings of the text. Just how well his intentions were carried out, and just what degree of fidelity to the 1591 design may properly be called for in a modern edition, are interesting questions, which deserve to be looked at quite closely.

I

The materials which enable us to form an assessment of Harington's design-intentions are quite plentiful. First of all, the advertisement to the reader, as D. F. McKenzie has written,[5] is "a remarkably informative statement about the range of reader responses that Harington thinks to elicit" (p. 104). In it, Harington explained "more plainely, then for the learned sort had haply bene requisite" the arrangement of the book, the functions of its component parts (or, rather, most of them, including the illustrations), and how and in what spirit he wished the book to be read. The language is plain and direct, admirably suited to its immediate purpose; yet it abounds, as so many old documents do, and like the rest of Harington's prose in this book, in fascinating half-expressed thoughts, which I shall be attempting to elucidate and amplify. If one reads the text of the advertisement straight through and counts up all the main verbs in the active voice and first person singular, one gains a very clear impression indeed of Harington's close personal involvement in every part of the book. So, too, the wording of its title suggests clearly enough Harington's concern with the book's presentation as well as with its substance:

In the 1591 edition, the advertisement occupies one full page (sig. A1) immediately preceding the plate for Book I. Its location there as part of the first regular gathering suggests that it may have been set and printed at or near the beginning of the printing-job, and not as the other preliminaries were at the end of it. This may account for the lack of any mention in it of the general allegory of the whole work and the life of Ariosto, both of which appear at the back of the book, after the poem and before the index, and may well have been late additions to the book's contents—just as most or all of the third part of the preface, the defence of Harington's own work, certainly was. If the general allegory and the life of Ariosto were indeed late additions, then Harington was willing to make a virtue of necessity, for he wrote them in such a way that they suit their location at the back of the book, balancing and complementing the preliminary material, and they are addressed (the first explicitly) to readers "that haue read the former Poeme". However, it is possible they were planned to go there from the beginning, and were not mentioned in the advertisement simply because Harington did not there wish to draw attention to them.

More light is shed on Harington's conception of the book's design, and on the chronological sequences of composition, design, and printing, by the

two surviving manuscripts, both of which ante-date the printed book. Bodleian MS Rawl. Poet. 125 is a scribal fair copy of the first twenty-four cantos, and is said to look like a "mock-up" for an elaborate edition to be illustrated both by engravings and also by woodcuts deriving from one or more of the Giolito editions.[6] Either that, or else perhaps it preserves two separate attempts at planning, one for woodcut illustrations and one for engravings. The inclusion of a Perseus and Andromeda engraving adapted to serve as an illustration of the Rogero and Angelica episode at the end of Book X shows that Harington at-one stage was considering an alternative plan to the straightforward copying and adaptation of the engravings by Girolamo Porro in the Franceschi edition. British Library MS Add. 18920 is a fair copy of Books XIV-XLVI, mostly in Harington's own hand, and has the special interest of being the manuscript that was used as printer's copy in Field's printing-house.[7] It is from this manuscript that we know of Harington's keen interest in details of the book's layout, ornament, and typography, expressed in what McNulty (p. xlii) called his "peremptory" instructions to Field; and it is by reference to this manuscript—to the layout of the material on the page, to the deletions and revisions and additions, to the printers' signature and page-number notations, to errors in the stanza-numbering, and so on—that many small puzzles about the making of the printed book can be resolved, wholly or in part. Missing from this manuscript are the end-of-canto notes on the moral, the history, the allegory, and the allusion, except those for Books XXVI, XXIX-XXXVII, XXXIX, and XLIV-XLVI, which are present. The presence in the manuscript of some of these end-of-canto notes may possibly be related to Harington's absence or irregular attendance at the printing-house while the book was being printed. Some of those present in the manuscript may have been written early, before Harington went away for a time; but those for Book XXXVI, at least, were written late, after the copy was cast off for the press, for they are written around a deleted signature-notation which is copied out again below.

Further information about Harington's design-intentions can be obtained by comparing the 1591 edition with other printed books, and especially with the various Italian editions of Ariosto's poem that served as its design-models. These include not only the 1584 Franceschi edition but also the Valgrisi edition and possibly earlier Italian editions as well.[8] In particular,

it is instructive to compare the treatment by the English printers and engravers of these Italian design-models with Harington's treatment of his literary sources for the poem and its apparatus.[9]

Reference was made earlier to some doubtful points of chronology. A good deal is clear, however, and before going on to consider the book's design in more detail, it will be worth while sketching out some of the salient facts that can be inferred. The poem itself was the first part of the British Library manuscript to be copied out, with room being left for the addition in single or double column of the verse-arguments; but the fair copy of the verse was not necessarily completed before the earlier cantos were revised, equipped with verse-arguments and marginal notes, and printed. The marginal notes were evidently entered in at various times. They are in handwriting of various sizes and degrees of neatness, now in secretary script, now in italic. Since several have been traced to their sources in Lavezuola's "Osservationi" and Ruscelli's "Annotationi" incorporated in the Franceschi edition, it seems likely that Harington, while using these and perhaps other commentaries, would turn from one place to another in the manuscript to enter his marginal notes. The end-of-canto notes (mostly absent from the manuscript) seem to have been written in sequence, beginning from the beginning; this is observable from the text, for some promises made about them are not fulfilled, or their fulfilment is deferred.[10] From Book IV on, most of them, like Ruscelli's canto-by-canto "Annotationi" in the Valgrisi and Franceschi editions, were written to fit the available space left over at the end of the verse; they usually end neatly at or near the foot of the page. The general allegory was written after the work of annotation was complete, as can be seen from its first paragraph; so also, presumably, was the life of Ariosto. Drafts or early versions of the preface and the index seem to have been in existence at an early stage of the work, although both were printed late. They are both referred to in the advertisement to the reader, somewhat inaccurately, and also in marginal notes and elsewhere.[11] In brief, the text of the work seems to have

been composed in a rather complicated chronological sequence.

The printers, however, seem to have worked through the book systematically, gathering by gathering, from A to ¶. The sequence of setting within each gathering does not appear to follow a regular pattern. There was no need to keep to one, since most of the text consisted of verse in regular stanzas, easy to cast off accurately. The running-titles never wander further than into an adjacent book or gathering, forward or back. Wrong page-numbers are few and far between, even though there are some lengthy sequences of them in the printer's copy, most but not all subsequently corrected. The major confusion in page-numbering occurs in Book XXIV, after an irregular gathering of four leaves at the end of Book XXIII—about which more later. An interesting improvement in the layout of the canto-openings is first made in Book XIV. These layouts are of three types. Book I, with a lengthier headtitle than the rest, has the first six stanzas on the first page. The remainder have the first seven, the break between columns coming after the first line of the third stanza in Books II-XIII, XV and XXVI, but after the second line of the third stanza (a decided improvement) in Books XIV, XVI-XXV, and XXVII-XLVI. It seems clear that the improvement in layout was made too late for Book XV but in time for Book XIV; or, in other words, that sig. K4 was printed before sig. I5. (The untypical break later on, at Book XXVI, was necessitated by an extra turned-over line, mistakenly indented below the initial in the first stanza.) From Book XIV on, also, and not before, there regularly appear below the notes of allusion such lines as "The end of the Annotations vpon the fourteenth booke". Questions of layout and presentation were evidently reconsidered at this stage of the job, and it may be more than sheer coincidence that, from Book XIV on, the printer's copy has survived—though it is of course not known when the copy for Books I-XIII and the verse-argument for Book XIV was lost. Harington's manuscript direction to Field at the end of the annotations to Book XLVI shows that only then was the type chosen for the preface, general allegory, and life of Ariosto. Pointedly, I think, these were to be—and are—in the same fount of pica roman as Field had used for printing The Arte of English Poesie (1589): "Mr Feeld . . . I would have the allegory (as allso the appollygy and all the prose that ys to come except the table [)] in the same printe that Putnams booke ys." Not only the fount of pica roman but also the initials and ornaments of "Putnams booke" were used again in Harington's.[12]

I have been unable to establish whether the sheets regularly went through the rolling press, for the printing of the engravings, before or after the printing of the text; it may or may not be significant that two of the wrong page-numbers, "208" at sig. S3v and "141" at sig. X5, are on the same side of the sheet as engravings. The printers regularly made allowance for these engravings in their casting-off, and in the 1591 edition (unlike the 1607 and 1634 reprints, and for that matter unlike the 1584 Franceschi edition)[13] they are all correctly printed, in their proper places and the right way up.

