CHAPTER XXXIVTHE FINITENESS OF THE UNIVERSE
IN classical science, space, the fundamental continuum, was thought to be Euclidean and flat throughout, hence infinite in extent. As soon, however, as Einstein was led to believe that the fundamental space-time continuum of the universe was not rigorously flat, but manifested various degrees of curvature from place to place around matter, the possibility of its turning out to be curled round on itself, and hence finite, had to be investigated further before any definite opinion could be expressed. True, this fundamental continuum was no longer three-dimensional space; it was four-dimensional space-time. But, apart from this difference, the same major problem still presented itself, and we were permitted to ask: “Is the universe of space-time infinite or is it finite?”
More precisely, if we appeal to a two-dimensional analogy, must we regard the universe of space-time as curved locally around the individual masses of the universe, but as flat or quasi-Euclidean on the whole over an infinite area, much like the rippled surface of an infinite pond of water? Or must we suppose that the rippled surface of the water is more akin to that of the rippled surface of a liquid sphere closing round on itself?
Now, if we consider the theory in its present state, we must agree that a finite universe is excluded. We can understand this point by considering the law of gravitation, which in Einstein’s theory is also a law of space-time curvature. We remember that the gravitational equations were given bywhere the’s represented the curvatures of space-time at a point and whererepresented the characteristics of matter at the same point. Outside mattervanished and the gravitational equations reduced toor, what comes to the same thing, tothis last equation being compatible with perfect flatness of space-time (). Einstein had assumed that this condition of perfect flatness existed at infinity, and it was when this supplementary condition was taken into consideration that Einstein’s law ofgravitation became determinate. This was the law which enabled him to anticipate the double bending of a ray of light, the Einstein shift-effect, as also to account for the curious motion of the planet Mercury.
It is obvious that a law of gravitation which entails the existence of perfectly flat space-time at infinity must of necessity lead to an unclosed space-time universe; and so we see that with Einstein’s original gravitational equations, a finite self-contained universe cannot exist. We are thus led to consider how it would be necessary to modify these equations for a finite universe to be realised.
Now we mentioned, when discussing the law of gravitation, that the inclusion of an additional tensor(being any arbitrary constant) in the equation of curvature would not destroy the tensor nature of the law; hence, so far as the requirements of all general laws are concerned, this modified law of space-time curvature is acceptable. If it is accepted, the universe can no longer be quasi-Euclidean and infinite; it will have to be closed, hence finite. In order to understand the reason for this statement, we must examine the law of curvature with theterm included. It will be convenient to write the term:;and in this case, the law of space-time curvature in the interior of matter becomesOutside matter, we obtainor its equivalent,Thus the inclusion of theterm changes the law of space-time curvature outside matter fromto.Now this new law of curvature is no longer compatible with perfect flatness (), but withwhich yields a closed spherical, space-time extension. We thus obtain a four-dimensional spherical Riemann space-time as a possible model of the universe. From these equations we may also deduce,which shows us thatis proportional to the curvature of the universe.
Inasmuch as in any case the universe, whether finite or infinite, is certainly of huge proportions, its curvature must be practically insignificant. We may infer, therefore, that,if itexists at all, is undoubtedly a small, a very small, quantity; so that in problems relating to the dynamics of the solar system, for example, we may continue to ignore its existence. Theoretically, of course, this omission would be unjustified, for the presence of theterm, by modifying the law of curvature, will inevitably modify the lay of the space-time geodesics. As a result, the orbits and motions of the planets round the sun should be affected by its existence, and it should be possible to settle the question of its presence empirically by observing the planetary motions with increased refinement. Practically, however, owing to the extreme minuteness of(assuming it exists at all), empirical observations of this sort would be insufficient to detect it. The only means we may have of establishing its existence directly (and, with it, the finiteness of the universe) will be to observe the motions of the stars in the globular clusters or in the Milky Way (as we shall explain later). All we can say at present is that the existence of a finite universe is proved to be a possibility, so far as the mathematical requirements of the theory of relativity are concerned.
We might attempt to settle the question by appealing to the principle of action. But, here again, both hypotheses, that of the infinite and of the finite universe, would reveal themselves as possible; hence we shall have to consider some other method of solving the problem.
All we have done up to this stage has been to show that Einstein’s gravitational equations (or space-time curvature equations) are incompatible with the existence of a finite universe. If, however, some very smallterm were assumed to be present, far too small to affect the observed planetary motions, a finite universe would ensue. In short, a finite universe is theoretically possible. But before giving any further consideration to the problem, we may well enquire what is to be gained by this additional complication.
The original gravitational equations, those entailing perfect flatness for space-time at infinity, appeared to have yielded satisfactory results. Furthermore, when Einstein assumed that space-time would degenerate to perfect flatness at infinity, he was merely expressing the validity of the principle of inertia in regions far removed from the influence of matter. In other words, at infinity, far from matter, free bodies would pursue straight courses with constant speeds (as measured in a Galilean frame). Of course, the principle of inertia might be incorrect, but still, until such time as its validity was questioned seriously, it was natural enough to accept it. With the finite universe, on the other hand, we should have to assume that space-time always retained a residual trace of curvature, so that the straight courses of free bodies would be impossible. Free motion would be in circles of huge radius.
Here, however, we must recall that Einstein’s theory is not one of loose, unsupported guesses. Had Einstein merely been interested in making the universe this or that, without taking into account the totality of facts at the scientist’s disposal, his cosmological speculations would be no more worthy of consideration than those of the Greeks, or of Dante, or of the philosophers of the seventeenthand eighteenth centuries. We may be quite certain, however, that with a conscientious scientist like Einstein the methodological procedure will be very different. And so, when we find that the theory he advocates entails a finite universe, we may anticipate that there will be very strong reasons for his choice. Indeed, as we shall see, a number of separate considerations lead to this solution. First of all, there is the apparent stability of configuration of the star distribution. Then again, we have such philosophical considerations as the probable physical relativity of all motion and of inertia in accordance with Mach’s ideas; the intrinsically amorphous nature of all empty continuous extensions as demanded by Riemann; the desire to discover unity in nature by proving that matter can be reduced to electricity and gravitation. All these facts, together with many others, can be satisfied only provided we assume that the universe is finite. Unfortunately, a proper understanding of the problems involved requires an appeal to mathematical analysis. However, we shall attempt to proceed in a very crude way without introducing mathematical symbols to any great extent.
