[163] A comprehensive search for other planetary systems must use ground- and space-based instruments. Indirect search can be done from the ground, but direct detection must be tried from space. Prospects for the latter are not clear at this time.
Whether life can originate and evolve in the absence of a planet is an open question, but the available evidence suggests that planets may be necessary to provide the conditions that are conducive to the formation and evolution of life. Thus we might regard planets as "cosmic petri dishes."
The role of planets in the origin and evolution of life in the Universe is sufficient rationale for considering the detection of other planetary systems. However, the rationale for undertaking such a search extends beyond the question of life in the Universe. Perhaps the strongest reason is that the results of the search, be they positive or negative, are essential if we are ever to understand the origin of the Solar System: only data concerning the frequency of occurrence of planetary systems and their general properties will allow us to test empirically the many working hypotheses on the origin of he Solar System. A comprehensive search may also yield valuable insight into the process of star formation, since the prevailing view on this process suggests that planetary systems should be a frequent by-product. Data from a search will either affirm this view or indicate the need for a new view of star formation.
A reasonable question to pose is whether any other planetary systems have been discovered. The answer is that there is no unambiguous evidence for the existence of other planetary systems. This answer may come as a surprise, given the tremendous increase in sophistication of astronomical observations. To be sure, advances in technology have had a significant [164] impact on astronomical instrumentation. Until very recently, however, none of these advances has taken a form that pertains directly to a search for other planetary systems.
There have been observations that have been interpreted as giving evidence for the existence of other planetary systems-the best-known example concerns a small and cool M-dwarf star known as Barnard's star. An extensive discussion of the controversy regarding the existence of companions to Barnard's star is beyond the scope of this paper; however, the essential elements are summarized below, and more detail can be found in Black (1980a).
Van de Kamp and coworkers at Sproul Observatory have conducted a prolonged study of Barnard's star and believe they have found some indication of a dark companion. The observational technique employed by the Sproul group was that of astrometry -the determination of the apparent position of a star as a function of time. Van de Kamp found that his data were consistent with either a single dark companion of mass comparable to that of Jupiter, revolving about Bernard's star in an eccentric orbit with a period of 24 years (van de Kamp, 1963), or two dark companions with masses comparable to that of Jupiter, moving in coplanar circular orbits with periods of 12 and 24 years, respectively (van de Kamp, 1969).
Hershey (1973) and Gatewood and Eichhorn (1973) have cast doubt on the validity of the earlier Sproul data. Hershey found evidence for small "sudden" changes in the apparent right ascension position of stars in a star field unrelated to the position of Barnard's star. The timing of these major discontinuities (1949 and 1957) coincide with times when there were adjustments and/or alterations to the Sproul telescope. They also correspond in both time and amplitude to the dominant apparent perturbation in the position of Barnard's star. Gatewood and Eichhorn analyzed data on Barnard's star taken with two other astrometric telescopes, one at Van Vleck Observatory and one at Allegheny Observatory. They were unable to confirm the perturbation that appears in the Sproul data. It appears likely that instrumental effects in the Sproul telescope gave rise to spurious data. It should be emphasized that these effects are at a very low level and do not vitiate other research done on the Sproul telescope. Further, these results on Barnard's star should not be taken as evidence that it does not possess a planetary system. Continued studies of Barnard's star, with increased accuracy, are currently underway at Sproul, Allegheny, and the U.S. Naval Observatories. However, the Barnard's star story emphasizes the difficulty of detecting other planetary systems. The expected magnitude of signals due to another planetary system is very small, generally at or well below the accuracy obtainable with existing telescopes.
The first detailed examination of techniques for detecting other planetary systems was a 1976 Workshop study sponsored by NASA-Ames [165] Research Center (Morrison et al., 1977). A second NASA-sponsored workshop on ground-based techniques for detecting other planetary systems was held in 1979 (Black and Brunk, 1980a,b). During the summer of 1976, a systems design study was sponsored jointly by NASA-Ames Research Center and Stanford. That study-Project Orion -centered on the design of a novel ground-based astrometric telescope that would be 30 to 50 times more accurate than existing systems (Black, 1980b). The material summarized here derives mainly from the findings of these three studies.
