The Search for Signals from Other Civilizations

Author(s): Sebastian Von Hoerner

Source: Science , Dec. 8, 1961, New Series, Vol. 134, No. 3493 (Dec. 8, 1961), pp. 1839-1843
Published by: American Association for the Advancement of Science

Stable URL: https://www.jstor.org/stable/1707703

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/1707703?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

American Association for the Advancement of Science is collaborating with JSTOR to digitize,
preserve and extend access to Science

This content downloaded from

78.190.114.155 on Fri, 03 Feb 2023 16:15:13 UTC
All use subject to https://about.jstor.org/terms
 the technical state has been reached. Sci-
ence and technology have been brought
forward (not entirely, but to a high
degree) by the fight for supremacy and
by the desire for an easy life. Both of

The Search for Signals

these driving forces tend to destroy if
they are not controlled in time: the first
one leads to total destruction and the

from Other Civilizations second one leads to biological or mental

degeneration. In summary, we assume
that a state of mind not too different

The waiting time for answers may be greater than from our own will have developed at
many places but will have only a lim-
the longevity of the technical state of mind. ited longevity.

Sebastian von Hoerner

Distance between Civilizations

All of the following quantities are

supposed to be average ones within a
A search for extraterrestrial signals knowledge of the diameter of solar the neighborhood
earth, of, say, 1000-parsec
from an intelligent source should would
be have given them a distance radius (2). to
We call vo the fraction of
guided by two estimates, one of the
thesun which is too large by a factor
all stars which have planets where life
probable nature of such signals and of
the2. They would have assumed that To the time needed to de-
can develop,
other of the distance from which the theysun had average absolute bright-
velop a technical civilization (defined,
might come. We cannot search for ness; comparison of the average
for appar-
example, by the presence of highly
something without at least a rough idea ent brightness of the ten brightest advancedstars
radio techniques), I the lon-
of what to look for, and we cannot with that of the sun, together with
gevity the
(3) of the technical civilization,
detect an object if the range of ourdistance of the sun as derived above, T the age of the oldest stars, and v the
means of perception is too short. The would have yielded a distance betweenfraction of all stars which at present
present article (1) is concerned in largeneighboring stars which is too small byhave a technical civilization. If we as-
part with the distance. The objective in 5 percent. sume, for the present purpose, that the
making such estimates is not to makeAll that is needed in this approach rate of star formation has been constant
is the right classification and one ab- over the time T, we then have
statements about other civilizations but
solute value to start with (in the fore-
solely to lead to a working hypothesis
which could guide a search. going examples, the diameter of the
/V (T - To)/T if I > T - To
Because we have no knowledge what- earth). The resulting estimate can be, v -
(2)
of course, completely wrong, but the
soever about other civilizations, we have \Vo (l/T) if I <T -To
to rely completely on assumptions. The probability that it will be is very small,
one basic assumption we want to make and the probability that the result will
If we call Do the mean distance be-
can be formulated in a general way: be right is high. This is the best we can
demand. tween neighboring stars, then the mean
Anything seemingly unique and pe- The basic assumption in the present distance between neighboring technical
culiar to us is actually one out (1) civilizations, D, is given by
of many and is probably average.
article is that our planetary system and
our civilization are about average and
D = Dov-"/3 (3)
As a demonstration of the power of that life and intelligence will develop
by the same rules of natural selection
this method one can show that even the
ancient Greeks could have estimated wherever the proper surroundings and In order to obtain the average longevity
the needed time are given. This includes 1, we have to go into some detail. We
the distance of the sun from the earth,
and even the distance between neigh-the assumption that the average civili- adopt the following five alternatives by
zation will reach our present level of in- which the longevity of a technical civil-
boring stars, if they had just applied
tellectual concern or state of mind (sci- ization (or of its technical state of mind)
the foregoing assumption to the earth,
ence, technology, search for interstellar might be limited: 1) complete destruc-
assuming it to be an average planet,
and to the sun, assuming it to be communication)
an after about the same tion of all life; 2) destruction of higher
length of time as we did and will face life only; 3) physical or mental degener-
average star. They would have assumed
about the same difficulties as we do.
that the earth was of average diameter, ation and decay; 4) loss of interest in
albedo, and distance from the sun;However,
com- we should also assume that science and technology; 5) no limitation
our present state of mind is just one ofat all. In cases 2 and 3, another civiliza-
parison of the average apparent bright-
many
ness of the five known planets with thatpossibilities and that it will be tion might develop on the same planet
succeeded by other interests and
of the sun, together with the Greeks' out of the unaffected lower forms of
activities. life, and we assume that the time needed
The author, who was affiliated with the Na-
tional Radio Astronomy Observatory, Green We should not underestimate thefor such recurrence is small compared
Bank, West Virginia, when this article was
written, is a member of the staff of the Astro- power of two critical factors thatwith
can To. Now, we call lh . . . the
nomisches Rechen-Institut, Heidelberg, Germany. terminate the life of a civilization once average longevity in the above five
8 DECEMBER 1961 1839