There exist two major solutions for the finite universe. The first is theSpherical,Elliptical, or, more rigorously, theHyperbolical universeof de Sitter; the second is theCylindrical universechampioned by Einstein. We will consider de Sitter’s universe first. But before discussing these difficult subjects, there are certain peculiarities about the laws of nature which it will be necessary to mention.
In the majority of cases, the laws of nature are expressed in the form of laws of contiguous variations in space and in time. Mathematically, this is equivalent to saying that the laws are expressed by differential equations. Now differential equations give us the law of variation of magnitudes from place to place or from time to time, but they do not yield us the precise values of these magnitudes.
Consider, for example, a differential assertion such as the following: “A train is moving away from us with a constant acceleration of one mile per second.” Now this differential law, though it specifies the nature of the motion, does not enable us to deduce the position the train will occupy at any specific instant. To solve the problem we must be informed of the train’s position, and also of its velocity, at some initial instant. This illustration gives us an idea of the limitations of differential assertions or equations. In order to render them determinate, we must always supplement them with additional information by stating the value of the variable quantity at some definite point or time. In particular, when we consider a magnitude which may vary according to some differential law from point to point throughout a region of three-dimensional space, we must state the value and other characteristics of the magnitude, or variable, over a closed surface bounding this region of space. In mathematical terminology, we must know the boundary conditions before the law can be regarded as perfectly determinate.
Let us apply these considerations to the differential law which expresses the Newtonian law of attraction. It is given, as we haveseen, by Poisson’s equation,and expresses the law of distribution of(the Newtonian potential) in a gravitational field around matter and in the interior of matter (whereis the density of matter from place to place). But, written in this form, the law is indeterminate; for it yields a number of laws differing from Newton’s law of the inverse square. Mathematicians have proved that in order to obtain Newton’s law, we must supplement our differential equation by specifying that the potentialtends to a constant finite limit at spatial infinity.
Now, in Einstein’s theory, the analogue of Poisson’s law is the law of space-time curvature,Just as Poisson’s law was the differential law of the Newtonian potential distribution, so now Einstein’s law is the differential law of distribution of the ten potentials, the ten’s; andis the energy and momentum of matter at every point. Again, as in the case of Poisson’s equation, we must supplement these differential equations of the gravitational field by specifying the values of the’s at the boundary at infinity.
Of course the’s are not numbers or scalars like;they are tensors. Hence their values at a point will vary with the mesh-system, or frame of reference, we may select. Nevertheless, if we specify our choice of a frame of reference and decide on the values the’s are to have in this frame, their values in all other frames will be determined automatically. This is, of course, owing to the fact that the’s are tensors.
In the theory, as we have described it up to this point, Einstein selected a Cartesian space-time mesh-system, that is to say, a Galilean frame, and he assumed that as measured in this mesh-system the’s at infinity were to be given byall other’s vanishing. These boundary conditions implied that space-time was perfectly flat at infinity, hence satisfied the equation.It may be realised that we are merely expressing in another way what was said at the beginning of the chapter; the boundary values written above and the flatness of space-time at infinity being equivalent statements. Under these circumstances we saw that the universe could not be finite; we should have the quasi-Euclidean infinite space-time universe.
Now, in the opinion of both Einstein and de Sitter, this assumption of perfect flatness at infinity was too drastic. For though this perfect flatness is in harmony with the law of inertia, it appears rather arbitrary to assume that the law must hold at infinity, even were we to agree that it holds in less distant regions, a conclusion which is by no means certain. In Einstein’s own words: “It is certainly unsatisfactory to postulate such a far-reaching limitation without any physical basis for it.”
In addition to this first reason for doubting the perfect flatness ofspace-time at infinity, we must remember that the’s do not retain their standard boundary values (1, 1, 1, -1) when we change our mesh-system. In any frame other than a Galilean one, say in a rotating frame, these boundary values at infinity would be modified; this is, of course, due to the fact that the’s are tensors. It follows that the extremely simple boundary values appear to single out a privileged frame of reference (a Galilean one), and this, in Einstein’s opinion, is contrary to the spirit of the theory of relativity. At any rate, whatever may be thought of the force of these arguments, it can easily be seen that if we wish to assign invariant values to the’s at infinity, values which will remain the same in all frames of reference, the simplest solution will be to assume that they all vanish at infinity. Then, of course, owing to their tensor nature, their vanishing in any one frame ensures their vanishing in all other frames. This was indeed de Sitter’s solution; namely, that all the’s vanished at infinity.
De Sitter presents another argument in defence of his solution. It deals with the relativity of mass or inertia, but as I have been unable to reconcile his views on this subject with the characteristics of his universe, I may have misunderstood his idea; hence I cannot discuss it. At all events, the relativity of inertia is a most important argument in favour of Einstein’s cylindrical universe, so we shall consider it when dealing with Einstein’s solution.
Now when we accept de Sitter’s vanishing values for the’s or potentials at infinity, the direct outcome is a finite space-time universe. We see, then, that in this way the boundary conditions have been obviated entirely, since in a finite universe, closed round on itself, there is no longer a boundary. The precise form of de Sitter’s universe is a spherical four-dimensional space-time. When empty of matter, it is given by the equation.In other words, it is the type of universe to which we are led when theterm is included in the gravitational equations; hence, as we have seen, it is a type which is theoretically possible.