Thus, with only one major exception (the irregular gathering of four leaves at the end of Book XXIII), the work of preparing and printing the book seems to have gone reasonably smoothly, once the methods of annotating and illustrating it had been chosen and devised. The only major alteration in the book's contents made in the course of printing seems to have been the addition of new material at the end of the preface and possibly also at the back of the book. The additions—if that is what they are—were intended principally but not solely to answer and forestall readers' criticisms of the translation, verse, and notes. These last, the end-of-canto notes in particular, seem to have given Harington a good deal of trouble, for in them he "strained" himself not only "to make mention of some of my kindred and frends" (as he wrote in the third part of the preface) but to organize them according to a preconceived fourfold division (moral, history, allegory, and allusion) and at the same time to make them fit neatly on the page. However, the chosen method was applied pretty consistently, and the description Harington gives in the advertisement to the reader of the functions of these end-of-canto notes is quite remarkably accurate.

II

It is time now to consider in more detail the "setting foorth" of the book as it is described in the advertisement to the reader and to try to form an assessment of Harington's authorial intentions in regard to the book's design (illustrations and typography). The first point to be noted is that the "setting foorth" is by no means merely derivative; it is adapted, not copied, from the Franceschi and Valgrisi editions, and has a number of interesting original features. Secondly, even its most conspicuously derivative features—that is to say, the illustrations—were borrowed for an artistic purpose and employed to calculated effect; or so it can be argued. They are examples of "honest Theft".

The first question to be asked concerns Harington's book in relation to his readers. What was Harington's purpose, his authorial intention, in bringing

out a book of this kind? On the one hand, it was a translation—and therefore presumably chiefly intended for readers ignorant of Italian; on the other, it was elaborately and expensively produced—and presumably beyond the pockets of many who most had need of it. No record seems to survive of the price at which Harington's Ariosto was sold to the Elizabethan public, but comparable books were sold in the 1590s for more than ten shillings, unbound; and binding would have cost another three or four shillings.[14] Harington recognized that "all that may reade this booke are not of equall capacities" and his translation and notes are accommodated to the needs and wishes of many different readers: the Queen, his "kindred and frends" (and among them especially, perhaps, his wife),[15] the book-trade authorities (no doubt), and others. In particular, many of the more elementary aids to comprehension with which he equipped the poem seem to have been addressed to the less learned readers of his book and here, I believe, he may have been thinking primarily of the needs of children, as well as of the less cultivated readers of the (wealthier) middle class. He appends his life of Ariosto: ". . . to satisfie such as are desirous to know who this Ariosto was, whom I haue so greatly extolled in my Apologie, as a benefactor of all studious mindes, and on whose worke I haue employed so much time, to put it into English verse, and to bestow so many notes as I haue done vpon the expounding of his Allegories, and whatsoeuer else I iudged fit for the readers of weaker capacities." We know that Harington's book was owned by families of the gentry and nobility and read by younger members of them, but just how widely the first edition of 1591 was distributed among the poorer literate classes is uncertain; not widely at all, one imagines.

Considering the readership, it is entirely understandable that the ancillary matter with which the poem is accompanied should tend, for all its elaboration, to be comparatively simple and straightforward (though it is graced with many details intended to give pleasure to the more discerning). In the preface, which is clearly intended to be a statement of some weight, Harington undertakes the initial task of introducing the translation to his English readers. He relates the work to the moral and literary concerns and controversies of contemporary English society, deprecating and ridiculing much that seems to him foolish, and dealing at length with the matters that seem to him to present serious threats to the success of his book. These are principally the fears of the pious but less cultivated reader that such "Italian toyes" as Ariosto's poem might be "hurtfull for his soule, or conscience", and the cant of the time that poetry was a "nurse of lies" and an "inticer to wantonnes". Harington defends his work in his own individual way, laying

particular stress on the poem's tendency to virtue, the true ground and basic credibility of Ariosto's narrative, and the pleasure as well as the profit to be taken from the poet's delightful fictions. Characteristically, he pokes fun at the hypocrisy of readers who blame and at the same time enjoy the poet's "lasciuious" stories, but for all that he takes the charge of "wantonnes" seriously and does his best to answer it. He has an equally individual point of view on the purely literary side of the question, arguing rather paradoxically that Ariosto, as a modern Christian poet, is at once superior to the ancient epic poets and free to depart from their example, yet good at imitating them and in following "verie strictly" Aristotle's "rules" of epic poetry. It is in accordance with both these lines of defence that in his marginal and end-of-canto annotations Harington interprets the poem and comments upon it. Then, at the end of the book, in the general allegory, he makes use of a not altogether like-minded treatise by Bononome[16] in order to present a serious, coherent summary of the good uses and profit to be had from reading the poem. Considering the readership, again, it is quite understandable that a great deal of Harington's apparatus should be intended to help the reader simply to follow and remember the story. The entire text of the advertisement to the reader is relevant here as a summary statement of Harington's authorial intentions, but it will be convenient to focus our attention, to begin with, on the remarks concerning the "Direction for continuing the diuers stories" and on the index: Then, later:

Ariosto's narrative method is too large a topic to be fully treated here. The particular points relevant to this discussion are the characteristic interruptions of the stories both within and at the end of the cantos (the two kinds of interruption are quite distinct from one another); the artistry in the succession of Ariosto's short narrative flights; the resumption of the preceding narratives at the beginning of each canto after the proemes; and the interposition in the narrative texture of a number of whole stories told consecutively from beginning to end. These last seem to be what Harington

is referring to in his remarks quoted above on the "seuerall tales . . . that may conueniently be read single", even though the list of the tales includes not only these but some others—for example, the stories told by newly introduced characters in order to explain their present predicaments and (hardly surprisingly) such major episodes as the "Tale of Rogeros comming to Alcina" and the "Tale of Orlandos madnes". By including these incomplete or interrupted narratives in the list of the tales, Harington clearly intended to make the list serviceable to those reading the book a second time, selectively; but his earlier remarks make it equally plain that he wished a first-time reader to begin at the beginning and to read the poem straight through, consecutively.

That key phrase—"the meane matter betweene the so deuided stories (vpon which commonly they depend)"—is strikingly reminiscent of the words of a modern critic, C. P. Brand, on Ariosto's "specific contribution to the romance tradition":

Whether Harington can be credited with a similar view of Ariosto's "invention" on the strength of one bare hint in the advertisement may be open to doubt; but at the very least he must have recognized that, when the threads of a story are picked up again, the reader is assumed to know about the developments in the plot that have taken place since it was broken off, and that without such knowledge the point would be missed of the many meetings, pursuits, vows, recognitions, exchanges and gifts and thefts of horses, armour, weapons, the magic ring, and so on, by which the plot is advanced. The memory of what is going on in all the various story-lines is an essential part of the poem's richness of texture. There are whole cantos in succession in which various major characters are never mentioned; they are off-stage, assembled in the wings or in the tiring-house, so to speak; or, more exactly, the reader knows, or should know, that Astolfo is still flying around the globe on the hippogriff (let us say), or that Fiordeliege is still seeking her beloved Brandimart. Harington's index permits the reader to refresh his memory on these points and, if need be, refer back to the text; so do those marginal directions that refer back; while those that refer forward not only are aids to selective reading but also, and perhaps more importantly, help the reader by telling him what to expect—how soon or how much later a story "abruptly broken off" will be resumed, how long there is to wait before knight recovers horse (say) or lover is reunited with lover. A similar concern to set the reader's expectations on the right course and to clarify the complexities of the plot, as well as to assist selective reading, is seen in the marginal notes on the stories' beginnings and endings, and in the practice of naming in the margin characters as they approach or appear, more often than not well before they are named in the text. It has to be admitted that all these aids to comprehension can sometimes be sharply at odds with Ariosto's narrative mystification (compare the text and marginal note at X: 47-48, for example).

