fruitful.

People sometimes say that scientific methods are inapplicable because
we cannot try experiments in social matters. I remember being long ago
struck by a remark of Dr. Arnold, which has some bearing upon this
assertion. He observed upon the great advantage possessed by Aristotle
in the vast number of little republics in his time, each of which was
virtually an experiment in politics. I always thought that this was
fallacious somehow, and I fancy that it is not hard to indicate the
general nature of the fallacy. Freeman, upon whose services to thorough
and accurate study of history I am unworthy to pronounce an eulogy,
fell into the same fallacy, I fancy, when he undertook to write a
history of Federal Governments. He fancied that because the Achaean
League and the Swiss Cantons and the United States of America all had
this point in common, and that they represented the combinations of
partially independent States, their history would be in a sense
continuous. The obvious consideration that the federations differed in
every possible way, in their religions and state of civilisation and
whole social structure, might be neglected. Freeman's tendency to be
indifferent to everything which was not in the narrowest sense
political led him to this--as it seems to me--pedantic conception. If
the prosperity of a nation depended exclusively upon the form of its
government, Aristotle, as Arnold remarks, would have had before him a
greater number of experiments than the modern observer. But the
assumption is obviously wrong. Every one of these ancient States
depended for its prosperity upon a vast number of conditions--its race,
its geographical position, its stage of development, and so forth,
quite impossible to tabulate or analyse; and the form of government
which suited one would be entirely inapplicable to another. To
extricate from all these conflicting elements the precise influence due
to any institutions would be a task beyond the powers of any number of
philosophers; and indeed the perplexity would probably be increased by
the very number of experiments. To make an experiment fruitful, it is
necessary to eliminate all the irrelevant elements which intrude into
the concrete cases spontaneously offered by nature, and, for example,
to obtain two cases differing only in one element, to which we may
therefore plausibly attribute other contrasts. Now, the history of a
hundred or a thousand small States would probably