The characteristics of de Sitter’s universe are thus those of a spherical four-dimensional space-time of radius.As for,its magnitude is proportional to the universal curvature. As a matter of fact, the appellationhyperbolical universewould be more correct, for it isspace-timewhich is curved spherically, not space and time. Here we must remember that in space-time, space is real and time imaginary; and if we wish to express the shape of the universe in terms ofrealtime instead ofimaginarytime, the hyperbolical form is the correct one.[106]
As we have explained in the note, de Sitter’s universe is open at both ends in the time direction, so that time does not come back as it would in a spherical universe. When we consider points farther and farther removed from where we stand, all the’s decrease and finally vanish, as was demanded by the requirements discussed previously. In particular, this decrease and this ultimate vanishing of the potentialindicate, respectively, the gradual slowing down and the complete arrest of time in the distant regions of the universe. This arrest of time, however, is fictitious, for if we transported ourselves to the point considered, we should find that things went on as usual. It would now be where we had stood formerly that the arrest of time would appear complete. Thus, for every observer there would exist a locus of distant points where nothing would appear to change or move. A ray of light could never circle round the universe, but would gradually slow down and ultimately be arrested when this passive horizon of the observer was reached.
We also see that the nearer an incandescent atom is situated to this passive horizon, the slower would its vibrations appear; hence, in the case of a sodium atom (emitting a yellow light under normal conditions), the redder would be its light, till finally the light would cease completely when the passive horizon was reached by the atom. Now it happens that among the most distant luminous bodies known, namely, the spiral nebulæ, the great majority actually do appear to emit slowed-down light vibrations. Of course this effect might be attributed to the general tendency of all spiral nebulæ to recede from the solar system. But as no satisfactory reason can be advanced to account for this general recession, Eddington considers that de Sitter’s hypothesis of the hyperbolical universe has much to commend it.
A few further remarks must be made with reference to de Sitter’s universe. We remember that in Einstein’s theory the perfect flatness of space-time entailed the complete absence of gravitational or inertial forces (so far as a Galilean observer is concerned). Therefore, were it not for the curvature that matter itself imposes upon space-time around it, two parcels of matter would never attract each other. In de Sitter’s universe, however, there exists a residual hyperbolical curvature of space-time at every point, regardless ofthe additional curvature that would be superimposed by matter. This universal curvature, as can be shown by calculation, would produce a mutual repulsion between bodies. Unless these bodies were of minute proportions, this universal repulsion would be counterbalanced and overcome by the more important curvature generated by the bodies themselves; so that, in spite of all, gravitational attraction would be accounted for. Nevertheless, we should be led to a most displeasing duality in our conception of gravitational action, since we should be in the presence of two separate phenomena:
1. The universal repulsion which is totally foreign to matter;
2. The gravitational attraction between matter and matter.
Thus, it would be impossible to escape a duality in our conception of gravitation, if by gravitation we imply the cumulative action of both effects, yielding the one observable attraction of matter for matter.
We may summarise the chief characteristics of de Sitter’s universe as follows: First, this universe can exist in the absence of all matter. It then possesses a definite spherical structure and a radius of curvature,where,which is proportional to the universal curvature, is some constant which bears no palpable relation to anything else in the universe.[107]We must conceive of it as a constant of nature posited once and for all by the Creator. Secondly, if matter is introduced into this empty universe, it will produce local space-time puckers or curvatures around it; and to these local curvatures the phenomenon of attraction between matter and matter will be due. In the presence of matter, de Sitter’s universe will not be stable, for the universal repulsion will tend to cause matter to scatter if the mutual attractions are too feeble to oppose it.
In the opinion of a number of thinkers, de Sitter’s universe presents certain aspects which are most unsatisfactory. We shall discuss these later. At all events, Einstein rejects it and is led, by following a somewhat different line of attack, to a very different type of finite universe, thecylindrical universe.
We shall now proceed to a more detailed study of Einstein’s universe. First of all, we must mention certain general cosmological considerations that directed Einstein to the formulation of a finite-universe hypothesis.
Quite independently of the relativity theory, in the days when space-time was unknown, astronomers had puzzled over certain difficulties dealing with the universe as a whole. Two possibilities appeared to suggest themselves. Either star-matter was present everywhere, distributed more or less homogeneously throughout infinite space, or else the stars were concentrated into a nucleus forming an island of matter in an infinite ocean of space.
The first alternative entailed certain difficulties when we considered the illumination of the heavens at night. Calculation showed that thenight sky should be many times more luminous than it is. Of course this difficulty might have been overcome by assuming that as we wandered farther and farther from the earth, the percentage of dark stars increased; or, again, that vast quantities of cosmic dust intercepted the light and obscured the heavens. It might also have been assumed that light was gradually absorbed in its long journey from the distant stars.
Over and above these first difficulties, it was shown that if the stars attracted one another according to Newton’s law it would be impossible for them to be spread more or less uniformly throughout infinite space; rather would they concentrate into a nucleus. Precise calculation shows that the density of matter, hence the number of stars per unit volume, would vanish at infinity, decreasing progressively from a centre more rapidly than,whererepresents the distance from the centre. It appears, then, that a nucleus of stars cannot be avoided if Newton’s law is to hold.[108]The astronomer Seeliger attempted to realise an infinitely extended universe of matter by modifying Newton’s law in an appropriate way. But, as Einstein remarks, the attempt had neither empirical nor theoretical foundation.
Finally, then, if we accept Newton’s law, we must assume that the universe presents the form of a nucleus of stars lost in infinite space. But here, again, a new difficulty awaits us. In the first place, astronomical observation reveals no trace of any such nucleus. Of course it might be contended that this was due to the limited range of our astronomical observations. Nevertheless, this first argument against the nucleus hypothesis cannot be passed over lightly, for we must remember that in these speculations on the form of the universe, we have to make the best of the flimsiest of clues; we cannot operate as in a laboratory.
