Werner in discussing this point, and also in discussing the Walden inversion and other topics, rejects the idea of localized attraction and prefers the notion of a general attraction of the atom as a whole. But a general attraction seems hard to reconcile with the fact that it is never satisfied by more or less than four univalent atoms or groups.

In 1887 Arrhenius proposed his theory of electrolytic dissociation, or, as it is better called, of ionization. The attention of chemists was once more directed to the intimate relation between electrical forces and chemical affinity. About ten years later J. J. Thomson demonstrated the existence of electrons as definite units of negative electricity having a mass eighteen hundred times smaller than that of the hydrogen atom. Almost at the same time the first discoveries of radioactive elements were made and as these were developed Rutherford brought out his theory of the disintegration of atoms. It soon became evident that atoms are not homogeneous, indi-, visible particles, as had been more or less tacitly supposed but that they must be complex aggregates of electrons and much more dense material with a positive charge. J. J. Thomson made an attempt to account for the properties of atoms by supposing a uniform positive sphere into which electrons found their way and distributed themselves in accordance with their attraction for the positive matter and their repulsion for each other. In contrast with this Bohr, Rutherford and others have assumed a very dense central nucleus, with a positive charge, about which electrons are in very rapid rotation. Experimental evidence based especially on the scattering of helium atoms which are shot through a thin film of gold or other metals has given very positive evidence in favor of atoms with a central nucleus. Some progress has been made in the discussion of the relation between the spectra of some of the elements and the revolving electrons of their atoms.

While physicists have been busy with studies of radioactivity and have reached some rather positive conclusions about atoms, which

are certainly dynamic, rather than static, chemists have been busy with the applications of the new electro-chemistry in its relation to the electron theory. It is evident that in ionization a positive univalent ion has lost an electron and a negative ion has gained one. Chemists have usually been very reticent about the fate of these electrons when ions unite but it seems rather certain that some atoms still retain their position or negative character in their compounds. This was first definitely pointed out for nitrogen trichloride, in which the chlorine is positive, as contrasted with nitrosyl chloride, NOC1, in which it is clearly negative.

The evidence for the existence of positive chlorine in nitrogen trichloride and of negative chlorine in nitrosyl chloride is usually stated to be that the trichloride hydrolyzes to ammonia and hypochlorous acid while nitrosyl chloride hydrolyzes to nitrous acid and hydrochloric acid. Recent experiments, not yet published, have shown that dry ammonia and chlorine react with the formation of the trichloride and also that a dry solution of the trichloride reacts with dry hydrochloric acid to form ammonium chloride and chlorine. These facts seem to prove that the trichloride is formed by the addition of chlorine to ammonia, forming chloroammonium chloride, NH,Cl+Cl-, followed by the splitting off of hydrochloric acid, and a repetition of the process till all of the hydrogen is gone. On the other hand, the decomposition of the trichloride by hydrochloric acid is caused by the addition of the acid, giving trichloroammonium chloride, NC1,H+Cl-, followed by the splitting off of a positive and a negative atom of chlorine, as free chlorine. Under suitable conditions these reactions may be practically quantitative in either direction.

The close analogy between these reactions and those which take place in the substitution of chlorine for hydrogen in methane or in benzene is very striking and recalls Michael's theory that addition compounds are at first formed in such cases. The idea which is partly new is that such additions are always of two atoms or groups, one of which is

positive and the other negative. Also that atoms may form unstable compounds of higher valence when one of the new valences is positive and the other negative. The instability of such compounds is due to the fact that they contain two atoms or groups of opposite sign.

This is closely related to Abegg's contravalences, according to which the non-metallic elements, in particular, exhibit maximum positive and negative valences the sum of which is eight. Thus sulfur has a negative valence of two in hydrogen sulfide and a positive valence of six in sulfur trioxide and in sulfuric acid. But sulfur very readily assumes three negative valences, in such compounds as triethyl sulfonium iodide, (C,H,),SI, in which it has also assumed one positive valence. We may assume that methane, in a similar manner forms with chlorine the compound CH,Cl+Cl-, which at once splits off hydrochloric acid.

