It would be endless to point out the oddities and incongruities which result from this classification. Whether we take the words wealth and production in the largest, or in the most restricted sense in which they have ever yet been employed, nobody will dispute that roads, bridges, and canals, contribute in an eminent degree, and in a very direct manner, to the increase of production and wealth. The labour and pecuniary resources employed in their construction would, according to the above theory, be considered productive, if every occupier of land were compelled by law to construct so much of the road, or canal, as passes through his own farm. If, instead of this, the government makes the road, and throws it open to the public toll-free, the labour and expenditure would be, on the above system, clearly unproductive. But if the government, or an association of individuals, made the road, and imposed a toll to defray the expense, we do not see how these writers could refuse to the outlay the title of productive expenditure. It would follow, that the very same labour and expense, if given gratuitously, must be called unproductive, which, if a charge had been made for it, would have been called productive. When these consequences of the purely arbitrary classification to which we allude have been pointed out and complained of, the only answer which we have ever seen made to the objection is, that the line of demarcation must be drawn somewhere, and that in every classification there are intermediate cases, which might have been included, with almost equal propriety, either in the one class or in the other. This answer appears to us to indicate the want of a sufficiently accurate and discriminating perception, what is the kind of inaccuracy which generally cannot be avoided in a classification, and what is that other kind of inaccuracy, from which it always may be, and should be, exempt. The classes themselves may be, mentally speaking, perfectly definite, though it may not always be easy to say to which of them a particular object belongs. When it is uncertain in which of two classes an object should be placed, if the classification be properly made, and properly expressed, the uncertainty can turn only upon a matter of fact. It is uncertain to which class the object belongs, because it is doubtful whether it possesses in a greater degree the characteristics of the one class or those of the other. But the characteristics themselves may be defined and distinguished with the nicest exactness, and always ought to be so. Especially ought they in a case like the present, because here it is only the distinction between the ideas which is of any importance. That we should be able with ease to portion out all employments between the two classes, does not happen to be of any particular consequence.