This content downloaded from

78.190.114.155 on Fri, 03 Feb 2023 16:15:13 UTC
All use subject to https://about.jstor.org/terms
alternative cases and pi . . . pa the If we adopt T = 101 years, v0.06 = 3) The value D = 360 parsec means
probability of their occurrence. If . . . [somewhat less than the estimate c )f Su that the antenna-receiver system to be
14 T - To, we have Shu Huang (4)], and Do = 2.3 arsec p; used for a search should be able to
(as the average distance of the ten reach a distance of at least, say, 400
v = v,o I [(M.)/T] + . . -(pJ /T)nearest stars from the sun), we>et g parsec, and this also demands an esti-
+ ps(T -To)/T t x [I + (p, + p3) (4) mate of the probable nature of the
v - 2.6 X 10-7
+ (p, -p3)2-+ ...] ) signals and the power emitted.
as the fraction of stars which
or have 4) The civilizations we find will very
technical civilizations at present,
nd ai
probably be much older than we are,
v voQ(1/T) (5)
1 and they will be more advanced. Our
Do = 360 parsec
with the average longevity I defined by chance of learning from them might be
4 as the average distance to the tenlear-
r considered the most important incen-
1 {=1
pili + p5(T - T) (6) est technical civilizations. Furthern lore, tive for our search.
we find from Eq. 8 that the firstivili-
c 5) Since 360 parsec is about 1000
zation we
and a recurrence factor Q, defined byreceive signals from willhave
I light years, the waiting time tw for an
a most probable "technical age"foanswer to a question will be about 2000
Q = l/[1 - (p.q+p)] (7) years on the average; this implies three
t= 1.2 X 10*yr
Another interesting question is the (14) consequences: (a) the contacting signals
following: At what stage are and
thethus will have weathered the first would already contain messages (in-
first
crisis
civilizations we meet most likely to (destruction)
be? a long time ago; eluding an introduction to a language);
We call t the time from the beginning and there is a probability of (b) there would be no need to hurry in
of their technical phase (defined by ad- "speaking," and relatively slow pulses
pr = 75 percent
vanced radio techniques) to the present. (15) might be used; (c) there might be
The probability that the first civiliza- that it will be the successor of ol Ider, some "speaking" and "listening," but
tions we contact will be of group i is inet. "mutual exchange" of ideas would be
extinct civilizations on the same pla
given by Pi - v v, and their average On the other hand, the chance ofieet-
n rather limited because of the long time
"technical age" at the moment of con- ing a civilization in exactly the ame
s scale involved.
tact is tf =- h/2. The most likely value phase that we are in [still confro nted
for their technical age, then, is with the crisis of destruction (grc
oups
:ent. Possible "Feedback" Effect
i = I and i = 2)] is only 0.4 perc
t -Pait p = (E p,il2) /2l (8) Finally, we define the average lon
gev-
s by
ity of the most frequent civilization The mutual exchange of ideas just
The probability is L = 2t and obtain mentioned leads to the consideration of
a feedback effect of the longevity, via
