The Lick Observatory Planet Search 1. Department of Physics and Astronomy, San Francisco State University. Electronic mail:

Transcription

1 The Lick Observatory Planet Search 1 R. Paul Butler 2 and Georey W. Marcy 2 Department of Physics and Astronomy, San Francisco State University San Francisco, CA, U.S.A., Electronic mail: paul@further.berkeley.edu To be submitted to: \Astronomical and Biochemical Origins and the Search for Life in the Universe" Proceedings of the 5th International Conference on Bioastronomy IAU Colloquium No. 161, Capri 1{5 July 1996 eds. C.B. Cosmovici, S. Bowyer, and D. Werthimer Version date = September 18, 1996 Received ; accepted 1 Based on observations obtained at Lick Observatory, which is operated by the University of California. 2 Also at Department of Astronomy, University of California, Berkeley, CA U.S.A

2 { 2 { ABSTRACT We have constructed a precision Doppler technique with which we have detected 6 extrasolar planets to date. Doppler precision is achieved by inserting an iodine absorption cell in the telescope, providing a ducial wavelength scale against which to measure Doppler shifts. Our current precision is 3 m s?1, which corresponds to one part in one hundred million in wavelength, or 1/1000th of a pixel on the CCD detector. As of 1996 August, our survey of 120 stars has revealed six stars that show velocity variations consistent with Jupiter{mass companions in Keplerian orbits. These objects span a greater range of orbital radii and eccentricity than the planets in our own solar system. Three of these objects are \51 Peg{type" planets with circular orbits having radii 0.15 AU or less. The other three objects have masses between 1.7 and 6.5 M JUP, eccentricities between 0.01 and 0.6, and semi{major axes between 0.4 and 2.1 AU. We have not found any objects having masses between 10 and 80 M JUP, the domain usually associated with brown dwarfs. These six new companions represent a new population of objects having extremely low masses, similar to that of Jupiter. We have begun a survey of an additional 400 stars using the Keck 10{m telescope, which will allow the detection of Neptune{mass planets. Subject headings: Stars: Planetary Systems Stars: Brown Dwarfs

3 { 3 { 1. Introduction During the last 15 years, a number of groups have initiated long term Doppler searches for planets orbiting solar{type stars (Campbell et al. 1988; Walker et al. 1995; Mayor & Queloz 1995; McMillan et al.1994; Cochran & Hatzes 1994; Brown et al. 1995). These groups have demonstrated precision of 15 m s?1, and have surveyed a total of 300 stars for 2 to 12 years. The reex motion of the Sun due to Jupiter is 12.5 m s?1, thus yielding a detection threshold of 2 M JUP at 5 AU for these studies. The spectacular detection of the rst planetary companion to a Solar-type star (Mayor & Queloz 1995) represents a historic plateau in technological achievement and scientic understanding. Within 9 months of Mayor & Queloz's discovery, an additional 6 planet{like companions have been found orbiting solar-type stars: 55 Cancri, Boo, And (Butler et al. 1996b), 47 UMa (Butler & Marcy 1996), 70 Vir (Marcy & Butler 1996), and 16 Cyg B (Cochran et al. 1996). These seven companions were all detected by precise Doppler monitoring of the host star. These companions have M sin i between 0.5 and 7 M JUP, suggesting that the actual masses reside within a range associated with extrasolar giant planets, whatever their origin. (Here, i is the unknown orbital inclination.) These rst \planets" around solar-like stars (shown schematically in Figure 1) exhibit a variety of characteristics. The \51 Peg{type" planets (51 Peg, 55 Cancri, Boo, and And) are distinguished by an orbital radii less than 0.15 AU, closer than expected from theory (Boss 1995; Lissauer 1995). 70 Vir has a relatively large mass (M sin i = 6.5) and a high eccentricity (e = 0.4), suggesting a less-dissipative formation history. 16 Cyg B has a smaller mass than 70 Vir (M sin i = 1.7 M JUP ), but an even larger eccentricity. The companion to 47 UMa resides in a circular orbit of radius 2.1 AU and has a mass of 2.4 M JUP /sin i, and thus most closely resembles Jupiter in our Solar System.

