Earth is even now advertising itself splendidly to the Universe. We cannot know whether the first artificial nonhuman signal detected would place us in the role of intended recipient or of eavesdropper, but in our explorations we should allow for both possibilities.
One can make either of two basic assumptions about our first contact with an extraterrestrial civilization: (1) that it will arise through a purposeful attempt, perhaps through the use of an interstellar radio beacon or (2) that a civilization will be detected through no special efforts of its own. The latter hypothesis, often called eavesdropping, is concerned with the extent to which a civilization can be unknowingly detected through the by-products of its daily activities. While much thought has gone into the idea of purposeful contact, eavesdropping has been somewhat neglected; we will argue that it deserves more attention.
The overall likelihood of contact through eavesdropping depends on the nature and intensity of the civilization's "leakage," as well as on how long that leakage continues. Very general arguments (Oliver and Billingham, 1973, and the other papers in this volume) show that radio waves provide the most economical and reliable means of contact at interstellar distances. This is true not only for intentional contact, but probably for eavesdropping as well. In any case, there can be no argument with the fact, first discussed in print by Webb (1963) and Shklovskii and Sagan (1966), that the presence of humans can already be detected at interstellar distances as a result of the complex communications and transportation network now spread over our globe. Of course, we do not know how applicable our present situation is to  the more general case of all galactic civilizations over all time. It may be that our present "leaky" state will soon be terminated by advancing technology, but on the other hand it may continue for a very long time, perhaps even longer than any period in which we might have the perseverance to send out purposeful messages. If we are at all typical, then we should perhaps also be looking at least as diligently for unintentional signals from others as for intentional ones. To explore the principles involved, we might start by asking about emissions from Earth as detected at interstellar distances.
Detailed consideration of all parts of the electromagnetic spectrum reveals that radio waves represent by far the most important "leakage" from Earth (see Bracewell, this volume). For instance, nothing that we do with visible light, not even exploding a hydrogen bomb, compares in the least with the Sun's output. But at wavelengths from 1 cm to 30 km, our society has organized a host of activities that give our planet an unnatural "radio signature": television and radio broadcasting, radars used for weather, navigational, and military purposes, "short-wave" communications ("hams," CBs, taxis, police), satellite communications, and so forth.
We now want to put ourselves in the "shoes" of an extraterrestrial radio astronomer on a planet revolving about a star far from our Sun. Which of these radio services would be best for our eavesdropper to tune in on? Which is detectable to the greatest distances? Which potentially carries the most information of use to the eavesdropper? To answer these questions one must study many factors, including the power of each service's transmitters, the frequencies and bandwidths involved, the types of antennas used, and the fraction of time spent transmitting. (Readers desiring technical details on these and other matters should consult Sullivan et al., 1978). One example of these factors is the general tradeoff between the information content (TV picture, spoken words, Morse code) of a transmitted signal and the farthest distance at which it can be detected. This can be understood by noting that one gets more range by concentrating transmitted power at the fewest number of frequencies possible. But the information content of a signal is contained in the arrangement of its power among a number of neighboring frequencies and increases as we spread the power over a greater bandwidth.
Three other important criteria in the evaluation of each radio service are: (1) that the signal should be exactly the same from day to day, (2) that the amount of sky illuminated by the transmitting antennas should be large, and (3) that the number of transmitters on Earth should be large. Regarding (2). remember that the radio waves from even a stationary antenna can  sweep out a large portion of the sky as a result of Earth's rotation. Furthermore, each antenna has a characteristic beam into which the transmitter power is directed. If an antenna is designed so that the power is concentrated into a relatively small region of the sky, the range of detection for the signal increases, but at the expense of excluding many potential listeners.
Keeping the above factors in mind, an examination of all radio services reveals two categories of strong signals escaping Earth that might be of interest to an extraterrestrial observer.
An acquisition signal merely announces our presence over a large region of space by its very existence but is not generally useful for careful study because it fails to meet one or more of the criteria given above. An information signal, however, satisfies all three criteria. At the present time on Earth, some of the most important acquisition signals originate from a half-dozen or so U.S. military radars (and their presumed Soviet counterparts). These Ballistic Missile Early Warning System (BMEWS) radars sweep out a large fraction of the local horizon with extraordinarily powerful transmitters. The result is that this "radio service" provides by far the most intense signals that leak from our planet to a large fraction of the sky.
