As we read in the newspaper that 'the educated Bengalis are disaffected' we either see, in the half-conscious substratum of visual images which accompanies our reading, a single Babu with a disaffected expression or the vague suggestion of a long row of identical Babus all equally disaffected.
These personifications and uniformities, in their turn, tempt us to employ in our political thinking that method of _a priori_ deduction from large and untried generalisations against which natural science from the days of Bacon has always protested.

No scientist now argues that the planets move in circles, because planets are perfect, and the circle is a perfect figure, or that any newly discovered plant must be a cure for some disease because nature has given healing properties to all plants.

But 'logical' democrats still argue in America that, because all men are equal, political offices ought to go by rotation, and 'logical' collectivists sometimes argue from the 'principle' that the State should own all the means of production to the conclusion that all railway managers should be elected by universal suffrage.
In natural science, again, the conception of the plurality and interaction of causes has become part of our habitual mental furniture; but in politics both the book-learned student and the man in the street may be heard to talk as if each result had only one cause.

If the question, for instance, of the Anglo-Japanese alliance is raised, any two politicians, whether they are tramps on the outskirts of a Hyde Park crowd or Heads of Colleges writing to the _Times_, are not unlikely to argue, one, that all nations are suspicious, and that therefore the alliance must certainly fail, and the other that all nations are guided by their interests, and that therefore the alliance must certainly succeed.

The Landlord of the 'Rainbow' in _Silas Marner_ had listened to many thousands of political discussions before he adopted his formula, 'The truth lies atween you: you're both right and both wrong, as I allays say.' In Economics the danger of treating abstract and uniform words as if they were equivalent to abstract and uniform things has now been recognised for the last half century.