The review is divided into four sections. Some background on the problem of planetary detection is given in the next section. Ground-based techniques for detecting other planetary systems are then discussed, followed by space-based techniques. The paper concludes with a summary of the prospects. Of necessity, we will not discuss details of the many techniques or of the possible instruments that could be used in a search for other planetary systems. Readers interested in such detail can refer to Black (1980a) and references cited therein.
Before embarking on a discussion of the prospects for detecting other planetary systems, it will be useful to examine the nature of the detection problem itself. What are the likely observable manifestations of another planetary system?
To appreciate the possible range of observables that might arise from a planetary system, consider a model of a simple system (fig. 1). The model....
[166]..... system is comprised of a planet of mass Mp revolving about a star of mass M* in a circular orbit of radius Rp. The instantaneous center of mass of the system is located on a line between the star and the planet and is depicted in figure 1 as Iying a distance R* from the star. In addition to this geometric characterization, we must also characterize the radiative properties of each member of the system. An important parameter in this regard is the temperature of each body (Tp and T* for the planet and star, respectively). Also important is size, and we designate by dp and d* the diameters of the respective bodies (taken to be spheres for simplicity). More complicated characterstics are needed to completely specify radiative properties. However, these additional characteristics (such as albedo and magnetic field strength) are not possible to specify a priori in any reasonable way, so we will begin by specifying only a limited number of parameters.
How might this simple planetary system be detected? If there were no planets, the center of mass of the star would coincide with the center of mass of the total system. The presence of a planetary companion has the effect of displacing the star with respect to the center of mass of the system (barycenter). Further, because the planet revolves about the barycenter, the star must also revolve about the barycenter. It is essentially a celestial teetertotter; the star constantly adjusts its position to "balance" the effect of the planet. For this simple system, the displacement of the star is simply
(1)
If the planet is revolving with an orbital
period of P, the star must also -revolve with that period. If this
orbital motion is superposed on a general linear motion of the entire
planetary system, a distant observer will see that the star does not
move at a constant rate in a straight line (the barycenter does);
rather it will appear to "wobble" as it moves across the sky. The
magnitude of the true wobble is given by equation (1), but the
magnitude of the apparent wobble depends
on the distance between the observer and the planetary system. If we
denote that separation distance by D, the apparent angular extent of
the wobble,
, is given by
= arctan(R*/D)
Because R* is always much smaller than D, we can use the approximation
R*/D radian (2)
Combining equations (1), (2), and Kepler's Third Law gives
= 9.8 x 10-4 KpMp arcsec (3)
[167] where
The units of P, Mp, M*, and D are,
respectively, years, Jovian masses (~1.9 x 1030 gm), solar masses
(~2 x 1033 gm), and parsecs (~ 3 x 1018 cm). This grouping
of parameters may seem strange, but note that Kp involves distance,
stellar mass, and orbital period. Each parameter is determined
observationally, as is ; M* is determined independently, based on the type of
star. This isolates Mp, the planet's mass, as the principal unknown.
We will consider some numerical examples later in this section. The
measurement of the angular displacement, , is accomplished through astrometric observations (such
as those by van de Kamp).
In addition to causing an angular displacement in the position of a star, the presence of a planet causes the star to orbit the barycenter of the system. The orbital speed, v*, is given by
29.8 K2Mp m/sec
where K2 
(PM*2)-1/3. Now, when a body
emitting or absorbing certain spectral features moves with respect to
an observer, the wavelength 
of that emission or absorption changes relative to the
wavelength NO it would have if the source were not moving. The change
in wavelength, 
, due to this Doppler shift is given by
(5)
where c = 3 x 108 m/sec is the speed of light and vLOS is the speed of the star toward (or away from) the observer, that is, along the line of sight. For our model planetary system, we have
(6)
where i is the inclination angle between an
observer's line of sight and the orbital plane of the planetary
system. Note that, if 
(observer looking normal to the
orbital plane), then 
and is independent of the value
of v*. The quantity that is measured is essentially 
, which can
then be related to the product v* cos i, which is
proportional to Mp cos i. Instruments that measure these wavelength shifts
are called radial velocity meters. As we see from a comparison of
equations (2) and (6), the amplitude of the apparent astrometric
observable is smaller for more distant stars than it is for nearby
stars (for identical R* values). This is not
the case for the radial [168] velocity; the
amplitude of is independent of the distance D. On the other hand, a
small almost always implies that a low-mass companion is
present' whereas a small does not necessarily imply that a small companion is
present (because of the cos i factor).