While BMEWS radars pass criterion (2) above, they fail (3) and partially fail (1) because they are so few and often change their frequency of operation to avoid being jammed. Nevertheless, if an external observer used equipment comparable to the most sensitive radio telescope on Earth (the 305-m diameter dish at Arecibo, Puerto Rico), we calculate that a BMEWS-type radar could be detected as far away as 15 light years. This distance includes only about 40 stars, but, of course, it is possible that our eavesdropper possesses a much more sensitive radio telescope than we do. If "he" had something like the largest one ever proposed for Earth, namely, the array of a thousand 100-m dishes called for by Project Cyclops (Oliver and Billingham, 1973), he could detect a BMEWS-type radar at a distance of 250 light years. In this case at least 100,000 stars are possible candidates for such an eavesdropper's location. But note that radio waves travel at the finite speed of 1 light year per year, and thus it will take until the 23rd century, or 250 years from now, before all these stars have had a chance to be bathed in the radiation of our defense system radars!
After picking up a BMEWS (or other) acquisition signal, the observer needs at least 100 times more sensitivity in his equipment to reach the rich lode of information signals emanating from Earth. It turns out that television broadcast antennas (or stations) are the most intense sources of such signals.
 All other services either have their transmitter power spread over too broad a Frequency band (for instance, FM broadcasting and most radars) or do not transmit continuously (ham radio operators) or from the same location on Earth each day (taxis, aircraft). Many signals, such as medium-wave AM broadcasting and almost all short-wave communications, never even penetrate the ionosphere, the reflective layer of charged particles that surrounds Earth. We thus concentrate on TV broadcasting; all other services that leak from Earth are less intense and merely add to the background noise a distant observer would measure in the direction of our Sun as seen in his sky.
But again note that TV broadcasting from Earth has been in existence for only 40 years. Figure 1 illustrates the phenomenal growth in intensity of the resultant ever-expanding "power bubble." On a cosmically infinitesimal time scale, Earth has indeed become a very bright planet, outshining the Sun by orders of magnitude in certain narrow frequency ranges.
To understand why television is so valuable to the eavesdropper as an information signal, we should discuss some of the characteristics of TV broadcasting signals. Perhaps the most important facts are that there is a large number of very powerful TV stations on Earth (fig. 2) and that about half a station's broadcast power resides in an extremely narrow band of frequencies, only about 0.1 Hz wide, called the video carrier signal. The other half of the power contains the picture information and is spread out in a complex manner over a far larger frequency range of about 5 MHz. Nowhere in this broader region is the power per Hertz even a thousandth that at the video carrier frequency. It would therefore be much more difficult (a factor of 2 x 104 is a good estimate) for the eavesdropper to receive full program material than to simply detect the presence of the carrier signal. (Given the quality of most TV programs, we find this fact very reassuring.) An observer near Barnard's star, third closest to the Sun at a distance of 6 light years from Earth, is thus about to receive television signals originating from the 1974 House Judiciary Committee hearings, but he probably cannot find out if Nixon was impeached. In the discussion below, we assume that only the video carrier signals of stations, not program material, are detected.
The combination of reasonably high power and small bandwidth means that the most powerful TV carrier signals can be detected (at optimum frequencies of 500-600 MHz) from distances as large as 1/10 those discussed for the BMEWS radars. The narrow-band nature of the signal also enables the observer to measure extremely accurate Doppler shifts in the frequency of the carrier signal, allowing a determination of the relative speed with which.....
 .....each station is moving to an accuracy of about 0.1 m/sec. Each station's signal thus contains information concerning the myriad motions in which its broadcast antenna participates while anchored to the rotating and revolving Earth (see fig. 3). Note that stations on a common channel will not fall precisely on top of each other's frequency because the combined effects of engineering sloppiness, deliberate frequency offsets, and Doppler motions all shift a station's video carrier frequency by much more than its width (fig. 4). This means that our hypothetical observer could not obtain a more favorable signal-to-noise ratio by trying to receive simultaneously all the Channel 5s, ....
 ....for example. The problem of detecting radio leakage from Earth as a whole is thus essentially identical to the problem of detecting its single strongest transmitter.