From the point of view of book design, however, the breaks between the cantos are the main thing, and they have their own special character. Ariosto's congedi at the end of each canto, followed by the proemi at the start of the next, serve quite a different function from the dropping within a canto of one narrative thread in order to pick up another. The congedi and the proemi enforce a pause, and at the same time, despite Ariosto's more sophisticated treatment of them, they hark back to the practice of his predecessors in the romance tradition, suspending the action at a particularly exciting point.[18] The proemi themselves are neatly made to fit their positions, facing as they do Janus-like to past and future, and the resumed narrative after the proemio invariably—or, rather, with only one exception[19] —takes up the story again at just the point where it was broken off by the preceding congedo. The narrative technique is time-worn and familiar, reminiscent of cliff-hanging endings in silent-film serials. Or, in more modern terms, it is as if the poet imposes a "freeze" (as the TV technicians call it) on the moving images of his story, and begins the next canto's narrative with the melting of the same "freeze". Or again, in perhaps more decorous terms, it is as if Shakespeare, by mingling together and modifying the principles of the romantic and classical drama in a way he did not, had permitted himself the grossest disunities of time, place and action from scene to scene, and yet had rigorously observed an unknown—and, of course, theatrically quite pointless —liaison des actes.[20]

The elaborate treatment of the canto-breaks in the printed books—Valgrisi's, Franceschi's, Harington's—clearly reinforces the poet's own requirement of a pause. Hence the positioning of the plates immediately before each canto; hence the elaborately bordered verse-arguments (and, in the Italian editions, the prose allegorie) at the top of the page above the beginning of the text; hence the ornamental initial in each first stanza, inviting the reader to start the new canto with the proeme. To be sure, a similar treatment of the canto-breaks (or such divisions) can be seen in other contemporary books of poetry where the same narrative technique is not used and where there is no regular liaison across the divide; but Ariosto's use of liaison was well understood by the illustrators of the Valgrisi and Franceschi editions that Harington used, and also by the author of the verse-arguments that Harington translated.

The plates and the arguments of Harington's book are to be considered more fully later. Here it is appropriate to point out a rhythmical regularity that has not previously been commented upon. The borders surrounding the verse-arguments in Harington's book are of two kinds: one, a frame of type-ornaments; the other, a compartment made up of four head-pieces, i.e. ornamental woodblocks. The two borders alternate in a regular sequence, with a change of step at Book XXIV. This feature of the design was imitated, not from the Franceschi edition, where six different engraved borders are used in no particular order, but from the Valgrisi edition of 1556 or one of its reprints. In the 1556 edition, according to Mrs Falaschi, "Each canto is preceded by an argomento framed in an ornamental woodcut for which there are two alternating patterns consisting of putti, garlands and masks . . .".[21] Although there is no positive evidence on the matter, we may take it that Harington either suggested or at least approved of the imitation. Significantly, the alternating sequence—still with its change of step at Book XXIV, which seems not to have had Italian precedent—was to be preserved in the 1607 edition, printed by Field "for Iohn Norton and Simon VVaterson" and published in Harington's life-time. It was not preserved in the posthumous 1634 edition, printed "by G. Miller for J. Parker".

III

Harington "gaue direction" for the making of the plates and was particularly proud that they were "all cut in brasse". However, it is a striking fact that in only three of the plates were significant changes made to the designs. The remaining forty-three were copied as closely as possible within the limits of the English engravers' abilities. Why was this so? Lack of ideas, lack of a skilled draughtsman, or some such reason? Was Harington merely taking the easy—though most costly—way out, and filching the best illustrations he could find in the Italian editions? Or did he really understand and

appreciate the effects achieved in Porro's plates and deliberately mean to transfer them into his own book? The truth, I think, embraces affirmative answers to all these questions. Harington intended his book to be a lavish production, but he also intended the illustrations to serve a practical purpose and to assist the reader in his understanding and enjoyment of the poem. He writes in the advertisement to the reader: The striking thing about the remarks on the perspective, and for that matter about Ruscelli's remarks in his "A i lettori", which Harington is here translating and abridging and which originally referred to the woodcuts in the Valgrisi edition, is that they give no hint of the use to which the perspective is put or of the other means employed by the unknown artist of the Valgrisi woodcuts to represent Ariosto's poem in picture. Yet the Valgrisi artist sought to illustrate not only the action of the poem but also its shape and spirit; so, in his own way, did Porro; and there are good grounds for believing that Harington appreciated and respected Pooro's artistic purposes, or at least came in the end to appreciate and respect them.

Porro's practice differs in a number of respects from that of his immediate predecessor, the Valgrisi artist, but the basic principles of illustration are the same. These Valgrisi woodcuts, too, had their models and precedents, as Mrs Falaschi has shown, but their unique character is what is of interest. They are the first full-page illustrations of Ariosto's poem and the first and last attempts to represent in full detail and in a single plate the entire substance of each canto. The perspective is the key to the interpretation of the multiple scenes, which usually (though not quite invariably) follow one another in narrative sequence upwards from the foot of the page, so that the figures—"cosi sfuggendosi, & diminuendo a poco a poco"—become reduced to minute proportions at the top. Where the perspective does not follow the narrative sequence, it is because the artist has sensibly chosen to separate Paris from Damascus, for example, and to arrange the illustrated actions geographically in different parts of the illustration (Plate XVIII). The problem of perspective diminution was compounded by the small size of the wood-blocks. For Valgrisi's quarto edition, they were inserted in a series of what look like quite unnecessarily broad and elaborate factotum borders. They were used without the borders in Valgrisi's octavo edition, also first published in 1556.

The task undertaken by the Valgrisi artist is indeed quite breathtaking, as well as eye-straining, in its ambition. The problems of illustration posed by Ariosto's poem each seem to have been systematically isolated and tackled. How to represent—say—a lady who vanishes, a palace which dissolves, a phantasm

of different appearance to each beholder? How to represent the variety of personages in the poem, from the historical Charlemagne to the allegorical and prosopopoeical "Silence", "Pride" and "Discord"? How to illustrate immense journeys undertaken in the course of a few stanzas? What to do about the many reported narratives, some of which are the most vivid and amusing stories in the poem? Valgrisi's artist found solutions to all these problems and more, ranging from invention to systematic omission; but his aim was clearly to make his illustrations as comprehensive as possible. He even illustrated the final proeme, showing a large Ariosto in a small sailing boat approaching a quayside thronged with his patrons and friends.

From our point of view, the important thing about the Valgrisi illustrations is that both the beginnings and the endings of the cantos are systematically represented: the first in the main scene of larger figures at the foot of the page, the latter diminutively somewhere at the top, very often in the top right-hand corner. There is thus a kind of pictorial liaison from one woodcut to the next, which parallels the poet's own liaison between the cantos. So, also, the points at which stories are "abruptly broken off" and later picked up again are illustrated in the body of the pictures, somewhere in the nearer or further middle distance.

Porro, taking these woodcuts as his model, took over many of the same technical solutions. He, too, but more consistently, used perspective to represent the narrative sequence. He, too, omitted reported narratives, had his problems with the phantasms, represented the naked and invisible Angelica all too visibly, the palace of illusions all too solidly. His designs usually borrow a good deal from the Valgrisi woodcuts, often containing details that are reversed adaptations of those in the woodcuts, so that (for example) the endings of the narrative sequences usually appear in the top left-hand corner. We may confidently presume that Porro worked with the Valgrisi edition open before him, but he also consulted the text, usually carefully, in order to make his own adaptations and improvements.[22] Allowing himself more space by including his own narrow borders as an integral part of the engravings, he gave much greater prominence to the foreground scenes and carefully strengthened and varied the composition of the designs. He enlarged the principal figures and made use of architecture, rocks, foliage, and so on, to separate the smaller groups of figures into discrete sequences. However detailed they may be, his plates do not become mere frames for a mêlée of minute figures—as some of the Valgrisi woodcuts certainly do.