Then there is a second argument adverse to the nucleus hypothesis. Calculation shows that the cumulative Newtonian gravitational attraction of the stars would be insufficient to retain any individual star and prevent it from escaping to infinity, leaving the nucleus forever. Similar conclusions would apply to rays of light. It would follow that our universe of stars was in a state of dissolution. It is of course possible that such is indeed the case; yet, to a number of thinkers, a universe in process of dissolution appeared most unlikely. With Newton’s law of gravitation there was no escape. This was thecosmological difficulty which classical science had to face.
Inasmuch as the justification for these statements does not involve any very complicated ideas, we may proceed to explain them briefly. Consider the hypothetical case of the earth situated in infinite space, and isolated from all other bodies. If, then, a mass is abandoned at spatial infinity, it will be attracted towards the earth, gradually gathering momentum, and may finally hit the earth’s surface with a certain velocity, which calculation proves to be about seven miles a second. This particular speedis given by the following formula:whererepresents the value of the Newtonian potential at infinity andits value on the earth’s surface.[109]Conversely, if a body is shot out radially from the earth’s surface with an initial speed smaller than this critical speed,it will always return to the vicinity of the earth. On the other hand, were the body’s initial speed to be greater than the critical speed,it would never return. We see, therefore, that the value of this critical speed depends on the difference in value of the Newtonian potential at infinity and at the earth’s surface.
Now consider the case of a nucleus of stars. The stars will be moving with various velocities under the influence of their mutual gravitational attractions. Statistical mechanics shows that every now and then the total energy of the stellar universe will be transferred to one single star; this star will then be moving at a tremendous speed. The total energy of the sidereal universe being finite, this speed will never be infinite, so that there will exist a maximum finite speed which no star of the nucleus can ever surpass. If, therefore, we consider the Newtonian potential due to the entire nucleus of stars, and if the difference in the value of this potential at infinity and near the centre of the nucleus is sufficiently great, no star will ever be able to tear away and be lost to infinity.
From this, it might appear that a nucleus of stars could endure permanently under Newton’s law. Theoretically, this would be possible, but astronomical observation proves that this possibility must be rejected. For if there existed very great variations in the value of the potential throughout the universe (as the preceding hypothesis would demand), enormous star velocities would ensue. Now the outstanding astronomical fact is that the star velocities are extremely low, proving that the potential varies but slightly from place to place throughout the universe even over vast astronomical distances. It appears, then, that the nucleus is held together very loosely, and a star would not have to be possessed of an inordinately high velocity to tear away and be lost forever.
In addition to this first difficulty, when Boltzmann’s law of gas equilibrium is applied to the stars, treating the latter as so many molecules, the nuclear form of the universe appears to be utterly impossible.
Now, thus far, we have been considering the problem of the universe from the standpoint of classical science and Newton’s law. No great change need be introduced into our arguments when we treat the same problem in terms of Einstein’s law of gravitation and of the infinite quasi-Euclidean universe. Once again, a continually impoverished nucleus would appear to be in order. There is, however, a novel point that may be mentioned. In classical science, we were concerned with the Newtonian potential,and we explored its value throughout the universe by observing the velocities of the stars. With the space-time theory, the potential(or, more precisely, a certain mathematical expression into whichenters) usurps the position of.Nowalso affects the flow of time, hence affects the rate of vibration of atoms (as in the Einstein shift-effect); and so we can detect variations in the value of the potentialfrom place to place through the medium of spectroscopic observations. But when we undertake our observations from a Galilean frame (the earth will do in this case), we notice that lines in the spectra of the most distant nebulæ occupy positions practically identical with those of terrestrial sources, whence we may conclude that as far as the telescope can explore, the value ofremains approximately constant. As for the velocities of the stars, they would be affected by the values of all the’s. So here, again, we see that owing to the low velocities of the stars, all the’s vary but slightly from place to place throughout the universe (when computed in a Galilean frame). Thus, with Einstein’s theory, just as with Newton’s, there appears to be nothing to prevent stars and rays of light from leaving the nucleus; the mutual gravitational attractions would be insufficient to retain them.
In short, if there were to be a self-contained universe,somethingwould have to be modifiedsomewhere.
We now come to Einstein’s final solution, that of thecylindrical universe. He remarked that the low velocities of the stars, together with their more or less uniform distribution throughout space, suggested that they were in a state ofstatistical equilibrium. A few lines will explain what is meant by this.
Suppose that on a billiard table containing a very large number of identical balls we impart a rapid motion to one of them. Our ball will eventually collide with some of the others. These others, in turn, will collide with yet others, and finally all the balls will be set in motion, if there is no friction of any kind, and no loss of momentum when the balls rebound from the cushions, they will go on moving forever. But as time progresses a gradual change will take place. Little by little we shall find that the energy we imparted to the first ball appears to have split itself up among all the balls on the table, so that finally, on an average, all the balls will be moving with approximately the same velocity. When this condition ofequipartition of energyis attained we shall have reachedstability of configuration, orstatistical equilibrium. This state of equilibrium will correspond to maximum entropy, and will endure forever.[110]We see that it is characterised by the fact that all the balls possess more or less the same energies or velocities.
Now, as we have said, astronomical observation would suggest that the stellar universe has reached this condition of statistical equilibrium. It is permissible, therefore, to neglect the trifling irregularities and agglomerations in the distribution of star-matter, and to assimilate the entire stellar universe to a vast cloud of cosmic dust at relative rest, everywhere of the same density. It is then found that for this cloud of dust to maintain a permanent, stable configuration, the time direction of the universe as a whole, for an observer who is non-rotating with respect to the stars,must be straight. It is easy to see why this condition should arise.