It seems worth while to remark that we have much need of revising our conception of the non-metallic elements as being negative elements. This is based on the fact that the halogens when univalent form their most stable compounds as negative atoms and also because nitrogen, sulfur and similar elements form very stable negative groups, such as SO1, NO,, etc. We are a little apt to forget that oxygen forms the negative part of these groups and that nitrogen and sulfur are positive in them. Some apparent anomalies in the laws for the orientation of substituting groups in benzene derivatives are easily explained when these facts are kept in mind.

It will have been noticed that very much importance is attached, above, to additions and decompositions in which the two groups or atoms added or split off are of opposite sign. Nef, Michael and others have often emphasized the importance of reactions of this type but at a time when an accurate consideration of the positive or negative character of the atoms or groups involved was not in vogue.

Some of our ideas need revision, as it seems to me, in this regard. Let us take the commonly accepted formula of nitrous oxide,

first and then O-N N. Such a structure takes account of the probability that one nitrogen atom remains positive and the other negative throughout the series while the formula usually given implies that at least one valence of one nitrogen atom changes from negative to positive. The formula here proposed also seems more in accord with the ease with which nitrous oxide gives up its oxygen. I can not take the time to consider the very interesting discussions of Falk and Nelson, of Fry and of Jones nor the valuable additions to our experimental knowledge made by the last two. As was remarked at the beginning, every important advance must combine experimental evidence with new explanations of natural phenomena.

G. N. Lewis has proposed the hypothesis that carbon compounds are not held together by polar valences because they do not readily ionize. It seems possible that this is true in some cases but it is difficult to believe that there is any very essential difference between the reaction of methyl iodide with silver nitrate and that of potassium iodide with the same reagent.

Werner distinguishes between primary valences which hold simple compounds together and also cause the binding of carbon atoms, and secondary valences, which are effective in the formation of complex, molecular compounds. Strong reasons have been given for thinking that the ammonium compounds are

genuine valence compounds and not molecular compounds, as Werner supposes them to be. If we assume rotating electrons, as we are almost compelled to do, molecules may well be held together by magnetic attractions and it seems possible that such attractions give rise to the secondary valences of Werner. He and his students have done magnificent work in the study of complex inorganic compounds, and some apparent contradictions between his ideas and those of organic chemists will doubtless be cleared up in the future.

To repeat, the most important recent advance in the theory of valence has been the interpretation of the theory in connection with the electron theory and a beginning toward the study of positive and negative atoms in organic compounds.

It is not very safe to prophesy, but I am impressed with the need of a more definite knowledge of the structure of atoms as a basis for a better understanding of valence. So much progress has been made in this direction during the past twenty years that further advance seems probable. Several different lines of study may be suggested, the coordination of which might lead to further knowledge.

The spectra of the elements must be intimately connected with the structure of the atoms. Professor Morley once suggested that the problem is something like that of a man who should endeavor to determine the mechanism of a grand piano with only sound waves to guide him. Yet the matter seems by no means hopeless and a beginning has already been made. It is noticeable that spectral lines are only given out by ionized gases and are possibly connected with the motion of valence electrons.

Certain forms of structure, especially those forms in which there are what Baeyer so aptly called fliesende doppelte Bindunge, absorb light of certain wave-lengths and give us colored compounds. This means that it is possible to calculate exactly the rate at which certain atoms, or parts of atoms, or of molecules vibrate.

The atomic numbers determined by X-ray

spectra must be due to some more rapid and very fundamental sort of vibration which changes as the nucleus of the atom increases in weight from one element to another of the periodic system.

By means of X-rays it has been possible to determine the actual arrangement of atoms in crystals. The conclusion has been drawn from some of this work that the results do not agree with our customary conceptions of valence. I am inclined to think that the achievements of organic chemistry are not to be discarded so easily.