pr -p p2 - P3 (Q - )/Q (9) L=2.4 X lOyr (16) radio communication, on itself. Suppose
that there will have been other civiliza- that the estimated value for the (unaf-
tions before them on the same planet. fected) longevity I (Eq. 11) is too large,
First Conclusions so that in reality the waiting time for
The foregoing analysis seems to be
fairly straightforward up to this point, answers, ti, is greater than the longev-
As mentioned earlier, this estir nate ity of the technical state of mind. Then
but it tends to become a matter of per-
should be regarded only as a worl king nobody will ever get an answer to his
sonal opinion when one begins to adopt
hypothesis for the purpose of guikding call. Some still-hopeful civilizations (af-
numerical values for the average lon-
a future search for extraterrestrial
sig- ter having made too optimistic an esti-
gevities, 14, and the probabilities of oc-
nals. If we assume that the va lues mate for 1) might, for a while, send
currence, pi, of the various alternative
adopted in Table 1 are not too ong,wrcsignals which might be picked up occa-
cases. As a justification for doing so at
and if we neglect the "feedbackect" eff
sionally by others. But if the search for
all, I mention two arguments. First, one
discussed in the next section, we
can signals, on the average, is not success-
cannot design an adequate receiving
draw the following conclusions. ful, then loss of interest will usually
system without some estimate of this
1) The value v = 2.6 X 10-' mn eans come soon. This we call case A. On
kind. Second, the uncertainty of I en-
that only one in 3 million stars
will the other hand, suppose that our esti-
ters Eq. 3 only with the power 1/3:
have a technical civilization, and this mated value of I is too small and that
D -1-1/S (10) implies that we cannot search for sig- a real exchange is possible. This will
nals from a certain number of indi ivid- have a tendency to keep interest alive
In Table 1 appear the values which in
ual, conspicuous stars; we must < over a very long period and might even
scan
my opinion are the most likely ones,
the whole sky continuously. lead to civilizations' helping one an-
and for the sake of brevity I omit all
2) This value also implies that no other to solve problems and weather
the long discussions which led to will
other civilization these
send contac :ting crises. This we call case B. Thus, in my
values (p5 = 0, for example, means that
signals (intended to attract atten ttion opinion, there is a high likelihood that
I do not believe in this one at all).
and to establish contact) in the di irec- there will be either no exchange or a
Maybe this very subjective guess seems tion of our sun as one of the conspficu- great deal, but a low likelihood of an
a little pessimistic, but I want to be on nals in-between situation. A small amount
ous stars. But such contacting sig
the safe side. From these values we find
might be sent from beacons in all di.rec- of exchange is, so to speak, a non-
1 =6500 years and Q= 4 (I1) tions over the whole sky. equilibrium state. Unfortunately, our
1840 SCIENCE, VOL. 134

