It will be convenient if I summarize very briefly the conclusions that have
just been reached. They are as follows.
One should expect a cosmological model based on Symmetrical Impermanence to contain some extensive clouds of very tenuous gas. In an
expanding universe these would begin to form as soon as the distance
between adjacent galaxies exceeded a certain critical value. At first these
clouds would grow rapidly in extent, while they still remained so tenuous
that the gas between stars within our galaxy would have to be called dense
in comparison. But when these extragalactic clouds had acquired a substantial gravitational field of their own, they would capture hydrogen from
their surroundings and would also shrink under the influence of their own
gravitational fields. Hence they would become more dense. But in spite of
their smaller volume they would also become more massive, for loss from
matter falhng out of the cloud would cease and be replaced by gain from
the matter that fell into it.
These clouds would bear but little resemblance to the spiral nebulae.
Nor will the closer study of their formation that is to follow shortly
improve the resemblance. On the contrary. It will be found that the
incipient clouds must differ from spiral nebulae quite significantly in shape,
in size, in the way matter within them is distributed, in their own intrinsic
Can it then be inferred that the cloud will eventually develop into a
spiral nebula, into a galaxy similar to those with which astronomers are
familiar? If one cannot infer this, the cosmological model that is based on
Symmetrical Impermanence does not resemble actuality and this basic
hypothesis must be abandoned and with it the even more basic Principle of Minimum Assumption. Much depends on whether one can infer from
established knowledge and this principle that the extragalactic cloud can
develop into a familiar spiral nebula.
For this reason it now becomes necessary to consider in detail what one
should expect to happen to an extragalactic cloud after it has begun to
form on an astronomical summit. But such consideration can with ad-
vantage be preceded by an appreciation of the task that it involves, and
this will be most easily obtained, be it at the cost of some repetition, if I
recall what has been said in Chapter 2 about scientific method.
To infer from established knowledge what happens to an extragalactic
cloud is comparable to what engineers do when they infer the performance
of machines that have been designed but not yet made machines that may
embody entirely new principles and that may be quite different from anything that has been made before.
To draw such inferences taxes the engineer's powers of logical reasoning
considerably. It calls for great care lest some decisive factor be overlooked;
it may require the use of some new mathematical tool; the design may have
to be preceded by much experimental research of which the purpose is to
extend the range of established knowledge. But in engineering the new
machine does not have to be manufactured and tried out before anything
is known of its performance. The designer knows, of course, that some
mistake or unforeseen circumstance may falsify his predictions, so his
performance figures include margins, technically known as tolerances.
When there is some doubt, a prototype may precede mass production.
But these precautions do not imply that the engineer lacks all faith in his
own powers of calculation and logical inference; and in the great majority
of cases experience shows a close agreement between inference and
Amateur scientists and philosophers often fail to appreciate this. The
technique of inference, which is a commonplace in engineering, appears
to them as unpardonable speculation. A consequence of this is that both
scientific and philosophical conclusions are too often assessed by the
wrong criterion. When a conclusion is arrived at by a process of inference,
the proper course is to examine the reasoning that has led to the conclusion.
This may have been faulty and can then be exposed by better reasoning.
But the history of science abounds in occasions when this obvious duty
has been neglected. For one critic of Darwin who examined his reasoning
in the early days of evolutionary theory a hundred dismissed the theory
by quite different, and far less sound, criteria.
There are sundry reasons for this and one is man's natural intellectual
indolence. The detection of faulty reasoning requires a substantial mental
effort, while other criteria can be applied with little or none. One of these
false criteria is applied when the result of inference from established
knowledge is dismissed as 'mere speculation'. If this happens more often
than one should expect, it is because the difference between inference and
speculation tends to be obscured.
Confusion sometimes arises from giving a false status to observation.
There are schools of philosophy in which the validity of any statements
about unobservables is denied. These schools find ready adherents, for
the notion seems superficially plausible that one cannot say anything
valid about something that one has not and cannot observe. However,
the plausibility evaporates when one remembers the everyday work of
scientists and engineers.
A machine that has never been made is certainly unobservable at the
time when the designer makes statements about its performance, and so
the philosophers of this school would have to say that such statements
could not be justified scientifically. These philosophers advocate a kind of
empiricism that would make the engineer's task impossible. It is fortunate
that most physicists, and all engineers, know better than to practise what
is sometimes preached to them.
Cosmology is typical, but by no means unique, among the fields of
study in which one cannot afford to listen to these schools. In cosmology
inferences have to be drawn about systems that are just as unobservable as
a machine that has not yet got past the blueprint stage. Predictions in this
field require the same disciplined care as in engineering; if this care is
exercised, they are just as valid; they are subject to the same kind of un-
certainty. As often happens in engineering, moreover, inferences cannot
be checked with the help of small-scale models.
All this applies to the present theme. We are here trying to draw
inferences about the behaviour of masses of gas so enormous that their
movement is largely governed by their own gravitational fields. Such
masses are not at the disposal of the laboratory technologist. The telescope
tells us a little about them, but by no means all that we need to know. If
we are to extend our knowledge in the field of cosmology, we must adopt
the engineer's technique of inference. There is no alternative and those who
deprecate inference do not suggest one. One must either use this technique
with the thoroughness and care of the machine designer, or abandon all
hope of gaming a better insight into cosmological processes.
Top of Page