The Walden inversion has been interpreted by Fischer and Werner in terms of general attraction between atoms as opposed to localized attraction. The subject deserves careful study.

The evidence from radioactive disintegrations that helium atoms and electrons form essential parts of some atoms and very probably of all, is certainly important. The speculations of W. D. Harkins in this connection may be mentioned. The enormous quantity of energy liberated by the disintegration of an atom is certainly significant.

Three suggestions as to the functions of electrons in holding atoms together may be mentioned. One is that one or two electrons rotate about a line joining the centers of two positive nuclei, at a point midway between them. In ionization these electrons must remain with the negative ion. A second suggestion is that the magnetic effect produced by rotating electrons may cause an attraction between atoms. This is the magneton theory of Parsons and something of the same sort was suggested by Sir William Ramsay in the last paper which he published. It would also account for the non-polar combinations of G. N. Lewis. A third suggestion is that an electron may rotate about positive nuclei situated in two atoms. The velocity of electrons must be so much greater than ordinary motions of atoms that the atoms might be kept together in this way.

Whatever conclusions may finally be reached with regard to the structure of atoms it seems very certain that the structure will be found

to be dynamic rather than static. It is hard to conceive of a quiescent electron.

WILLIAM ALBERT NOYES

SOME RECENT CONTRIBUTIONS TO THE PHYSICS OF THE AIR. II

BAROMETRIC FLUCTUATIONS

Another meteorological phenomenon that has been the subject of numerous investigations, is the pressure of the atmosphere. This pressure undergoes many changes, and the causes of some of them, such as its variations between summer and winter, and between ocean and continent, have long been known. But the causes of certain others, notably the great changes that accompany cyclones and anticyclones, have not yet been fully determined. Neither of these classes of changes, that is, those whose causes have long been known, and those whose causes still remain partially obscure, will be discussed here, though each is important and full of interest. But

there are two other classes, namely, barometric ripples, and the pressure wave of semidiurnal period, that, because of their obvious interest to physicists, will be briefly considered.

Barometric "Ripples." Small pressure changes having amplitudes usually of 0.1 mm. to 0.3 mm. and periods of 5 minutes to 10 minutes, and continuing for hours or even days together, are very common during cold weather. They are not greatly different in magnitude from the well known wind effects on the barometer, but obviously of different origin since their amplitude has no relation to the local wind velocity.

Their explanation appears to lie in the fact that whenever layers of air that differ in density at their interface flow over each other, long billows, analogous to water waves, and generated in the same way, are produced. If, now, the under layer is colder than the upper, as it is during the radiation or surface inversions of winter, and rather shallow, 100

During the summer, when air billows rarely form near the surface, though frequently at greater altitudes, especially that of the cirrus cloud, barometric ripples do not appear. This, doubtless, is because wave disturbances in air, as in water, do not penetrate far beneath the wave level.

Semidiurnal Pressure Changes.-It has been known now for more than two and a half centuries that there are approximately regular daily variations in the height of the barometer that culminate in two maxima and two minima in the course of twenty-four hours. During particularly calm weather these fluctuations are conspicuous on the current barogram, as shown in Fig. 5, obtained at Grand Turk Island, West Indies, and always revealed by averages, no matter how masked by storm conditions, as shown by Fig. 6, based on records, covering fourteen years, at Key West, Fla.

Although, as stated, this phenomenon of the semidiurnal variation of the atmospheric pressure has been familiar almost since the invention of the mercurial barometer, and although repeatedly studied, its cause or causes remained until quite recently wholly unknown.

Some of the observed facts in regard to this twelve-hour cyclic change of pressure are:

(a) The maxima occur at, roughly, 10 o'clock A. M. and 10 o'clock P.M.

(b) The minima occur at, roughly, 4 o'clock A.M. and 4 o'clock P.M.

(c) The amplitude, when other things are substantially equal, varies with place approximately as the square of the cosine of the latitude.