Uninscribed Meaning:

VRML Applications in Electronic Editions of Medieval Texts

Stephen H. A. Shepherd


It seems to me that much of the greatest potential in electronic editions lies in the way in which they can present and manipulate images. I have spoken to representatives of a few publishers about the possibility of their publishing electronic editions of some of their more popular or useful texts, and their responses tend to be rather uniform: most inevitably respond that text is harder to read on a computer screen than it is on paper, that books are still more portable than computers, and that computer hardware simply is not keeping up with developments in software (and never will catch up). And then there is that little problem of copyright.... Rarely in such discussions, moreover, do you find publishers willing to acknowledge the true advantages of tag-ability and the consequent search-, concordance-, and collation-compiling ability of electronic texts. 
The one thing they usually do not dismiss, however, is a recognition that the electronic medium can present more color images more cheaply, more variously, and often with better quality than any photograph in a book ever could. CRT and active-matrix LCD screens provide as much as a tenfold increase in luminance and color depth over a printed image--it's like viewing slides (and slides scanned into computer graphics formats simply glow on-screen). 1 And, as Hoyt N. Duggan notes in his on-line prolegomena to the Piers Plowman Electronic Archive at the University of Virginia, a digitized image of a MS can offer a reader "access to a kind of textual information once available only to readers with access to major manuscript collections." 2 With the ability to magnify portions of digitized images, to alter contrast, brightness, color-balance, to enhance the representation of edges and textures, it is possible "to recuperate erased letters and in some cases to determine what lies beneath passages of overwriting" (ibid.). 
A great advantage of the digitized photographs of MSS is thus clearly their potential for improved paleographical and textual analysis; but this need not be where our conception of MS imaging stops. For some year s now one MS in particular has struck me as crying out for a number of different forms of electronic imaging; only recently (since just before Christmas of 1995) has the technology--that is to say accessible and reasonably intuitive software--start ed to offer what look like viable solutions. By "accessible" I mean that the software will run on a MAC or PC and works with established file formats and mark-up languages; I think it is important that we seek out software which can be operated on portable platforms--such as the 486 laptop I have used for demonstrations of the software I discuss later in this essay--so that the digital "proxy," so to speak, can be taken, if need be, back into libraries and used alongside the very MSS it represents and analyzes . Given the regrettably declining status of evidence in current literary studies, I think it important, however liberating and demystifying electronic texts and their images may be, that the new medium have built into it some mechanism which invite s the test of its own authority--what are we going to do if we end up with a "Kane-Donaldson" digitizing contesting a "Schmidt" digitizing? 
The manuscript in question is Bodleian Library MS Ashmole 33. It contains one text, that of the Middle English metrical romance of Sir Ferumbras. This is not a particularly distinguished poem from a literary point of view, but the conditions of its preservation in manuscript are fascinating. To quote H.M. Smyser, 

the manuscript... is of considerable interest. It is in one hand throughout and is an autograph, like the Ormulum, the A3enbite of Inwyt, the Peterhouse Equatorie of the Planetis ... and a few if any other important Middle English manuscripts. It is bound in a triple envelope of parchment consisting of two [Latin] documents of minor importance concerning the diocese of Exeter; the outer is dated 1357, the inner (folded double...) is dated 1377. On the back of the inner cover is an original draft, in the same hand as the manuscript proper, of lines 331-759 of the Ferumbras, and elsewhere on the covers an original draft also of some thirty scattered other lines. [H.M. Smy ser "Charlemagne Legends," in J. Burke Severs, ed., A Manual of the Writings in Middle English, Fascicle 1: Romances (Cambridge, Mass., 1967), p. 85] 