This content downloaded from

78.190.114.155 on Fri, 03 Feb 2023 16:15:13 UTC
All use subject to https://about.jstor.org/terms
estimate for the unaffected longevity power, while the remaining some, what the different quantities with the longev
just happens to fall into this unstable t en- ity L, I have made some calculations,
less difficult quantities Q, v,, and ph
region, so it is hard to tell which one ter only with the power 1/3. Therel fore, with the results given in Table 2. Usi
of the two alternatives is the one which we throw all uncertainty into L and the values that we used for Eq. 23,
will prevail. write we get
The governing quantity in this prob-
K = (L/Lo)'/
lem is the ratio of the average longevity (21) v = voQpkK/T = L/(1.25x l" yr);
of the most frequently occurring civili- with a critical longevity Lo definee
I by D = Do/v'/3 = 2.3 parsec/v"/';
zations, L, to the average waiting time
for answers, t,,. This ratio is Lo = (8DoT/ c3vQpk) "' (22) K = (L/L,,)/3 - (L/4500 yr)/3; (24)
If we use the values Do, = 2.3 pa; rsec, t,{ = 2D/c = 15 yr/v'/3
K = L/t1, (17)
T = 10 yr, c = 3 X 10' cm/ /sec,
where L = 2t, t is given byQEq.
v. = 0.06, 8, pk
= 4, and and
= 1/3, Eq. I wish to emphasize that the longevity
t,o = 2D/c (c is the velocity of light). 22 finally gives L is the one extremely uncertain quan-
From the values of our estimate (Eqs. tity, while all the other quantities in
13, 14, and 16) we obtain Lo = 4500 yr
Eqs. 24 may be trusted to a fair degree.
Because of the exponent 1/4 in22,
Eq.
Thus, whatever L we choose requires
K = 10.2 (18) I think the value for the critical Ion gev- that the answers of Table 2 for this L
Ten conversations per lifetime of a ity (Eq. 23) should not be too far then be accepted.
civilization would mean some exchange,wrong. Everything depends, then, on Especially, we must accept very long
but not much, and the question still re-
the question of whether the unaffe
cted waiting times t,. (of at least 1000 years
mains open whether this is enough to longevity is great compared with
50005 and probably more) if L is to stay with-
trigger, through feedback, case B. Inyears. If it is, the feedback will be in reasonable limits, and this is a rather
trig-
order to show more clearly in what gered, generating case B, and this will disappointing result. It means that the
way the quantity K depends on the as- increase the longevity L considera ibly, feedback and case B can be triggered
sumptions made, we write up to some limiting value L,,. only if the more highly advanced civili-
At this point, however, we I have zations are able to think, to plan, and
K c/2D,,[vo0Q) '('pp,lI/T'Ir )] (19)reached the limit of our method of
esti- to act in terms of thousands of years.
and for simplicity we will assume that,mate, because, having had no com mu- This is extremely different from our
from the various alternate groups ofnication with other civilizations, we own situation, in which we would be
Table 1, only one group, index k, has a have nothing to start with and wecan- happy if we could solve the problems
high value of the product pili as com- not know how strong the effect ofthe of the next 5 years. But it is not impos-
pared with the values for the other feedback will be. It is purely a pers{
onal sible, either, that natural selection and
groups. We then have L = h,: and belief if we think that L,, will nol t be increasing good sense might work in
higher than, say, a million years and this way. Furthermore, even if the av-
K = (c/2D,,)(v,,Qp.).'(L'/:'T:') (20)
that it will probably be much lessthan
t erage value for K is too small to trigger
We see that, unfortunately, the most that. case B, there might be large fluctuations
uncertain quantity, L, enters with a high In order to illustrate the variatioi n of of K in time and space; if the feedback
has triggered case B once somewhere,
the resulting effect then would tend to
Table 1. The most likely values for i, and p, (see text). go on in time and to expand in space.
Estimated Value adopted It is amazing how similar this seems to
Alternative range for 1 -- - --? - ptl, be to the problem of the origin of life
(yr) It (yr) Pt in general (and one might feel a strong
Complete destruction 0-200 100 0.05 5 temptation to draw some more paral-
Destruction of higher life 0-50 30 0.60 18 lels).
Degeneration 10'-10' 3X10' 0.15 4500 In summary, I think that the feed-
Loss of interest 103-10: 10t 0.20
2000 back effect will play an important role,
No limitation T - T T -To 0.00
0 in one direction or the other. The basic
idea is just that of giving up if one is
disappointed and of increasing the ef-
Table 2. Variation of 1/,, D, t,o, and Kone
fort if with longevity
is successful. But I am L (se
L 1t unable to say in which of the two di-
(yr)
_(yr) 1 /
____ 1/V___ l, (yr) K
(parsec)
rections the effect is more likely to be
100 1.3 X 10' 2,480 16,200 0 .006 felt. We expect to find either a high
300 4.2 X 10 1,720 11,200 .027 activity in communication at shorter
1,000 1.3 X 109 1,150 7,500 .13 distances (200 to 300 parsec) between
3,000 4.2 X 107 796 5,190 .58 civilizations of extremely long time
10,000 1.3 X 107 534 3,480 2
30,000 4.2 X 106 370 2,420 12
.9 scales (case B) or very little if any
100,000 1.2 X 106 248 1,620 61 4 activity at greater distances (600 to
300,000 4.2 X 10' 172 1,120 268 7 1000 parsec) from civilizations similar
1,000,000 1.2 X 106 115 750 1,330 to our own (case A). We should be
prepared for both possibilities.
8 DECEMBER 1961
1841

