the spark was passed through the air, which consists chiefly of a mixture of nitrogen and oxygen, and the resultant oxid of nitrogen was absorbed in caustic potash. More oxygen was added from time to time until the last of the nitrogen was used up. Now Cavendish noticed, as have many chemists since his day, that it was always impossible to remove all the nitrogen; in every case about one per cent, of gas remained. There is no record that any one ever suspected that this residue was not nitrogen; such is, however, the fact, and the gases, argon, helium and, perhaps, others are present. These can be best recognized by passing an electric spark through the rarefied gas and examining the spectrum. It seems now very strange to us that an element so abundant that an ordinary sized room contains no less than a thousand liters should have so long escaped discovery. The reason is not far to seek. Argon and its congeners are distinguished by a most remarkable exhibition of properties, in that they have apparently no chemical affinity, and no compounds of them are known. From this fact it has been argued by some that these gases cannot be considered chemical elements, for all elements hitherto known do form compounds with other elements. It is, however, a curious fact that in the periodic table we find, in the eighth group, place for several just such elements, as we have seen, without affinity, and neither positive nor negative in electrochemical character. It may well be that helium, neon, argon and xenon belong in these vacant spaces.
If this be the case, there is still a difficulty which confronts us, and this is that argon possesses an atomic weight slightly higher than the next element in order, potassium, instead of lower. This would not, however, be a unique instance of such a difficulty in the table. It was formerly thought that the two metals, nickel and cobalt, had identical atomic weights, and though the salts of nickel are generally green, and those of cobalt red, in other respects these metals and their compounds are very much alike. After the discovery of the Periodic Law, when it was seen that cobalt belonged in the second series of the eighth group and nickel in the third, it was supposed that further study would necessarily show nickel to have an appreciably higher atomic weight than cobalt. We have already seen that in this same group, before the appearance of the periodic table, the accepted atomic weights of osmium, iridium and platinum were incorrect, and it was the fact of their mis-arrangement in the table which caused Seubert to revise their weights. Very much labor has been spent upon the revision of the atomic weights of nickel and cobalt. Gerhardt Krüss supposed he had found a new and heavier metal, hitherto unknown, in ordinary cobalt, and that this caused the atomic weight to be estimated too high. He called the new metal gnomium, but it was soon shown that gnomium has no real existence. The more accurate the
