may be ordered in any way that fits it, and the complex of relations connecting such data is controlled solely by the resulting errors. The problem is to make these as small as possible by continuous adjustment, and experience itself furnishes no ground for belief that these outstanding errors themselves constitute an ordered group. They stand always, a chaotic totality, between the physical theory and the world of experience, preventing us from regarding that world as subject to law in the sense that relations are rigorously satisfied by its phenomena. Suppose, then, that we replace this world of experience by a fictitious world, governed by a causal principle and osculating the world of experience during the present. This fictitious world will possess both a past and a future rigorously deducible from its present, and the postulate of causality identifies the past and future of this fictitious world with the past and future of the world of experience, and so creates both for the world of experience. The principle of causality so interpreted postulates a temporal but not a spatial validity for the physical theory. That our physical laws in general are applicable to the stellar system is extremely doubtful,25 and that the planetary theory in particular is applicable outside the Solar system is uncertain. Duner's 26 conclusion, drawn from the determination of double star orbits, that: es wäre kaum zu rechtfertigen, irgend eine der bisherigen Bahnbestimmungen als Bestätigung der Gültigkeit des Newtonschen Gesetzes auserhalb des Sonnensystems anzuführen, may be regarded as valid today. The problem of the dynamical behavior of double and multiple star systems may be stated exactly as for the Solar system: Is it possible to choose, for the motions of the multiple star system, measures of length and time, a set of masses, a trihedron of reference, and a law of force, such that the resulting equations represent the behavior of the system to within an error 25 Seeliger, Über die Anwendung der Naturgesetze auf das Universum, Rivista di Scienza, Vol. VI (1909). Chwolson, Dürfen wir die physikalischen Gesetze auf das Universum anwenden? Rivista di Scienza, Vol. VIII (1910). 26 Einige Bemerkungen über verschiedene Berechnungen von Doppelsternbahnen, Astr. Nachr., 132 (1893). less than the probable error of observation in the neighborhood of a given epoch? It should be emphasized that the choice of no one of these undetermined elements in the representation need be the same as for the Solar system, and it is a priori improbable that the same choice of all of them will serve for both. The investigation of Bertrand's Problem: 27 the determination of the law of force controlling the motions of a double star system under the conditions that the law of areas be satisfied in the true orbit, and that the apparent orbit be an ellipse, seems to lead, under certain admissible assumptions, to the Newtonian law of attraction, with the same inertial time and trihedron as for the Solar system, although the observational verification of the elliptic character of the apparent orbit is uncertain. An inertial time and trihedron can always be so selected that the motion of a double star will be Newtonian, but that it should appear so to move when referred to the inertial time and trihedron of the Solar system, can be nothing more than an approximation, and that a multiple star should so move is improbable. The triple system Cancri, studied in detail by Seeliger, 28 presents certain anomalies in its motions hard to adjust; and whether these would disappear under a different choice of time and trihedron of reference, does not appear from the discussion-indeed cannot well be determined until much more extended and accurate observational material is at hand. The apparent accord between observed and computed double star orbits, even, is of so rough a character that all that may be concluded is that the planetary time and trihedron will serve as a first approximation. While the evidence does not permit us to deny spatial validity to the planetary theory, it certainly does not justify us in asserting it, and this is the only part of the general physical theory which can be put to observational test for spatial extension. As long, however, as we do not fall on contradiction in the application of the general physical theory to the stellar universe, 27 See Appell, Mécanique Rationelle, I (1909), p. 400. 