Annurev Aa 20 090182 000245

ANNUAL
REVIEWS Further
Quick links to online content
Ann. Rev. Astron. Astrophys. 1982. 20:1-35
THE UNIVERSE: PAST AND

PRESENT REFLECTIONSl
Fred Hoyle
Cumbria, Great Britain
Defining the Universe to be everything there is, manifestly we cannot be

expected to understand it exactly, since to do so would require a complete
command of the laws of physics and the fantastic calculating power to work
through the detailed properties of assemblies containing very large numbers of
particles . What then are astronomers doing in their studies of cosmology?
Obviously, we are beavering away to give an imperfect answer to an imper
fectly defined problem. The issue is the quality of our approximations. I
suppose nine astronomers out of ten work on the presumption that our approx
imations are quite meritoriously good . In this essay I hint that it may be
otherwise.
Let me begin autobiographically. I did my undergraduate work in mathe
matics and my graduate work in theoretical physics , the latter in the closing
stages of the golden age of quantum mechanics and in the opening phase of
the golden age of nuclear physics, a lucky time in which to receive an
education. I was lucky too to have spent five wartime years at what was then
the frontier of electronics. So a little by deliberate choice, but mostly by
accident, I found myself tolerably equipped in 1945 for tackling some of the
rewarding problems with which astronomy was then strewn, like the treasures
of Ali Baba's cave. I hesitated to plunge ahead into cosmology , however, for
reasons substantially the same as those explained above. It seemed highly
unlikely that the knowledge-of-the-day could possibly be adequate to deal with
the whole Universe; and yet if one did not proceed as if it was , it would be
necessary to think outside what was already known, and how is it possible to
think outside one's knowledge?
To think completely outside is of course impossible. Nevertheless, one can
go beyond the detail of what is currently known, provided one maintains the
'Portions reprinted from Engineering & Science Magazine, November, 1981. Published at the
California Institute of Technology, Pasadena, California.
2 HOYLE
"style" of physics, and indeed some of the most profitable adventures in

science have come in just this way. The special theory of relativity had been
in the air a long time before Einstein in 1 905, certainly from the Michelson
Morley experiment of 1887. There had been contradictions between the old
details of Newtonian theory and new details contained in work by Fitzgerald,
Lorentz, Poincare, and Planck. What Einstein did was to put an issue of
style-all observers in uniform motion measure the same speed of light
ahead of the confusion of detail. When this was done consistently, a lot of the
old detail went out of the window, to be replaced by deductions that were
subsequently shown by experiment to be correct. Of course, physicists never
admit to "style," because the word brings an image of Beau Brummel to mind.
Instead they talk loftily about "principles," a stuffed-duck posture. Style it is,
and principle it is not.
Two more bits of autobiography. In 1938-39 I had the good fortune to
become a research student of Paul Dirac, who impressed on my young mind
the preeminence of style over detail. And in 1939, when I first became a
Fellow of St. John's College, Cambridge, I had Ebenezer Cunningham as a
colleague. Ebenezer was of the older generation of Cambridge mathema
ticians. He had been young himself at the time of Einstein's work of 1905. He
told me that immediately after Einstein' s paper was published, he had sat
down one afternoon to read it, and in only a few hours he passed from an acute
perception of the previous difficulties to seeing that Einstein had resolved
those difficulties. I think it is generally agreed that of all British scientists in
those days Cunningham was the nearest to discovering the special theory of
relativity himself. I asked him if he had been disappointed to find another
young man had got there ahead of him. He replied in the following general
terms: "In some degree , naturally, especially as the mathematics itself was
within my powers . But the disappointment was compensated by an extreme
sense of clarity . In any case, I could see that I had just not been ruthless
enough. "
So it came about, again b y good luck, that I had learned the importance of
style from Paul Dirac, and the importance of ruthlessness from Eb Cunning
ham, that most Christian of gentlemen. But what could style and ruthlessness
do for one in cosmology? This was the question I asked myself in the imme
diate postwar years.
To answer this question I must tum for a moment to the game of chess.
Expert players assert that in a high-level game only two or three moves really
matter, the rest are more or less automatic . In one case known to me this was
certainly true. In 1972, Bobby Fischer wrested the world championship from
Boris Spas sky by a rather overwhelming margin .. Yet the outstanding moment
of the match belonged to Spas sky . In the 1 1 th game, in a variation extensively
analyzed by Fischer, Spassky on his 14th move played the queen's knight
REFLECTIONS ON THE UNIVERSE 3
backward from c3 to its home square of b l. Overlooked for many years by the
whole international chess fraternity , the move instantly converted a probable
win for Fischer into an essentially certain loss. So much can tum on a single
move , not only in chess but also in cosmology.
The corresponding decisive move in cosmology is given by a yes-or-no
answer to the following question:
Did the whole Universe come into being , all in a moment , about ten
billion years ago?
In 1947, when I first began to ponder this question, the geologist Arthur
Holmes was asserting with immense courage that the measured age of the
Earth was greater than the age of the entire Universe as then given with
confidence by Edwin Hubble. One therefore needed little in the way of an
enquiring mind to start pondering. Although astronomers have since increased
Hubble's age estimate about fivefold , much the same situation exists today.
Thus the currently favored value of about 100 km S- l mpc-1 for the "Hubble
constant" runs foul of the age of our galaxy , whether the latter is determined
by nuclear methods or from the lifetimes of the oldest stars.
The importance of age estimates is to force the above question seriously into
the mind, not to determine the answer to it-the age estimates are too uncer
tain to be relied on for so crucial a decision. By considering the question
seriously I mean contemplating the possibility that the .answer to it might be
no. Many believe there is a given answer, like a problem in a student exam
ination , and that the ex cathedra answer is yes. It then seems axiomatic that
discrepancies of the kind mentioned in the previous paragraph must inevitably
be resolvable, a point of view that invites the shading of calculations and the
slanting of data. Besides which , the attribution of a definite age to the Uni
verse , whatever it might be , is to exalt the concept of time above the Universe,
and since the Universe is everything this is crackpot in itself.
I would regard the need for the Universe to take precedence over time as a
knockout argument in favor of a negative answer to the above question , if we
could be certain our ever-present idea of particles of various sorts existing in
four-dimensional space-time was correct. One could then dismiss cosmologies
of finite age because they were offensive to basic logical consistency. But one
has to contemplate surprises; indeed , the following speculation would be a
surprise if it were true.
Nowadays , the particles of the 1 930s , which we blissfully thought to be just
four in number-electrons , protons , neutrons , and neutrinos-are known to
be,.complex aggregates of more elementary particles-quarks, God save us.
The wave function of a quark has a triple hierarchy: the "spin" multiplicity of
Dirac , a multiplicity of five or more called "flavor ," and a multiplicity of three
or more called "color." Each multiplicity has its own group of mathematical
4 HOYLE
transformations , with "spin" determined by general complex transformations

