It seems to me, therefore, that we must accept the alternative which is only mentioned to be repudiated by Jevons, namely, that Political Economy, if not a "mathematical science," must be part of sociology. I should say that it clearly is so; for if we wish to investigate the cause of any of the phenomena concerned, and not simply to tabulate from observations, we are at once concerned with the social structure and with the underlying psychology. The mathematical methods are quite in their place when dealing with statistics. The rise and fall of prices, and so forth, can be stated precisely in figures; and, whenever we can discover some approximation to a mathematical law (as in the cases I have noticed) we may work out the results. If, for example, the price of a commodity under certain conditions bears a certain relation to its scarcity, we can discover the one fact when the other fact is given, remembering only that our conclusions are not more certain than our premisses, and that the observed law depends upon unknown and most imperfectly knowable conditions. Such results, again, may be very useful in various ways, as illustrative of the way in which certain laws will work if they hold good; and, again, as testing many of our general theories. If you have argued that the price of gold or silver cannot be fixed, the fact that it has been fixed under certain conditions will of course lead to a revision of your arguments. But I cannot help thinking that it is an illusion to suppose that such methods can justify the assertion that the science as a whole is "mathematical". Nothing, indeed, is easier than to speak as if you had got a mathematical theory. Let _x_ mean the desire for marriage and _y_ the fear of want, then the number of marriages is a function of _x_ and _y_, and I can express this by symbols as well as by ordinary words. But there is no magic about the use of symbols. Mathematical inquiries are not fruitful because symbols are used, but because the symbols represent something absolutely precise and assignable. The highest mathematical inquiries are simply ingenious methods of counting; and till you have got something precise to count, they can take you no further. I cannot help thinking that this fallacy imposes upon some modern reasoners; that they assume that they have got the data because they have put together the formulæ which would be useful if they had the data; and, in short, that you can get more out of a mill than you put into it; or, in other words, that more conclusions than really follow can be got out of premisses, simply because you show what would follow if you had the required knowledge. When the attempt is made, as it seems to me to be made sometimes, to deduce economical laws from some law of human desire--as from the simple theorem that equal increments of a commodity imply diminishing amounts of utility--I should reply not only that the numerical data are vaguely defined and incapable of being accurately