Another possible observable manifestation of a planetary system involves the transit of a star by a planetary companion. During a transit, the planet passes between the star and an observer and blocks out part of the starlight. The associated apparent dimming of the star could, in principle, be detected with very accurate photometric observations. Because this technique for detecting other planetary systems is somewhat limited, we will not discuss it further here.
Astrometry and radial-velocity studies involve observations of the stellar member of a planetary system; the presence of planetary companions is inferred from some observable effect that those companions have on the star. For that reason we refer to astrometry and radial velocity as indirect detection techniques.
Both the planet and the star depicted in
figure 1 would be sources of radiation. Generally, the total
radiation flux 
from the planet will consist of a sum of three
components: thermal, reflected, and nonthermal (reflected radiation
is nonthermal, but it is convenient to treat it separately); that
is,
(7)
where v is the frequency of the radiation.
Although there are many known examples of nonthermal planetary
radiation (e.g., Jupiter's radio bursts), it is virtually impossible
to describe in a general way the range of phenomena that could give
rise to nonthermal emission from a planet. In this paper we will
ignore this contribution to and concentrate on the thermal and
reflected components.
If we assume that the intrinsic thermal flux from a planet is that of a black body of temperature Tp, then
(8)
where 
is the power
per unit frequency interval, and the constants c, h, and k are the
speed of light, Planck's constant, and Boltzmann's constant,
respectively. The reflected component is given by
(9)
[169] where f denotes the
fraction of the total stellar radiation flux, 
, that is incident upon a planet, and the parameter 
denotes the
fraction of the incident flux that is reflected. In the context of
figure 1,
(dp/4Rp)2 (10)
The parameter is generally a function of frequency and of the
composition of the reflecting medium; it also depends on the position
of an observer relative to the star-planet pair. Clearly, the
radiated flux from a planet depends on a number of factors, and we
consider next a specific example to obtain a better feeling not only
for the magnitude of the problem, but also for the frequencies that
are best from a detection standpoint.
Figure 2 shows as a function of frequency for the Sun and for Jupiter.
In deriving these curves, it was assumed that the Sun radiates as a
black body and hence its radiation is described by equation (8) with
the solar diameter and temperature replacing dp and
Tp,
respectively. The Jupiter curve was obtained by taking 
The contrast ratio between....
[170] .....the Sun and
Jupiter is lowest at low frequencies, approaching a limiting value of
The contrast
ratio at the frequency of peak emission by Jupiter is somewhat
larger, about 104. By comparison, the ratio in the visual part of the
spectrum, where Jupiter is simply reflecting sunlight, is about 5 x
108.
The contrast ratio between star and planet is only part of the detection problem. Another major aspect of the problem is the angular separation between the star and planet. Pursuing the model of a Jupiter-Sun pair, we note that at a distance of 10 parsecs (32.6 light years), the linear separation between Jupiter and the Sun corresponds to only 0.5 arcsec angular separation. It is this combination of a relatively bright object located very close (in angular measure) to a relatively dim object that makes direct detection of planets revolving around other stars so difficult.
We next examine the indirect detection of a
model Jupiter-Sun system as viewed from a distance of 10 parsecs. The
maximum angular extent of the wobble in the motion of the Sun due to
Jupiter alone is given by equation (3). In this case,
Mp =
1, M* = 1, D = 10, and P = 11.8, giving 
5 x 10-4 arcsec. Some
appreciation for the smallness of can be had by
realizing that 5 x 10-4 arcsec corresponds
to only a few angstroms (1 Å = 10-8 cm) as viewed from a
distance of 15 inches. Thus our planetary detection problem is akin
to measuring the displacement of atoms in this page of paper due to
thermally induced vibrations! If we were to consider the Earth-Sun
pair alone, we would find ~ 3
x 10-7 arcsec (again as viewed from a distance of 10
parsecs), smaller than the Jupiter-Sun example by a factor of nearly
1600. What are the expected radial velocity effects for our model
planetary system? Using equation (4) with Mp = 1,
M* =
1, P = 11.8 (for Jupiter), we find that v* 13 m/sec. By way of comparison,
the Earth-Sun system would lead to v* 0.09 m/sec.