in two dimensions (unimodular), and with both "flavor" and "color" deter
mined by higher-dimensional complex transformations that are, however, less
than general-they are restricted to be unitary in order to maintain conser
vation laws among the particles.
It was my intention on leaving Cambridge in 1 972 to try removing the
separated categories of these three groups of transformations to permit them
to become mixed together by complex transformations. One would then be
unable in general to speak about such-and-such a form of particle existing in
space-time, because space-time (contained in the two dimensions of a general
complex space) would be amalgamated with the spaces determining the nature
of the particles . As things fell out I did not carry through this program.
Publicly, my reason was that I became involved in other things. Privately, I
became scared by the mathematical difficulties .
On such a point o f view, the Universe would be described by a single
abstract space of high dimensionality in which the general transformations
were permitted. Our everyday description of the world as a set of particles "in"
space-time would be applicable only to a subspace, or subspaces, of the
general space. One could then suspect that it is the special projection into a
particular subspace of the state-vector of the portion of the Universe we
observe that determines the detailed numerical values of the so-called coupling
constants of physics.
When astronomers refer to the "age" of the Universe, could it be that the
"age" is really the time interval since the state-vector went into such a sub
space? If so, it would be invalid to ask what happened "before" in a temporal
sense, because the state-vector would belong to the general complex space to
which the space-time concept could not be applied. To be sure, one might still
give meaning to "before" in an abstract sequential way, but not in the usual
way with respect to "time" as a real parameter.
I find this concept attractive, but I do not find it attractive to suppose that
the portion of the Universe we observe experienced such a drastic trans
formation only ten billion years ago. To explain my position I must tum to
biology. To most astronomers the thought of information crucial to cosmology
being derived from biology will , I suppose, appear ludicrous. But the Uni
verse is everything, and to omit information from any source, especially
biology with its vast store of information, would be truly ludicrous.
In Steady-State Cosmology Revisited (University College Cardiff Press ,
1 980) I estimated (on a very conservative basis) the chance of a random
shuffling of amino acids producing a workable set of enzymes to be less than
1 0-40000• Since the minuteness of this probability wipes out any thought of life
having originated on the Earth, many whose thoughts are irreversibly pro-
grammed to believe in a terrestrial origin of life argue that the enzyme estimate
is wrong. It is-in the sense of being too conservative.
Could the vast store of information necessary for the development of biol
ogy have been accumulated in only ten billion years? If you are inclined to
think that it could, take a look at what we know of the most recent four billion
years, and what many people believe to be the beginning' of the Universe in
a big-bang cosmology . Such a beginning occurs in a holocaust of radiation
little suited to harboring the delicate organization of biology, while the past
three to four billion years on the Earth have yielded no change in the intricate
biochemical complexity of life. The enzymes go essentially unchanged from
the cells of a human to the most primitive single cells, which are thought to
be typical of life as it existed in the early days of the Earth. Hence we have
a situation without a promising beginning and with no change of the crucial
aspects of the life system over the last one third to one half of the ten billion
year time interval. Where then did the miracle of information contained in
biological systems arise? How does one deal with a probability as small as one
part in 1040000? In my view only by giving the Universe a very long history ,
much longer than ten billion years.
Using this 1 980 argument, one arrives at the same rejection of a short age
for the Universe that I started from in 1 947-48, a position which meant
changing the details of cosmological theory but not changing the style (a
condition demanding that one keep to Einstein's general theory of relativity).
Einstein's theory equates a set of quantities (tensor) determined by the geo
metrical structure of space-time to another set of quantities of a physical nature
known as the energy-momentum tensor. Nobody had ever pronounced an edict
as to exactly what the energy-momentum tensor must be, except that it be
derivable from an "action principle." Hence it followed that whatever one
could do to cosmology had to be done in the action formula.
Although several kinds of "field" appeared in the action formula, nobody to
that time had introduced into cosmology the simplest field of all, a scalar
function of position. So it was obvious that a scalar should be tried (the C -field
as I called it). There was little freedom in how a scalar could be so employed,
either as a pure field term in the action or in coupling to the particles (classi
cal) . So the theory more or less ran itself.
The consequences were surprisingly far-reaching, like those of Spassky's
apparently simple move against Fischer. But unlike Spassky's move , which
immediately drew approval, the new theory was soon in trouble with both
astronomers and physicists . The new equations required matter to be created,
which was said to be impossible. Although this opinion did not impress me
unduly-I thought it only a guess, which it was-the criticism persuaded most
astronomers to treat the theory in a cavalier manner. On the pretext that any
6 HOYLE
stick is good enough to beat a dog with , the theory was assailed by obser
vations with low signal-to-noise ratios that were claimed to be disproofs. Yet
throughout this criticism one could hold fast to the critical point that , without
departing from the style of physics, the short "age" of the Universe had been
banished; the Universe was everlasting , into the past as well as into the future .
The term in the action formula that coupled particles to the C -field, and the
pure C -field itself , contained an ambiguity of sign. This was not unexpected
because choices of sign arise with other widely accepted terms in the action.
For example, the term giving rise to ordinary gravity would , if its sign were
switched, make gravity a repulsive force instead of an attractive one. The
choice required by the new cosmological theory made the energy density of
the C -field negative, a condition which in my student days I had come to think
impossible (at least when there was a field-to-particle coupling , as there was
in this case). The argument went as follows: The creation of particles with
positive energy would make the field energy more negative , causing the
strength of the C -field to increase. This would have the effect of creating more
particles at an increased rate , making the field energy still more negative and
the C -field still stronger. And so on , into a catastrophic instability .
The argument is a poor one , however , because it assumes space-time to
remain flat. A field of negative energy density acts like negative gravity ,
causing expansion , or explosion if the situation is sufficiently drastic. It was
indeed the negative energy density of the C -field that produced the recession
of the galaxies in the new cosmology , thereby explaining the expansion of the
Universe in physical terms instead of assuming it ad hoc, as was done in other
theories.
In the early days of the new theory , 1950 or thereabouts, we knew little yet
of the violent local explosions that are all the rage in astronomy nowadays.
Otherwise it would have seemed natural to attribute them to creational in
stabilities caused by local condensations of the C - field , and it would have been
hard then for the older theories to have survived in the face of such evidence.
True, astronomers today have convinced themselves of other ideas for explain
ing the now-observed local explosions , but these ideas are also unattractively
ad hoc.
New ideas have never come to me by winging their way down from the
clouds , but from calculating orthodox positions , and then finding that situ
ations did not work out as they were supposed to do. The following is an
example that began in a very innocent way.
In the 1 950s astronomers thought the interstellar grains to be water-ice.
Nothing at all was riding on this issue for me , and I would gladly have believed
the conventional point of view if calculation had shown it to be viable. The
trouble was that interstellar grains are constantly changing their positions in
relation to the stars , and even the briefest sojourn of a water-ice grain in a
region where the temperature rises to, say, -150°C will cause evaporation. I
did not find that, once evaporated, water-ice grains would recondense by
themselves under any conditions that seemed plausible. I mentioned this
difficulty in a general astronomy book!, suggesting the grains must ,consist of
a more refractory substance than ice, as for instance gJ;aphite. The matter
rested so lightly with me, however, that I did nothing about it until some five
years later when I had a research student (Chandra Wickramasinghe) in need
of a problem.
One encouraging indication was that the physical properties (optical con
stants) of graphite happened to be such as would give a reasonable approxi
mation to the observed 1/ A law of extinction that the grains produce in the
visual light of distant stars. Furthermore, the behavior of the optical constants
with respect to frequency enabled us to predict that graphite would produce
enormous extinction in an ultraviolet waveband centered at about 2000 A.
When this predicted large extinction was actually discovered approximately
one year later from rocket firings, it seemed certain that there must be some
thing right with the graphite idea.
There are those who are so uncomfortable with new situations that their
practice, on hearing a new idea, is to search for an immediately overriding
objection to it. I work in the opposite way. To begin with, I search for the good
things to be said about a new idea. If some emerge, and especially if they look
strong, I then tum to criticism. And the stronger an idea becomes the more
relentlessly I search for objections to it. In accord with this methodology, the
time had come by 1965 to put the graphite idea through the wringer. By this
time, enough was known of the reflectivity of interstellar grains for Wick
ramasinghe and I to see that the reflectivity of graphite was too low. Graphite
was too absorptive, too black. There were also difficulties with the technical
details of what is known as the "polarization" produced by the grains. It
seemed therefore that while there was something right about graphite, the
graphite theory could' not be totally correct.
Water-ice has a very strong infrared absorption band near 3 . 1 micrometers,
but attempts to observe this band had not yet shown in the mid-1960s that
water-ice is almost completely absent from grains in the general interstellar
medium. We were not debarred therefore from considering a composite grain
model-grains with graphite cores and water-ice mantles . Of course there was
still an evaporation problem for the ice, but at least it was more tolerable to
have water-vapor condensing around already existing graphite cores than to
have ice grains condense de novo.
This composite core-mantle theory was a parameter-fitting enthusiast's de
light. The shapes and sizes of the particles could be varied, as well as the
'Frontiers of Astronomy, Wm. Heinemann, 1955.

8 HOYLE
relative proportions of graphite and ice. By now, the Institute of Theoretical

Astronomy at Cambridge had a fair-sized IBM computer on which Wick
ramasinghe calculated an immense number of cases. With so many parameters
available, a moderate correspondence with all the data was inevitable and
could not therefore be considered much of an achievement. The important
thing was to obtain a really good correspondence, and this holy grail eluded
us with maddening persistence. Starting from a moderate agreement with all
the data, we would tune-up the parameters to get some particular feature
almost exactly right, only to find the correspondence with the rest of the data
had become worse. Gradually it was born in on us that we had a wrong theory
on our hands. To add to our troubles, observations were showing grain prop
erties to be remarkably uniform over the whole galaxy, and it would be
impossible to obtain uniformity with the many floating parameters we were
using.
So it came about that in the later 1960s we began thinking of what other
kinds of particles there might be. As the search intensified we became in
creasingly desperate, to a point where we even tried grains of solid hydrogen,
although at best such grains could exist only in the coolest dense clouds and
not at all in the distributed interstellar medium. From this escapade something
important emerged, however. Certain details, which had never been right
before, fitted immediately into place for solid hydrogen. This limited success
proved to be a consequence of the very low refractive index of solid hydrogen,
much lower than we had thought to use before. Just as the ultraviolet ex
tinction showed there was something right about graphite, so these later
calculations for solid hydrogen showed there was something right about a low
refractive index.
Accepting that the ultraviolet extinction near 2000 A was caused by graphite
particles, the observations were now good enough for the sizes and shapes of
the particles to be inferred. The result was another shock. The particles had
to be spheres with diameters not greater than a few hundred Angstroms. No
method of forming graphite particles we could then think of would produce
such particles-we expected either long whiskers or sooty plate-like struc
tures.
We considered grains composed of magnesium Gxide (MgO) and silica
(Si02), and when about a year later infrared observations near 10 micrometers
indicated the existence of circumstellar "silicates" the moment again seemed
right for hats to be thrown into the air. But true to our previous experience
there were troubles. Silica crystals have two enormously strong infrared fea
2
tures (absorption coefficient � 30 ,000 cm g-I) and neither was seen, even
as minor effects. While these strong narrow bands could be broadened into
more indefinite features by combining Si02 with some other molecule, as with
MgO in enstatite (MgSi03) or forsterite (Mg2Si04), we could not believe this
would be the case for every last scrap of silica, such as would be necessary
for the strong features not to show up at all. Besides which, I was convinced
that the astronomical community had been misled by a wrong calculation,
namely a calculation for a strictly thermodynamic situation, when MgSi03 and
Mg2Si04 are slightly more stable than MgO and Si02 taken separately. But
grains are not formed in a strictly thermodynamic situation. Such grains were
thought to form in outward flows of gas from stars, and such flows are so
markedly nonthermodynamic that MgO and SiOz would remain separate, if
indeed Si02 forms at all.
In 1969, E. M. Purcell published an interesting calculation that showed how
the minimum amount of matter required to produce the observed extinction of
starlight could be calculated from general physical principles. Purcell's work
showed that, relative to the observationally determined amount of interstellar
gas (mostly hydrogen and helium), the condensable materials are remarkably
efficient. Provided the condensed grains are of optimum shapes and sizes, the
amount of the condensable materials is sufficient to explain the observed
extinction of starlight, but not by any great margin, and then only if an
appreciable fraction of interstellar carbon, nitrogen , and oxygen is condensed.
If one were to omit the C , N, and 0 from the grains , the amount of the rest,
such as MgO, SiOz, CaO, and Fe would be insufficient to explain the ex
tinction by a factor of at least 3 .
Precisely because of the far-reaching implicatiori of this result, there i s a
disposition among astronomers to deny it. Indeed, one should deny it, but not
in the sense that many would like. The factor 3 by which such materials as
MgO, Si02, CaO , and Fe fail to explain the observed extinction is calculated
using the assumption that the grains into which these materials are condensed
are optimum in their size distribution for producing extinction over the whole
sweep of wavelengths from one micrometer to 1000 A. For inorganic grains ,
which have no strong size-determining property, this is an implausible condi
tion. It becomes even more implausible when one recalls the remarkable
uniformity of the extinction over the whole galaxy. The prudent conclusion is
that were C, N, and 0 excluded, the grain-forming materials in the interstellar
medium would be deficient by a factor of at least 5 . A corollary is that the
grains responsible for most of the interstellar extinction must be largely com
posed of C, N, and 0, with the possibility that hydrogen is associated with
these elements . Since inorganic solids built from H, C, N, and 0 evaporate
much too readily, the inference is that the grains are organic, an inescapable
conclusion unless some serious mistake has been made in estimating the total
quantity of interstellar material. Wickramasinghe and I reached this conclu
sion with an initial sense of bewilderment, for how could organic grains be
formed in great quantity throughout the interstellar medium, and similarly
everywhere with a size distribution centered at about 0 . 7 micrometers?
10 HOYLE
After leaving Cambridge in 1972 I had other things beside interstellar grains
to think about, and it was not until 1 976 that I returned to this question.
Meanwhile, Wickramasinghe had considered polyformaldehyde as a possible
grain-forming material, (CO H2)n, built from the two commonest molecules in
the Universe, H2 and CO. A more complex, but more stable, substance built
from the same elementary ingredients is obtained by forming rings, usually
from five or six COH2 groups , with some adjustments of atoms between the
groups, and by then linking the rings through oxygen atoms into aHnear chain,
with the elimination of an H20 molecule at each link, rather than 'having .the
carbon atoms joined directly one to another along the chain. This is :the
difference between polyformaldehyde and a polysaccharide.
My father was a wool merchant and in my early youth he taught me a simple
way to distinguish a strand of real wool from imitations. Imitations burned
leaving a trail of ash; wool burned by shriveling, with a little ball of free
carbon accumulating at the burning end. I remembered this observation from
days long gone by. Here at last was a way to obtain the small carbon spheres
demanded by the ultraviolet data, from the degradation of a suitable organic
polymer, as for instance the keratin in wool.
Wickramasinghe dug out from the literature an infrared transmittance curve
for the common biopolymer cellulose. A glance at the curve showed it to have
properties of greater interest in the infrared than anything we had seen before.
At its longwave end the curve was like that which astronomers had christened
"Trapezium material," while at the shortwave end there was a broad absorp
tion similar to that due to water. A . H. Olavesen kindly obtained a carefully
calibrated cellulose spectrum for us, and he also showed that a representative
sample of other polysaccharides all had spectra very much like cellulose.
This was sufficient for a number of interesting calculations to be done, with
more satisfactory results than anything achieved in the 1960s. The way ahead
seemed to be to press the calculations to finer and finer limits. Conscious that
there might be a charring of organic material toward the inner regions of our
sources, with a consequent variation of transmittance properties forcing rather
complex calculations, I felt the need for a more sophisticated computer than
my little hand-held Hewlett-Packard . I therefore applied to the Science Re
search Council for a modest grant wherewith to purchase a suitable mini
computer. It is a matter of history that the application was refused, not just
once but for a second time upon appeal. I mention this affair not to s�ggest
that the Science Research Council be summarily dismantled (whieh , of course,
it should be), but to.e�plain,why the line of research pursued so far had to be
abandoned. It was now necessary to adopt what Americans call an end-around
play.
An interesting question forced itself on one's attention. With the realization
that the interstellar grains are largely organic, one sees that the material of the
early solar system must have contained an enormous quantity of organics, at
least 3,000 Earth masses. Much organic material would be destroyed by the
heat of the solar nebula, but much would survive in the comparatively cool
outer regions, especially in the regions of the distant comets. And since at
subsequent times a fraction of comets have developed orbits of high eccen
tricity, bringing them to the inner regions of the solar system, with a part of
their evaporated material enmeshing the terrestrial atmosphere, there was a
known process for transferring organic material from the outer distant regions
of the solar system to the Earth. Could this potentially very large and con
tinuing source of organic material have formed the basis for the origin of life,
rather than the comparatively trifling quantity of organics generated in terres
trial thunderstorms and other small-scale events?
Wickramasinghe and I answered this question affirmatively, and so arrived
at a temporary equilibrium point in our thinking: the organic basis of life was
interstellar, a position that others are now favoring. It was at this stage that we
began our technical readings in biology, fully expecting the usual picture of
the terrestrial origin and evolution of life to be amply confirmed by the facts.
Unlike the situation in astronomy, where one has to struggle against a paucity
of facts, in biology one has to struggle against being swept away in an
avalanche of information. However, if one can avoid being overwhelmed, and
if the many facts can be fitted into a consistent picture, then one can have
considerable confidence in the result, a major advantage over many situations
in astronomy.
This first resting point did not survive our early readings in biology, as it
was quickly apparent that the facts point overwhelmingly against life being of
terrestrial origin, which would require happenings every bit as miraculous as
the views of religious fundamentalists. Although released from this conceptual
millstone, my brain made no leap to freedom. It plodded its way, small step
by small step, first to the comets. Because comets must have experienced
break-up and reformation, with material interchanged between them, and
because the material would be organic, it was possible to think of the whole
ensemble of comets as a life-generating unit. And because a few comets are
breaking-up and scattering their contents all the time, the process was not
relegated to the remote past. This was a big plus, since theories that relate to
current events stand much above those concerned only with situations long
dead and done with. There was much of interest to be worked through in this
first shift from the Earth to comets, and the investigation of a number of side
issues deluded us for a while into thinking that the main problem had been
faced.
It had not, of course. One does not face the factor 104000°, discussed above
in connection with the enzymes, simply by going from the Earth to the comets,
a move which yields a gain factor of about 106• Nor does one face 1040000 even
by venturing from the solar system to all the other star systems of our galaxy,
a further step that yields an additional gain factor of 1011. Yet such was our
12 HOYLE
next move, to a galaxy-wide ensemble in which life originated and evolved.

