The Computer's Tale

Tuesday, December 01, 1998
Biochemist Christopher Howe of Cambridge University took a break from his lab one afternoon to attend a lecture on medieval manuscripts. "We were shown a slide of something called a stemma, which looks at the changes in manuscripts as they're copied, to produce a tree of relationships. And I suddenly thought, ëGosh, this is exactly what I've been doing all day long.'"

Although Howe studies the evolution of various types of bacteria, algae, and plants, he realized that the computer programs he uses to construct bacterial family trees might be perfect for tracing the origin of old texts. The errors made by scribes copying texts by hand, Howe reasoned, would be analogous to mutations that accumulate as species evolve.

A friend put Howe in touch with Peter Robinson at De Montfort University in Leicester. Robinson, a manuscript scholar, was trying to determine which of the surviving complete fifteenth-century versions of Geoffrey Chaucer's Canterbury Tales are closest to what Chaucer originally wrote. He also wondered if Chaucer left a finished version before he died in 1400. In general, the greater the variation from text to text, the more frequently they have been copied.

Scholars had attempted, with little success, to answer such questions without the help of computers. Howe expected that the program he used for DNA analysis would be much more accurate than the unassisted eyes of scholars. So he and Robinson decided to feed the program 850 lines taken from the Canterbury Tales' ribald "Wife of Bath's Prologue."

Of the 58 extant complete manuscripts of the prologue, 11 turned out to have the fewest variations, a sign that they had been copied fewer times than the other manuscripts had. But even these 11, says Robinson, contain significant differences, which leads him to believe that Chaucer's original text was probably not a finished product but a working draft, complete with deletions, word changes, and alternate passages. The 11 copies in Howe and Robinson's study probably each incorporate different versions of that rough draft.

"In the ëWife of Bath's Prologue' there are some 26 lines, from five different passages, that occur in some manuscripts but not others," says Robinson. "That suggests that Chaucer wrote these originally, and then he changed his mind and decided that he would delete them."
Comment on this article