28 Consult article, Doppelsterner, Valentiner's Handwörterbuch der Astronomie, p. the spatial system so constructed is a logically admissible picture of that universe. There is, however, this essential difference between a temporal exterpolation and a spatial extension. The first is entirely outside of observational control, while the second is always subject to possible contradiction, and may fall at any moment when the necessary material for testing it is at hand. Temporal exterpolation is then an application of the principle of causality, but spatial extension is not directly referable to any postulate of knowledge, but remains an empirical problem. A question of some interest arises here. Why is the principle of causality so conceived as to contain the exterpolation of one only of the variables entering the physical theory? Is it because of the structure of the theory itself, depending on the way in which time enters? This question falls in un monde Bergsonien where the representation is conceived in terms of time alone; but in a Minkowskian world with a symmetric timespace representation,29 exterpolation in time alone would seem to need some sort of justification, even though it have its origin in lack of observational control. The agreement of the physical theory with observation throughout historic time may be regarded as the same sort of check as the agreement throughout planetary space, and the resulting time-space representation may be looked upon as having just exactly that time-space validity which includes historic time and planetary space. The absence of control gives no a priori justification for extension in time, although this extension must be regarded as the direct result of that lack of control. Finally, does a past or future created in accordance with the Postulate of Causality possess reality? The answer depends of course on the criterion of reality adopted, and, if we agree with Enriques,30 that: Qualunque sia il motivo che genera l'ipotesi, la scienza assume come criterio fondamentale la verifica di questa al lume dell'esperienza ragionata. Di guisa 29 Compare Langevin, L'Évolution de l'éspace et du temps. Rivista di Scienza, Vol. X (1911). 30 Il problema della realtà, Rivistà di Scienza, Vol. IX (1911), p. 262. che la definizione della realtà scientifica sembra ridursi non tanto ai caratteri di ciò che forma oggetto dell'immaginazione creatrice, quanto alle rigide norme di una verificazione indipendente dal volere e dal sentimento. -then, from the point of view which regards empirical verification as the proof of reality, we must deny reality to the past as constructed, since it is inaccessible to verification. But this older point of view, which required empirical verification of an objective configuration, which, nevertheless, did not cease to exist when all connection between the external world and its representation was broken, has given way under a severe gnoseologic critique to a modern conception of reality, of which invariance is the criterion, but which regards this invariance as relative and approximate. To the same extent, therefore, to which the present is real-that is, to the extent to which its representation presents invariant relations-the past is real, being conditioned by the same invariance. STUDIES IN HYDROGENATION1 LEROY MCMASTER Associate Professor of Chemistry Catalytic reductions have been carried out for a long time in order to prepare new substances from the unsaturated organic compounds by the addition of hydrogen. Palladium and platinum have frequently been used along with the hydrogen. The unsaturated hydrocarbons can be converted into paraffins by passing a mixture of their vapors and hydrogen over the heated metal. The oxides of nitrogen are converted into ammonia by the same process. Saytzeff, as early as 1871, reduced benzoyl chloride to benzyl alcohol and nitrobenzene to aniline by passing their vapors, mixed with hydrogen, over finely divided palladium. Acetylene and ethylene have been reduced by Wilde' by this method, using platinum-black as the catalyzer. Colloidal solutions of platinum and palladium in water and alcohol are very active, and in the presence of hydrogen reduce the unsaturated acids, such as oleic, fumaric, maleic, and cinnamic, to the saturated condition. The colloidal solutions of these metals and also some other metals can be prepared by passing a current between electrodes of the metal under water. Colloidal solutions of metals can also be prepared by reducing the salts of the metals by hydrazine salts. Thus prepared, they seem to be more stable than when prepared by the electrical method. A large amount of colloidal copper has been prepared in this laboratory by the electrical method. Fokin' has found that those metals which occlude hydrogen have the strongest reducing action upon unsaturated carbon compounds. They also act at a lower temperature. Fokin was studying the activity of different metals as electrodes for 1 From the Chemical Laboratory of Washington University. 'J. prakt. Chem. 4, 418 (1871); Chem. Zentr., 19 (1872). * Ber., 7, 352. J. Russ. Phys. Chem. Ges., 38, 419, 855; Chem. Zentr. (1906), vol. 2, 758; (1907), vol. 1, 324. |