These numerical examples are not meant to represent the expected run of observable parameters for nearby planetary systems (if any exist), but they should give the reader a feeling for the magnitude of the detection problem. Most nearby stars are less massive and fainter than the Sun. The former characteristic is generally helpful from the standpoint of indirect detection techniques, while the latter is either helpful or not, depending on the wavelength used, for direct detection techniques. Against this backdrop of the obvious difficulties inherent in trying to detect other planetary systems, we turn next to a brief review of the capability of existing instruments.
Earth's atmosphere presents a number of difficulties that effectively preclude the use of direct detection techniques from the ground. In contrast, although the atmosphere does present problems in varying degree for [171] indirect detection techniques, these problems can be overcome to a great extent. In discussing the various ground-based techniques, we first briefly review the current status of a technique and then the prospects for improvement.
Astrometry is the technique most widely used
in searches for dark companions to stars. The accuracy of a single
observation, using existing telescopes and detectors (i.e.,
photographic plates) is typically around 0.020.03 arcsec. By
observing a given star on many nights during a year, the accuracy of
a so-called yearly normal point can be increased to 0.005 arcsec or
better. The maximum expected value of for the
Sun-Jupiter system as seen from 10 parsecs is 0.0005 arcsec, an order
of magnitude below the accuracy limit of typical astrometric
systems.
What are the prospects for increased accuracy in astrometric systems? In the near term (the next 4 years), it appears that studies either currently underway or soon to be started will provide a thorough examination of three general ways to make high-accuracy astrometric observations. Each of these would employ photoelectric detectors in lieu of photographic plates, thereby minimizing a major current limitation to accuracy. Two of the three techniques require the construction of new telescopes; the third does not. Finally, the limitations on accuracy arising from Earth's atmosphere appear to be no worse than about 10-4 arcsec, and in some cases as low as 10-5 arcsec.
The three general techniques to be explored
are (1) classical narrowfield astrometry, (2) single-aperture
interferometry, and (3) two-aperture interferometry The first adopts
the technique currently used in astrometry but introduces modern
technology in both the detector and the telescope so as to reduce the
magnitude of the errors they contribute. The second technique uses a
single-aperture telescope (no new telescope is needed) either as a
speckle interferometer (see Worden, this volume) or as an amplitude
interferometer. Astrometric studies done with this technique differ
from those done with the first technique in that only close binary
systems are observed. One does not look for a wobble in the motion of
a star as measured against a reference frame defined by other stars;
one looks instead for changes in the separation of the binary
pair. Use of an internal reference frame like this means that one
cannot tell which of the stars has a planetary companion (if one is
detected). The third technique uses two relatively widely separated
(
10 m)
apertures. The basic observing scheme is identical to that of the
Classical method (i.e., measurements of the position of a star
relative to a reference frame defined by other stars), but one works
with fringe patterns rather than the usual Airy disk type of
image.
[172] Single-aperture interferometry involves a rather modest expenditure of money and will almost certainly be developed for use in planetary searches. This technique should routinely yield observations accurate to 10-4 arcsec and, in special cases, to 10-5 arcsec. The two-aperture interferometry technique is currently under study by Shao at MIT. The theoretical accuracy of such a system depends on the properties (e.g., aperture size and separation) of the telescope; the currently envisioned system should have errors no larger than 10-4 arcsec. However, the hardware and data acquisition aspects of two-aperture systems are complex, and it will take 2 or 3 years before we know whether an operating system can be constructed that will yield the predicted level of accuracy. By way of contrast, the classical technique involves relatively simple hardware and data acquisition; the principal question is whether the accuracy of a "modern" system is significantly better than that of current systems. Preliminary tests indicate that the gain in accuracy will be significant (i.e., more than an order of magnitude).
The long-term prospects for ground-based astrometry in a search for other planetary systems are excellent. One or more of the techniques discussed above are likely to be developed to the point where a routine observational program will be conducted. It is too early to tell which technique will prove best suited for the problem, but the near-term activity should provide an indication in that regard. It is clear that the critical assessment of the capability of ground-based astrometry which has occurred as a consequence of the interest in detecting other planetary systems will give rise to a new generation of astrometric systems-systems of vastly improved accuracy for use in a wide range of astronomical studies.