A few scientists had speculated in the nineteenth century on life as a
galaxy-wide phenomenon, and in the early years of the present century Svante
Arrhenius had added flesh to the bones of those earlier speculations. Our views
had come by a different route, with many later facts to guide them, and so with
2
differences from Arrhenius. But there was also much that was the same.
Lack of a suitable computer had forced me out of the interstellar grain
problem into a different line of thought, but now the new line suggested a
renewed attack on the grain problem. Could the interstellar grains be biolog
ical cells, some perhaps alive and others in various stages of degradation, with
the graphite particles needed to explain the ultraviolet extinction coming at the
last stage of the degradation process? I don't know how others go about
dealing with questions like these, but I suspect that many cough violently, go
purple in the face, and that by the time breathing returns to normal the question
is safely gone. I myself go off for a long walk on the mountains , preferably
over a route with the harsh reality of a bit of crag on it, so that, if I don 't fall
off the crag, I can still have a little assurance that my wits are still within
earshot.
Unlike inorganic cells, which can have any size at all, biological cells have
well-defined size distributions, as for example the diameter distribution of
spore-forming bacteria shown in Figure 1. Notice the clustering around 0.7
micrometers. This value, typical for bacteria, is the same as the known sizes
of the interstellar grains (in the case of rod-shaped particles, it is the rod
diameter that matters, not the rod length) . Under freeze-!iried conditions, as
in interstellar space, bacteria normally develop cavities within themselves, and
small partially hollow particles scatter light according to a volume-averaged
value of the refractive index. Hence bacteria in interstellar space would scatter
radiation in the visible spectrum like solid particles of unusually low refractive
index, the property we had found from the investigation of solid hydrogen to
fit certain tricky details of the observations so well.
When hard-pressed by adverse conditions some bacteria break-up into
smaller, more-or-Iess spherical wall-less cells called mycoplasmas . The curve
of Figure 2 is a calculation due to Wickramasinghe of the extinction of
starlight produced by a mixture of mycoplasmas, of graphite spheres produced
by the degradation of a fraction of the mycoplasmas, and of the bacterial
distribution shown in Figure 1. The points of Figure 2 are observations, with
both observations and the curve normalized to a valuc of 1.8 magnitudes per
kiloparsec at a wavelength of 0.55 micrometers . Unlike the complex calcu
lations of the late 1960s, which went on for years, the calculation leading to
Figure 2 was carried through in a few days. Einstein is reported to have
'This situation is discussed at length in Space-Travellers, the Origins of Life, University

College Cardiff Press, 1981.
remarked that, while God may be subtle, He is not malicious. If the grains
were not organic, it would surely be incorrigibly malicious to have given us
such poor results in the 1960s with basically the correct theory and such an
excellent result now with a wrong theory!
But this was only an entertaining diversion from the main issue. How is the
factor 1040000 really to be faced? Not by a galaxy-wide ensemble of living cells.
Not even by adding other nearby galaxies to the ensemble, or even the totality
of galaxies observable with the largest telescopes. To face 1040000 the ensemble
of life must be hugely cosmological in its scale, and our cosmology has to
extend into the past by a time interval exceeding ten billion years by an
enormous factor.
So we are back to the starting point, but now with more substance to the
argument. It will of course be in the reader's mind to ask if 1040000 is really
inevitable. The answer is yes, if life is to originate by what are called the
"blind" forces of nature, which is to say without initial information. Nothing
is to be gained by attempting to shake the calculation of 1040000 . The issue you
will recall was the probability of a set of amino acids randomly falling together
into a workable aggregate of enzymes. Certainly it is easy to frame a deceitful
argument, in the following way for example. Start with much simpler, much
smaller, enzymes that are sufficiently elementary to be discoverable by
chance. Then let evolution in some chemical environment cause the simple
enzymes to change gradually into the complex ones we have today. The first
retort to this mental deception is that an appeal to initial simplicity has been
c...
<IJ
.Q
§C
20
o 1.0 1.5
diameter (/J.m)
Figure 1 Size distribution of spore-fonning bacteria.

