require causation; yet not even the first causal hypothesis can be formu-
lated without a very thorough description of the phenomenon in ques-
tion-including the history of its becoming what it is described to be. One
of the most unfortunate developments in the evolution, as it were, of the
scientific study of politics is the proliferation of independent variables.
With only the most primitive accounting given for the dependent variable
(that is, the phenomenon to be explained), we confront an enormous ap-
paratus of independent variables, which do not represent an improved
science but merely a shift of focus away from the initial problem -- as de-
fined by the dependent variable. In contrast, when the dependent variable
has been thoroughly described and its salient features have been identi-
fied, then the relevant independent variables almost present themselves.
A good case in point is the phenomenon of critical elections and realign-
ment. This is no more and no less than a description of what is alleged to
be a recurring phenomenon, yet it energized the study of parties and elec-
tions for many years and was instrumental in bringing history to contem-
porary political science. It may now be energizing the field once again,
with a new description. What else can Walter Dean Burnham mean by
"pattern recognition" in Chapter 3 of this volume?

My affinity for political history and the study of the evolution and de-
velopment of political institutions and practices is thus based upon their
commitment to description as the most sophisticated thing we are called
upon to do in political science. Description requires theory, because to
render a complex whole into its essential parts demands the use of stan-
dards and criteria, which must come from somewhere. Description also
requires theory because it involves categorization and classification -- in a
word, taxonomy -- according to structure and function and according to
time or sequence: periodization, in fact. This is what analysis is all about:
breaking a thing down into its parts. But what parts? Any parts, as long as
the logic imposed upon the breakdown permits you to put the parts back
together again in a more meaningful way. That is definitely theory, and it
is a lot of theory, even if implicit, before ever getting to questions of causa-
tion. I do not want to be misunderstood about causation: Causation is
okay, so long as it knows its place. Description comes first. Development
is little more than description of a particular state of affairs at more than
one point in time. And causation involves the description of one thing in
terms of something that precedes it.

All this may qualify me for formal affiliation with political history and
the new institutionalism. But that also makes me a bit uncomfortable. To
quote Groucho Marx, I generally don't like to belong to any club that
would have me as a member. As I see it, the main weakness of history, as
evolution or development, is that it fails to recognize its own limitations.
Specifically, development is inconceivable without nondevelopment; de-

-x-