If the stars can be considered at relative rest, inasmuch as their world lines constitute geodesics of space-time, it is necessary that these geodesics should not taper together but should lie along parallel time directions through space-time. This is possible only provided the structure of space-time is such that one same time direction endures at every point of space for an observer at rest with respect to the star distribution. We thus obtain one same straight time direction for the universe. Of course, this does not mean that around each individual star the space-time curvature, and hence the time direction, do not suffer modifications. Far from it; since it is these local curvatures which are responsible for gravitational attraction between matter and matter. In the present case, however, we are concerned solely with the structure of space-time as a whole, on a grand universal scale, so that the star distribution can be treated as uniform and likened to a gigantic cloud of dust. Stability, therefore, requires that this ultra-macroscopic structure should be conditioned by a straight time direction. This eliminates de Sitter’s universe, since in his model the direction of time was curved, and his universe, in consequence, not stable.
The net result of Einstein’s calculations was to show that if the universe were a closed three-dimensional spherical (or elliptical)space, not space-time, on which a straight time axis was erected, stability would be possible, the star velocities would remain low, and their distribution uniform. We should thus obtain theCylindrical Universe, where the time direction proceeding from minus to plus infinity would represent the axis of the cylinder.
But here a most important point must be noted. For stability to endure, a certain condition would have to be satisfied. Calculations, based on theformula mentioned onp. 314, prove that the mean densityof matter would have to be equal to.In other words, the total amount of matter in the world would have to be proportional to the radius of the universe.
Assuming these conditions to be satisfied, it would be possible to calculate the volume of the universe, provided we knew the mean densityof the matter distribution. For since,and since the Gaussian curvature of space depends on,a knowledge of this curvature would yield us the total spatial extent of the universe. Now the value of the mean density of star-matterhas been computed by astronomers. If we accept this value of,we obtain for the universe dimensions which are certainly too small, for they conflict with the enormous distances credited to the spiral nebulæ. It would appear, then, as though the cylindrical universe was incompatible with the facts of observation. But, as Einstein remarks, no reliance can be placed on the value ofcommonly accepted by astronomers. In his own words:
“The distribution of the visible stars is extremely irregular, so that we on no account may venture to set down the mean density of star-matter in the universe as equal, let us say, to the mean density in the Milky Way. In any case, however great the space examined may be, we could not feel convinced that there were no more stars beyond that space. So it seems impossible to estimate the mean density.”
Einstein then suggests an indirect method for determining the value of,hence ofand of the size of the universe. We shall discuss this method on a later page; it refers to the motions of the stars in the Milky Way.
It is easy to understand how it comes that in the cylindrical world no nucleus of stars need exist. The two-dimensional analogue of a star distribution spread out uniformly throughout the entire volume of a three-dimensional spherical space would be given by a homogeneous distribution of stars over the two-dimensional surface of an ordinary sphere situated in three-dimensional Euclidean space.[111]It would be the actual surface of the sphere which would be the analogue of the space of the universe, andnotthe volume enclosed within thesphere. Reasons of symmetry would show us at first sight that a uniform distribution of matter on the spherical surface would be stable, there being no reason for a nucleus to formhererather thanthere.
In the same way, an observer moving through the three-dimensional space of the universe would be represented by a perfectly flat being moving over the surface of the two-dimensional sphere. Wherever this imaginary being might move, the star distribution on the sphere’s surface would appear to be the same. Likewise, in the spherical space of the universe, wherever we might move, the distribution of the stars around us would remain homogeneous.[112]Thus the cylindrical universe obviates the displeasing necessity of believing in the existence of an island of star-matter in an ocean of emptiness.
In Einstein’s universe, a ray of light, following a geodesic, would circle round the universe and return to its starting point. This constitutes a difference as against de Sitter’s universe in which the ray of light would be arrested at the passive horizon. In the cylindrical universe, owing to the straightness of the time direction, there is no horizon. For the same reason there will be no slowing down of time in the remote regions of space, so that the general reddening of the light emitted by the spiral nebulæ would appear difficult to account for in Einstein’s universe. De Sitter’s scheme, as we have noted, affords an immediate explanation of this remarkable effect.
It has sometimes been held that the light rays emitted from a star would circle round the universe of space in all directions, and then meet again at the antipodes of space, forming the ghost of a star. It would be well to point out that though theoretically this would be possible, its chances of being realised in practice would be extremely slight, to say the least. The fact is that the perfect sphericity of the universal space can hold only provided matter is homogeneously distributed and everywhere at rest. As this is only approximately the case in the real universe, the spherical spatial universe will be rippled and will deviate more or less from this simple form. As a result, the rays of light emitted from a star will not converge accurately at the antipodes; hence, ghosts of stars need not be considered.
Next we must consider the restrictive condition,which had to be complied with if Einstein’s universe was to be stable. We remember thatis proportional to the space-curvature of the universe, whereasrepresents the mean density of matter. Now it appears extraordinary that the density of matter in the universe should just happen to be equal to.Einstein, however, urges that this equality ofandis no coincidence at all; according to him, there exists a causal connection between these two magnitudes. The density of matter creates,or, again, the total amount of matter in the universe creates the universe of space.Were more matter to be created, the universe would expand; were matter to be annihilated in part, the universe would contract. Indeed, were no matter to be in existence, there would be no space; hence, a totally empty universe would be quite impossible. Thus there is a marked difference between Einstein’s universe and de Sitter’s universe, which can exist without matter.
We may also mention another reason which directed Einstein to the cylindrical universe. We remember that when discussing the boundary values of the’s we saw that de Sitter considered it necessary to obtain boundary values which would remain invariant to a change of mesh-system. In accordance with this requirement, he assumed that all the’s vanished at the boundary. As a result he obtained his hyperbolical universe. Similar considerations prompted Einstein. Here, however, it must be noted that in Einstein’s model, while all the space-’s ()vanish at infinity in our mesh-system, the time-(), owing to the straightness of the time direction, does not vanish. For an observer at rest with respect to the stars it assumes the value +1. It follows that when, instead of merely changing our space mesh-system, we change our space-time mesh-system, the values of the’s are subjected to variations, so that de Sitter’s condition of invariance is not completely satisfied. De Sitter maintains, in consequence, that Einstein’s universe does not lead to a true theory of relativity.