4 { 4 { This array of characteristics of planetary companions challenges current theories of the formation of planetary systems (Lissauer 1995; Wetherill 1996; Boss 1995). Some new disk dynamics have been proposed to explain the small orbital distances and the large eccentricities (Rasio & Ford 1996; Lin et al. 1996; Artymowicz et al. 1991). The mass densities, the chemical composition, and the environments of proto{planetary disks may span a wider range than had been included in Solar{System disk models. 2. The Doppler Technique Errors in Doppler velocity stem from spatial and temporal dierences in the way that stellar and reference spectra are obtained. Reference lamps generally do not travel through the telescope optics, and they are seldom taken simultaneously with stellar observations. These reference lamps therefore do not precisely convey the wavelength zero point, dispersion, or the point{spread{function (PSF) of a spectrometer at the time of a stellar observation (Valenti et al. 1995; Butler et al. 1996a). Due to these long standing problems, standard astronomical Doppler velocity measurements typically have errors of 500 m s?1 or larger. In contrast, Jupiter induces a velocity of 12.5 m s?1 in the Sun. It is possible to achieve much higher Doppler precision by simultaneously obtaining the stellar and reference spectra (Grin & Grin 1973a). A number of possibilities have been investigated to provide a simultaneous reference spectrum, among them telluric lines, chemical absorption cells, and Fabry{Perot interferometers (Grin & Grin 1973b; Campbell & Walker 1988; McMillan et al. 1994; Cochran & Hatzes 1994; Brown et al. 1995). After an extensive literature and laboratory search, we determined that an iodine absorption cell oered the best set of compromises for precision Doppler work. The chief advantages of iodine are: 1) large absorption coecient, 2) broad wavelength coverage,

5 { 5 { and 3) long term stability. A large absorption coecient allows the construction of a short absorption cell, which facilitates mounting and temperature stabilization. Finally, iodine is relatively non{toxic. The construction of the Iodine cell is described elsewhere in detail (Marcy & Butler 1992; Butler 1987). The cell is mounted directly in front of the spectrograph slit in the converging f/36 telescope beam. Starlight passes through the cell just prior to entering the spectrograph slit. The iodine cell thus acts as a transmission lter, imposing thousands of extremely sharp lines on the starlight. A photon-limited Doppler analysis must consist of a full model of the spectroscopic observation. Instrumental eects that must be considered include shifts in the spectrograph wavelength zero point, changes in the spectrograph dispersion and distortions of the instrumental prole or 1-D PSF of the spectrometer. The observation is modeled as the product of two functions, the intrinsic stellar spectrum, I s, and the transmission function of the iodine absorption cell, T I2. This product is then convolved with the spectrograph PSF and binned to the wavelength extent of the CCD pixels. This construction of the synthetic spectrum is carried out as described by Valenti et al. (1995) and Butler et al. (1996a). An example of the modeling process is shown in Figure 2. The top panel shows the transmission function of the iodine absorption cell (T I2 ) as determined by the Fourier Transform Spectrometer at the McMath Solar Telescope. The second panel shows the intrinsic spectrum (I s ) of the bright star Ceti (HR 509). The dotted points in the third panel show an actual observation of Ceti taken through the iodine absorption cell with the "Hamilton Spectrograph". The solid line in the third panel is the model of the observation constructed from the two input functions, T I2 and I s, convolved by the derived spectrograph PSF. The bottom plot shows ten times the dierences between the model and the observation. The RMS dierence between model and the observation is 0.4%, consistent

6 { 6 { with the photon-shot noise in the observation of 0.3%. We routinely achieve a precision of 3 m s?1, which corresponds to wavelength shifts of one part in one hundred million. The pixels in the CCD detector are 15 microns wide. A 3 m s?1 shift is pixels, or about 100 silicon atoms on the CCD. 3. Doppler Velocity Results During the past nine years we have monitored 120 F, G, K, & M dwarf stars with our precise Doppler technique. The 0.6 m CAT and 3 m Shane telescopes at Lick Observatory are both used to feed the \Hamilton", a coude echelle spectrometer (Vogt 1987). The measurement errors had been 10 { 15 m s?1 until 1994 Nov when the Hamilton optics were refurbished and the wavelength coverage was doubled. As a result, measurement errors have decreased to 3 m s?1. As of 1996 August, six stars have shown velocity variations consistent with planetary companions in Keplerian orbits. Three of these objects have orbital periods less of two weeks or less, similar to the 51 Peg system, and three have orbital periods of 4 months to 3 years. The information from Doppler velocities is sucient to determine all the orbital parameters of the system, except the inclination angle of the orbital plane. Without knowing the inclination angle, it is not possible to precisely determine the true mass of the companion. Instead, a minimum mass, equal to the true mass times the sine of the inclination angle, is calculated Peg{type Planets Although 51 Peg{type planets were theoretically unexpected, they are observationally the easiest types of planets to detect with the Doppler velocity technique. These planets