This content downloaded from

78.190.114.155 on Fri, 03 Feb 2023 16:15:13 UTC
All use subject to https://about.jstor.org/terms
Second Conclusions probability of success if it is guided bya large number of signals at the same
the best guess we can make, but almostfrequencies in both scans. This is the
If the feedback plays the role we none if it is made without a definite first thing to ask.
think it will, then some of our first plan. The following considerationsAsare
con- for long-distance calls, we have
clusions must be modified. Because we very incomplete and tentative; my estimated
main the probability of the earth's
cannot decide as to the direction of the purpose in proposing them is to stimu- being hit by one. Because the answer is
feedback, both possibilities must be late the formulation of better ones that not very encouraging I shall skip the
considered. finally could be used. details and just give the result. If each
1) No essential change is needed for As to the purpose, we can think of civilization speaks, on a permanent
Nos. 1, 2, 4, 5a, and 5b of our first three general possibilities: local commu- basis (and listens, as well), to a number
conclusions; we cannot search for singlenication on the other planet, interstellar n of its neighbors, if the messages are
stars, and nobody sends messages to uscommunication with certain distinct sent with beamwidth f/, and if we are
especially; the civilizations we meet partners, and a desire to attract the able to detect these signals at q times
at-
will be much more advanced than we tention of unknown future partners. the distance to which they are sent (q
are, and contacting signals will alreadyThus, the things we should look for we > 1 because detecting is easier than
contain messages which might usemight rela- call local broadcast, long-distance understanding), then the probability of
tively slow pulses. calls, and contacting signals. The local our being hit by chance is about
2) Our receiving system should be broadcast has the highest likelihood of
able to reach a distance of either 200
P (7r/120) q' p n2 .... (25)
existing but may be extremely difficult
to 300 parsec in case B or of 600 to to detect because of its relative weak- a value which is independent of L
1000 parsec in case A. ness. Long-distance calls would not be and D.
3) In case A there would be little or intended for us but might hit us just by If we regard P = 1/2 as sufficiently
no interstellar communication. In case chance; the probability is small, how- large to warrant a search and regard
B we should expect a highly developed ever. Contacting signals would be in- q = 5 and / = 1 minute of arc as likely
communication system and much ac- tended for exactly the kind of search values, this would require that each
tivity. we plan to make, and therefore they civilization should speak, on a perma-
4) According to the law of natural should have the highest probability of nent basis, with n - 1300 others, and
selection, a variety either has the will detection, provided they do exist. Local it seems very unlikely that this is the
and the ability to maintain itself or it broadcasts would exist in both cases case. But if we regard n = 50 as a
soon dies out. Thus, if the feedback A and B, as defined earlier, whilereasonable
long- value, we would then need
effect has triggered case B and still distance calls and contacting signals the somewhat extreme values q = 10
maintains case B, this implies that somewould exist in case B only. and / = 10 minutes of arc, which,
effective means exist for "beginners" to As to the frequencies used, those for again, are unlikely. Because all the un-
establish contact with other civilizations the local broadcast might be not too known quantities enter Eq. 25 at high
(contacting signals). different from our own, but for com- powers, we think that a chance hit is
munication over interstellar distances, highly unlikely (though not impossible).
the range of frequencies would be lim- Another difficulty is that the bandwidth
Nature of the Signals ited by considerations of economy. probably would be extremely narrow
Drake (5) calculated the combined in- and that we have no way of guessing
In order to have a reasonable hope fluence of galactic and atmospheric the exact frequency used.
of success, we should be guided in our noise and found a broad minimum be- The contacting signals form a fasci-
search by a definite idea of what to tween 1000 and 10,000 megacycles. In nating problem. Provided they do exist,
look for. This idea might turn out to a recent paper (6) Drake finds, even they are intended to attract the atten-
be wrong, and we would then have tofor sending and receiving from above tion of any new civilization. If we were
start with a better one. But it seems planetary atmospheres, a very general able to guess the most economical
hopeless to search the whole sky,rule all for defining the most economical method of doing this, we would know
the time, over all frequencies and withfrequency. It should lie in the range exactly what to look for; this would
extremely narrow bandwidth, just from
for1000 to 30,000 megacycles per greatly increase the probability of de-
"something." second, most probably at about 10,000tection and thus (to close the circle)
I suggest that we assume that the na-megacycles per second (X = 3 cm), a would make this method the most eco-
ture of the signals will be defined frequency
en- which still could be observed nomical. There is just one problem:
tirely by two things: (a) the purpose from within our atmosphere. how to define precisely the word eco-
they serve; (b) the most economical Drake (6) has pointed out that the nomical? I have to admit that I have
way to achieve it. Both of theselocal we broadcast would occupy a large not found a definition worth writing
might be able to guess. The argument number of narrow channels, distributed down, and I must, at present, leave this
that other civilizations could be com- over a larger frequency range. He hasproblem open.
pletely different does not help at all worked
in out a very effective method of Suppose we had found the right def-
guiding a search, even if it is true, detection, a cross-correlation between inition. This would enable us to calcu-
whereas the foregoing assumption will two independent frequency scans. This late for each method suggested the
lead to a definite program, even if itmethod
is is not concerned with the single price C (or whatever we call meas-
invalid; only by trying can we tellsignal, with its frequency or its strength, ure of the effort on our side, on the
whether it is valid or invalid. To sum- but answers with increased sensitivity other side, or on both) which has to be
the question of whether or not there are paid in order to yield a probability of
marize, I think that a search has a fair
1842 SCIENCE, VOL. 134