14 HOYLE
allowed for already. Thus the number 1040000 was obtained from a calculation
in which less than twenty amino acids were required to be in specific se
quential positions for each of two thousand enzymes. If the calculation is to
be criticized it should be on the grounds of being much too conservative. But
the real deceit comes from ignoring the problem of what it was in the environ
ment that caused simple enzymes to evolve into complex ones. If the environ
ment contained information, what was its source? If not, then an improbability
of the order of 10-40000 has been concealed in the behavior of the environment.
To face 1040000 one must think unthinkable thoughts, which means any
thought with a chance greater than 1 in 1040000 of being right, a condition that
permits a wide class of possibilities! One such possibility is that the enzymes
were put together in accordance with instructions. Given a knowledge of the
appropriate ordering of amino acids, it would need only a slightly superhuman
chemist to construct the enzymes with one hundred percent accuracy. It would
need a somewhat more superhuman scientist (again given appropriate in
structions) to assemble a livip.g cell, but not a level of skill outside our
comprehension. Rather than accept a probability less than 1 in 1040000 of life
having arisen through the "blind" forces of nature, it seems better to suppose
that the origin of life was a deliberate intellectual act. By "better" I mean less
likely to be wrong.
--
u
Q. 7
�
';;-
�
- 6
�
'-'"
Z 5
0
...
u 4
Z
j:
>< 3
....
0
0 1 2 3 4 6 7 8 9
5
A-I (�-I)
Figure 2 Wavelength dependence of interstellar extinction normalized to 1.8 mag/kpc at
-I
,\ 1.8 Mm . Points are astronomical observations; curve is for grain: modeL (Circles are
-I
=
average extinction compiled from many sour<;es.by Sapar & Kuusik 1978. Triangles are ESA data
:
c ' given by Jamas et al. 1976. Squares ar�Jroll.l Bless & Savage 1972.)
A spaceship approaches the Earth; but not close enough' for its imaginary
inhabitants to distinguish individual"terrestriakanimals. They see growing
crops, roads, bridges, and a debate ensues. ;Are these chance formations or are
they the products of an intelligence?
It is not at alMifficult'td formulate examples of events with exceedingly low
probabilities. A'roulette wheel operates in a casino. A bystander notes the
sequence of numbeFs' thrown by the wheel over the course of a whole year.
What is the chance that this particular sequence should have turned·up? Well,
not as small as 1 in 104000°, but extremely small nonetheless.- ,So·there is
nothing especially remarkable in a tiny probability. Yet it surel)! would be
exceedingly remarkable if the sequence thrown by the roulette wheel in the
course of a year should have an explicit mathematical significance, as for
instance if the numbers turned out to form the digits of 1T to an enormous
number of decimal places. This is just the situation with a living cell, which
is not any old random j umble of chemicals.
Taking the view, palatable to' most ordinary folk but exceedingly un
palatable to scientists3, that there is' an, enormous intelligence abroad in the
Universe, it becomes necessary to write blind forces out of astronomy. Inter
stellar grains, living cells, are to be regarded as powerful tools, every bit as
purposeful if you like as a garden spade. We :know from astronomical studies
that the grains are mysteriously connected with a whole range of phenomena:
the rate of condensation of stars; the mass function of stars; magnetic fields;
spiral arms of galaxies; and quite probably with the formation of planetary
systems_ Not one of these phenomena has been explained in better than fuzzy
terms, just as the vie'ws of the imaginary travellers in the spaceship would be
fuzzy if they attempted to explain terrestrial fields, walls, and ditches as
products of the blind forces of nature.
It would be necessary to calculate in 'full detail the properties of complex
biopolymers in order to obtain the information required for the construction of
a living cell, Such a project would be quite beyond our practical ability, but
not beyond our comprehension. Indeed we are nearer to understanding what
would be involved in it than a dog is to understanding the' construction of a
power station.
Edward Blyth, who wrote on natural selection as early as 1835-37, re
marked that when the idea first occurred to him "a variety of important
considerations crowded on the mind." So it is here. Suppose you were a
superintellect working through possibilities in polymer chemistry. Would you
not be astonished that polymers based�'oll'the carbon atom turned out in your
calculations to have the remarkable properties of the enzymes and other
biomolecules? Would you not be bowled over in surprise to find that a living
'Because, of course, scientists delight in seeing themselves as the only Johannes Factotums in
the whole Universe,
16 HOYLE
cell was a feasible construct? Would you not say to yourself, in whatever
language supercalculating intellects use, "Some supercalculating intellect
must have designed the properties of the carbon atom, otherwise the chance
of my finding such an atom through the blind forces of nature would be less
than I part in 1040000 ." Of course you would, and if you were a sensible
superintellect you would conclude that the carbon atom is a fix.
From 1953 onward, Willy Fowler and I have always been intrigued by the
remarkable relation of the 7.65 Mev energy level in the nucleus of 12C to the
7. 12 Mev level in 160. If you wanted to produce carbon and oxygen in roughly
equal quantities by stellar nucleosynthesis, these are the two levels you would
have to fix, and your fixing would have to be just where these levels are
actually found to be. Another put-up job? Following the above argument, I am
inclined to think so. A common sense interpretation of the facts suggests that
a superintellect has monkeyed with physics, as well as with chemistry and
biology, and that there are no blind forces worth speaking about in nature.
This problem of the energy levels of 12C and 160 is by no means the most
puzzling I have accumulated in a lifetime of research. One particular problem
proved so difficult that I had long since put it aside as hopelessly intractable.
With an entirely new outlook now available, however, I took a fresh look at
this old problem, with results described technically in the Appendix, and in
more colloquial terms below. As always , I introduce this further topic with a
bit of autobiography (permitting myself an idiosyncratic run-to-the-wicket),
beginning with a remark or two on the educational process .
Like entropy, which perpetually increases, educational standards per
petually worsen. And like entropy, which increases inevitably because of the
policies of physics, education worsens inevitably because of the policies of
educators . Instead of teaching being properly confined to the rote-learning of
facts and well-proven techniques , pupils are confused nowadays by the teach
ing of meanings that they cannot comprehend.
You can see what I mean by attending a few rehearsals of amateur the
atricals. You will find the players attempting to perform the play long before
they know their parts. They strive to give meaning to their lines while still
reading from script, and they even strut the stage while still reading from
script. The outcome is that, despite an unconscionable number of rehearsals,
the players reach the actual performance still insecure, just as nowadays the
school population reaches the age of college entrance in a considerable mea
sure unable to read efficiently and insecure in elementary mathematical pro
cesses.
There was none of this in the old days of rote-learning, preferably done by
chanting. Well-proven techniques acquired by chanting stood their recipients
in good stead for a whole lifetime . And the higher the standard one seeks to
achieve, the more necessary does rote-learning become. The situation is the
same in mathematics and science as it is in learning a musical instrument, or

learning to play difficult parts like Macbeth and King Lear on the stage. The
golden rule is to learn the lines, or the scales, or the mathematical processes,
to a point where they can be reproduced with extreme facility. Only then
should one worry seriously about meanings and interpretations .
I have never learned the lines of Macbeth or King Lear. but I will bet that
those who have will have gone through something like the following experi
ence. After continuing for days, weeks, or even months thinking they knew
the part thoroughly, a sudden perception of several lines will have hit them all
in a moment, and they will then exclaim in delight to themselves "Well, surely
this is what old-man Shakespeare must really have meant!" This is certainly
what happens in science and mathematics. True understanding comes from
just such perceptions. Since these perceptions involve the fine-processing of
highly ordered information in the brain, carefully learned technique is an
essential prerequisite. True understanding cannot be picked-up casually from
a lecturer on the rostrum, any more than one can learn to ride a bicycle by
watching someone else riding a bicycle. Facts, techniques, and meanings
thrown higgledy-piggledy into the brain produce as flat a situation as all kinds
of food thrown at random into a stewpot.
Worse, there is a somber aspect to the matter. Experience shows that
teachers giving instruction in well-proven techniques stay honestly within their
understanding, but experience also shows that teachers-encouraged or re
quired by educational policies to discuss meanings with their pupils
inevitably go much outside their own understanding. This explains why so
much Old-style school teaching, especially at sixth-form level, was good, and
why nowadays so much teaching at universities, especially in the humanities,
is bad.
The reader will properly ask in what respect this present article differs from
the system I am condemning. It differs in that I exercise no blackmail over the
reader, such as is given by the teacher-pupil relationship, and by public and
university examinations. The reader is free to take it or leave it, which is
exactly the way things should be with all discussions of meaning and under
standings. This was the way it was in the old days when undergraduates in the
ancient universities "read" their subjects literally, and where obtaining a "first
class" in university examinations did not depend on regurgitating the opinions
of lecturers. The situation would not be half as sinister as it is if rubbish came
easily out of the brain. The trouble is that, whereas rubbish easily goes in, like
a Japanese harpoon into the body of a whale, rubbish comes out not at all .
The fine-ordering of technical information into the brain carries penalties as
well as rewards. The greater the perceptions, the greater the stress. Percep
tions imply a reordering in the brain, not j u st of a few details, but of consid
erable blocks of information, and the larger the blocks the bigger the mental
18 HOYLE
disturbance . Nothiqg of.rthis;l:kind··can .happen when the storage is of the

higgledy-piggledy stewpot kind " . so ;lhat a modem education may be said to
give protection against "brainstonn" situations . This is presumably why the
advocates of modem education assert that it produces "rounded" personalities ,
which is a little like arguing that by taking a sledgehammer to the engine of
a car one ensures that the car will 'not-be involved in an acCident .
. thad two '�brainstorms" in my student years , the circumstances of which I
- can recall rather >precisely. The first occurred about a month before my final
undergraduate: examinations in late May, 1936. True to ' my- belief in rote
learning , I cotlld .reproduce, like the lines of Macbeth or;.J(ing Lear, most of
the details given in Eddington' s The Mathematical Theory ·oj Relativity. One
day, while pondering the relation of certain of the mathematical symbols to
actual physical measurements of ,space and time, I decided my understandiIlg
of the meaning of "clocks"· and " measuring rods" was defective. I had cQme
upon the difficulty in the, general the�ry of relativity, but it was soon. �pparent
" that the same problem occurred also in the special theory , which to this point
- I had thought easy meat. To.my horror, I now found I couldn 't understand the
apparently simple special theory . So how could I possibly take an examination
in the far more difficult general theory? I seriously contemplated postponing
the examination for a further year (an option open to me) ; ·but I managed to
resolve. the crisis before the month was out.
The, second crisis occurred in the spring of 1 9 3 8 , and this one I was not able
· :10 resolve within a month, a year , .or even a generation . I was sitting on the
banks of the River Cam when it happened. I had just finished swimming and
was waiting for the Red Lion in Granchester to open for afternoon tea .. Until
this moment I believed I understood quantum mechanics. By now I had won
a sought-for research, prize, ithe- subject of my entry for .it being in quantum
mechanics. Besides which, I was then in the process of becoming a research
student of the great'Paul Dirac . So what terrors could quantum,mechanics hold
for me? Plenty . As I sat , waiting for the afternoon tea, I suddenly,realized that
I didn't understand it at all.
To this moment I had accepted Without critical,. appraisal the explanation of
uncertainty in quantum mechanics which had been suggested by Werner
Heisenbe�g and Neils Bohr-the Copenhagen school as it became called
according to),which'uncertainty in a quantum system is taken to be a necessary
consequence of interference by the observer with the system. I now, saw that
this explanation was inadequate, and quite possibly incorrect. Unknown- lto
me, Erwin Schrodinger had run into .the same situation and had -been , so
. appalled that he had been led to exclaim "I don't like it (quantum mechanics) ,
and I'm sorry to think L everhadlanything to do with it!"
. I suppose I should have:: g.one·(to .talk to Dirac , but like most students I had
. not learned to put my thoughts coherently into words , ; and with Dirac you had
t o b e clear in your u s e o f words . S o I tried discussing the difficulty with my

fellow students , only to find them unable to see any problem. This caused me
a year later to quit theoretical physics for astronomy . My brain patterns were
seriously disturbed, and I could not see how to reshuffle them in a satisfactory
way. Instead of accepting open defeat, I followed the military strategy of
retreating, hopefully to fight another day .
Almost twenty years later, the same problem resurfaced in an article by
Hugh Everett III (Reviews of Modern Physics, Vol. 29, 1957 , pp 454-62).
Since the lay-out of Everett's argument is to my knowledge the clearest to have
appeared , I have adopted it in the Appendix . To avoid technicalities here I
content myself now with a more informal discussion, involving an example I
constructed in 1 938-39 (an example similar to the so-called cat paradox of
Schrodinger) .
Everybody who becomes involved with quantum mechanics is likely to be
uncomfortably aware that the dynamics of the theory leads to a spreading
vagueness in the world. If you begin with a reasonably well-condensed par
ticle wavefunction, a so-called wavepacket, the packet spreads with time, and
pretty soon the thing is all over the Universe. Yet our observation of the world
does not suggest any such spreading vagueness . Hence one infers that in some
way there must be a compensating sharpening of the picture . Let us see from
the example of 1 938-39 how this sharpening occ urs .
A city has an inner citadel that can be sealed-off completely from the people
of the city, who in tum can be sealed-off by a wall from the surrounding
countryside (in the fashion of ancient Troy) . The citadel contains a nuclear
bomb4 with a trigger controlled by a device with quantum uncertainty, con
structed in the following way . Count n similar radioactive nuclei with a
reasonably long half-life T. Over a specified time interval Tin, electric power
is supplied to the system; otherwise the power is off and nothing can happen.
The trigger is arranged to be tripped if one or more of the n nuclei disintegrate
in the specified interval. No experimental physicist would have difficulty in
actually making such a device , and nobody would doubt that for sufficiently
large n, the probability of the trigger being activated would be �.
Now consider the grand-ensemble wavefunction for the whole city, people
and all. Some physicists , foreseeing trouble ahead , have sought to deny this
step, claiming that quantum mechanics cannot be applied to macroscopic
systems. But this is worse than a brainstorm. It is a kind of scientific nervous
breakdown, for it would require a maximum number of particles up to which
quantum mechanics was applicable and beyond which it was not-an absurd
position . Although the set of base states for the grand-ensemble wavefunction
is huge, the base states can be separated into two mutually exclusive catego-
"The possibility of constructing a nuclear bomb was apparent as early as 1939.