Furthermore, it will be seen that the cylindrical universe introduces a species of absolute time, and this again has been used as an argument against its acceptance. But it should be noticed that this absolute time appears only when we view the universe as a whole. For ordinary problems restricted to smaller areas, the fusion of space and time remains complete, as before.
It is, of course, apparent that by changing the boundary conditions, as Einstein has done, by substituting a cylindrical universe for the quasi-Euclidean one, the gravitational equations will be somewhat modified. They will no longer be those which yielded the double bending of a ray of light. For now the additionalterm will be included in their expression; though, so far as the planetary motions are concerned, the presence of this additional term will not modify our results. In de Sitter’s universe this introduction of the foreign magnitudeinto the law of gravitation was somewhat unsatisfactory, marring the simple beauty of the equations;was some independent constant posited by the Creator. The same argument could scarcely be directed against Einstein’s equations; for it should be remembered that if, as Einstein suggests,is to be attributed to the matter of the universe, the circle is closed, since indirectly, at least, nothing but matter enters into the law of gravitation. According to this view, we might say that the total amount of matter in the universe conspired to create a gigantic law of gravitation. It is, then, under the cumulative action of matter that rays of light and free bodies describe circles.
From a philosophical standpoint, the cylindrical universe is of greatinterest; it justifies Riemann’s premonitions that matter must create the metrics of the fundamental continuum. Riemann, of course, was thinking of space alone, for space-time was unknown to him; but, apart from this difference, Einstein’s results confirm Riemann’s profound views. Indeed, quite aside from these speculations on the universe, the general theory had already brought a partial confirmation of Riemann’s ideas by proving that matter could influence the space-time structure. Riemann’s attitude was, however, more radical. It was not merely a question of matter modifying a pre-existing spatial structure, but of its creating itin toto.
Both in the quasi-Euclidean infinite universe and in de Sitter’s, the rôle of matter, though important locally, was of secondary importance when the world was viewed on a cosmic scale, since the space-time structure could exist independently of all matter. It followed that matter would be accidental, and not essential. Eddington sees no objection to these ideas. He suggests that it may well be in the nature of empty space-time to curl round on itself and manifest a definite metrics, curvature and size. But this attitude leads to a dualistic conception when we remember that matter influences the structure of space-time locally. On this account, both in de Sitter’s and in the infinite quasi-Euclidean universe, the physical properties of space-time, while conditioned in the main by characteristics immanent in the extension itself, would still depend partly upon the presence of matter. We could then scarcely avoid a dualistic conception of space, which to many would be intolerable.
Eddington obviates this dualism by reversing the problem, and by assuming that the physical properties and geometry of space-time arise from the requirements of space-time itself. According to this view, matter is not an active cause, but a symptom, and reduces to mere regions of greater curvature or puckers in space-time or the metrical field. These our senses would interpret as connoting the presence of matter. However, it appears difficult to reconcile these ideas with the complexity of matter, for matter is constituted, in large measure at all events, by electric charges. Hence, to reduce matter to variations in the geometry of space-time would be equivalent to reducing electricity to the’s. Thus far, at least, such attempts have been unsuccessful. At any rate, Einstein rejects this view. He argues that a perfectly empty continuum, a void—call it space-time, or space, or anything else—must be completely amorphous; that is, can possess no inherent metrics or structure of its own, for a void with a structure is not a void at all.[113]
Furthermore, even if it were admitted that space was not really empty,but was filled with an ether that conferred a structure upon it and was responsible for its dynamical properties, it would appear incredible that this structure should be other than homogeneous. Hence, it was logical to assume that the local variations in curvature accompanying the presence of lumps of matter should be attributed to the presence of this matter itself, so that matter would again be a cause, and not a symptom. The simplest way to escape the dual conception which we mentioned previously was, then, to admit that Riemann was in the right, and that the metrics or metrical field of the space-time of the universe was conditioned in its totality, and not merely modified by the presence of matter. At any rate, if these views are adhered to, it is easy to see that the cylindrical universe is the only one that can be accepted, since it is the only one in which the metrics of space-time may be conditioned solely by the action of matter.[114]
It is interesting to note that it is within the realm of possibility that Riemann’s views may be finally submitted to a test. For suppose that astronomical observations should succeed in proving that de Sitter’s universe or the quasi-Euclidean infinite universe was the correct one: Riemann’s ideas would then have to be abandoned. We should have to conclude that a metrics was inherent in an empty continuous extension; though, of course, we might always avoid this attitude, either by assuming that space was atomic or, again, by identifying the metrical field with some subtle ether floating in space.
We now come to still another important consideration which drove Einstein towards the cylindrical universe. We are referring to therelativity of inertia. In the general theory of relativity, prior to Einstein’s investigations on the form of the universe, there existed a number of most displeasing dualities. In the first place, the theory of relativity had succeeded in proving that velocity through space was relative, but it had failed to extend this relativity to accelerated motion. After having made a promising start towards the relativisation of all motion, it had abandoned the attempt halfway before the final goal was reached.
Then again, a similar duality existed in the case of mass and weight. Here we haveweightdepending on the distribution of surrounding matter, so that weight would be inexistent in a world containing but one body. On the other hand, here was the inertial mass of the body which would subsist even in an otherwise empty world. Thus, were we to be transported far from all large masses, we could hold a ton of bricks above our heads without the slightest effort, since the bricks would have no weight. And yet, when we tried to push these same bricks aside, we should encounter the same inertial resistance as on earth. What rendered this duality particularly displeasing was the fact that the general theory had proved that even mass was influenced in a minute degree by the proximity of matter, so that the mass of a body appeared to be due in part to the distribution of surrounding matter and in part to the body itself. In other words, there existed a partial relativityof inertia, just as there existed a partial relativity of motion.