For anyone who has had to struggle with the only published edition of the poem, done by Sidney J. Herrtage for the Early English Text Society in 1879, an electronic edition is a thing devoutly to be wished. The quality of the transcription is very high, but the editorial apparatus is unequal to the complexities of the MS: not only is there a draft to edit as well as the fair copy, but both draft and fair copy contain erasures, palimpsests, strikeouts, and interlinear and marginal additions and amendments. Herrtage prints the text of the draft below that of the fair copy rather than beside it, which would be the more useful location for comparison. Erasures, palimpsests, and strikeouts almost always go unremarked; the actual location of interlinear and marginal additions or amendments is not given (and in some cases Herrtage inserts such corrections in the wrong place, despite the scribe's very precise use of pointers and joining lines; this is probably due to the fact that Herrtage made his edition from a transcription made by "Mr. George Parker, of the Bodleian Library" (EETS, Extra Series No. 34, p. xxvi). 
In an electronic edition, the draft and fair copy would of course be hypertextually linked (on a large enough monitor both texts could be viewed simultaneously in separate scrolling frames) and different flags (eve n different colors) could be inserted for erasures, palimpsests, strikeouts, and interlinear or marginal amendments and/or additions. The advantages of these features would be not just to aid clarity of reference, but would perhaps also offer insights into the scribe's compositional methods. For not only is the scribe a copyist, he is also a translator working from a French original, as well as an author, constantly (and sometimes very subtly) making changes to the received story (though sometimes he will make his text closer to the French in his fair copy than it was in the draft).3 He also changes spellings from draft to fair copy--and to my knowledge linguists have yet to look into this; no doubt the only available edition has put them off. A few years ago, I made a brief foray into looking at this MS and its text from a computerized point of view. Malcolm Parkes had posed the question of whether the scribe's corrections conformed to a concerted pattern o f revision or whether they were merely "roving" revisions done on a whim and in no particular order. Looking carefully at the MS I created a database (in Lotus) of the number of revisions which occurred on each folio side of the fair copy and generated a graph. My conclusions constitute the standard academic cop-out; there do seem to be patterns of concerted revision centering on certain kinds of passages and increasing as the text nears its end, but this is done against a background of widespread roving revision. The analysis can be improved upon; the graph does not distinguish between erasure, palimpsest, interlinear and/or marginal emendation and/or addition, while a properly marked-up electronic edition would be able to generate a more sophisticated series of graphs for each of the scribe's revising entities. It would be an advantage to have such an analysis because of the fact that the scribe seems to have revised what was apparently at one point meant to be a fair copy in and of itself. At what point did it become yet another draft? Presumably erasures and palimpsests, even interlinear corrections, belong to a concern to keep the fair copy looking (reasonably) "fair," whereas marginal additions and corrections might belong to a period after the decision was made to produce another fair copy. 
The scribe's revisions are not the main topic of this essay, but I did want to introduce the notion that we need to be as specific as possible about characterizing the way in which a scribe marks up the space provided him on each page of a manuscript. I have set as the main title of this paper Uninscribed Meaning; that is something of a misnomer inasmuch as all space in a manuscript (unless it is completely blank) is in effect "inscribed"; that space which is written on determines the space which is not--and we ought perhaps to include here the orientation of the written and unwritten space, whether it is in columns or rows, vertical or horizontal, and whether text is written at a certain angle in opposition to other text (as in a line written vertically in a margin). I would like also to include the notion of what might be described as "dis-inscribed" space, regions of the manuscript where text has been removed or blotted out. There may also be occasions when we ought to take note of folds or creases in the vellum or parchment and additional inserts (the Ormulum in particular comes to mind) if it seems possible that such information might tell us something about the "life" of the book, the conditions of its construction and use. 
With several of the kinds of features I have identified, special mark-up flags will be of little use--about as much use as written descriptions of, say, how to tie a shoelace. Recall Smyser's description of the binding of Ashmole 33; just what is a "triple envelope" and how do you get one from only two parchment sheets? If the inner cover is folded, which side is its "back"? And just how are the "thirty scattered other lines" actually "scattered" across the two parchment sheets? My solution to this problem when talking about the manuscript to others is to use the drawings as well as a essay facsimile (upon which the drawings are based; the drawings are reproduced in the article in The Medieval Translator, cited above). I did not, however, make the facsimile only for this purpose. Ashmole 33 is now preserved in three parts; the original binding was disassembled sometime in the 19th century and the paper book that is the manuscript proper (the "f air copy") was given a new hardcover binding. The parchment wrappers (no doubt kept not because they had Middle English verse on them, but because they had Latin text on them) are now kept in a separate box. So preserved, no care has been taken that the original direction of their folds be retained or that they be kept supple, with the result that they are now hardened, prone to cracking, and are curled up in precisely the opposite direction to that which they originally took when enclosing the paper book. Not surprisingly, moreover, most of the surfaces of the wrappers have been subject to considerable rubbing and wear, and so, even if one is able gingerly to pry their folds apart, it can be very difficult straightaway to locate the "odd thirty scattered lines" of Middle English draft which exist beside the main four columns of draft. The drawings help to establish the organization of the "triple envelope" but they do not account for the relative location of all portions the Middle English draft. With the paper facsimile, however, the location of the portions of draft is clearly laid out, and the relative location is evident in three dimensions
With the paper facsimile I have since been able to make a number of additional findings about the scribe/author/translator's compositional methods. The so-called "scattered lines" of Middle English draft occupy twelve areas, all of which exist in the margins of the Latin documents and all of which in fact contain the Middle English lines in orderly columns of between two and nine lines' length. Each column is separated by a vertical line (the edges of some of the columns have been trimmed in the assembly of the binding and would originally have been longer; there are seven columns on the outer wrapper and five on the inner wrapper). Encouraged by the impression of orderliness which was emerging from the reconstruction afforded by the facsimile, I decided to try to establish some readings. Most of these short columns are very badly worn, and sometimes only as few as two or three letters per line are discernible. Two of the seven columns on the outer wrap per have proved illegible, but from the other five, and the five on the outer wrapper, a pattern emerges. Making allowances for the trimming of some columns, it appears that each is contiguous with its neighbor. Thus the five legible columns on the outer cover provide draft corresponding to lines 232-278 of the fair copy. The five short columns of the inner wrapper provide draft corresponding to lines 746-825 of the fair copy. At the tops of these documents the elongated decorative ascenders of the letters of the first Latin line have been erased to make way for the English text; this is an important fact, since it suggests that the scribe did not have an unlimited supply even of used parchment. 
As I mentioned earlier, the main body of Middle English draft is found on the inside of the inner parchment covers; the four columns of draft there would, before the sheet was tailored to fit the binding, correspond very nearly to ll. 331-759 of the fair copy. If one takes into account a quantity of draft lines lost to the missing portions of the covers, and the fact that all of the ten legible columns stand in a contiguous relationship to their neighbors, it is highly probable that the two parchment sheets represent nothing more than consecutive sheets in the draft. They certainly are not just "scattered other lines". That there is some overlap (of lines 746-759--just 13 lines) with the draft of the other side o f the same inner parchment sheet, combined with the evidence of a miserly use of parchment, suggests that there was little more draft of this kind than what has survived. This argument is reinforced if we look at the graph I discussed earlier, where it is evident that less revision is going on near the beginning of the fair copy than nearer the end. Perhaps, then, what the wrappers reveal is that the scribe/author/translator just needed draft to get the ball rolling and then rendered the rest of his work as fair copy without need of a draft (although, to judge from his later revisions, it may have dawned on him that this method was less than satisfactory). 
To return to the issue of creating digitized representations of manuscript space, my experience with Ashmole 33 suggests that an accessible, shareable, and portable electronic means of constructing a three-dimensional facsimile would be very useful. The facsimile would in the process of its construction be useful as a research tool in its own right as well as subsequently supplying a guide for other scholars. In order to view the facsimile a "plug-in" for one's Web browser is required. I recommend Cortona by Parallel Graphics; it is a self-installing freeware product which handles newer and older formats of VRML code. With such a plug-in one can use the mouse and its buttons to move up to, away from, and around the virtual objects; if you encounter, say, a virtual room, you can move around inside it.