20 HOYLE
ries , one in which the city goes about its normal business, and the other in
which everything goes up in a mushroom-shaped cloud. At the beginning of
the specified time interval the amplitudes of the states in the mushroom
category are zero, but as the time interval proceeds the amplitudes become
nonzero, and by the end of the time interval the sum of the squares of the
moduli of the amplitudes of all the base states in the mushroom category add
to � . And of course the sum of the squares of the moduli of the amplitudes of
the base states in the normal category falls from unity at the beginning of the
time interval to � at its end. This is just a complicated way of saying that the
chances of the city surviving and of it being annihilated are even steven .
By good fortune you, the observer, are not incarcerated in the city. You are
out in the surrounding countryside at a safe distance, from which position you
will be able to see the mushroom cloud, if it goes up. However, because you
are unbearably fretted by the situation, you take a stiff dose of a drug that
causes you to sleep through the critical time interval, and for long enough
afterward so that you cannot tell what happened from the condition of the sky.
Nor can you search for radioactive fallout, because you don't have a Geiger
counter. But with cunning you have arranged for a camera to take pictures
throughout the critical interval. You retrieve the film from the camera, think
ing that a decision on the fate of the city is contained in the emulsion of the
film. Yet according to quantum mechanics , the grand-ensemble wavefunction
is such that the chances of the silver grains in the emulsion being arranged to
form a mushroom cloud and of them being clear of such an arrangement
remain even. There is no way in which the camera could have made a decision
on the fate of the city.
Now proceed to develop the film. In the dark room as you apply developer
and fixer there is no light. Satisfied at last that you have done a good job, you
take the developing tray outside and hold the prints up to a light. And then at
last you know what happened to the city. B y "know" I mean you condense the
wavefunction for the city. Instead of continuing any longer with amplitudes
giving even chances for the two categories of states, you now set all the
amplitudes of one category to zero, and the other category t<,\kes unit proba
bility. All subsequent experience will be consistent with this drastic shift in the
wavefunction. If you make a journey to the site of the city you will either find
the people alive there, going about their business, or you will find a scene of
woeful devastation. It will all fit exactly to what you decided in your first
glance at the prints in the developing tray .
I will not pretend that I saw all this in a flash on the spring day in 1 938. To
that moment I had followed standard texts in which quantum systems were
thought of as minute in comparison with the observer, and so it appeared
reasonable to think of a huge hulking observer interfering with, and intro
ducing uncertainty into, quantum systems. What I now saw was that quantum
,uncertainty could occur i n a system huge compared to the observer, and indeed
that one could have situations in which the perturbations of the quantum
system by the observer were quite negligible . It was so for most cases of
decaying radioactive nuclei. From the time of Rutherford' s experiments in the
early years of the century, it had always been emphasized that decaying nuclei
went their own sweet way, irrespective of the experimenter (this was before
the era of particle accelerators) . What I could not understand, as I sat in 1938
on the banks of the River Cam, was how, in view of this known situation in
nuclear phy s i c s scientists had nevertheless thought of interference by the
,
observer as the cause of uncertainty. The uncertainty was inherent, and yet it
was somehow the observer who contrived to resolve the uncertainty. How was
this done from a theoretical point, from within quantum mechanics itself?
This was the question I failed to resolve, the question which led me in 1939
to leave theoretical physics for astronomy, thinking with youthful idealism to
be entering a more rational subject. Although an almost forbidden question in
1 939, it has been asked more often by the younger postwar generation of
physicists. But apart from the article by Everett, not much has been published
on it, the general attitude being summed up by a distinguished younger
physicist who remarked "This is a matter on which we must each have our own
private thoughts . "
I have returned twice to the question; in the years 1 964-1 970, when for the
first time I felt I could see a chink of light, and now, very recently. Let me
begin with the 1964-1 970 period. It was then that I became a dyed-in-the
wool believer in the time symmetry of basic physics. Of course there are
aspects of our experience that are not time symmetric-thermodynamics, the
past-to-future propagation of radiation fields, and certain features of p articl e
physics . In my view such asymmetries are cosmological manifestations, how
ever, not blisic physics. Here I have space only to discuss the past-to-future
propagation of the electromagnetic field. In a famous demonstration, Wheeler
and Feynman showed more than thirty years ago that one could have a
time-symmetric electrodynamics augmented by a cosmological response from
the future that reproduced exactly the same results as the classical Maxwell
Lorentz theory. For some years it was thought that a similar demonstration
could not be given in quantum physics, but in the late 1 960s Jayant Narlikar
and I showed that, just as in the classical case of Wheeler and Feynman, it was
possible to have a time-symmetric local quantum theory augmented by a
cosmological response from the future that reproduced exactly all the practical
results of normal quantum electrodynamics . Although there was no difference
at all in its statistical predictions, the time-symmetric theory was interestingly
different in its details . Unlike normal quantum mechanics, no pure-amplitude
theory could be formulated because the cosmological response involved both
the wavefunction and its conjugate complex. This I saw as an advantage. The
22 HOYLE
pure-amplitude aspect of normal quantum mechanics involves a redundancy,