Until recently, radial-velocity studies of
stars yielded measurements accurate to about 1 km/sec. The
introduction of new techniques, based on ideas first advanced in the
early 1950s, has led to stellar radial-velocity observations accurate
to a few hundred meters per second. Much higher accuracies (on the
order of meters per second) have been obtained from radial-velocity
studies of planets in the Solar System, but to date no stellar
velocity system routinely yields observations sufficiently accurate
for planetary detection (i.e., 
10-20m/sec).
The near-term prospects for obtaining the required high-accuracy radial-velocity measurements are good. A major source of error in stellar radial-velocity studies has been the nonuniformity of the stellar image itself. Recent work has shown that the use of high-quality optical fibers eliminates this source of error. The basic idea is to image a star on an optical fiber (100-200 µm in diameter), and to use the output from the fiber as the input the radial-velocity meter. The effect of the fiber is to scramble the input [173] image so as to yield a uniform, steady output. The fiber also makes it possible to physically decouple a radial-velocity system from a telescope. The system can sit on the observatory floor and be fed a signal through a flexible fiber. This eliminates errors due to flexure arising from changes in the orientation of a telescope with respect to Earth's gravity field.
One of the principal uncertainties with this technique is that we do not yet have any data at the expected level of accuracy for the intrinsic radial-velocity variations of stars. A significant step in this area can be made by studying the radial-velocity variations of the nearest and brightest star, the Sun. Most astronomers feel that virtually all main-sequence stars will undergo intrinsic velocity variations at the 10 to 20 m/sec level, but data are needed. The major concern is that such variations may be quasi-periodic and could thereby give a spurious indication for the presence of a planetary companion.
The projected near-term activities in the area of radial-velocity systems should provide a clear indication of which system(s) are best suited for the search for other planetary systems, and they should also provide valuable data concerning intrinsic radial-velocity variation for main-sequence stars. The knowledge gained from these activities will be used to guide a long-term, radial-velocity program. There is little question that stellar radial-velocity observations can be conducted that will have errors on the order of 1 m/see, sufficiently small to permit discovery of other planetary systems.
All known methods for direct detection of planetary systems involve space-based systems. The indirect detection methods discussed in the previous section can also be employed with space-based systems. As Earth's atmosphere does not appear to be a limiting factor in the obtainable accuracy of radial-velocity systems, there is little point in conducting a radial velocity search from space. The accuracy of astrometric methods is limited ultimately by Earth's atmosphere (particularly the narrow-field astrometric method), so one might consider doing astrometry from space.
There are no current space-based systems that could be employed m a search for other planetary systems. NASA is planning to launch two systems in the 1980s that might be used for such a search. The first of these, the infrared astronomical satellite (IRAS), is a joint venture between the Netherlands and the United States. The IRAS is basically an infrared survey satellite that will scan the sky using a variety of instruments and will operate for approximately 1 year. Although the IRAS will not be used specifically to look for other planetary systems, the significantly increased sensitivity and wavelength coverage of the IRAS over that of existing infrared systems might provide unexpected results. The second NASA space-based telescope is [174] the space telescope (ST). As with the IRAS, the ST was not designed for the purpose of searching for other planetary systems. Although the imaging capability of the ST will far exceed that of ground-based telescopes, it appears unlikely that it will be able to cope with the severe imaging problems discussed above. The ST will also have subsystems with which astrometric observations can be made. Preliminary estimates indicate that the accuracy of the ST's astrometric systems will be about 10-3 arcsec, better than that of current ground-based astrometric work but one or more orders of magnitude worse than the accuracy of ground-based systems that will be operating when the ST is launched. The European Space Agency (ESA) is currently considering construction of a special-purpose astrometric space system. The primary function of the ESA satellite would be to obtain parallax (distance) measurements on a large number of stars. Estimates indicate that the accuracy of this system, for the type of observation needed in a search for other planetary systems, is comparable to that of the ST's astrometric systems (~10-3 arcsec). Thus, the prospects for meaningful contributions to a search for other planetary systems with currently planned or considered spacebased systems are not good.
