Passing on, between the atomic weights of potassium, rubidium, and caesium there is a difference of about 16 × 3; a similar difference between calcium, strontium, and barium; between scandium and yttrium; between titanium, zirconium, and cerium, and so on; but with wider and wider divergence from the supposed constant, 48 = 16 × 3. In short, we have a seeming regularity, but only a very approximate one—a regularity, in fact, in which a vivid imagination must play a conspicuous part in order to detect it.
Now, up to the present, no reason has been suggested to account for the divergence from this irregular regularity, which a little expenditure of time will enable any one to trace through all these numbers. But one thing has been remarked; there is the same seeming regularity between certain physical properties of elements and their compounds: their specific volumes, their melting-points, their refractive indices, and other properties vary from member to member of the same column in a manner bearing more or less similarity to the periodic variation of the atomic weights.
It happens that among compounds of carbon we are acquainted with series of compounds which, in variation of molecular weights and gradation of properties, bear a striking resemblance to the elements thus arranged. Thus we have the series:—
| CH4 | Methane | 16 |
| C2H6 | Ethane | 30 |
| C3H8 | Propane | 44 |
| C4H10 | Butane | 58 |
| C5H12 | Pentane | 72 |
and a host of others up to a compound of the formula C30H62; in each case there is a constant difference of 14 between the molecular weight of any one hydrocarbon and that immediately preceding or succeeding it in the column. Such a series is termed a homologous series. The analogy is very tempting; to suppose that a similar constant difference should exist in the relations of the atomic weights of the elements, and that they too are undecomposable compounds of two unknown elements, is an attractive hypothesis, but one for which there exists no proof; indeed, it is rendered improbable by the irregularities just pointed out.
But there is one noticeable feature in the periodic arrangement of the elements. It is, that although the differences are irregular (e.g. between B = 11 and C = 12 the difference is 1, while between O = 16 and F = 19 the difference is 3), yet there is no marked displacement in the order of arrangements of the elements, inasmuch as no element has an atomic weight lower than that preceding it in the horizontal line. It was for some time supposed that tellurium and iodine were thus misplaced; and indeed it is even now not quite established that they are not, but the balance of evidence is in favour of tellurium having a lower atomic weight than iodine.
Argon, however, is a marked exception. With an atomic weight of 39·88, its natural position would lie between those of potassium and calcium; but there is no room for it. And for this reason considerable doubts have been thrown on the validity of the conclusion to be drawn from the found ratio of its specific heats, 1⅔, viz. that its molecule and its atom are identical. If it were a diatomic gas, like chlorine or hydrogen, its atomic weight would be 19·94, and it would find a fitting position after fluorine and before sodium. And the difference between its atomic weight and that of helium, to which the atomic weight 2·1 would for the same reasons then attach, would be 17·84, one not incomparable with 16. But, as before remarked, it is difficult, if not altogether impossible, to conceive of a diatomic structure to which all energy imparted in the form of heat should result in translational motion, and as a matter of fact none such is known.
There are two methods of escape from this dilemma. If the gases termed argon and helium are not single elements, but mixtures of monatomic elements, then what has been termed their atomic weights will represent the mean of the atomic weights of two or more elements, taken in the proportion in which they occur. For example, supposing that argon is a mixture of an element of atomic weight 37 with one of atomic weight 82, the found atomic weight, nearly 40, would imply a mixture of 93·3 per cent of the lighter, with 6·7 per cent of the heavier element. We must therefore carefully examine all evidence for or against the supposition that argon is a mixture of elements.
It is well known that elements with high atomic weights have, as a rule, higher boiling-points than those with low atomic weights in the same columns. Perhaps the most striking case is that of the elements fluorine, chlorine, bromine, and iodine. Whereas fluorine has never been liquefied (chiefly owing to difficulties of manipulation, due to its extraordinarily energetic action on almost every element and compound), chlorine boils at -102°, bromine at 59°, and iodine at 184°. And if a mixture of chlorine and bromine gases be cooled, the bromine, if present in sufficient amount, will condense first, in a fairly pure state, little chlorine condensing with it. But in a mixture containing only 7 per cent of bromine with 93 per cent of chlorine (analogous to a mixture of the two supposed constituents of the argon mixture) the pressure of the bromine gas in the mixture would be only 7⁄100ths of the normal pressure, or 53·2 millimetres. At this pressure the boiling-point of bromine is about -5°, so that, on cooling to that temperature, bromine would begin to show signs of liquefaction. This is, however, still nearly 100° above the boiling-point of chlorine; and there would therefore be no difficulty whatever in detecting such a percentage of bromine in a mixture of chlorine and bromine gases on cooling the mixture to a moderately low temperature.[30]
Argon has been liquefied. A sample of pure argon was sent by Professor Ramsay to Professor Olszewski of Cracow, well known for his accurate researches at low temperatures; and he found the boiling-point of argon at atmospheric pressure to be -186·9°, and its melting-point to be -189·6°. There was no appearance of liquid before the boiling-point was reached, nor was there any alteration of temperature as the argon boiled away, and these are signs of a single substance, not of a mixture; moreover, the melting-point was a definite one; and here again, mixtures never melt suddenly, but always show signs of softening before melting. So far as this evidence goes, therefore, it points to the conclusion that argon is not a mixture of two elements.