After becoming familiar with computers and establishing a re-
search plan, a scholar faces the challenge of transforming an
unwieldy source into a usable data set. Historians often become
interested in formal analysis when discovering a mass record such
as a shipping list, marriage register, poll book, or personnel file
which promises exciting insights--if it could only be analyzed! Her-
meneutic researchers react predictably, but problematically: Some
scholars simply ignore an unmanageable source, declaring it unim-
portant and thereby impoverishing research. Others browse so as to
"get a feel" for the document, but their impressions risk grave er-
rors if the data are not uniform. More conscientious historians tend
to select a record segment, small enough to be probed in detail but
often quite unrepresentative of the whole. Finally, well-meaning
researchers sometimes employ inadequate techniques such as end-
less card files or pencil lists which require laborious sorting and
limit the exploration of variables. Such questionable procedures
might have been inevitable as long as there were no alternatives,
but with the maturation of historical computing, they have become
inexcusable. Once convinced of the potential benefits of a more
stringent approach, a researcher will wonder about how to make a
source machine-readable, how to organize a codebook, and how to
enter the data into the computer. These seemingly practical ques-
tions have important interpretative implications as well.

In order to use the computer in analysis, historians must make
their sources machine-readable. While textual data bases can pre-

-37-