view is strongly supported by some remarkably interesting experiments made by Mr. Moseley. If we could be sure that we had a complete list of the elements, that few, if any, had escaped the vigilance of the chemist, and that all the elements were members of one family, the atomic number would be the quantity with which we should naturally connect the number of electrons in the atom: for we may regard each element as derived from the preceding one by the addition of a primordial atom containing one electron. There may, however, be more than one family of elements, the successive members in each family growing by a common unit, though the members of one family cannot be changed into those of the other by the addition or subtraction of this unit. I think there are reasons for believing that there are two families of elements; for if there were only one family we should expect that the atomic weight of the lighter elements would increase by a common difference. This is not so. If, however, we divide the lighter elements into two families, those with even and those with odd atomic weights, we find that in each of these families the atomic weights do, with very few exceptions, increase by the common difference 4, and that in fact we get much greater simplicity and order when we arrange them in two series than when we regard them as successive members of a single series. This is illustrated by the following table, which contains the elements whose atomic weight is not greater than 40:
| He | 4 | Li | 7 | ||
| Be | 9 | B | 11 | ||
| C | 12 | N | 14 | ||
| O | 16 | F | 19 | ||
| Ne | 20 | Na | 23 | ||
| Mg | 24 | Al | 27 | ||