Porro's main concern was to make his designs bolder and more dramatic. To do this, he occasionally omitted or conversely added or enlarged the illustration of the very first narrative stanzas of a canto. If these were omitted, the next few stanzas supplied the principal subject. He clarified the action, either by separating successive phases of an episode combined in one

of the Valgrisi scenes or—conversely again—by combining in one bold design actions represented successively in the woodcuts. He altered the perspective, nearly always in order to bring it into closer accord with the narrative sequence (and if this meant mingling Paris and Damascus together in one continuous architectural composition, so it had to be). Porro's designs are more representational and less schematic than those of the woodcuts. He does not imitate the Valgrisi artist in illustrating some of the cantos from a bird's eye view; nor does he use maps to illustrate journeys; he makes no attempt to portray allegorical personages, except quite inconspicuously, or to find a way of representing in the illustrations such a passage as Saint John's praise of writers in canto XXXV, illustrated in the Valgrisi edition by a statue of "Immortality". More particularly, in the interests of artistic clarity and simplicity—a relative simplicity, to be sure—Porro frequently reduces the amount of background detail illustrating the latter parts of a canto, or he even omits to illustrate them altogether. Narrative sequence, not narrative significance, is the guiding principle, together with the search for opportunities in the first few pages of a canto for a good, striking foreground scene.

The results of this procedure are at once admirable and curiously unsatisfactory. To take the most obvious example, Ruggiero's rescue of Angelica takes place towards the very end of canto X and is therefore illustrated only by tiny figures in the top left-hand corner of Plate X. Similarly, the pastoral idyll of Angelica and Medoro is shown only in the background of Plate XIX. The foreground of Plate XIX, at least, gives prominent treatment to Cloridan and Medoro, thanks to the continuation of that episode at the start of the canto; but, even there, it is a later part of the episode that is illustrated, not the famous passage in canto XVIII describing their nocturnal search for the body of Dardanello, hidden in darkness and then suddenly illuminated by the moon. (This is illustrated in the background of Valgrisi's Plate XVIII, but not in Porro's.) At their worst, Porro's methods bring about some quite ludicrous distortions of emphasis. Thus, in Plate IV, a conspicuous detail in the foreground of an otherwise admirable design illustrates Bradamante's purchase of a horse from an inn-keeper—a mere narrative iota (Ariosto, IV: 10. 3-5) and omitted by Harington from his translation—but Ruggiero's conversion and baptism (Ariosto, XLI:59) are not illustrated at all in Plate XLI, though the deficiency is partially made good in Plate XLIV. The lack of illustration of reported narrative is also unsatisfying. It means the omission, for example, of Pinabello's vivid description of Atlante and his flying horse in canto II, an important passage because it gives the reader his introduction to the hippogriff.

Yet Porro's methods do peculiarly and no doubt deliberately befit and exploit the exact location of his plates in the Franceschi edition, coming as they do regularly between the end of one canto and the beginning of the next. The principal scenes represent, if not precisely the moments of narrative "freeze" (and some of them do exactly that), then at least the immediately following or closely following action. Their subjects are usually

intelligible, at least in a general way, to the reader who has finished the previous canto and not yet started the following one. No more than the principal scenes are thus broadly intelligible, and even they can only be fully interpreted by reading on. Moreover Porro departs from his usual practice, for perfectly defensible artistic reasons, in Plates XXII, XL and XLI, as well as in Plate XXXII where Ariosto's narrative required him to do so. Nevertheless, it may fairly be said that Porro's method of illustration reinforces the poet's own liaison at the canto-breaks rather more emphatically than does that of the Valgrisi artist. However Porro did not attempt to maintain any consistent kind of liaison from plate to plate. That kind of liaison is present or not in his plates, or sometimes only tenuously suggested, according to the exigencies of the designs.

The English engravers' copies of the Italian plates now need to be compared with Harington's translation itself. Philip Hofer remarked that the English engravers "copied both subject and style, while the translator . . . went to the opposite extreme, and took bold liberties with the text!"[23] This, I believe, gives quite the wrong impression. The generally good accord between the plates and the translation is what needs to be emphasized, not the quite numerous small discrepancies. The principal scenes are all perfectly intelligible by reference to the translation, and apart from Plate IV, already mentioned, there are only two plates out of the forty-six where details given prominent or conspicuous treatment by Porro are inadequately explained in the text. These discrepancies occur in Books XIII and XVII. In both, it may be said that they work ultimately to the benefit of the book as a whole—faute de mieux, that is, and assuming thorough revision of the verse was impracticable. Both, interestingly, are the result of Harington's systematic abridgement in the translation of scenes of cruelty and suffering. In Book XIII, Orlando's brutal punishment of the outlaws who had held Isabella captive in their cave is condensed to a single undescriptive final couplet:

Plate XIII represents conspicuously, in the upper left-hand quarter, just how this was done. Orlando pulls at a rope to drag his prisoners out of the cave, and above, the outlaws' corpses hang by their chins from the lopped branches of a dead-looking tree. The outlaws' fate is clear to the English reader, even though some of the details in the plate may be puzzling. The other major discrepancy is similar. In Book XVII, Rodomont's martial exploits within the walls of Paris are much abbreviated in the translation, which preserves the vehicle of an epic simile of a snake shedding its skin but omits the tenor, the description of Rodomont's shining helmet and armour, together with all the narrative describing his Pyrrhus-like assault on the palace with fire and sword and the terror of the trapped Christians within (Ariosto, XVII:9-13; Harington, XVII:6-9). All this is carefully illustrated in the plate, and although the details are not explained in the text, the scene is readily comprehensible by an English reader as both illustrating and elaborating Harington's stanzas.

A few words also need to be said about the alterations made to Porro's designs as they were copied for Harington's book.[24] Plate XXVIII, as is well known, was altered by the removal of an architectural background to make room for a pictorial summary of the "Tale of mine Host" about Jocundo, the story which—as tradition has it—first involved Harington in his work of translation and for which he incurred his penance of having to carry it through to completion before being allowed back to court. The light-hearted spirit in which the penance was probably imposed may be gauged from the inclusion in the plate of three scenes of copulation as background vignettes. Plate V was also altered, no doubt by Harington's direction, to include scenes of love. Dalinda's tale of Polinesso's amorous "sleights" is represented at bottom left and the traveller's tale of Ariodante's "saltus amatorius" at top right. There was also a minor change, with nothing (for once) overtly amorous about it, in Plate III. Atlante's castle is shown at top left, presumably to illustrate Melissa's description of it (although according to the narrative the castle does not come into view until the next canto), and a mountain goat at top right is presumably there to suggest the dizzy height of the precipice. Otherwise, the only changes are just as Mrs Falaschi has reported them: two plates reversed (Plate IV and XVII), one copied perforce not from Porro's engraving but from the Valgrisi woodcut and enlarged and reversed in the process (Plate XXXIV), some minor but sensitive improvements (Plates XIII and XXIII), some errors of copying in the labelling, and a number of small omissions and simplifications of detail in the designs and in the borders. Plates XXXIX and XLVI, particularly, are rather crude and simplified by comparison with the originals. None of Porro's various errors of interpretation, large or small, was corrected.[25]

The alterations in Plates III, V and XXVIII all violate Porro's principle of not illustrating reported narrative. We may assume that Harington disliked that principle, or at least was not going to let it stand in his way in his treatment of two favourite cantos. He seems also to have been dissatisfied by the lack of prominence given in Porro's Plate X to Rogero's rescue of Angelica. The Bodleian manuscript shows that he had toyed with the idea of including a full-page plate illustrating that episode, though just where it might have been fitted into the printed book is not at all clear.

