jmcg (10/29/82)
On our campus there's a medieval historian for whom computers, specifically using `troff' under UNIX, have begun to make an important impact in his work. An important activity for historians is tracking down various surviving versions of texts and "merging" them into a definitive edition. This has several benefits, aside from the purely intellectual value for the person doing it: the contents of the texts, once published, are now available to other scholars for MUCH less than would be required to, say, travel to Europe to visit the several libraries that have extant copies. The texts themselves are now likely to survive longer, since most scholars can get what they need from the published text rather than having to handle the originals. And the historian who publishes it stands to have an enhanced reputation. Historians seem to have a general fondness for well-produced books. It is not unusual for a book to spend 6-12 years "in press" (thus not serving any of the beneficial purposes mentioned above). Part of the slowness is the fact that letterpress is slow, but most of the time appears to be taken up in extraordinarily detailed typesetting (different types of commentary on the text are displayed in separate footnote areas at the bottom of each page) and painstaking proofreading. Using troff, this historian has achieved some rather dramatic results. He was able to produce camera-ready copy (well, he had to paste-up a couple of things) for his book in about a year. The typesetting was accurate enough to satisfy the (very fussy) people at the Vatican Library (it has many fewer typographical errors than they normally must be satisfied with even after 10 years of proofreading) and the published book is available for considerably less than half the price of comparable works. Its main drawback is that it had to be reproduced by photo-offset. Two other areas in which computers are applied to history are probably also worth mentioning: the first is the general area of text analysis. Producing a useful concordance for a corpus of text is a lot more difficult than just running `index': there are a lot of words that you don't care very much about, and many of the ones that you DO care about have multiple meanings whose occurrences must be separated out. The word list must be coordinated with the typesetting so you can give the reader a way to find instances in context (title, section, line number). The other area is probably familiar--using statistical packages like SPSS to crunch unwieldy collections of data into something from which conclusions might be drawn. ------------------------------- Let's hear something from UCSD's Center for Music Experiment (sdcarl)! Jim McGinness UCSD Chemistry (714) 452-4016 sdcsvax!jmcg (Remember, San Diego's area code changes to 619 on 6Nov82.)