The beam patterns into which TV broadcast antennas radiate are important in such an analysis. It turns out that these antennas (whose purpose, after all, is to broadcast to Earth and not to the stars) confine the transmitter power to within a few degrees of the horizon, but distribute it about equally in all compass directions. Those radio waves directed above the horizon completely escape Earth's atmosphere, and even about half those below the horizon manage to escape by bouncing off the ground. (Only a negligible portion ever reach any TV set.) Since most of the power is broadcast near the horizon, only when a star is rising or setting, that is, when it is on the horizon as seen from a given antenna location, will it be illuminated with radio power (fig. 3).
After his initial discovery of these radio waves from the direction of our Sun, our eavesdropper would undoubtedly first ask, "Is this some kind of strange natural radio emission or has some form of civilization produced it?" It would seem that the narrow-band nature of the signals would be one of the best clues that the signal is artificial in nature, since no astrophysical process known to us can channel comparable amounts of energy into such small-frequency intervals. Other clues, such as polarization of the signals, also exist. And yet, who knows? Perhaps the theorists of another planet are clever enough to come up with a substance whose emission spectrum matches that of the observed radio waves! Clever theorists notwithstanding, for this discussion we assume that the signals from Earth would be recognized as artificial.
As shown in figure 3, when a star is near the horizon and thus illuminated by a particular station, the station must be near the edge of Earth as seen from the direction of the star. The result is that Earth has a very "bright" edge or limb, when observed with a receiver for TV frequencies (40 to 800 MHz), but the great distance to our eavesdropper's radio telescope means that he is unable to discern the disk of Earth. Nevertheless, the Doppler shift of each station due to the rotation of our planet can tell him not only whether the station is on the approaching or receding side of Earth, but also whether its latitude is near the fast-spinning equator or the more slowly moving polar regions. Furthermore, he could discover a station's longitude from the times of the twice-daily appearance of the signal from each station. Thus he could construct a map (just like fig. 2, but of course without the outlines of the continents) of all detected stations, each located to an accuracy of a few kilometers.
Because of the extremely nonuniform distribution of stations on Earth, the total number of stations visible at any one time to an outsider will vary with a period of 24 hr (sidereal). The situation as it would be measured from  Barnard's star is shown in figure 5. The peaks correspond to the times when population centers with concentrations of television transmitters are on Earth's limb. By combining data on these intensity variations and Doppler shifts in a straightforward fashion, any eavesdropper could deduce his position relative to our equator (we would say his declination), the radius of Earth (6000 km), and the rotational velocity at the equator (0.5 km/sec).
With this information in hand, the observer is likely to suspect that he is dealing with a planetlike body. His next step might be to study Earth's annual motion about the Sun (at a rate of 30 km/see), which causes very large Doppler shifts in the signals of all the individual stations. By tracking these shifts over a year or more, the Earth-Sun system can then be investigated exactly as astronomers here study what they call single-line....
 ....spectroscopic binaries. In such a system, two bodies (usually two stars) are orbiting about each other, but only the Doppler shifts in the spectral lines of one member (usually the brighter of the two) can be measured. In the present case, the "spectral lines" are the TV carrier signals and the "bright" member is Earth, far outshining the Sun at the radio frequencies we are discussing. It can be shown that radio observations of Earth, together with standard optical observations sufficient to give an estimate of the mass of the associated G2 dwarf star, which we call our Sun, would yield all the vital orbital data for Earth -its orbital period, its eccentricity, the true Sun-Earth distance, etc. The eavesdropper would then be able to provide his colleagues in the Exobiology Department with a good estimate for Earth's surface temperature, allowing them to place constraints on the possible forms of life responsible for the radio signals. It also turns out that from the duration of each station's daily appearances, one can readily deduce the beam size and thus the dimensions of the transmitting antennas (typically 15 to 20 m), yielding vital clues to the size scales of terrestrial engineering.
There are also more subtle effects contained in the TV carrier signals, effects which may or may not remain ambiguous to the observer. For instance, seasonal variations in vegetation, weather, and the ionosphere will leave their mark in each station's signal. Vegetation has an influence on the amount of power reflected from the surface, as does the choppiness of the sea for coastal stations. Weather and ionosphere affect the direction and intensity of the radiated power, either through winds flexing the antenna structure or through our upper atmosphere bending and absorbing the radio waves on their way out. These conditions will cause the observed power levels and times of station appearance to vary slightly and, at first, inexplicably from those predicted. Detailed study may nevertheless allow a few basic conclusions; for example, the presence of an ionized gas around the planet might be deduced from the clue that the lowest frequency stations are much more affected than those at higher frequencies.