Considering the plates by themselves, the eventual dropping of the Rogero and Angelica illustration is probably the strongest piece of evidence we have that Harington came in the end to understand and respect Porro's principle of using perspective to indicate narrative sequence. What of the effects of liaison from canto to canto? Harington understood and respected these, too, as can be seen from his treatment of the Italian verse-arguments and from the effects of their juxtaposition with the plates.

IV

Since the Italian "Argomenti" and the Valgrisi woodcuts were first published together, it seems probable on purely a priori grounds that there would have been a certain measure of collaboration between the author of the arguments and the artist of the woodcuts—if only to the extent of one consulting the other's work, when it was possible to do so, as well as the text of the poem itself. They seem in fact to have agreed on a common approach, each working towards the same ends according to the potentialities of his own medium. However, neither Ruscelli in his "A i lettori" nor Harington in his advertisement to the reader say very much about the function of the arguments in the context of the book as a whole. Ruscelli writes merely: "Gli Argomenti in ottaua rima, che habbiamo posti in questo Libro di Canto in Canto, sono del Signor Scipione Ammirato, giouane di belle lettere, di felicissima vena, & di molti studij." Harington, true to form, is a little more forthcoming, but not much: "Also (according to the Italian maner) I haue in a staffe of eight verses comprehended the contents of euery booke or canto, in the beginning thereof, which hath two good vses, one, to vnderstand the picture the perfecter, the other, to remember the storie the better." What is common to these two editorial comments about the arguments is that they draw attention to the fitness of their form and their location. The arguments are written in the same verse-form as the poem itself (Italian or English), and they are "posti . . . di Canto in Canto" "in the beginning thereof". Like the plates with which they are associated, the arguments serve the double purposes of introduction and recapitulation, whetting the reader's appetite to continue and at the same time helping him to remember and consider what he has read. However the forty-six stanzas (Italian or English) may also be read in sequence, as a miniature version of the whole poem. The sequence bears a certain resemblance to a "corona" of sonnets, and one suspects that readers may sometimes have attempted to use them as epitomes, as a substitute for reading the poem itself at length.

In the Italian arguments, all the main-springs of the plot—characters, motives, actions, consequences—are neatly abstracted and fixed in the reader's memory by a variety of means: crisp characterization (especially of the villains in the story), repetition of key words and names, and a complex syntax in which much use is made of participles, phrases in apposition, and so on, in order to give at least some idea of the connections between the various stages of the plot. The final couplet of one stanza is regularly echoed in some

way in the next one's opening lines, so that the stanzas are linked together in sequence in a way that imitates and represents Ariosto's own effects of liaison between cantos. Harington's practice of breaking up Ammirato's long, complex sentences into a number of shorter and simpler ones does introduce some obscurities of time-sequence, motivation, and the like, and for the sake of brevity Harington rather more frequently relies on bare proper names, lacking characterization or with the characterization much reduced. Thus "Pinabel di Maganza, traditore" (II:vii), for example, becomes plain "Pinnabel" and Gabrina, instead of being called "la vecchia odiata e vilipesa" (XXI:iv), becomes merely a "most unvvorthy wight". Sometimes, also, the connection between the final couplet of one argument and the opening lines of the next is lost or weakened by absence of repetition. It is reasonable to assume that all such changes were forced on Harington by technical difficulty, but there are also other changes in the English arguments which seem to be clearly deliberate departures from the Italian, made in order to present the poem in a different light or to offer different interpretations. The argument to Book VII, for example, is one of several made to conform with the moral and allegorical interpretations of the story given in Harington's end-of-canto notes, and there the English wording departs so radically from Ammirato's Italian that it can almost be considered free composition.

In the Valgrisi edition, more often than not, the first and last lines of the "Argomenti" correspond closely with the beginnings and endings of the narrative sequences as they are represented in the woodcuts; indeed, the artist's comprehensive method of illustration ensured that practically everything mentioned in the "Argomenti" would be represented somewhere or other in the woodcuts—if only one has the patience to find it. In the Franceschi edition, the correspondences between the plates and the "Argomenti" are not so close, for reasons that have already been explained: the different choice of principal scenes, the greater prominence given to them, the lack of illustration of later episodes in the cantos, and so on. Generally speaking, Harington's book has much the same measure of correspondence between plates and arguments as Franceschi's book has—less than that of the Valgrisi edition, but still present in varying degrees from canto to canto. In some cantos practically everything in the argument is illustrated; more often, the beginning but not the end of the argument is illustrated; in only one quite exceptional instance—Book XXVI—nothing mentioned in the argument is represented in the plate (except "Merlins well") and nothing illustrated in the plate, except the well, is mentioned in the argument. In what way, then, did Harington mean the arguments to help the reader "to vnderstand the picture the perfecter"? When it comes to close reading of the plates, the arguments are clearly of no help, but with the principal scenes it is different. A number of Harington's alterations in the arguments seem quite evidently to have been made in order to bring them into better accord with the principal scenes of the corresponding or following plates. His practice is by no means consistent, and one could wish that a few more such alterations had

been made; but there are enough of them to be sure that Harington deliberately sought by this means to achieve or to restore in the context of his own book not only a good correspondence between arguments and plates but also a clear sense of liaison from canto to canto. It is necessary to list and briefly to describe a fair range of examples: Porro redesigned Plate I so that Charlemagne and the battle are brought to the foreground. Moreover Harington seems to have taken the Christian defeat at "Burdells", and not Angelica's flight, as the true starting-point of the poem: cf. I:5-10; VIII:64.8 (Ariosto, VIII:72.8); and XIV:1 mn. These lines, for which there is no equivalent in the Italian, match the illustration of Dalinda's story in Plate V. Porro in Plate XIV gave much greater emphasis than the Valgrisi artist did to Agramant's musters. Harington reinforces the emphasis by repetition, and characteristically chooses not to emphasize Orlando's single-handed feat of arms. Harington's alteration in line vii brings the argument into better accord with the proeme to Book XXI, where Zerbino is presented as an example of good faith and the keeping of promises, however difficult (cf. XXI: Moral). The alteration in line viii is more significant, since it shows Harington's characteristic treatment of a canto-break handled untypically by Ariosto and by Porro. In the final stanza of canto XXI Zerbino hears the sounds of fighting and screaming—they are Pinabello's dying screams. The resumed narrative in canto XXII gives only a brief anticipation of Zerbino's discovery of the as yet unnamed corpse; the story will be told fully in canto XXIII. Porro's Plate XXII, unlike the Valgrisi woodcut, omits the corpse and has for its principal scene the next episode of Astolfo at the palace of illusions; Porro is reserving the corpse for the foreground of Plate XXIII. Harington, tender-hearted as usual, omitted Pinabello's dying screams from the translation. The eighth line of his argument thus removes a source of considerable potential embarrassment in translation, narrative, picture and design. This is an example of a change made in order to bring the argument more into line with the interpretations given in the notes. In writing the notes of allusion to Book XXIII, Harington borrowed what he wanted from Fornari and Lavezuola, but he deliberately omitted Fornari's description of Orlando's madness as something altogether uncommon, grand, and marvellous (cf. McNulty, p. xxviii, where the passage from Fornari is quoted). His treatment of the argument is precisely similar. Plate XXVII has Doralice's bolting horse as its most conspicuous foreground feature. The incident is interpreted allegorically in the notes to Book XXVII. Porro's Plate XXIX illustrates Rodomont's rage and, unlike the Valgrisi woodcut, does not illustrate the manner of the hermit's death—Rodomont hurls him into the sky and out of the story (Ariosto and Harington: XXIX: 6-7). Porro's Plate XXXVIII, again unlike the Valgrisi woodcut, omits Rogero's departure. Harington was no doubt glad to follow suit for the sake of some extra syllables, and could feel free to do so since Rogero's departure to serve his King is lengthily praised in the proeme. Harington here brings forward the "poco lieto" from Ammirato's argument for canto XLIII. The earlier Italian arguments make no mention of Brandimart's death-wound but emphasize Orlando's victory and the defeat of the infidel champions. In Plate XLII, Brandimart lies wounded in the foreground, but the motive for the change is almost certainly literary, not pictorial. Harington chooses to prepare the ground for the sad and solemn ending of the love-story of Brandimart and Fiordeliege.