7 { 7 { are about 100 times closer to their host stars than Jupiter is to the Sun. This increases the gravitational tug on the host star, and hence the magnitude of the Doppler velocity signal. The short orbital period allows many orbits to be observed over a few months, rather than the 12 years required to follow Jupiter through a single orbit. Figure 3 shows 3 months of data (1995 October through 1996 January) on the archetype 51 Peg system (Marcy et al. 1996). The Doppler velocity variations are t by a sine wave. The period is 4.23 days, the semi{amplitude (K) is 57 m s?1, and the RMS t to the sine is 5.3 m s?1, consistent with expected errors. Residuals to the t are shown at the bottom. The minimum (m sin i ) mass of the companion is 0.45 M JUP. Most of these spectra were taken with the 0.6 m Coude Auxiliary Telescope (CAT) and have S/N 70. Observations made on the 3 m telescope have S/N 200 and yield an RMS of 3 m s?1 to a sine t. Non{orbital explanations for this variation (and for the other \51 Peg{type" systems), such as pulsation and rotating spots, have been ruled out by precise photometry (Guinan 1995; Henry et al. 1996; Perryman et al. 1996; Baliunas et al. 1997). The orbital characteristics of the four \51 Peg{type" systems are shown in Table 1. Surface gravity measurements for all four stars indicate that they reside on or just above the main sequence. They span the spectral range from F7 through G8. All four stars appear to be metal{rich, which may provide a clue to their formation. The possibility exists that we may be viewing these systems at an extreme orbital inclination, thus increasing the true mass of the companions by an order of magnitude or more over the minimum \m sin i " mass. Assuming that the orbital inclinations of planetary systems are randomly distributed, there is a 0.1% (1 chance in 1,000) that the true mass of a companion will be larger by a factor of 10 or more than the minimum m sin i value. For every 10 to 40 M JUP \brown dwarf" masquerading as a Jupiter mass planet, there should be hundreds of objects detected with m sin i 10 M JUP. Of the 300

8 { 8 { stars currently being surveyed at high precision, 4 \51 Peg{type" objects have been found, none with m sin i larger than 4 M JUP. This argues that there is no large population of closely orbiting brown dwarfs fron which to draw systems with extreme orbital inclinations. The Doppler velocity semi{amplitude of a brown dwarf in a 4 day orbit would typically be greater than 1 km s?1, which would have been detectable for at least the last 80 years. No such brown dwarf companion has ever been found. Theoretical models of gas giant planets at small orbital distances have been computed by Burrows et al. (1995), Saumon et al. (1996) and Guillot et al. (1996). Assuming a Jupiter{like composition, the radii of all four 51 Peg{type companions is 1.2 R JUP, enlarged relative to Jupiter due to the absorbed stellar radiation, which also controls the eective temperatures of the planets. The derived temperature for the companion to 55 Cancri is 700K, Boo is 1400K, And is 1300K, and 51 Peg is 1300K (Burrows et al. 1995; Guillot et al. 1996) Outer Planets In addition to the four 51 Peg{type planets, three planets with more distant orbits have been discovered. The orbital elements for these planets are given in Table 2. Only one of these planets, 47 Ursae Majoris, has both a circular orbit and an orbital distance similar to giant planets in the solar system. As Figure 4 shows, the orbital period is 3 years, and the semi{amplitude of the velocity variations are 47 m s?1. Residuals to the Keplerian t are shown at the bottom of the gure. The companion has a minimum mass (m sin i ) of 2.4 M JUP, and orbits at a distance of 2.1 AU. The surface temperature of such an object would be at least 180K due simply to absorbed stellar radiation, and probably slightly higher due to intrinsic heating from gravitational contraction (Guillot et al. 1996). The companion is separated from the primary star by 0.2 arcsec, which portends astrometric