I have constructed two VRML models, one of the inner cover and one of the outer cover of Ashmole 33: 

View the VRML model of the Inner Cover 

View the VRML model of the Outer Cover 

One thing which becomes clearer with the VRML facsimile of the inner cover is that the scribe has oriented the top of his longest columns to coincide with the top of the Latin document on the other side; he is nothing if not regular and methodical. 
Now to the creation of such VRML models. First, you can see the source code by selecting the "View/ Document Source" menu in Netscape. For the inner cover facsimile, you will see reference to a VRML file named INNERCOV.WRL. This file is named and called into the HTML file with an "EMBED SRC" command, which behaves very much like an "IMAGE SRC" command and which can thus make the window just about any size you like. Click here to see the source for INNERCOV.WRL. To generate that complex code you need a VRML authoring tool; the one I used  was called "Fountain", produced by Caligari Software;  it was later called "Caligari Pioneer" and is now called TrueSpace.  One crucial feature of  the program is that it allows you to place imported graphic files--they can contain photographs or drawings--onto the surfaces of VRML objects as "textures". My facsimilies were made by mapping drawings I had done in Microsoft's PC-Paintbrush onto the surfaces of the planes which form the model. It is conceivable, then, to construct a facsimile of a book or MS which actually has pictures of the book as its surfaces. It is also possible to have transparent or translucent surfaces in VRML, in which case 3D representations of MS collation can replace the two-dimensional nested "V's" of traditional collation diagrams; and, if the collation is of a paper MS, then the precise location of watermarks for each folio can be depicted. 
One promising development in VRML is the provision for animations; that is, a VRML object can be programmed to move through its world in a certain way. Conceivably a book could be shown opening and then turning to another page, though the animation may have little practical use to someone wanting to "handle" the model in their own way. What may be better is to configure different surfaces of a VRML object as "links" with other files; I can, for instance, have a model of a closed book upon which one can click with the mouse and which will then replace itself with another VRML model of the book opened. Animations can, moreover, be rendered in only two dimensions in simpler ways; there are any number of programs that will allow you to do this. Both of the pages above with VRML models of the covers of Ashmole 33 have a link to an "animated GIF" of the disassembly of the binding. The animation employs 35mm photographs I took of my paper facsimile. In this instance the animation is somewhat redundant against the drawings I have published, but, in the absence of someone with the talent to do such drawings, a series of animated photographs seems more desirable. 
I remember that one of my favorite programs when I had just gotten started with PC's was "Chuck Yeager's Advanced Flight Trainer"; at the time it was a fairly advanced reworking of the concept of the flight simulator. I suppose what I am talking about now is a kind of "Book Simulator"--perhaps it could be developed as the "N. R. Ker Book Simulator" or the "E.T. Donaldson advanced codex simulator." When you crashed in Chuck's simulator (I used to forget to lower the gear of my F16 upon landing), an image of Chuck appeared and said something like "Well, you really bought the farm on that one, rookie." I'm not sure that a book simulator will reach that level of sophistication (if we forget to distinguish between recto and verso, an image of Langland's Holy Church pops up and says "Thow doted daffe!...dulle are thi wittes!") but I think that it is just this kind of programming application that can advance the unique status of electronic editions. 


Stephen Shepherd, 
Associate Professor, 
Department of English, 
Loyola Marymount University, Los Angeles