One last example may be given here, out of sequence:

Is it too fanciful to suggest that the translations here are intended to make up in some measure for the loss of the full-page plate illustrating Rogero's rescue of Angelica? That Harington was determined to give prominence to the "naked" and "bound" Angelica, somehow or other? The lines at least do something to repair the botched translation of the stanzas in the narrative: Harington, X:81-85; Ariosto, X:95-99.

At their best, the canto divisions in Harington's book are extraordinarily well handled, especially where plate and argument appear on facing pages. A canto-beginning is carefully prepared for in the endings of the previous argument and of the previous canto itself, and on turning over the leaf a wealth of information is displayed through several media: pictorial, poetical, typographical, even marginal. Plate complements argument, proeme complements both (and is the reader's entry-point into the new canto), the topic of the proeme is picked up and developed in the notes (and there may well be a marginal reference to the notes),[26] and with any luck the very stanzas illustrated in the principal scene of the plate will be there on the facing first page. Particularly good examples of this kind of collaboration among all or nearly all of these elements may be seen in Books I, IV, VII, X, XIV, XVI, XVIII, XXIII, XXVII and XXX, and there are many others which, though less good all round, still offer many delights—particularly in the plates. However,

a good many of the plates do not in fact face the canto-beginnings, but appear overleaf on the preceding recto. This is a puzzling aspect of the design of Harington's book and of the Italian editions as well, even though the puzzle may largely be a twentieth-century one. Sixteenth-century readers may have found the arrangement more tolerable than we do. (McNulty in his critical edition very properly, but anachronistically, had all the plates arranged to face the canto-beginnings.) Since the plates do have their retrospective qualities, their appearance on the preceding rectos can no doubt be considered a fair second-best; but second-best it certainly is, and so also it seems to have been considered by the sixteenth-century editors. Of the forty-six plates, thirty-three appear on versos facing the canto-beginnings in the Valgrisi and Franceschi editions. In Harington's book, only twenty-seven of them do—leaving nineteen that appear overleaf. Before examining this question further, it is necessary to consider more closely than we have yet done Harington's method of annotating each canto.

V

The critical purposes served by the work of annotation are not our present concern, but rather the use of space. From the point of view of book design, Harington's method of annotation can be seen as an ingenious solution to a practical difficulty. He sought to imitate his Italian sources by publishing an annotated edition, but Ruscelli's canto-by-canto "Annotationi" could supply him with very little material towards his own notes: they are mostly comments on the fine points of Ariosto's use of the Italian language. Harington's solution was to borrow what he wanted from his other sources—the commentaries and the Valvassori prose allegorie [27] —and to transfer all this borrowed material into his end-of-canto notes and also (sometimes) into his marginal notes. All, that is, except the material borrowed for the general allegory and the life of Ariosto, which as we have seen were probably later additions to the book's contents. The omission of prose allegorie at the head of Harington's cantos makes sense as a consequence of this method. He needed the material for the end-of-canto notes, and to transfer it there had the additional advantage of bringing forward the first seven stanzas of a canto on to the first page, instead of four as in the Franceschi and Valgrisi editions, so that—whether consequently or, possibly, of set purpose—there was a better chance that the principal scene of an illustration would face the very stanza or stanzas it illustrated.[28]

Ideally both sets of annotations, Italian and English, should have regularly come to an end at the foot of a recto page, so that the following plate

would appear on the next verso. However, this ideal could not always be achieved, for quite often the annotations end neatly at the foot of a verso, facing the next plate, and sometimes they end too soon, well above the foot of the page. In Harington's book, if the notes ended rather high on the page, a tailpiece was regularly used to help fill up the remaining blank space. Where the verse of a canto happened to come to an end on the page —recto or verso, high up or low down or in the middle—made a good deal of difference to the ease or difficulty of achieving or approximating the desired effect. Where the verse ended high on a verso, the notes would need to be greatly extended to reach to the foot of the next recto; where the verse ended low on a recto, they would need to be either greatly compressed in order to fit, or else greatly extended to fill up the next two pages as well. Where the ideal was unattainable, it was usually possible to achieve at least one of the ideal conditions—a neat finish at the foot of the page, or alternatively the plate correctly imposed on a verso so that it faced the canto-beginning.

The method used by Ruscelli and Valgrisi to adjust the length of the "Annotationi" and to work out their relationships with the following plates is not known. Whatever it was, it worked. Franceschi simply adopted the same layout, but improved on it by adding new copy (presumably also written by Ruscelli) to the notes to cantos XXVII and XXXVIII, so that they too end at the foot of the page in his edition, as they do not in Valgrisi's.[29] In difficult cases, at least one or other of the two conditions for ideal layout is satisfied. There is only one exception to this, at the break following canto XXXVII, and even there the annotations fill up a good two-thirds of the page. In all other cases where much space is left blank, the "Annotationi" finish on a recto, followed by a plate properly facing its canto-beginning. The treatment of the "Annotationi" to canto XXXVI is of particular interest. There the very brief notes could quite easily have fitted below the verse, and were probably originally intended to go there; but they were imposed on the following recto, leaving much blank space at the bottom of both pages (pp. 410-411), presumably so that the following plate would appear on a verso.

In Harington's book, there are three canto-breaks where neither of the conditions for ideal layout are met. In the first two of these, the breaks following Books XVIII and XXV, the notes do fill up most of the remainder of the verso page even though some space is left blank. The third, following Book XXX, is a striking exception to the general rule and conspicuously fails to meet either condition. One suspects a blunder somewhere, and a blunder can be found. It will be discussed in some detail shortly.

There is no doubt at all that Harington concerned himself personally with these questions of layout. We have the evidence of remarks made in

the text of the annotations themselves—mostly grumbles about the lack of space available to him, e.g. at II: Allegory, IV: History, XIV: History, XV: History, and elsewhere—and also of several of his instructions to the printer. Moreover Harington's recorded words can be set in a wider context by examining the results achieved in the book itself and also, where possible, the means used to achieve these results. From all this we can be sure that Harington, as a rule, liked his annotations to finish neatly at the foot of a page, even if it meant making cross-references from one part of his apparatus to another in order to avoid running over the space available for any particular set of notes. How much he cared about the location of the plates (recto or verso) is harder to determine. His recorded words are silent on the topic, so that one is driven to relying on inferences from the book itself and from the printer's copy.

Consider, for example, Harington's instruction to the printer given at the end of the notes to Book XXIX. These notes include two verse-quotations, one Latin, one English, which in the manuscript are both written out in long lines, two lines of verse to one of script. The four lines of Latin verse are each numbered, but not the lines of English verse—it is probable that Harington thought of these latter as three lines of verse with internal rhyme as well as end rhyme, for they are also set out as three lines in Ruth Hughey's edition of The Arundel Harington Manuscript of Tudor Poetry (1960; no. 177). The instruction to Field reads: "Yf thear bee roome enowgh in the page print the verses at length yf not then print them as they are heer written. and yf need bee leave out the latten verse quyte." Harington did not wish the notes to run over on to the next printed page. At it turned out, there was "roome" for the Latin verses to be printed "at length", but the six (or three) lines of English verse remained set out as three. Plate XXX appears properly on the next verso, but there is no means of knowing if this is what Harington was aiming for, or if he merely wished to have a neat finish at the foot of the page.