These problems were intimately connected with that other type of duality we mentioned in the preceding pages. There again we saw that space-time was modified by the presence of matter, and yet that even in the absence of matter it maintained a structure. Of course, all these dualities are not equally objectionable to all thinkers; but it must be admitted that for those who crave unity in nature, every effort to remove them will be welcomed.
Einstein had attempted to render radical this relativity of inertia and motion even before discovering his cylindrical universe. He had solved the problem by postulating certain boundary conditions on the fringe of the nucleus of stars in infinite space. Thanks to these boundary conditions, the mass of a body would vanish at infinity, far from matter. But these solutions were abandoned because they appeared to conflict with the low star-velocities. With the cylindrical universe, however, all these difficulties were overcome, and for this reason, among others, it satisfies a natural philosophical craving. Inasmuch as all these considerations present an intimate connection with Mach’s mechanics, we will discuss the significance of Mach’s ideas at some length.
As is well known, on account of the existence of centrifugal force, of the inertial frame and of the rectilinear paths followed by free bodies according to the law of inertia, Newton felt himself compelled to postulate the existence of absolute space. Einstein summarises Newton’s attitude in the following lines:
“In order to be able to look upon the rotation of the system, at least formally, as something real, Newton objectivises space.... What is essential is merely that besides observable objects, another thing, which is not perceptible, must be looked upon as real, to enable acceleration or rotation to be looked upon as something real.”
The opposite stand was taken by Mach, who endeavoured to account for the dynamical manifestations of acceleration and rotation without appealing to that suprasensible entity, “absolute space.” Instead of holding, then, that centrifugal force was generated by a rotation in absolute space, he urged that we should attribute it to a rotation with respect to the material universe as a whole. Inasmuch as the vastest agglomeration of cosmic matter is given by the totality of the star-masses, it would follow that the dynamical effects would be generated by the rotation of an object relatively to the stars, or, what amounts to the same thing, by a rotation of the stars relatively to the object. In fact, these two different alternatives would represent but a difference in phraseology, since with absolute space banished, apart from stars and matter generally, there would exist no other terms of comparison.
To this rotation relatively to the star-masses would be due the protuberance of the equator, the splashing of water in Newton’s bucket, the bursting of a rapidly revolving flywheel, etc. In short, all those dynamical effects which in Newton’s opinion betrayed rotation in absolute space, are held by Mach to betray a rotationrelative to the star-masses of the universe. Now, it should be apparent that Mach’s relativity of rotation entails more than a mere kinematical representation. As he is careful to specify, this active rôle of the stars is due not to the accidental circumstance that they happen to be visible, but to universal gravitational actions which would be generated between matter and matter when a state of relative acceleration or rotation was present. In this way, forces of inertia and centrifugal force are no longer to be attributed to the intrinsic structure of absolute space; they now become akin to gravitational forces. Were the stars invisible, no change in the dynamical manifestations would be observed, but if their masses were to be annihilated, if the stars ceased to exist (not merely ceased to shine), all the telltale dynamical effects would disappear; a flywheel would not burst, water would not splash, and so on. In much the same way, the elliptical orbits of the planets are due to the gravitational attraction of the sun’s mass;not, of course, to the accidental circumstance that the sun happens to be luminous rather than dark.
From these conclusions further important consequences follow; in particular, we must assume that the inertial mass of a body must be generated solely by the mutual actions existing between this body and the other bodies of the universe. We can understand the reason for this statement when we realise that were all the stars to be annihilated, all inertial forces would disappear; but it is precisely the existence of inertial force which is responsible for the effort required to set a body in motion or to arrest it when started. For this reason the disappearance of inertial forces would automatically entail that of inertia or mass. We may note the contrast with Newton’s views, for in classical science the inertial forces arose from the structure of absolute space itself, and mass was regarded as an intrinsic property of matter. In the present case, mass joins weight in being a relative; for just as a body has no weight in the absence of other bodies, so now we see that, according to Mach, it would also present no inertial mass if situated all alone in an otherwise empty world.
When we consider Mach’s mechanics, we see that its net result has been to identify Newton’s absolute space with the space defined by the stars; absolute space has become materialised, so to speak. We may obtain a more concrete representation of the new ideas by imagining a crisscross of threads extending between each individual star and all the others. Rotation, which develops dynamical effects, is, then, the rotation of an object with respect to this network of threads; and it is obviously a matter of indifference to suppose that the object is rotating among the threads and stars, or that threads and stars are rotating around the object. Thus the inertial frame is any frame in which the threads appear to be non-rotating.
Mach’s mechanics might seem to draw too much on our credulity. We can grant that the sun exerts a gravitational effect on the earth, but to assume that the bursting of a flywheel is due to the presence of the stellar universe would appear to be stretching matters too far. Nevertheless, it should be remembered that Newton’s introductionof that suprasensible entity, absolute space, was also extremely distasteful to a number of thinkers, so that it is a question of choosing the lesser of two evils.
These ideas were upheld by Mach for a number of years; and although they can lay claim to no special originality, they are always referred to under his name. Mach’s stand has been so consistently misrepresented by philosophers that we feel justified in warning the reader by mentioning some of the erroneous assertions that have been made with respect to it.