9 { 9 { and direct IR follow-up work. The remaining two planets have surprisingly large orbital eccentricities. The Doppler velocities for the rst of these, 70 Vir, is shown in Figure 5. These velocities have been folded (phased) into the orbital period of days. The velocity curve is clearly not sinusoidal. Residuals to the Keplerian t are shown at the bottom of the gure. The orbital eccentricity (e = 0.4) and the large mass (m sin i = 6.5 M JUP ), have led to some speculation that this companion would best be described as a \brown dwarf". As \failed stars", brown dwarfs are usually though to have masses in the range of 20 to 80 M JUP, just below the hydrogen burning limit. No objects in this mass range have been detected in any of the precision velocity surveys. If 70 Vir is in the low mass tail of the brown dwarf mass distribution, then we have entirely missed the main body of brown dwarfs. The eective surface temperature of the companion to 70 Vir is 82 C, permitting water in liquid form. The planetary companion to 16 Cyg B was independently discovered by William Cochran and Artie Hatzes of the University of Texas and by the Lick Observatory Planet Search. The orbital period is 2.2 years, and the semi{major axis is 1.7 AU. The orbit of this companion is even more eccentric than 70 Vir (e = 0.57), but the minimum mass of this companion is only 1.65 M JUP. With a probable mass between 1.65 and 3.3 M JUP, it would be dicult to justify the label \brown dwarf" in describing this object. This suggests that objects of planetary mass do exist in highly eccentric orbits. The eective surface temperature of this object is about 190 C, similar to the companion around 47 Ursae Majoris. 4. Discussion A total of 300 stars have been surveyed from 2 to 12 years by Doppler velocity groups that have demonstrated a precision of 15 m s?1 or better. Figure 6 shows a mass

10 { 10 { distribution histogram of the seven objects that these groups have detected. The one other known low mass companion, HD , has not been included as it was detected in a low precision survey that could not have detected 1 M JUP companions. Perhaps the most surprising result of the high Doppler precision surveys (15 m s?1 ) is that no companion objects to Sun{like stars have been found with masses in the brown dwarf range of 10 to 80 M JUP. Brown dwarfs are observationally much easier to detect than the seven lower mass objects that have been found. These seven objects are therefore members of a separate class of objects, distinct from low mass stars and brown dwarfs. In 1980 David Black wrote, \It is perhaps surprising that there is no generally accepted modern denition of `planet.' " This is perhaps even more true today than it was in Various researchers have argued that such a denition should be based on orbital eccentricity and formation history, as well as mass. In providing an operational denition, David Black wrote, \The view taken here is that the term `planet' refers to any object whose mass is comparable to or less than the mass of Jupiter. Adoption of this operational denition does not imply that bodies which are more massive than Jupiter are not planets." Prior to the discovery of 51 Peg, most theories of planet formation suggested that extrasolar planetary systems would resemble our own solar system, with giant planets orbiting several AU from a central star, and terrestrial planets orbiting closer in. With the exception of the companion to 47 Ursae Majoris, none of the discoveries to date are similar to objects found in the solar system. A number of new theories have been generated which augment existing theories to explain the diversity of the observed systems. Current theory demands that Jupiter{like planets form beyond the ice boundary in the circumstellar disks (about 3 AU). It has been proposed that these Jupiter{like planets can then be viscously dragged inward by gravitational interactions with the disk to form 51 Peg{type systems (Lin et al. 1996; Boss 1996). Another scenario suggests that gravitational interactions

11 { 11 { between two massive planets can throw one planet into a 51 Peg-type orbit, while leaving the other planet in a distant eccentric orbit (Rasio & Ford 1996). This theory can plausibly explain both the 51 Peg{type planets and the eccentric planets. The Doppler velocity data for two 51 Peg{type planets show residuals that are consistent with longer period planets. It may be possible that in the case of 70 Vir and 16 Cyg B, the expected 51 Peg{type planet was thrown in too close to the star and did not survive. Out of 120 stars, we have found 6 companions with planetary mass. The nine year length of our survey limits us to objects that lie in the within 4 AU of their host stars. We therefore estimate that 5% of Sun{like stars either have 51 Peg{type planets with m sin i > 0.4 M JUP, or more distant planets (within 4 AU) with m sin i > 1.5 M JUP. As Figure 6 shows, there is already a hint that as the mass of the companions decrease, the number of such objects increase, consistent with theory (Lissauer 1995). In 1994 November our precision improved from 10 m s?1 to 3 m s?1. With the new higher precision, we expect to nd companions with masses in the range of Neptune to Saturn within the next few years. In 1996 July we began a parallel project using an iodine absorption cell on the worlds largest telescope, the Keck 10 meter, to survey an additional 400 stars. By 1998 we expect other 10 m class telescopes, including the Hobby{Eberly Telescope in Texas and the VLT in Chile, to begin surveying many hundred additional stars at a precision of 3 m s?1 or better. By the end of the decade, optical interferometers capable of astrometrically detecting extrasolar planets should also be in operation (Colavita & Shao 1995). Much of the excitement about the recent discoveries is centered on the possibility of life. As David Black wrote in 1995, \Planets can be thought of as cosmic petri dishes." A Jupiter mass object in a roughly 1 AU orbit (such as 70 Vir and 16 Cyg B) would make the existence of a terrestrial planet in the habitable zone dicult because of dynamic instabilities. Gehman et al. (1996) calculated that terrestrial planets in the habitable zone

