Re-editing an inscription from a poor-quality scanned estampage
Ideally, digital re-editions of inscriptions should be based on a fresh look at the original or a high-quality surrogate. But what do you do when all you have at hand is a previous edition and a facsimile that has been scanned in rather poor quality? To put it another way, how important is a superb surrogate for a previously edited text?
There seem to be three levels on which the creation of digital editions is advantageous. On the most exalted level, a novel edition by someone who fancies himself a competent and diligent epigraphist may be able to achieve new insights and to correct old mistakes in the way a text has been read. On a much more mundane level, encoding a text in EpiDoc affords us a granularity of representation that is rarely achieved in printed editions. And on the most down-to-earth level, the mere fact that a text is encoded in TEI and published on the Web makes it possible for that text to be accessed and studied far more easily than a printed one, and to be processed by computers in far more meaningful and sophisticated ways than a mere e-text replicating the printed edition.
The lowest of these levels can be achieved simply by encoding a printed edition as given, replacing bracket-based markup with XML encoding but without comparison to any kind of visual representation. On the higher levels, even a poorly scanned inked estampage helps improve the detail and accuracy of the edition to some extent, and to catch and correct some oversights of the previous editor. But obviously, the more detailed and accurate the visual representation available to you, the higher your chance to do so. As for the holy grail of new insights, such things are rarely found by looking closely at something others have looked at closely before.
To be sure, it is a quintessentially scholarly experience, and a rewarding one, to spend hours looking at a badly damaged inscription and finally to puzzle out one character that nobody else has read in centuries, or to be able to pronounce authoritatively that a great predecessor in fact misread a character. But such highs are uncommon, and ultimately most of them contribute less to the sum total of human knowledge than the simple drudgery of digitising someone else’s edition to make it accessible to number-crunchers. After all, most of the scholars who published the editions now available in print were themselves competent and diligent epigraphists who, moreover, normally had at hand either the original item itself, or the inked estampage in its physical reality. Exceptions do occur, of course. We have all seen published work that failed to meet expected standards. Such work can often be recognised even before collation with a facsimile, and even a scan of modest quality can prove most of the rest wanting.
So, let us assume you are working from a published edition of high standard, coupled with a poor, but not altogether useless, visual representation of an inscription in less than pristine, but reasonably decent condition. What, then, do you do if the prior edition prints a certain string of, let’s say, eight characters as clearly read, but in your scanned rubbing you can only see two of those clearly, along with some indications of six more, and no discernible trace of the remaining two? There’s a fair likelihood, probably upwards of fifty percent, that the chap who wrote that edition could really discern those characters clearly. Perhaps another forty percent that he exaggerated his confidence, and at least some of those characters are in fact unclear in the original, even if their reading is confident in the context. And there is a sliver of a chance that the previous chap was both overconfident and wrong about one detail or another. Now what can you do about this? The reason why you can’t make out all the previous chap could, might be the lack of resolution, jpeg noise, or aggressive image enhancement in your scanned image. If so, scrutinising the printed estampage may help you honestly encode a reading as unclear even though the previous chap did not admit that forty percent of the time, and may even give you a shot at catching one of those wrong details and correcting it. But if that print is not already on your bookshelf, then the effort to obtain it, even in times not afflicted by Corona, may not be worth that slight chance of improvement. The law of diminishing returns clearly applies. Honesty and accuracy in encoding the fact that some parts of some akṣaras are unclear in the original is well and good, but very rarely a crucial point in the usefulness of the digital corpus, and the chance that a fundamentally reliable original edition was wrong in a particular detail is generally minuscule. And of course, it may be the case that even the printed facsimile is far inferior to the tangible and textured estampage (or the original object) that the original editor worked from. What you then need to be able to verify his claim (or call his bluff) is not an interlibrary loan but a field trip, an investment at a whole new level of magnitude.
All in all, to achieve the best profit-investment ratio, I would recommend as a rule to accept the claims of a competent previous editor even if they are not confirmed by the facsimile available to you. Ultimately, it usually matters little whether the visarga your predecessor printed as clear in the text is in fact unclear, or not there at all. That said, caution is of course due in some cases. The previous chap’s reading may be counter-indicated by your facsimile; some insight of yours, based on your understanding of the context and the advances made since the previous editor’s time, may imply a different reading that seems likewise possible from your facsimile; or the locus itself may be crucial for one theory or another. In any of these latter cases, a note can be added to your file next to the adopted reading, pointing out the need for further verification. After all, a digital edition lives and evolves, and unlike a printed one, can be improved at any later time when you have made that interlibrary loan or that field trip.