because information is discarded in passing to practical results. The time
symmetric theory yields the practical results without redundancy.
While these considerations did nothing directly to answer the quantum
puzzle, they served to lift a dark cloud upon it. One of the unremitting
struggles of my life had been to read Mathematical Foundations of Quantum
Mechanics by 1. von Neumann (for the full flavor try the Springer edition,
Berlin 1 932, but if you want it in less excruciating form try Princeton Univer
sity Press, 1955). In the later chapters of his book, von Neumann claims to
demonstrate that no theory with a greater measure of predictability than
present quantum mechanics can be found, which would seem to close-out any
possibility of answering the critical question posed above. Everett makes a
courageous attempt to accept this position, as I describe in the Appendix, but
the better route seemed to me to contemplate the possibility that von Neumann
might have been wrong in his assertion . From someone with such slender
mathematical ability as myself this might seem a great conceit, but I was
encouraged in it by the conviction that in his challenge to Dirac over the
mathematical validity of the famous delta-function, von Neumann had been
mistaken. What I saw in 1 970 or thereabouts was that von Neumann had been
concerned with a finite local system. If cosmology was involved, with a
response from the future, the dynamical variables in the system could be
infinite, and the situation could then be different. This was the chink of light.
Even so, the problem remained acutely pUzzling. The future imposes a
condition on a local system because a signal goes out from the local system
to other material systems in the future, which respond with a return signal on
account of the time symmetry. One would like a situation in which the return
signal imposed a deterministic reality on the local system, forcing an explicit
decision to be made in all situations of an A or not-A kind, as in the example
discussed above (mushroom cloud or no-mushroom cloud) . The trouble is that
so long as one calculates the return signal from within quantum mechanics this
does not happen, just as von Neumann claimed. One is faced by a chicken
and-egg situation. The initial local system does not have deterministic reality
because the systems in its future with which it interacts do not have deter
ministic reality, and this is because the systems in the further future with which
the second systems interact do not have deterministic reality, and so on along
an infinite chain of interactions. Yet somewhere the gordian knot has to be
cut-it must be, since our everyday experience tells us that it is! The mathe
matical loophole lies at the limit of the infinite chain of interactions. True, we
cannot establish deterministic reality by starting within the chain and by
attempting to argue in a past-to-future direction toward the limit. But if we
were to start with deterministic reality at the limit, arguing backwards from
future to past, there would be deterministic reality at every link of the chain.
In other words, the troubJ� may well come from arguing the problem back-to
front instead of front-to-back.
This was the stage of my thinking following the work of the 1964-1970
period, before it became apparent (from the arguments given earlier) that an
enormous intelligence must be abroad in the Universe. As the Americans say,
this instantly created a new ball game. An intelligence of a strictly finite kind,
such as might calculate the properties of the enzymes, would not suffice to
resolve the quantum mechanical dilemma, however. For this , it would be
necessary to control the infinite limit discussed above , or some other process
of equivalent significance. At first, one might think the tremendous scope of
such 1j. thing would take it entirely outside the range of our oomprehension. But
remarkably this is not so. It is possible to see in rather: precise mathematical
. .terms how such a control could establish intelligence'throughout the Universe
by imposing information sequences on finite material systems . The technical
details of,how this might be done are given at the end of the Appendix . Here
I will jump the issue of how it might be done, to ask is it actually done? Is
information impressed in our brains from outside? Obviously yes, from the
five senses . But is there a subtle further component arising from an external
control of quantum uncertainty? The evidence is not of a kind that one is
obliged to consider compelling, but it is not negligible either.
I have always thought it curious that, while most scientists claim to eschew
religion, it actually dominates theirthoughts more than it does the clergy . The
passionate frenzy with ,which the big-bang cosmology is clutched to the cor
porate scientific bosom evidently arises from a deep-rooted attachment to the
first page of Genesis, religious fundamentalism at its strongest. A little should
be said in favor of this mania. Let us think of every animal as a computer
terminal equipped with a certain measure of backing storage, which has been
established partly through the animal' s genetic heritage and partly through
. inputs from the five senses. Each computer terminal receives. vestigial signals
arising from the phenomenon of the condensation of the ,wavefunction (see the
end of Appendix for details). The information content·()f these signals, ranging
from the simple to the complex, has to be interpreted against the available
backing storage. Where the information falls well within the capacity of the
backing storage we have a clear consistent picture, as in science. Where the
information falls at the limit (or outside) of the backing storage we have · a
1 ,mud!:lled illogical picture, a s i n religion. I n both cases the signals , are valid
'enough. Limitations arise in the interpretation, not in the signals themselves .
A dog cannot understand the operation o f a power station · because o f lim
itations in the scope of its backing storage , and in a like fashion we have
trouble with problems of religion, even if one is incorrigibly attracted to them
like moths to a candle (as scientists are) .
Perhaps because it remains, a cultural thread in the Yorkshire valleys where
24 HOYLE
I was born and brought up, I have always been an admirer of the music of
Messiah, latterly in the fonn in which Handel actually wrote it. Yet in my
earlier years I could make little or nothing of most of the words . A particularly
obscure passage comes from the bass soloist in the third part, just before the
famous trumpet passage: "Behold, I tell you a mystery: we shall not all sleep ,
but we shall all be changed, in a moment, in the twinkling of an eye, at the
last trumpet. The trumpet shall sound . . . . "
It is curious to contemplate that there could be a connection between
quantum mechanics and this apparent gibberish. Nevertheless, the persistent
religious conviction that the pattern of our lives is stored in the future looks
as if it could quite well be correct. At the mathematical limit discussed above .
At the last trumpet! What an extraordinary way to describe the outcome of a
sequence of arguments involving the condensation of the wavefunction , the
need to avoid von Neumann's mathematical result for finite systems , and
time-symmetric electrodynamics. Of course one can argue that the correspon
dences are fortuitous. Notice, however, that in time-symmetric theory
influences are indeed felt "in (less than) a moment, in (even less than) the
twinkling of an eye," and that all finite events are brought together at the
mathematical limit in the future. Fortuitous or not, it is curious that so many
people without scientific knowledge have believed in the idea, as if they had
caught a glimpse of a difficult message that they could only express in tenns
of an everyday analogy.
Religion is an interesting but not really convincing example of the computer
terminal data. Some years ago I had a graphic description from Dick Feynman
of what a moment of inspiration feels like, and of it being followed by an
enormous sense of euphoria, lasting for maybe two or three days. I asked how
often had it happened, to which Feynman replied "four. " We both agreed that
twelve days of euphoria was not a great reward for a lifetime' s work.
Actually, Feynman was lucky with his four times . Only once have I had a
similar experience . The circumstances were extraordinary and far outside any
other perceptions I have ever had. Rather as the relevation occurred to Paul on
the road to Damascus, mine occurred on the road over Bowes Moor. The time
was the late 1 960s, when Narlikar and I were struggling with the problem of
the quantum mechanical signal from the future. We were tackling the non
relativistic theory at that stage, and were seeking a way to evaluate a multiple
integral with a complicated integrand determined by a system of equations.
A small party of summer visitors to the Institute of Theoretical Astronomy
at Cambridge was spending a few days in the Scottish Highlands. Because of
a committee meeting I was late in joining them. I started alone from Cam
bridge, driving north by way of Scotch Comer, Penrith, Carlisle, and Stirling .
As the miles slipped by I turned the quantum mechanical problem mentioned
above over in my mind, in the hazy way I normally have in thinking mathe-
maties in my head . Normally, I have to write things on paper, and then fiddle
with the equations and integrals as best I can. But somewhere on Bowes Moor
my awareness of the mathematics clarified, not a little, not even a lot, but as
if a huge brilliant light had suddenly been switched on . How long did it take
to become totally convinced that the problem was solved? Less than five
seconds. It only remained to make sure that before the clarity faded I had
enough of the essential steps safely stored in my recallable memory. It is
indicative of the measure of certainty I felt that in the ensuing days I didn't
trouble to commit anything to paper. When I returned to Cambridge ten days
or so later, I found it possible to write the thing down without difficulty.
Many will smile if I say that such an incident was triggered by the deci
phering of a cosmic signal. It will be agreed that a sudden reordering of
substantial blocks of information in the brain must have been involved, but it
will be said that the initiating signal happened by chance, from a random firing
of neurons. Perhaps. I have no means of calculating the probability of random
brain processes just happening to trigger so complex an affair, but if I had, I
suspect I would arrive at 1 part in 1040000 , or less.
My last example is not exposed to this criticism, since it involves an output
too vast and too long sustained to be attributable to chance. Before the late
works of Beethoven became a fashion they were thought difficult, whieh they
are. My friend Leo Smit explained to me one of the difficulties, namely that
Beethoven was apt to combine two works into one, two sonatas in one, two
symphonies in one. There is no difficulty in following the famous Ninth
Symphony on a bar-by-bar basis, the difficulty lies in the overall structure.
While Beethoven admirers will tolerate no words of criticism against the
Ninth, critics, including Giuseppi Verdi, have described its fourth movement
as a failure. After Leo Smit's remark, I realized that "failure" is the wrong
word; "misjudgment" possibly, but not "failure." The point is that the first and
third movements together form a symphony of cosmic proportions and gran
deur (akin to the two movements of the Opus I I I piano sonata) , while in the
second and fourth movements Beethoven is down-to-earth, addressing us
more or less in our own terms . So one should think of two symphonies
interlaced with each other. For my example I want the first and third move
ments taken as a unity.
On a visit to the Kitt Peak National Observatory , I made the acquaintance
of the big 1. B . Lansing loudspeakers, which will put out 100 watts without
distortion. I want those speakers , a good hi-fi amplifier, a room large enough
to propagate the lowest register, and a first-rate modem recording. The volume
should be set so that the long-sustained drumroll in the middle of the first
movement sounds as if Homer himself had caught the thunder of Zeus on Mt.
Olympus.
But before we listen , let us enquire a little into the history of the composer.
26 HOYLE
A poor boy, tough, hard-working; determined to force himself to the top. Add
great -ability to determination, and no crackpot educators to deflect him from
"speCializing. " With these formidable advantages we find Beethoven in his
early thirties as· the greatest keyboard artist yet known, an artist gradually
establishing a name� as: a Gomposer. Now disaster strikes. Although still a
comparatively young man, Beethoven begins to go deaf. His deafness is a long
drawn-out affair, with loud hissing in the ears that must have introduced
distortions in the aural memories of earlier years . The situation was surely
much worse than if Beethoven had gone deaf all in a moment. Long before the
Ninth was written, however, B eethoven' s hearing had become negligible, so
that at its first performance he could hear neither the orchestra nor the applause
of the audience .
Now listen and ponder how those sounds were conceived. Did Beethoven
simply permute and combine memories for sound he had acquired in his
youth? At best, discounting distortion, those memories represented a stage of
development illustrated by the First and Second Symphonies, a universe apart .
from the Ninth. Remember too that it is hard to find anything in the past .
evolution of our species where the ability of a deaf man, beyond the prime of
life, to rearrange patterns of sound from far-distant memories would have
conferred a significant selective advantage .
The alternate view is that the deaf Beethoven , decisively cut-off from the
distractions of the world of men, equipped as a terminal with unusual backing
storage, was able to receive a particular component of the cosmic signals, and
with sharply increasing clarity as the years passed by . This view would be my
choice, but each of us must listen and decide. Perhaps the decision turns on
whether we ourselves hear the thunder of Zeus on Mt. Olympus.
ApPENDIX: ON THE CONDENSATION OF THE WA VEFUNCTION
Let cfJ;, i = 1 , 2, . . . , be a complete orthonormal set of wavefunctions, each

satisfying exactly the dynamical equations of a quantum mechanical system in
the absence of interaction with the observer, who to begin with has a grand
ensemble wavefunction denoted by <l> . Suppose for the moment that the
system is initially in the particular state cfJi. So long as system and observer
remain independent of each other the total wavefunction for system plus
observer is the simple' product cfJi <l>.
An interaction between system and observer, operating over a specified time
interval , in general , produces a complicated entangled total wavefunction at
the end of the interval,
where not only have states of the system at other suffix values appeared, but
the initial observer-state has "branched" into the set of observer-states <l>j'
j = 1 , 2, . . . .
The key aspect of the article by Hugh Everett III (Reviews of Modern
Physics, Vol. 29, 1 957, p . 454) lies in restricting the discussion to interactions
that are less than general, namely to interactions with respect to the base set
cPj, i = 1 , 2, . . . , that have the effect over a time interval of giving
cPi cI> - cPi <l>i, i = 1 , 2, . . . ,
where each observer-state cI>i involves only the corresponding system-state cPi '
Interactions with this special property are said to be "good. " By means of an
example due to von Neumann (from the latter' s book Mathematical Founda
tions of Quantum Mechanics), Everett shows that a good interaction occurs in
a particular special case. He then goes on, in what seems to me a gap in the
argument to ass um e the possib ility of a good interaction whatever the quan
,
tum system and whatever the observer. The idea is that the observer is free to
adjust the interaction by an appropriate design of experiment in such a way that
it is good. To establish whether or not this is true, one would need to examine
each case separately in detail. But because this issue is mathematical, not
conceptual, let us proceed (for the moment) taking it as axiomatic that the
observer can indeed find a good interaction whenever it is required for the
argument.
Suppose next that the system-state before interaction with the observer is
mixed with respect to the base set cPi, i 1 , 2, . . . , viz � i ai cPi' The
=
coefficients ai before the interaction are constants , since each state <Pi satisfies
the dynamical equation of the system in the absence of interaction. The total
wavefunction before interaction for observer + sy stem is thus � i ai cPi' For
a good interaction over a specified time interval, this initially separated total
wavefunction is changed to the mixed state � i ai cPi <l>i. Although the system
and observer have become entangled, it is the simplifying property of a good
interaction that the initial coefficients ai have been preserved, and that each <1>;
refers only to the corresponding system-state <Pi' Each term satisfies the cou
pled dynamical equations of system + observer, with each initial cPi evolv
ing separately to <1>; cPi, i 1, 2 , . . . .
=
Should a good observation of the system again be made, the total wave
function at the end of the second interaction period would be 2 j aj cPi <1>;. ;. No
additional mixing between the system-states and the observer-states would
occur, but the latter would again acquire information that served to identify the
corresponding system-state. If we were to think of ;, i 1 , 2, . . . , as
=
wavefunctions belonging to a set of :��.ubobservers ," each �ubobserver wO!lld

consider himself as being identified with a definite state of the quantum
mechanical system, cI>i with <Pi for each value of i. Every additional good
observation made by a particular subobserver on the system would confirm
that the system was "in" the corresponding system-state. Thus a series of
28 HOYLE
observations by good interactions would lead subobserver <Pi to <Pi,i . . . , which

would seem to the subobserver in question as repeated confirmation that the
system was really "in" the state <Pi'
Now let the outcome of the first good interaction of system and observer,
L j aj <pj <l>j, experience interaction with a second identical quantum system .
That is to say, each subobserver <l>i experiences a good interaction with L j aj
<pj' giving L j ai <Pj ;'j' with the subsubobserver-state <l>i,j containing, in se
quence, records of both <Pi and <pj. The total wavefunction would be L i L j aj
aj <Pi <Pj <Pi,j' Generalizing to a long sequence of good interactions between the
observer and a set of initially similar quantum systems, the total wavefunction
at the end of the interactions would be L i Lj , . . ai aj . . . <Pi <pj . . . <Pi,j, . . . ,
where each multi-subobserver <Pi,j , • , has a record in sequence of the states
.
<pj, <Pj' . . . . Now choose a particular <l>j,j • , subject to weights given by

