While textual critics of medieval and classic texts require expertise in paleography, those studying contemporary works, and in particular contemporary multimodal works such as musical theater, require a similar sort of facility with digital forensics. In this paper I will detail the digital forensic techniques (and a few dumb hacks) that I have used to create a detailed biography of the musical, RENT, from it’s earliest its earliest incarnation in 1989 to the final version edited by Larson and saved just two weeks before his death in 1996. I will also suggest a set of skills those attempting similar work may wish to acquire.
Jonathan Larson, the Broadway composer and lyricist, used his personal computer as a creative tool throughout his brief career from the mid-1980s to his premature death at 36 in January of 1996. After his death, around 180 of the floppy disks he used to save his work were donated to the Library of Congress. In 2007, I was permitted to create complete, bit-for-bit, data images of these disks and study any file related to the creation of Larson’s most famous musical, RENT. These disks contain several dozen drafts of RENT.
In some cases, multiple drafts are recoverable from a single file. In 1992, when most personal computers ran only about 5% as fast as an iPhone 4, saving an entire file to a floppy disk was a potentially time consuming activity. While writing RENT, Larson mostly used Microsoft Word 5.1, which had a feature called ‘fast save.’ The fast save feature sped the saving process somewhat by replacing the entire file only once every 14 saves or so. In most cases, it would simply append changes to the end of a file with information about where they belonged in the original document. When the file was opened in Word 5.1, the software would integrate these changes back into main text; however, by opening this file with a simple text editor, it is possible to see the text of the last full save along with all the emendations made during fast saves in an apparently uninterrupted text stream stored at the bottom of the file.
Unfortunately, the Word 5.1 standard has no openly published documentation. Thus, while additional text stored by a ‘fast save’ are visible, the way in which this text maps onto the ‘base text’ is not immediately obvious. Fortunately, with the help of Microsoft Research, I have obtained documentation that describes the standard (unfortunately provided under a non-disclosure agreement), which describes how this mapping was accomplished. Therefore, although this paper will not include a complete description of the specification for legal reasons, it will demonstrate how several versions of a Larson’s text might be reconstructed from a single Word file.
For example, throughout 1992 there are several files and folders on Larson’s disks that contain versions specially formatted for the laser printer. In 1992 Larson generally used a ‘dot-matrix’ printer that worked by pushing ink from a small typewriter-like ribbon onto paper in a series of small dots that composed the desired letters or the image. The printer would advance the paper through a spool in increments as tall as the printer ribbon as each line of dots was printed. Although this method of printing was relatively inexpensive and was capable of representing image that could be composed out of the dots (rather than, say, a fixed set of letters and numbers), it produced copies that, even in the 1990s, appeared amateurish. There were a finite (and relatively small) number of dots available on any line which made text appear jagged, and ink would transfer from the ribbon inconsistently, creating a faded, uneven look. Larson was content with this technology for rough drafts, but when he needed a copy to give to a potential producer, he would print the text on a laser printer (perhaps at a copy service store).
There was a complication, however. Text formatting conveys a good deal of meaning in musical theater libretti. Stage directions are often represented alongside spoken text as typed marginalia; overlapping lines sung in unison or counterpoint are represented in different columns. At this period, the relatively low resolution of the screen Changes in typeface or margins can change the line lengths and create wrap-around text where a hard break was intended creating an amateurish appearance that the use of the laser-printer was supposed to prevent. Larson’s usual font ‘New York,’ like many of those designed for early versions of the Macintosh operating system, was a ‘bitmap’ typeface, meaning that each character was stored as a small image. Just as a small ‘thumbnail’ image downloaded from a website becomes fuzzy or ‘pixelated’ when it’s size is increased much beyond the original size, the size of a bitmap font cannot be attractively increased beyond the original size of the character images. Moreover, the quality of the characters themselves – how smooth or jagged they appear – will not change even if the monitor or the printer quality improves. ‘Vector’ fonts, like the ‘Courier’ font used in Larson’s ‘laser’ drafts, on the other hand, are generated not from an image file but from an equation that describes the lines and curves of the characters. The result is that the characters can be smooth as the monitor or printer can make them and can be scaled up to any size without loss of quality. Moreover, ‘Courier’ was designed as a ‘monospace’ font – that is, each character occupies the same amount of space (an ‘M’ is the same width as an ‘I’). ‘Monospacing’ makes the process of aligning text much easier than it would be with fonts of variable width as one can be confident how many characters will fit on a line (regardless of what the characters might be). Thus, for those drafts which needed to look especially good on the page as well as on the screen, Larson would reformat his text to meet the requirements of the printer.
Thus, there are folders on Larson’s disks with titles like ‘Rent Laser’ that contain versions of the script formatted for higher quality printing. By consequence revising the script during the rehearsal process, when all copies should be printed on a laser printer, was difficult. In some cases, probably during the rehearsal or preview period when it was necessary to distribute frequent changes to a cast, Larson found it preferable to print only revised passages on a laser printer and then to literally cut and paste these revisions onto an older laser-printed version and photocopy the result. These pasted-together drafts, preserved in the Library’s paper collection, exemplify the awkward hybrid of centuries old print traditions with the advantages and quirks of new media and serve as a strikingly appropriate metaphor the history of the text to which they bear witness.
Of course, RENT, is hardly the only text for which both digital forensic and bibliographic expertise will be needed to conduct a contemporary critique génétique. Indeed, editors and textual scholars of most work composed after (at the latest) 1995, are likely to encounter digital drafts among their witnesses. This paper is intended to serve as a kind of early exemplar for the kind of techniques a contemporary digital textual scholar must learn and practice.