If elements were able to combine in only one way,--one atom of one with one atom of another,--the problem of determining the atomic weights would be very simple. We should merely have to take some one convenient element as a standard, and find by experiment how much of each other element would combine with a fixed weight of it. The ratios thus found would be the same ratios as those between the atoms of the elements, and thus we should have their relative atomic weights. The law of multiple proportion calls attention to the fact that the atoms combine in other ratios than 1: 1, and there is no direct way of telling which one, if any, of the several compounds in a given case is the one consisting of a single atom of each element. If some way were to be found of telling how much heavier the entire molecule of a compound is than the atom chosen as a standard,--that is, of determining the molecular weights of compounds,--the problem could be solved, though its solution would not be an entirely simple matter. There are ways of determining the molecular weights of compounds, and there are other experiments which throw light directly upon the relative weights of the atoms. These methods cannot be described until the facts upon which they rest have been studied. It will be sufficient for the present to assume that these methods are trustworthy. ~Standard for atomic weights.~ Since the atomic weights are merely relative to some one element chosen as a standard, it is evident that any one of the elements may serve as this standard and that any convenient value may be assigned to its atom. At one time oxygen was taken as this standard, with the value 100, and the atomic weights of the other elements were expressed in terms of this standard. It would seem more rational to take the element of smallest atomic weight as the standard and give it unit value; accordingly hydrogen was taken as the standard with an atomic weight of 1. Very recently, however, this unit has been replaced by oxygen, with an atomic weight of 16.