12 { 12 { are dynamically stable for the case of 51 Peg{type planets. Since precision Doppler searches are most sensitive to massive planets in small orbits, detections of giant eccentric planets found to date may rule out the possibility of habitable terrestrial planets in these systems. The stars that have not yet revealed planets may paradoxically be better suited to harbor life. We acknowledge generous support from NASA (NAGW 3182), NSF (AST ), and SUN Microsystems.

13 { 13 { Table 1. Orbital Parameters of 51 Peg{type Systems Param 55 Cancri Boo And 51 Peg P (d) T p (JD) e (0.014) (0.013) 0.15 (0.04) (0.009)! (deg) K 1 (m s?1 ) a 1 sin i (AU) ?6 f 1 (m) (M ) ? ? ? ?11 N (O-C) (m s?1 )

14 { 14 { Table 2. Orbital Parameters of Distant Planets Param 47 UMa 70 Vir 16 Cyg B P (d) T p (JD) e 0.01 (0.1) (0.007) 0.57 (0.1)! (deg) K 1 (m s?1 ) a 1 sin i (AU) f 1 (m) (M ) ? ? ?9 N (O-C) (m s?1 )

15 { 15 { REFERENCES Artymowicz, P., Clarke, C.J.,Lubow, S.H., & Pringle, J.E. 1991, ApJL, 370, 35. Baliunas, S., Henry, G.W., Donahue, R.A., Fekel, F.C., & Soon, W.H. 1997, \Properties of Sun-Like Stars with Planets II:... ", submitted to ApJL Black, D.C. 1980, \Project Orion: A Design Study of a System for Detecting Extrasolar Planets", NASA Publication SP{436. Black, D.C. 1995, in Annual Review of Astronomy and Astrophysics: \Completing the Copernican Revolution: The Search for Other Planetary Systems", (Annual Reviews Inc., G. Burbidge and A. Sandage eds), vol. 33, p Boss, A.P. 1995, Science, 267, 360. Boss, A.P. 1996, Lunar and Planetary Science, 27, 139. Brown, T.M., Noyes, R.W., Nisenson, P., Korzennik, S.G., & Horner, S. 1995, PASP, 106,1285. Burrows, A., Saumon,D., Guillot,T., Hubbard,W.B., & Lunine,J.I. 1995, Nature, 375, 299. Butler, R.P. 1987, Masters Thesis, San Francisco State University. Butler, R.P. & Marcy, G.W. 1996, ApJL, 464, L153. Butler, R.P., Marcy, G.W., Williams, E., McCarthy, C. & Vogt, S.S. 1996a, PASP, 108,500. Butler, R.P., Marcy, G.W., Williams, E., Hauser, H., & Shirts, P. 1996b, \Three New 51 Peg{type Planets", submitted to ApJL. Campbell, B., Walker, G. A. H. & Yang, S. 1988, ApJ, 331, 902. Cochran, W.D. & Hatzes, A.P. 1994, in \Planetary systems: Formation, Evolution, and Detection", (Kluwer Academic, B.F.Burke, J.H.Rahe and E.E.Roettger, eds.), p281. Cochran, W.D., Hatzes, A.P., Butler, R.P., & Marcy, G.W. 1996, to be submitted to Science