This bleak outlook derives mainly from the need for special instrumentation, and there is no reason why that instrumentation could not be realized. We now know enough about the sources of error in astrometric telescopes to construct a space-based device capable of conducting an astrometric search for other planetary systems with an accuracy approaching 10-6 arcsec. The prospects for direct detection systems are less clear. Studies indicate that a system to search for other planetary systems at visual wavelengths would push technology, particularly in the area of mirror construction, to or beyond its current capability. Direct detection at infrared wavelengths looks more promising, mainly because the brightness ratio between star and planet is more favorable in the infrared than in the visual part of the spectrum. The IRAS may provide evidence for other planetary systems, but its short lifetime (about 1 year) coupled with its modest spatial resolution suggest that we can only hope for tentative results. A special-purpose infrared system, perhaps an interferometer, will be required. Although there would be major technological problems with such a space-based infrared system, they do not appear to be as severe as those facing a visual direct detection system.
Observational evidence concerning the existence of other planetary systems, their frequency of occurrence, and general properties of the planetary members of those systems is an important ingredient in considerations of life in the Universe. Planets apparently are the cosmic petri dishes of nature's [175] experiments with life, but to date there is no unequivocal evidence for the existence of any planetary system other than our own.
We have briefly reviewed the nature of the planetary detection problem, some of the techniques that might be used in a search for other planetary Systems, and the prospects for success. Almost certainly, the initial stages of any comprehensive program to search for other planetary systems will involve ground-based instrumentation. The prospects for vastly improved accuracy in both astrometric and radial-velocity observations are good; measurements accurate to 10-4 arcsec and 1 m/sec, respectively, are feasible. Observations at this level of accuracy are sufficient to conduct a meaningful search.
For a search program to be fully comprehensive, it must involve both ground-based and space-based instruments. Techniques for the indirect detection of other planetary systems can be conducted from the ground, while direct detection techniques require space-based instruments. The prospects for a space-based search are less clear than are those for a ground-based search. None of the existing or currently planned space systems has the capability to mount a meaningful search for other planetary systems. A special purpose telescope, but one with the capability to perform other useful observations, will be required.
A search for other planetary systems is, in my view, one of the most important and timely problems for astronomy in the coming decade. Knowledge about planetary systems is essential not only to the subject of this conference, but also to understanding the origin of the Solar System. Further, and no less importantly, whether we find that planetary systems are ubiquitous or rare, that finding will have a significant philosophical impact on all mankind. That we will find an answer is certain, for we are looking for physical effects, the absence of which is as telling as their presence. We are limited in our quest only by our willingness to invest time, thought, and money.
- Black, D. C.: In Search of Other Planetary Systems. Space Sci. Rev., vol. 25, Jan. 1980a, pp. 35-81.
- Black, David C., ed.: Project Orion- a Design Study of a System for Detecting Extrasolar Planets. NASA SP-436, 1980b.
- Black, David C.; and Brunk, William E., eds.: An Assessment of Ground-Based Techniques for Detecting Other Planetary Systems, Vol. I: An Overview. NASA CP-2124, 1980a.
[176] - Black, David C.; and Brunk, William E., eds.: An Assessment of Ground-Based Techniques for Detecting Other Planetary Systems, Vol. II Position Papers. NASA CP-2124,1980b.
- Gatewood, George; and Eichhorn, Heinrich: An Unsuccessful Search for a Planetary Companion of Barnard's Star (BD + 4 °3561). Astronom. J., vol. 78, no. 8, Oct.1973, pp. 769-776.
- Hershey, J.: Astrometric Analysis of the Field of AC + 65 °6955 from Plates Taken with the Sproul 24-inch Refractor. Astronom. J., vol. 78, no. 5, June 1973, pp. 421-425.
- Morrison, Philip; Billingham, John; and Wolfe, John, eds.: The Search for Extraterrestrial Intelligence-SETI. NASA SP-419,1977.
- van de Kamp, Peter: Astrometric Study of Barnard's Star from Plates Taken with the 24-inch Sproul Refractor. Astronom. J., vol. 68, no. 7, Sept. 1963, pp.515-521.
- van de Kamp, P.: Alternate Dynamical Analysis of Barnard's Star. Astron. J., vol. 74, Aug.1969, pp. 757-759.