A second type of complexity results from such factors as a station's daily sign-off hour and the specific frequency and antenna conventions that it follows. These generally vary from one country to another, but can be the same even for countries that are widely separated but otherwise cooperative in trade, politics, or technology. For example, frequency assignments and other conventions are very similar in Japan and the United States. We can interpret these diverse patterns with our detailed cultural and historical knowledge, whereas the extraterrestrial probably cannot, unless his social theory is advanced far beyond our own. The overall problem is not unlike that confronting an archeologist trying to understand an ancient city with a knowledge of only its street plan. It can only be hoped that the many unsolved puzzles would not hinder the eavesdropper from understanding the more regular and straightforward features of Earth's radio spectrum.
In order to sample the radio signature of Earth from an external site and thus test whether TV broadcasting is in fact the principal component, S. H. Knowles and the author used the Moon as a handy and objective reflector of Earth's leakage. Using the 305-m Arecibo radio telescope, we scanned a wide range of frequencies between 100 and 400 MHz and found that, once local interference was eliminated (using an on-Moon, off-Moon technique), the frequencies of most observed signals could indeed be identified with the video carriers of various nationalities (fig. 6). Not all of the countries mentioned in figure 6 were on the limb of Earth as seen from the Moon, however, illustrating that some power leaks also from relatively high elevation angles at the transmitters, although of lesser intensity.
The above discussion is, of course, relevant to the larger issue of our own attempts to contact extraterrestrial civilizations. We note that the one civilization we know something about (our own) has sent out virtually no purposeful signals, yet has been leaking radiation for several decades. How typical this situation will prove to be in our own future or for other galactic civilizations is impossible to say. Cable television may replace the present system of broadcasting antennas, but new forms of radio leakage may appear as well. For example, even the slightest bit of back-lobe leakage from the giant transmitting antennas proposed for the solar power satellite would create extremely intense, repeatable, narrow-band microwave signals. It is true that the range of detection of a purposeful beacon is probably much larger than for leaking signals, but another civilization might be leaking prodigious amounts compared to us, or they might have even set up powerful navigational beacons for interstellar travel. Furthermore, purposeful signals require decoding of any received message, while we have seen that unintentional signals, so long as they are narrow-band and periodic, yield a great deal of information using only standard astronomical techniques. Not only that, but in a sense the information gained may be a more accurate reflection of the society's major concerns. At least this seems to be true for the case of our own civilization with its military and TV leakage, although we might not wish to admit it.
In terms of actual search strategies at the radio telescope, the eavesdropping hypothesis does not come into direct conflict with that of purposeful signals, but rather it suggests that a well-designed system must allow for....
....more possibilities. In particular, extremely narrow-band signals drifting in frequency and variable in intensity must not be automatically filtered out. And, of course, the rationale behind special "natural" frequencies fails; on the other hand, leakage will occur at such frequencies as well as at any others, so they should still be used, although perhaps not exclusively.
 In summary, then, we should keep two possibilities in mind when searching for extraterrestrial signals. We cannot know whether the most likely signals to be detected would place us in the role of intended recipient or of eavesdropper.
- Oliver Bernard M; and Billingham, John, eds. Project Cyclops a Design Study if a System for Detecting Extraterrestrial Intelligent Life. NASA CR-114445, revised edition, 1973.
- Shklovskii, I. S.; and Sagan, C., eds.: Intelligent Life in the Universe. Delta, N.Y., 1966, pp. 255-257.
- Sullivan W. T., III; Brown, 5.; and Wetherill, C.: Eavesdropping The Radio Signature of the Earth. Science, vol. 199, Jan.27, 1978, pp. 377-388 (related correspondence in Science, vol. 202, p. 374 ff.).
- Webb, J. A: Detection of Intelligent Signals from Space. In: Interstellar Communication, A.G.W. Cameron, ed., W.A. Benjamin, N.Y., 1963, p.178.