All real intervals of duration regard the past; for in the past alone can we find a real _before_ and a real _after_. The present gives no interval, as we have just stated, but only simultaneousness. The future is real only potentially--that is, it will be real, but it is not yet. What has never been, and never will be, is merely imaginary. To this last class belong all the intervals of duration corresponding to those conditional events which did not happen, owing to the non-fulfilment of the conditions on which their reality depended. As to the measurement of flowing duration a few words will suffice. The _when_ considered absolutely is incapable of measuring an interval of duration, for the reason that the _when_ is unextended, and therefore unproportionate to the mensuration of a continuous interval; for the measure must be of the same kind with the thing to be measured. Just as a continuous line cannot be made up of unextended points, so cannot a continuous interval be made up of indivisible instants; hence, as a line is divisible only into smaller and smaller lines, by which it can be measured, so also an interval of duration is divisible only into smaller and smaller intervals, and is measured by the same. These smaller intervals, being continuous, are themselves divisible and mensurable by other intervals of less duration, and these other intervals are again divisible and mensurable; so that, from the nature of the thing, it is impossible to reach an absolute measure of duration, and we must rest satisfied with a relative one, just as in the case of a line and of any other continuous quantity. The smallest unit or measure of duration commonly used is the second, or sixtieth part of a minute. But, since continuous quantities are divisible _in infinitum_, it may be asked, what prevents us from considering a finite interval of duration as containing an infinite multitude of infinitesimal units of duration? If nothing prevents us, then in the infinitesimal unit we shall have the true and absolute measure of duration. We answer that nothing prevents such a conception; but the mensuration of a finite interval by infinitesimal units would never supply us the means of determining the relative lengths of two intervals of duration. For, if every interval is a sum of infinite terms, and is so represented, how can we decide which of those intervals is the greater, since we cannot count the infinite?