In general relativity, which, though less firmly established than the special theory, is intended to explain gravitational phenomena, a more complicated metric of variable curvature is employed, which approximates to the Minkowski metric in empty space far from material bodies. Cosmologists who have based their theories on general relativity have sometimes postulated a finite but unbounded space-time (analogous, in four dimensions, to the surface of a sphere) as far as spacelike directions are concerned, but practically all cosmologists have assumed that space-time is infinite in its timelike directions. Kurt Gödel, a contemporary mathematical logician, however, has proposed solutions to the equations of general relativity whereby timelike world lines can bend back on themselves. Unless one accepts a process philosophy and thinks of the flow of time as going around and around such closed timelike world lines, it is not necessary to think that Gödel’s idea implies eternal recurrence. Events can be arranged in a circle and still occur only once.

The general theory of relativity predicts a time dilatation in a gravitational field, so that, relative to someone outside of the field, clocks (or atomic processes) go slowly. This retardation is a consequence of the curvature of space-time with which the theory identifies the gravitational field. As a very rough analogy, a road may be considered that, after crossing a plain, goes over a mountain. Clearly, one mile as measured on the humpbacked surface of the mountain is less than one mile as measured horizontally. Similarly—if “less” is replaced by “more” because of the negative signs in the expression for the metric of space-time—one second as measured in the curved region of space-time is more than one second as measured in a flat region.

Strange things can happen if the gravitational field is very intense. It has been deduced that so-called black holes in space may occur in places where extraordinarily massive or dense aggregates of matter exist, as in the gravitational collapse of a star. Nothing, not even radiation, can emerge from such a black hole. A critical point is the so-called Schwarzschild radius measured outward from the centre of the collapsed star—a distance, perhaps, of the order of 10 kilometres. Something falling into the hole would take an infinite time to reach this critical radius, according to the space-time frame of reference of a distant observer, but only a finite time in the frame of reference of the falling body itself. From the outside standpoint, the fall has become frozen. But from the point of view of the frame of the falling object, the fall continues to zero radius in a very short time indeed—of the order of only 10 or 100 microseconds. Within the black hole, spacelike and timelike directions change over, so that to escape again from the black hole is impossible for reasons analogous to those that, in ordinary space-time, make it impossible to travel faster than light. (To travel faster than light a body would have to lie—as a four-dimensional object—in a spacelike direction instead of a timelike one.)

As a rough analogy, two country roads may be considered, both of which go at first in a northerly direction. But road A bends round asymptotically toward the east; i.e., it approaches ever closer to a line of latitude. Soon road B crosses this latitude and is thus to the north of all parts of road A. Disregarding Earth’s curvature, it takes infinite space for road A to get as far north as that latitude on road B; i.e., near that latitude an infinite number of “road A northerly units” (say, miles) correspond to a finite number of road B units. Soon road B gets “beyond infinity” in road A units, though it need be only a finite road.

Rather similarly, if a body should fall into a black hole, it would fall for only a finite time, even though it were “beyond infinite” time by external standards. This analogy does not do justice, however, to the real situation in the black hole—the fact that the curvature becomes infinite as the star collapses toward a point. It should, however, help to alleviate the mystery of how a finite time in one reference frame can go “beyond infinity” in another frame.

Most cosmological theories imply that the universe is expanding, with the galaxies receding from one another (as is made plausible by observations of the red shifts of their spectra), and that the universe as it is known originated in a primeval explosion at a date of the order of 15 × 10⁹ years ago. Though this date is often loosely called “the creation of the universe,” there is no reason to deny that the universe (in the philosophical sense of “everything that there is”) existed at an earlier time, even though it may be impossible to know anything of what happened then. (There have been cosmologies, however, that suggest an oscillating universe, with explosion, expansion, contraction, explosion, etc., ad infinitum.) And a fortiori, there is no need to say—as Augustine did in his Confessions as early as the 5th century ce—that time itself was created along with the creation of the universe, though it should not too hastily be assumed that this would lead to absurdity, because common sense could well be misleading at this point.

A British cosmologist, E.A. Milne, however, proposed a theory according to which time in a sense could not extend backward beyond the creation time. According to him, there are two scales of time, “τ time” and “t time.” The former is a time scale within which the laws of mechanics and gravitation are invariant, and the latter is a scale within which those of electromagnetic and atomic phenomena are invariant. According to Milne, τ is proportional to the logarithm of t (taking the zero of t to be the creation time); thus, by τ time the creation is infinitely far in the past. The logarithmic relationship implies that the constant of gravitation G would increase throughout cosmic history. (This increase might have been expected to show up in certain geological data, but apparently the evidence is against it.)