This content downloaded from

78.190.114.155 on Fri, 03 Feb 2023 16:15:13 UTC
All use subject to https://about.jstor.org/terms
 detection Pd over a distance Dd within noncritical parameters systematically pose. We should try to guess both, in
a time td. For Pa we might take /2; over their possible range. These, then, order to increase the probability of de-
for Dd, the average distance D (200 to are the rules of this fascinating game. tection.
300 parsec); and for td, half of the We should mention two more rules 2) We have considered three kinds
number of years after which most new which possibly could play a role. First, of signals, with different purposes,
civilizations would consider giving up because of the long waiting times, the which we called local broadcast, long-
their search if it had not been success- contacting signals would probably con- distance calls, and contacting signals.
ful (some hundred years, perhaps). We tain messages. There are two possibili- The local broadcast has the highest like-
calculate the value of C for all methods ties: either the whole contacting signal lihood of existing but would be ex-
suggested and conclude that the one would vary in the manner of a coded tremely difficult to detect. Long-distance
with the lowest value of C will be message, or its plan would be devised calls would exist in case B only, and
exactly the method used by the others, in such a way that it would direct our the probability of their hitting us by
with one condition. The methods we attention to the exact frequency where chance is very small. Contacting signals
can think of, as well as our definition of the messages were being sent. Because would exist in case B only, and these
economy, depend on our present state detection has higher priority than the have the highest probability of detec-
of advancement. The other civilizations message and detection is easier if no tion because they would be devised for
will be much more advanced than we irregularity (code) is involved, and be- that very purpose. For this reason and
are but will have had experience with cause there would be no requirement because of conclusion No. 4 of my
beginners, and they will have set a for haste, in "speaking," I think that the "second conclusions," I recommend
certain standard of what a beginner second case is the more likely, and that that we begin the search under the as-
should know and how much he should a few channels would be enough for sumption of case B and look for con-
be able to guess in order to be con- the message. Second, the contacting tacting signals. If this should fail, we
sidered a future partner. The condition, signal should not interfere with other might then increase our effort by
then, is that we already meet this activities, such as already existing com- searching, under the assumption of case
standard. But whether we do or do not, munication: this means, for example, A, for local broadcasts.
we shall find out only if we try. that it should not occupy too much of 3) At present, no definite program
The value of C should be lowest the whole frequency spectrum. (Con- can be given for the search for con-
when all power is sent in a single tacting signals and long-distance calls tacting signals. But the general reason-
narrow channel at a certain frequency would have about the same range of ing required to arrive at such a pro-
which can be guessed by the listener. most economical frequencies.) To give gram is given: to guess and estimate as
As Cocconi and Morrison pointed out an idea of the way in which the con- much as we can about the nature of