Two centuries ago, Berkeley had defended views which present a certain superficial similarity to those of Mach. He argued that if there were no stars, no permanent points of comparison in the heavens, we could not even imagine the earth’s rotation, hence this rotation would be deprived of all meaning. To this argument the Newtonian would reply that it was, after all, not impossible to imagine absolute rotation as a rotation in a rigid jelly-like substance called space. Furthermore, he would add that it was the dynamical facts, the appearance of unsymmetrically distributed forces, that seemed to confirm the absolute nature of rotation. Absolute rotation was, then, totally irrelevant to our perception of luminous reference points in the heavens, or even to our ability to imagine their existence. So far as the Newtonian was concerned, even were the stars to be annihilated entirely, it would still be quite easy to determine whether or not the earth was in rotation. A number of experiments, such as that of Foucault’s pendulum, the gyroscope, and, more generally, the presence of centrifugal forces on the earth’s surface, would demonstrate the existence of this absolute motion. As against these arguments, Berkeley’s stand offers no reply. Only when, with Mach, we maintain that were the stars to be annihilated, these telltale effects would also disappear, only when we conceive of the stars not merely in a visual capacity or as a support for the imagination, but as producing causal gravitational influences, can an argument be presented against Newton’s absolute space. In short, according to Mach, it would be the gravitational action of the totality of the material universe, including the stars, whether visible or dark, the nebulæ, our sun, the planets and moons, that were responsible for centrifugal force.
Before we proceed farther, a number of objections must be discussed. It might be claimed that whereas the earth could be conceived of as rotating among the stars, it would be quite impossible to consider the stars as rotating round the earth, since in this case centrifugal force would cause the stars to fly apart to infinity. A criticism of this sort arises from a confusion between the Newtonian attitude and the new one. With Newton’s ideas the argument would hold, for the stars would indeed be describing gigantic circles in absolute space. But in the present instance the space of the stars themselves has usurped the position of absolute space, since it is in this star-conditioned space that circular motion generates centrifugal forces. And in their own space the stars are obviously at rest whether or not we regard them as rotating, together with their space, around the earth. Hence, centrifugal force does not arise on the stars, though it will befelt on the earth’s surface. An equivalent way of explaining the same point would be to say: The stars as rotating round the earth are not subjected to centrifugal force, for they are rotating with respect to nothing; the mass of the earth being insignificant in comparison with the totality of the star-masses.
There exists also another type of objection which is based on causality and on our consciousness of effort. For example, we may claim that when, through an act of our will, we spin round on our heel, we know perfectly well that it is our body that is rotating among the stars, and not the stars that are rotating around us. In support of our claim, we should point out that no feeble efforts of ours could ever have started the stars and nebulæ rotating backwards. Curiously enough, where objections of this sort are made, the critic is often fully prepared to accept the relativity of velocity. But, as Einstein points out, arguments of a very similar sort might also be advanced against this simpler type of relative motion. For instance, would not the engineer claim that it must be his train that is moving, and not the earth that is gliding backwards? Could he not also support his contention by pointing out that when he stops feeding the furnace, all motion eventually ceases?
Possibly a more lucid illustration of the relativity of velocity would be afforded by considering the case of a bird in flight. The bird may be conscious of its efforts and realise that when it ceases to beat its wings, its motion through the atmosphere is arrested. It would therefore be entitled to assert that its velocity was maintained through its own efforts, hence was of its own making.
But it is obvious that this velocity created by the bird would be, not absolute velocity through space, but velocity through or relative to the atmosphere. Knowing nothing of the pre-existing velocity of the atmosphere through space, the bird could know nothing of its own absolute motion; and this is all that the principle of the relativity of velocity asserts. And so, were the velocity of the atmosphere through space equal and opposite to the relative velocity of the bird through the atmosphere, the bird’s efforts would result in maintaining its body at absolute rest. Hence, the awareness of effort is no criterion of the existence of absolute motion; relative motion alone is involved.
Exactly the same arguments can be extended to the relativity of rotation, except that in this case we are confronted with the peculiar difficulty arising from the existence of centrifugal force. Mach’s stand obviates this difficulty by introducing a definite physical assumption. Our purpose is, of course, not to justify Mach’s mechanics (for this can be done only through precise experiment, not by a general line of talk), but merely to show that it does not lead to absurdities when the argument based on causality is applied.
Let us consider, then, what is implied in the statement: “We know that it iswewho are rotating on our heel among the stars; not the stars that are rotating around us.” When this distinction is made, the critic is assuming tacitly that there exists a super-space acting as an absolute frame of reference with respect to which rotation acquiresa meaning; for in the absence of an assumption of this sort, the distinction he wishes to defend would, of course, be meaningless.
The statement, then, reduces to the following: “We know that the material universe is non-rotating in the super-space; hence, when we start rotating on our heel with respect to the stars, we know that it is we who are rotating in the super-space, the stars remaining at rest.” This was essentially Newton’s stand when he identified the super-space with absolute space. But Newton at least had given physical significance to this stand by ascribing centrifugal force to a rotation in this absolute space. When, however, we follow Mach and reject Newton’s physical assumption, the absolute space is deprived of all possibility of manifesting itself physically. And so the critic’s assertion loses all physical significance, since henceforth the only space in which rotation can manifest itself physically is the star-space.
The metaphysician might argue, however, that Mach’s interpretation of centrifugal force does not obviate the existence of a super-space, but merely denies us the ability to establish its existence by physical means. Let us grant this contention and see where it leads us. When we assert that it is we who are rotating on our heel, we are assuming implicitly that prior to our rotation the universe and our body were non-rotating in the super-space. Were we to cast any doubt on the correctness of this initial assumption, no conclusions could be drawn. For suppose that initially the universe, together with our body, was rotating clockwise in the super-space; then, as a result of an anti-clockwise rotation on our heel, it would now be we who were non-rotating in the super-space, while the universe of stars persisted in its state of rotation. Hence, the stars would be rotating around us. So we see that according to our initial assumptions as to the state of rotation of our universe in the super-space, we could derive any conclusions we pleased. The problem reduces, then, to determining the precise rotationary motion of our universe in the super-space. Now, inasmuch as this rotation is precluded from manifesting itself in any physical way, since centrifugal force has already been accounted for, the problem can of course never be solved by the physicist. As for the metaphysician, unless he can furnish us with some supra-physical means of determining the true situation, he also will be incapable of supplying us with the required information. It follows that if Mach’s premises are accepted, the entire super-space conception becomes meaningless, for any assumption whatever can be made in regard to it with impunity.