In the manuscript, the tail end of the notes to Book XXXV is squeezed in above the heading to Book XXXVI. The notes end with a reference to Hudson's Du Bartas, but there is room for only one line of the poem to be quoted. In the printed book the quotation has been lengthened, presumably in proof, by a further five lines. The effect is to strengthen Harington's point and at the same time to bring the end of the notes nearer to the foot of the page. Plate XXXVI, alas, appears opposite, but at least one of the conditions for ideal layout has been met. Similarly, the notes of allusion to Book XXVIII (the tale of Jocundo) are made to fit: "Historie nor Allegorie, nor scant any thing that is good, can be picked out of this bad booke: but for Allusions, they come in my mynd so plentifully, as I can scant tell how to make an end, when I am once enterd into them: Only I will touch one or two, (to fill up this page withall). . . . Plate XXIX, alas, once again appears on a recto.

Where Harington had much space to fill, verse-quotation was a convenient

means of filling it—for example, of whole sonnets by Constable (Book XXXIV) and by Sidney (Books XI and XVI). In such places, also (e.g. Book XVII) Harington will expatiate in lengthy passages of free composition on the favourite topics of his social criticism. However, neither of these techniques seems to have been regularly used for the sake of ensuring that the following plates would appear on versos.

So far, in fact, there is little to suggest that Harington really cared whether the plates appeared on versos or rectos; but the blunder at the cantobreak following Book XXX, I believe, tips the balance in favour of the suggestion that he did. Here moreover we have a closer glimpse into Harington's working methods. The blunder seems to have been caused by reliance on faulty stanza-numbering in the manuscript. Stanza numbers 35-39 in Book XXX had been repeated, so that the last stanza of the canto was originally misnumbered "84" instead of "89"—though, like the others, it was later corrected. When he wrote the end-of-canto notes to Book XXX, Harington must have thought he had more space than in fact was available to him. The notes take up twenty lines of type and would have fitted comfortably under seven final stanzas of verse (78-84); but in fact there were five more stanzas, and here Harington had not left instructions to the printer that the notes were to be cut "yf need bee". The printers duly cast off and followed their copy, and the text necessarily runs over to the top of the next page, ending high on sig. Y1v after only twelve lines of type. It is followed by a tailpiece—the only tailpiece in all the annotations to appear on a verso. Here, at least, Harington seems to have calculated the length of his notes without reference to the not-yet-printed sheets, relying on the stanza-numbering and on his knowledge that the verse was to be set out in regular equal lengths of seven stanzas at a time. If his calculations had been sound, all would have been well and both conditions for ideal layout would have been satisfied.

The miscalculation in Book XXX, as is the way with these things, led to another one in Book XXXI. There, under the twelve final stanzas, Harington tucked in a very brief note on the moral. It is the shortest end-of-canto note in the whole book, a mere four lines of type, and ends: ". . . The rest of the booke hath no new matter, but such as hath bin noted before: and therefore I will end this little space with this short note." Plate XXXII appears opposite, but if the notes to Book XXX had been calculated aright, the whole sequence from Plate XXXI onwards would have been imposed one page earlier, and Harington's "short note" would have appeared properly at the foot of the preceding recto. That, I suggest, was the real reason why he kept the note so untypically short. These two examples, in which it is possible to trace failures to achieve ideal layout back to an observable miscalculation, lead me to think that, in fact, Harington did prefer the plates to appear facing the canto-beginnings. He must have been disappointed that in the finished book so many of his plates appear on rectos, by comparison with those of Franceschi's book.

Just how much further one can trace the consequences of this miscalculation

is unclear. Many of the following plates appear on rectos (Plates XXXIII, XXXV, XXXVI, XXXVII, XXXIX, etc.) but the notes to Book XXXIII are very long and one doubts whether Harington would have attempted to plan ahead beyond them.[30] Several of these canto-breaks are difficult cases where the verse itself ends at the foot of the page, and Harington has contented himself with supplying one full page of annotations. If only they could somehow have been imposed one page earlier or later, the results would have been very much better. However, in Book XXXVI, at least, Harington must have been aware that the following plate was to appear on a recto, for the notes to that book were written after the verse had been cast off. Harington must have seen the notation indicating the position of the plate on sig. 2C6: for he deleted the top line of it and wrote the last few words of his notes around it. Perhaps—who knows?—the following pages had already been printed, and Harington was late in supplying the copy for these notes. Possibly also the notes to Book XXXII were supplied late, for they stop abruptly at the foot of the verso printed page, and a tentative continuation of them in the manuscript is heavily deleted. One suspects that all these imperfections of layout may be related to Harington's absence or irregular attendance at the printing-house during this time. All these cantos are among those for which end-of-canto notes survive with the printer's copy.[31] Such an explanation would account for the much higher success-rate in achieving ideal layout of the plates in the first half of the book.

Book XLVI, of course, is a special case. There the Italian "Annotationi" end neatly at the foot of a verso page, facing the title-page of the "Cinque Canti", not translated by Harington. Harington's notes, unconstrained by plate or title-page, continue to the top of the next recto, where they take up six lines of type. They are followed by the line: "Here end the notes of the 46, and last Canto of Orlando Furioso"—a phrase that recalls the running-titles of that canto in the Italian editions, "CANTO QVARANTESIMOSESTO, ET VLTIMO". In the manuscript there is the most famous of all Harington's instructions to the printer:

(The remainder, concerning choice of type for preface, general allegory, and "all the prose that ys to come", was quoted earlier.) What exactly Harington had in mind when he asked for "some prety knotte" is impossible to determine. It sounds as if he were perhaps thinking of another tailpiece. At all events, Field's workmen used their common-sense and, very appropriately, chose a large headpiece to mark this major division in the text—the same headpiece as would be used for the dedication to the Queen.

The canto-break following Book XXIII is also a special case, and rather a remarkable one. It deserves to be discussed at some length. This cantobreak marks the mid-point of the poem, for Book XXIII ends with the story of the onset of Orlando's madness, the centre-piece of Ariosto's poem and the heart of what Harington called "the chiefe Allegorie of all the booke, and where-vpon the booke taketh his name Orlando Furioso" (XXXIX: Allegory). Harington made a number of special adjustments in his apparatus to make sure that his readers understood the importance of the episode. This was necessary because the madness of Orlando was not by any means the principal topic of Italian critical debate about the poem. Fornari in his Spositione, the critic from whom Harington borrowed most extensively, devoted the second of his two volumes to an explanation of the allegory, but far from interpreting the whole poem he concentrated his attention on the story of Ruggiero and Alcina and its ramifications. The same emphasis on Ruggiero and Alcina is seen in Bononome's "Allegoria", and also in Harington's general allegory, despite the adjustments made to the text in an attempt to give prominence to Orlando. The end-of-canto notes to Book XXIII are rather compressed by lack of space, and allegorical interpretation of the episode is deferred "till I come to restoring of his wit againe: which I count more proper for this subiect", but Harington did find space in the notes of allusion for an interesting comment on the artistic, literary, indeed Aristotelian qualities of the episode, on its psychological truth and its power to move. As we saw earlier, this note was taken—adapted, that is to say, not merely borrowed—partly from Fornari and partly from Lavezuola, and the argument to Book XXIII was also adjusted in exactly the same spirit. In all these various small ways, then, Harington can be seen to be altering his critical material in an attempt to redress the balance, and to express his own reading or vision of the poem; but commentary alone, however teased and twisted, was not enough.

In the Italian editions, the design-treatment of the canto-break following canto XXIII is not distinguished in any way from the others. However, Field was given the highly significant authorial instruction: "Between the xxiijth booke and the xxiiijth I would have a spare leafe . . ."—maddeningly, the sentence is unfinished because the rest has been trimmed away in binding.