(7), the only "milestone" we know of tacting signal might direct attention to the signals and to assume that the
in the interesting range of frequencies the message, I shall give one example. sender knows how much we can guess,
which might be used for this purpose is We distribute a large number of signals because this approach leads to the most-
the 21-centimeter line. I suggest a over the economical frequency range in economical kind of contacting signals.
modification. The background of a a pattern which is symmetrical with 4) The search for other civilizations
signal would be much stronger within respect to the center of this range. will have either a tremendous result or
this line than beside it-so strong as Toward this center we decrease the none at all. Thus I recommend, hoping
to drown out a small signal-and the spacing between the signals (and their that the first case obtains, that we begin
boundaries of the line are not well bandwidth) in proportion to the dis- as soon as possible and try as hard as
enough defined for us to place the tance from the center, until we arrive at we can. But to be prepared for the
signal exactly beside it. The next sug- an extremely narrow channel at the second case, I recommend the design of
gestion, then, might be to use, for very center of the pattern (all other de- a receiving system which can be used
example, exactly twice the frequency of tails of the arrangement are defined for ordinary astronomy as well, since,
the 21-centimeter line. If this should entirely by minimizing C). In this center because of the size and sensitivity
fail we would have to look for more channel an introduction to the language needed for its prime task, it will be
sophisticated methods of producing con- is repeated every 10 years, say, and at extremely powerful. The observing
tacting signals. the end of this time the listener is told time should then be allocated in equal
Each method will consist of a general at which frequency to findparts
the nextbetween the two projects assigned
plan for distributing the transmitted message, and so on, all of these mes-
to the instrument.
power over space, time, and frequency sages being sent simultaneously but
and of a number of parameters govern- being read in the right order. Finally,References and Notes
ing these distributions. We should be the listener is told at what frequency
1. It is a pleasure to thank F. D. Drake for
able to guess this plan, and to evaluate and with what power he shouldmany stimulating and helpful discussions and
answer.
for reading the manuscript.
those parameters which minimize the
2. One parsec = 3.26 light-years = 3.086 X 1018
value of C. As for the parameters cm = 1.92 X 1018 mi.
3. The importance of this quantity and its con-
which do not influence C, we will just Third Conclusions nection with the distance was first pointed
have to try them out until we hit the out by R. N. Bracewell [Nature 186, 670
(1960)].
right values. Thus, the probability of 1) Our search should be guided by
4. S. S. Huang, Publs. Astron. Soc. Pacific 71,
detection has to be calculated under the the assumption that the nature421of
(1959).
the
5. F. D. Drake, Sky and Telescope 19, 140
assumption that the plan and the min- signals will be defined entirely (1959).
by the
imizing parameters are known on our purpose they serve and by the 6. -most
, in preparation.
7. G. Cocconi and P. Morrison, Nature 184, 844
side and that we vary the remaining economical way to achieve this pur-
(1959).
8 DECEMBER 1961
1843

This content downloaded from

78.190.114.155 on Fri, 03 Feb 2023 16:15:13 UTC
All use subject to https://about.jstor.org/terms