16 { 16 { Colavita, M. & Shao, M. 1995, in \Planetary systems: Formation, Evolution, and Detection", (Kluwer Academic, B.F.Burke, J.H.Rahe and E.E.Roettger, eds.), p385. Gehman, C.S., Adams, F.C., & Laughlin, G. 1996, \The Prospects for Earth{Like Planets Within Known ExtraSolar Planetary Systems", submitted to ApJ?? Grin, R. & Grin, R. 1973a, MNRAS, 162, 243. Grin, R. & Grin, R. 1973b, MNRAS, 162, 255. Guillot, T., Burrows, A., Hubbard, W.B., Lunine, J.I., Saumon, D. 1996, ApJL, 459, L35. Guinan, E. 1995, IAU Circ Henry, G.W., Baliunas, S.L., Donahue, R.A., Soon, W.H., & Saar, S.H. 1996, submitted to ApJL. Lin, D.N.C., Bodenheimer, P., & Richardson, D.C. 1996, Nature, 380, 606. Lissauer, J.J. 1995, Icarus, 114, 217 Marcy, G.W. & Butler R.P. 1996, ApJL, 464,L151 Marcy, G.W., Butler, R.P., Williams, E., Bildsten, L., Graham, J. 1996, \The Planet Around 51 Peg", submitted to ApJ. Marcy, G.W. & Butler R.P. 1992, PASP, 104, 270. Mayor, M. & Queloz, D. 1995, Nature, 378, 355. McMillan,R.S., Moore,T.L., Perry,M.L., Smith,P.H. 1994, in \Planetary Systems: Formation, Evolution, and Detection", (Kluwer Academic, B.F.Burke, J.H.Rahe and E.E.Roettger, eds.), p271. Perryman, M.A.C. et al., A&A 310,L21. Rasio, F.A. and Ford, E.B. 1996, \A Dynamical Formation Process for Planets in Short{Period Orbits", submitted to Science.

17 { 17 { Saumon,D., Hubbard,W.B., Burrows,A., Guillot,T., Lunine,J.I., and Chabrier,G. 1996, preprint, accepted in ApJ Valenti, J.A., Butler, R.P., & Marcy, G.W. 1995, PASP, 107, 966. Vogt, S.S. (1987), PASP, 99, 1214 Walker, G.A.H., Walker, A.R., Irwin, A.W., Larson, A.M., Yang, S.L.S., Richardson, D.C. 1995, Icarus, 116, 359 Wetherill, G.W., Icarus, 119, 219. This manuscript was prepared with the AAS L A TEX macros v3.0.

18 { 18 { FIGURE CAPTIONS Fig. 1. Schematic diagram of the recently discovered extrasolar planets.

19 { 19 { Fig. 2. The Modeling Process. Top) The template Iodine cell spectrum. Second) The template stellar spectrum ( Ceti, G8V) Third) The binned points are an observation of Ceti made through the Iodine absorption. The solid line is a model of the observation. The model is composed of the template iodine and stellar spectra. The free parameters consist of the spectrograph PSF and the Doppler shift of the template star relative to the template iodine. Bottom) 10 times the dierence between the model and the observation. The model and observation dier by 0.4% rms. Fig. 3. Doppler Velocities for 51 Pegasi from 1995 Oct through 1996 Jan. A Keplerian t yields an orbital period of 4.23 d and a semi{amplitude of 57 m s?1, indicating a companion having m sin i = 0.45 M JUP at 0.05 AU. The residuals to the orbital t are shown at the bottom. The error bars of the individual observations and the standard deviation of the residuals are both 5 m s?1. Fig. 4. Doppler velocities for 47 Ursae Majoris from 1987 through The Keplerian orbit is circular, the period is 3 years and the semi{amplitude is 48 m s?1. The residuals to the orbital t are shown at the bottom. The companion has m sin i 2.4 M JUP, and a semi{major axis of 2.1 AU. Fig. 5. Phased Doppler Velocities for 70 Vir. Eight years of data has been folded into a single orbital cycle. The Keplerian orbital t has a period of d and a semi{amplitude of 315 m s?1. The residuals to the orbital t are shown at the bottom. The minimum mass of the companion is 6.5 M JUP, and the semi{major axis is 0.43 AU.

20 { 20 { Fig. 6. Mass distribution histogram of the seven substellar objects found via precision Doppler surveys. No objects have been found with m sin i greater than 7 M JUP. The startling mass gap between 7 and 80 M JUP demonstrates that the population of objects that have been detected are distinctly dierent from low mass stars and brown dwarfs. There appears to be an increase in the number of objects at lower masses. Selection eects favor the detection of high mass objects.