• •
the squares of the moduli of the coefficients ai aj • , but otherwise at

• •
random. After the choice has been made, let this particular subobserver count
how many records he has of <Ph of 4h, and so on, in the sequence i, j, . . . .
For a sufficiently long sequence of interactions, and for most choices of

<l>j,j, . . . , the resulting counts have the same ratios as do the squares of the
moduli of at. a2, . . . (see mathematical note at end) .
We have arrived at a somewhat amazing situation. Although theory requires
the observer to be associated with the totality ;,j, . . for all i, j, . . . , the
properties of just one typical subobserver-state corresponds to subjective ex
perience. A subobserver with wavefunction <Pi,j, . . . would find the first system
to be repeatedly "in" the state <pj, the second system to be "in" the state <pj ' and
so on . Moreover, for a sufficiently long sequence, the subobserver would find
<Ph 4h, . . . to be recorded with frequencies in the same ratios as the squares
of the moduli of the coefficients in the initial system-state, L j aj <pj, and so the
subobserver would arrive at the usual statistical results of quantum mechanics.
This brings us to the critical issue. Starting with the single wavefunction <1> ,
the observer's experience has gone through a series of branching, first i,j (j 1 , 2 , . . . ) and so on. This
= =
generation of an enormous tree through repeated branchings is the increasing

vagueness of quantum mechanics discussed in the main article. In the opposite
direction, it is the correspondence of subjective experience of a subobserver
wavefunction, <Pi,j, • • , that represents the condensation of the wave
•
function. The critical question is whether the Universe constitutes the whole
enormously complex tree of quantum mechanics, or is the Universe confined
to the particular route through the tree represented by a particular choice for
the subobserver state <l>j,j' . . . ?
Everett considered that it is the whole tree that constitutes the Universe (see
footnote on page 459 of his article) , We would then have no means within the
theory of specifying the particular route of which we are consciously aware.

To treat all routes equally, one would need to postulate an ensemble of alter
egos who are consciously aware of the other routes. Since each route satisfies
the dynamical equations (including the interactions) quite independently of the
other routes , there would be no way to compare notes with an alter ego , so
at least in this respect the situation would be free from contradiction.
The alter ego concept was not a new topic in Everett's article of 1 957. I
recall debating it with my fellow students in the years 1 937-39, and no doubt
it had been speculated about from the earliest days of quantum mechanics.
Opinion was, and I think still is, largely against it. One can object, not very
cogently, that it assumes an ensemble of existences of which we have no
evidence. More to the point, to make sense of the idea it would be essential
that the many routes through the tree be uniquely defined, and while the total
wavefunction � j Lj . • aj aj . , . o/j o/j
. . • <l>ij
. • is unique, the individ
• •
ual terms within the multiple summation , which determine the individual
routes , are not unique. Thus one could replace 0/1 , 0/2 by (0/1 ± 0/2)/Y2 in the
set of base states of the quantum mechanical system and then all routes
involving suffix values 1 and 2 would be changed.
To have any hope of countering this difficulty one must return to the axiom
according to which good interactions are always considered to exist at the
behest of the observer. In the actual universe there are only specific inter
actions, which may or may not be good. Instead of forcing the property of
"goodness" with respect to a preordained base set 0/;. i I , 2, . . , one
= .
might attempt an inversion of the situation . Take the interactions as they

actually are, and try choosing a base set with respect to which the interactions
are good . However, unless the interactions happen to be of special forms, with
the observer wavefunction <l> chosen to have special properties, such a project
fails. Indeed the conditions needed for interactions to be "good" are mathe
matically so remarkable that one wonders how they could ever occur in
practice. My own explanation , given in the main essay, is through controlling
signals from the future . The concept in Everett' s argument (as in von Neu
mann's theory of measurement) is that interactions are made good through the
observer's decision to make them so, but this surely begs the question, since
conscious decisions by the observer are not themselves explained by the
theory. Even so, I have agreed in the above to bypass the problem of the nature
of the interactions, and therefore press on to another difficulty present in the
argument at its very beginning .
While I am ready to admit that base states 0/;. i 1, 2 .
= , exist in an
. .
abstract sense as solutions of the dynamical equations of the system, nobody

to my knowledge has ever seen such beasts explicitly. What is done is to obtain
approximate solutions 0/;. i = I , 2, . . . , by omitting troublesome small
terms from the dynamical equations (of which there are plenty quite apart from
30 HOYLE
interaction with the observer!) . Usually ¢i . i 1 , 2 , . . . , are stationary =
states , which is to say the eigenstates of an approximate energy operator. The

omitted terms are then taken into account through a slow change with time of
the coefficients a; in the exact wavefunction L ; a; ¢;. As the coefficients
change, one says there are transitions between the states ¢;.
Everett's argument can be modified to fit this more realistic situation in the
following way. Let there be a time interval short enough for the coefficients
a;. i =1 , 2, . . . , to be taken as constants, but long enough for the observer
to establish interactions that are made good in some way , as before by taking
this possibility to be axiomatic . The result of applying good interactions
progressively to a sequence of similar systems is again given by
2:2: . . . a; aj . . . ¢; ¢j . . . <Pij • • • •
i j
Subjective consciousness now picks out a particular subobserver wave
function, <Pij say. Subjectively the first system is regarded as being in the
. . .
state ¢;. the second system in the state ¢i' and so on. This apparently definitive
situation becomes the initial condition for calculating the transitions that occur
in a succeeding longer time interval during which the quantum systems and the
observer are uncoupled.
As a matter of curiosity, consult any text on quantum mechanics. I will bet
(nearly my bottom dollar) you will find the author(s) specifying initial condi
tions for perturbation calculations by assigning their systems to explicit initial
states , but rarely, if ever, will the author(s) tell you how the systems got that
way in the first place. Unless one appeals to subjective consciousness in the
manner of the preceding paragraph, such calculations are a pretense. Since
most authors do not like to appeal to subjective consciousness , or to admit that
their work is a pretense, it is understandable that they say nothing!
I will sketch the transition calculation in two ways, first as it is done after
specifying initial conditions in the manner of the textbooks, and then accord
Jng to the general method of Everett.
At the end of the interaction period, the subobserver representing one's
conscious state has the wavefunction <P;,j . . . Now let a period of time .
elapse sufficient for the states ¢;. ¢j , to evolve subject to the exact
• . . .
dynamical equation of the quantum system, and let the observer be uncoupled
during this interval . Writing L k gik ¢k for the evolution of ¢;, i I , 2, . . , = .
the wavefunction becomes
<p;,}, . . . 2:2: ' . . . g;k g;f . . . ¢k ¢e .

k e
where the subobserver-state <Pi,j, evolves over the time interval according
.
. .
to its own now-independent dynamical variables.

Next, let there be a second interaction period, again short enough for the
coefficients gik not to change appreciably, but long enough for <Pj,j, . . to be .
coupled sequentially by good interactions to the sequence of quantum systems.

The result is
k t
in which each multi-subobserver-state <Pik,jf, . has two suffixes for each
quantum system in its record . (We have to contemplate that the details of the
interactions in this second pcriod may be different from those of the first
period, and that to make them "good" a linear transformation of the base states
of the quantum mechanical system may be necessary, If so, the transformation
can be absorbed into the coefficients gik without loss of generality.)
As always, we choose a particular <P ik j( , . . . to represent our conscious
,
experience, weighting our choice by the square of the moduli of the products
�ik �t . . , but otherwise making the choice at random. Our judgment is that
'
the first quantum system has changed from <Pi to <Pk (i can be the same as , or
different from, k), the second system has changed from <Pj to <Pt , and so on
along the sequence, Our typical conscious observer counts the numbers of the
various transitions, and obtains results which agree (for a long-enough
sequence-see the mathematical note) with the usual textbook calculations of
perturbation theory ,
Now for the general point of view. The total wavefunction before the first
interaction is L ) :j . . . aj aj . . . <Pi <Pj . . . . After the first interaction pe
riod the total wavefunction has, become Lj Lj , aj aj <Pi <pj
• • <Pj,j, . , • • • • • • . .
and after the evolution of 4>i to L k tik 4>b the total wavefunction is
i j k t
The second interaction period then leads to
LL . . , ai aj • • • LL . . . tik �t • • . 4>k <Pc • , • <l>ik,jC, • , • •
i j k f
Weighting the choice of subobserver for each set of numerical values of the
indices i, j, . . . , k, e , . . . by the square of the modulus of aj aj , • •
tik �f . . . we can arrive at a typical choice <Pjk, jf, . . . to represent our conscious
state, and we can use just the same counting procedure for determining the
transition probabilities. The outcome for a sufficiently long sequence is the
same.
Although ths ,general method appears at first sight only trivially different
from the, textbook- case, notice that it does not require a specification ,of '0ur
conscious state forthe firstinteraction period. Of course our actual experience
32 HOYLE
for the first interaction period will define a subobserver, <l>i',j"

• • say, but
.
there is no requirement for i ' , j I ,

• to be the same numbers as i, j, . . . in
• •
<l>ik,jC,
• • Thus there is no requirement for the state of consciousness in the
• •
second interaction period to have evolved from the state of consciousness in

the first period. In the second period we may be remembering the experience
of an alter ego in the first period.
From a practical standpoint one might think that such jumps in our state of
consciousness would involve issues of only minute detail, just because it is
usual for the observer to be enormous compared to the quantum system under
observation. The example discussed in the main article of a city that survives
or is annihilated shows such a supposition to be unwarranted. The differences
among the <l>ik,jC , • can be very great indeed. It is probable that every so
• •
called snap decision we make depends on only a few quantum transitions in

the brain. In the ensemble of our lives, with repeated switches to the memory
sequences of our alter egos there would be all the existences we would have
experienced if our snap decisions had been made differently. There could be
snap decisions affecting our behavior in moments of danger, the places we
visit, the people we meet, and perhaps even the people we marry. There is the
possibility of waking each morning beside a different spouse, although our
memory each morning will always be consistent with the spouse-of-the-day,
and we will therefore be entirely unaware of the other possibilities.
One tends to feel such a point of view is more suited to fiction than science,
and indeed I once used it as the basis of a novelS . One's instinctive prejudices
are not usually a sound guide to what is true or false, however. As Everett
pointed out, in the days of Copernicus people argued that the Earth could not
be moving around the Sun because we would surely feel the motion if it were.
Such opinions, based on prejudicial judgments rather than experience, tend to
be self-deceiving.
On the other hand, we must be careful not to attach false weight to the
apparent generality of the fantastic multiple-picture just described. For if this
multiple-picture is truly general, nothing from outside itself can be permitted.
How then is the special subobserver wavefunction representing our
consciousness-of-the-moment to be chosen? Subject to certain weighting fac
tors , but otherwise at random, we argue. How then is randomness to be
ensured? There can be no external throwing of dice, no external random
number generator. Without anything at all outside the general quantum me
chanical tree it is hard to see how to define the particularities of our con
sciousness. In any case what is our consciousness? The trouble with the
generality of the multiple-picture is that its claim for generality promises more
than it delivers . Indeed, the most relevant aspect of our experience, our highly
explicit consciousness, remains unexplained.
'October the First is too Late. Wm. Heinemann. 1966