Yet, instead of a "spare leafe", the printed book has a blank page-opening. Harington had forgotten that the plate for Book XXIV should appear on a verso, facing the start of the next canto, and that therefore his "spare leafe" would necessarily have been accompanied by further blank pages to either side; unless, that is, he chose to find some means of filling them—say, by extending the notes to Book XXIII on to the following verso and by concocting some sort of a title-page for the recto preceding the second half of the poem. Harington chose otherwise, and accepted the blank page-opening.[33]

That there were some uncertainties in the printing-house about what was going on is clear from the irregular four-leaf gathering at the end of Book XXIII and from the confused page-numbers throughout Book XXIV and at the start of Book XXV (i.e. all the next regular gathering). In the manuscript, the entire sequence of printers' notations between P.12/178 at XXIII: 8 and S.11./209 at XXVI:36 is in one way or another deficient or incorrect. The confusion in the printed book, however, is much greater here than in other places, where there are also lengthy sequences of missing or erroneous printers' notations in the manuscript. The picture that emerges is a pretty illustration of the relations between one printer and his author at the end of the sixteenth century. Harington knew what he wanted; Field knew what was practicable; together they reached a compromise solution, and in doing so caused a good deal of trouble to the workmen.

In other places, somewhat similar means are used to draw attention to a fundamental division at the mid-point of the poem and to the importance of Orlando's heroical—or perhaps we should call it heroi-comical—loss and recovery of his wits. Below the notes of allusion in Book XXIII the customary line appears, but the wording is different. Instead of something like "The end of the annotations vpon the xxiij. booke", which would have been typical, we read "Here end the first xxiij. bookes of Ariosto", and there is adequate but not excessive room below for a tailpiece. The contents-list similarly divides the poem into two halves, leaving a large gap, a thick white line, between the two entries, thus:

One hesitates to place much emphasis on such small points as these, but they are clearly done with a purpose—the same purpose as prompted Harington to ask for his "spare leafe".

It might show less than perfect balance on my own part to descant at any great length on the significance of these two blank pages at the heart of

Harington's book. Yet they are rich in significance—multiple rather than single. On consideration, the blank page-opening can be seen to be an improvement upon the "spare leafe" Harington originally asked for, rather than a mere compromise solution. Here the reader does indeed bend his eye on vacancy. The vacancy within Orlando's skull is matched by the vacancies in the ranks of Charlemagne's paladins, and just as Orlando loses his wits at the end of Book XXIII, so Bradamant at the start of the canto loses her way and is unable to return to Rogero. The blank double-spread accentuates the canto-break so that the "first xxiij Cantos" are distinguished from the "other xxiij Cantos" that are to follow; and the break in the narrative texture affects all three of the poem's main story-lines. The pause has become an interval, a deep breath at the mid-point of the poem.

The interesting thing about all this is that Harington had no Italian precedents for these features of the typographical design, or rather that he imitated Italian precedents and then took them one stage further. He followed Ruscelli, Valgrisi and Franceschi in making the annotations fit the space available for them on the page, so that the great majority of the pages in his book are printed from top to bottom of the printing surface. Only if this norm was clearly discernible would the two blank pages at the centre—for which there was no Italian precedent—have significance. Similarly he followed the 1556 Valgrisi edition (or one of its reprints) in maintaining an alternating sequence in the ornament of the canto-beginnings, and only in the context of that regularity does the change of step at Book XXIV also acquire significance. In the arrangement of the front and end matter and in the choice of type for the preface, general allegory, and life of Ariosto, he contrived a consistent design and sense of orderly sequence in the English book that are quite lacking in the Italian editions, yet wholly in keeping with the rhythmical presentation of the poem itself and its accompanying material; and the printing of the poem itself in roman, not italics, was just one of the many ways in which Ariosto's stanzas were tactfully "tempered to the obstinate humours of the island".[34] There could hardly be a better example than Harington's Ariosto of "triumphal form" in which the "centre" is "finessed", i.e. treated negatively; but the means used to achieve the effect were not the line-counting methods of numerology—which (it seems) interested neither Ariosto nor Harington—but rather the silent poetry of book design.[35] Harington expressly admired the shaping of Ariosto's poem, praising among much else its first and last lines and remarking in his life of Ariosto how "absolute a peece of worke" the poem is and how "euery matter" is "brought . . . to a good and well pleasing conclusion". He equally admired Ariosto's skills as a story-teller, recognizing his ability to "draw a

man with a continuall thirst to reade out the whole worke" (Preface). He expressly declared the story of Orlando's loss and recovery of his wits to be the "chiefe Allegorie of all the booke", and he retained Ariosto's title as part of his own. We may fairly conclude that in the shaping of his book he wished to do justice to the shaping of Ariosto's poem as he understood it.

No doubt, it is also a fair assumption that Harington wished to dress his book in the best contemporary Italian manner—to be sophisticated and up-to-date. Here, however, he was less successful. Considering together the nineteen plates that appear on rectos, the layout of the verse in seven stanzas to a full column instead of six, the omission of the prose allegorie at the head of each canto, the use of printer's ornaments in lieu of engraved designs in the borders for the arguments, the substitution of the sometimes troublesome "broken . . . discourse" (XVII:Moral) of annotation for commentary, and even the very choice of a folio rather than quarto format, one must acknowledge that Harington's book is a more thrifty production than Franceschi's—however lavish it must appear compared with other contemporary English books of poetry. Certain design problems were less successfully resolved in his book, and indeed, in one respect, the design of Harington's book was already out-of-date when it appeared in 1591, for in the previous year the Italian publisher Bartoli had brought out an edition of Tasso, also modelled on Franceschi's Ariosto, but with the design modified so that all the plates were made to face the canto-beginnings (Mortimer, no. 494). Of contemporary and early seventeenth-century English books, only Speght's Chaucer of 1598 and 1602 and Sandys' Ovid of 1632 may possibly owe certain features of their design to Harington's example.

To what extent—finally—should a modern edition follow the 1591 design? I hope I have shown that McNulty's edition, admirable though it is in giving us an entire version of the text (poem and apparatus) and in reproducing all the plates, nevertheless obscures Harington's meaning in certain small but crucial ways by failing to respect his design-intentions. The arrangement, typography, layout, and ornament, are all freely altered. His edition has been much discussed recently, but the question of book-design has not been seen to be part of the editorial problem, and it is worth considering what design-treatment of a modern edition would best fulfil Harington's intentions. Not, I suggest, a facsimile reproduction: Harington wished his book to be read and enjoyed. A re-set edition is called for, and let us assume (for argument's sake) that it will be in two volumes in paperback. There is room for considering a variety of design-treatments, but this is what I would recommend. The sequence of the contents from start to finish should be respected. The typography should follow the 1591 edition in its use of roman and (a calligraphic) italic, and in making appropriate discriminations in type-size, but the verse might well be in the same size of type as the arguments and the notes smaller. The page-layouts would necessarily be different from those Harington refers to in his text. The ornament of a paperback edition would need to be modest, but present; in particular, it should follow the system in Harington's book of an alternating

pattern in the ornament of the canto-beginnings, with a change of step at Book XXIV. The small page-size of a paperback edition would probably make it impossible to reproduce the English engravings; the most appropriate substitute would be the small Giolito woodcuts, which Harington is known to have borrowed for the Bodleian Library manuscript. Some, no doubt, might argue that the proposed treatment would result in a mere pastiche, a guess-work adaptation from an imaginary Elizabethan quarto; but I believe it would be closer to the spirit of Harington's enterprise than a plain text. Moreover, it would be fun to read, and books ought to promise the reader fun.

In all this, we need to remember what Greg well knew, and what any poet or playwright knows, that the substance of literary invention lies in the "Idea or fore-conceite of the work, and not in the work it selfe".[36] An editor's duty is to communicate that "Idea or fore-conceite" as fully and as faithfully as he can. His procedures will be determined by the particular circumstances of each case, and may according to those circumstances require close adherence to a single model or radical reconstruction from diverse sources. Among the sources that need to be considered is the design of the printed editions (or, for that matter, of a carefully-prepared presentation manuscript—but that is another story); and editorial treatment of what we may call the "accidentals" of book design needs to be informed by the same good sense, rational method and historical understanding as we expect in the procedures of textual criticism.

Notes