Turning now to the single-picture theory, in which the wavefunction is

condensed through the lopping of those branches of the quantum mechanical
tree that lie outside our experience, we can say with advantage that it is
precisely the lopping process that makes possible the phenomenon of con
sciousness, an immediately interesting concept that lies outside the capacity of
the multiple-picture. What then is the point of the general tree? Is its function
merely to be lopped? No. The general tree is really a reference tree, a
backcloth, that serves to define statistical relationships that must be preserved
in the lopping process.
The single-picture theory frankly admits the need for defining a particular
route through the reference tree. The speculation considered in the main essay
required the defining process to be through signals that propagate future-to
past, opposite to the branching of the tree itself, which goes past-to-future. It
is through this difference of time sense that one can contemplate going
"outside" the usual theory. In analogy to a two-stroke engine, quantum me
chanics is just one of the cylinders, stroking from past-to-future. The other
cylinder serves to condense the wavefunction , and it strokes from future-to
past.
I also suggest in the main article that the cylinder which strokes from
future-to-past is directed by a superintelligence, and that through the conden
sation of the wavefunction our thoughts are controlled. Doubtless this concept
may have seemed to the younger generation as the vaporings of an aging
scientist, rather like the spiritualism of Oliver Lodge and William Crookes, or
the "fundamental theory" of Eddington. So let me counter with a final blow
or two.
The outcome, as we saw above, of a sequence of good interactions between
the observer and a set of similar quantum systems each with the wavefunction
� ; a; cp; is to produce the mixed wavefunction � ; � j . . . a; aj . . . cP; cPj . . .
;j . , The condensation of this highly complex expression consists in the
, '
choosing of a particular subobserver wavefunction <fIij , " subject to the choice

,
satisfying a broad statistical requirement that ensures that the numbers 1 , 2 ,

. . . , tum up among the indices i, j, . . . , of a sufficiently long sequence in
the ratios lad2: la 12 : la312 . . . . This requirement ensures that in our con
2
sciousness, represented by <l>ij . , " we find the ratios of the number of systems
with wavefunctions <PI , <P , . . . , to be the usual statistical ratios of quantum
2
mechanics.
Subject to this restriction, suppose you were free to choose the numbers
appearing in the sequence i, j, . . . . What could you achieve thereby? To
understand the extraordinary power such a situation would give you, consider
a quantum mechanical system with only two states, <PI and cPz. Then the
sequence i, j, . . . , contains only ones and twos. Think of the ones as dots
and the twos as dashes . Although you must maintain a specified ratio of
dots-to-dashes , you will nevertheless have the freedom in a long sequence to
34 HOYLE
convey a great deal of information in Morse code . It follows therefore that if

the condensation of the wavefunction is written structurally into our brains,
information sequences can be implanted in our memory, as I speculated was
the case in the last part of the main essay. This is not vaporing; it is a
consequence of the nature of quantum mechanics. It is a consequence of the
cylinder working from past-to-future being insufficient to establish a com
pletely deterministic theory. There would be no such possibility if this cylinder
was Newtonian in its character, since the Newtonian theory, being wholly
deterministic, would have no room for the operation of the second cylinder
from future-to-past.
It is insufficient for information sequences merely to exist. To be effective
they would have to be read and acted upon , just as an unread book in a library
is an arid thing. We can contemplate that the information sequences present
in a finite material system may or may not be read and acted upon, and we can
contemplate that this is the difference between animate and inanimate systems.
Starting from the known biochemical facts, we have grown accustomed to
think of animate systems as being somehow connected with DNA and with a
highly complex aggregate of biopolymers . But there is no reason why DNA,
or hemoglobin, or cytochrome-c, should in themselves guarantee conscious
thought any more than a lump of rock or metal does, or a silicon chip.
Something of a drastically different nature is needed. The difference I suggest
lies in the ability to read the sequences i, j, . .Whereas lumps of metal and
. .
rock are illiterate, biological structures (at least at a certain measure of multi
cellular complexity) begin to become literate . Various animals are literate in
various degrees. A dog has trouble in understanding a power station, and man
has trouble in understanding religi'on . And so back to the concluding passages
of the main essay.
MATHEMATICAL NOTE
To discuss the choice of a subobserver wavefunction <Pij from � j �j . . .

. . .
aj aj . . . cPj CPj . . . <Pij begin with the case of a system having only two
. . .
states, CP l and 1>z. Writing p =ladz, q l az lz, we have p + q 1 .

= =
For a sequence i , j, . . . , with n places there are n ! / (n - r) !r! possible

arrangements with I appearing r times and 2 appearing (n - r) times. Since
each such possibility for <Pij has to be weighted by p r q n
. . . the chance of
- r,
choosing a subobserver who finds r systems to be in the state CPl , and (n - r)

systems in the state CPl, is the rth term in the binomial expansion of (p + q) n.
Now the root-mean-square deviation of the binomial distribution from its
mean values of np and nq is known to be v;;pq , which becomes negligible
as n becomes large. The mean values np and nq are the statistical results of
quantum mechanics.
For three states CPl , cfJ , CP3, write p = la l l2 , q l azl 2 , r = la3lz , p + q +
=
2
r = 1 . Lumping � and CP3 together, use the previous result. A typical choice
of <l>ik will thus have np suffixes corresponding to CPt and nO - p) suffixes
. . .
corresponding to either CP2 or CP3. The ratio of the weighting factors associated
with � and CP3 is q:r, and since CPl does not appear in n( 1 - p) of the suffixes,
the weighting factors for CP2 and CP3 in these suffixes are q /(1 - p) and
r / ( 1 - p) respectively. Applying the previous result to these n O - p)
suffixes , one finds � appearing n ( I p) . q /(1 - p) nq times, and qh
- =
appearing nr times . One can progress in the same way to four states , five
states, . . . . Hence the general result.
The coefficients �ik appearing in the discussion of transitions must be calcu
lated from the explicit dynamical equation of the system in question. Once
these coefficients have been obtained, the discussion proceeds similarly to the
above.
Literature Cited
Bless, R. c . , Savage, B. D. 1972. Ap. J.

1 7 1 :293
Jamas , C . , et aL 1976. Ultraviolet Bright-Star
Spectrophotometric Catalogue. Paris: Eur.
Space Agency
Sapar, A . , Kuusik, 1. 1978 . Publ. Tartu Astro
phys. Obs. 46:71

Annurev Aa 20 090182 000245

Uploaded by

Copyright:

Available Formats

Annurev Aa 20 090182 000245

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Annurev Aa 20 090182 000245

Uploaded by

Copyright:

Available Formats

ANNUAL

THE UNIVERSE: PAST AND

Cumbria, Great Britain

Defining the Universe to be everything there is, manifestly we cannot be

"style" of physics, and indeed some of the most profitable adventures in

transformations , with "spin" determined by general complex transformations

'Frontiers of Astronomy, Wm. Heinemann, 1955.

relative proportions of graphite and ice. By now, the Institute of Theoretical

next move, to a galaxy-wide ensemble in which life originated and evolved.

'This situation is discussed at length in Space-Travellers, the Origins of Life, University

Figure 1 Size distribution of spore-fonning bacteria.

same in mathematics and science as it is in learning a musical instrument, or

disturbance . Nothiqg of.rthis;l:kind··can .happen when the storage is of the

t o b e clear in your u s e o f words . S o I tried discussing the difficulty with my

"The possibility of constructing a nuclear bomb was apparent as early as 1939.

pure-amplitude aspect of normal quantum mechanics involves a redundancy,

ApPENDIX: ON THE CONDENSATION OF THE WA VEFUNCTION

Let cfJ;, i = 1 , 2, . . . , be a complete orthonormal set of wavefunctions, each

wavefunctions belonging to a set of :��.ubobservers ," each �ubobserver wO!lld

observations by good interactions would lead subobserver <Pi to <Pi,i . . . , which

<pj, <Pj' . . . . Now choose a particular <l>j,j • , subject to weights given by

the squares of the moduli of the coefficients ai aj • , but otherwise at

For a sufficiently long sequence of interactions, and for most choices of

generation of an enormous tree through repeated branchings is the increasing

theory of specifying the particular route of which we are consciously aware.

might attempt an inversion of the situation . Take the interactions as they

abstract sense as solutions of the dynamical equations of the system, nobody

interaction with the observer!) . Usually ¢i . i 1 , 2 , . . . , are stationary =

states , which is to say the eigenstates of an approximate energy operator. The

the wavefunction becomes

<p;,}, . . . 2:2: ' . . . g;k g;f . . . ¢k ¢e .

to its own now-independent dynamical variables.

coupled sequentially by good interactions to the sequence of quantum systems.

The second interaction period then leads to

LL . . , ai aj • • • LL . . . tik �t • • . 4>k <Pc • , • <l>ik,jC, • , • •

for the first interaction period will define a subobserver, <l>i',j"

there is no requirement for i ' , j I ,

second interaction period to have evolved from the state of consciousness in

called snap decision we make depends on only a few quantum transitions in

'October the First is too Late. Wm. Heinemann. 1966

Turning now to the single-picture theory, in which the wavefunction is

choosing of a particular subobserver wavefunction <fIij , " subject to the choice

satisfying a broad statistical requirement that ensures that the numbers 1 , 2 ,

convey a great deal of information in Morse code . It follows therefore that if

To discuss the choice of a subobserver wavefunction <Pij from � j �j . . .

states, CP l and 1>z. Writing p =ladz, q l az lz, we have p + q 1 .

For a sequence i , j, . . . , with n places there are n ! / (n - r) !r! possible

choosing a subobserver who finds r systems to be in the state CPl , and (n - r)

Bless, R. c . , Savage, B. D. 1972. Ap. J.

You might also like