The Local Group and Galactic Evolution I The Local Group I Satellite Galaxies I Cepheid Variables I Tides and the Roche Limit I Local Spirals I Chemical Evolution I Dwarf Galaxies I Future of the Local Group J.M. Lattimer AST 346, Galaxies, Part 5 The Local Group J.M. Lattimer AST 346, Galaxies, Part 5 The Local Group J.M. Lattimer AST 346, Galaxies, Part 5 The Local Group J.M. Lattimer AST 346, Galaxies, Part 5 The Large Magellanic Cloud S Dor HI gas ⇐= Large Magellanic Cloud Hα 10◦ or 8.5 kpc =⇒ ⇐= 7◦ or 6.0 kpc IR - 24µm optical J.M. Lattimer AST 346, Galaxies, Part 5 =⇒ Bulge Large Magellanic Cloud J.M. Lattimer AST 346, Galaxies, Part 5 Magellanic Clouds J.M. Lattimer AST 346, Galaxies, Part 5 Cepheid Variables MV = −3.525 log(P/d) + 2.88(V − Ic )0 −2.80 − 1.05AV J.M. Lattimer AST 346, Galaxies, Part 5 Pulsational Frequency The Euler equation of motion for a surface layer of mass m and radius R m G Mm 1 ∂p d 2R =− − . dt 2 R2 ρ ∂r Use V = 1/ρ and −Vdp = pdV − pV = pdV = 4πR 2 dr (p = 0 at the top of the layer (surface) and V = 0 at the bottom). In equilibrium, one has G Mm/R02 = 4πR02 p0 . With perturbation R → R0 + δR; p → p0 + δp m d 2 (R + δR) G Mm =− + 4π(R + δR)2 (p + δp) dt 2 (R + δR)2 d 2 δR 2G Mm 2 = m + 8πR p 0 0 δR + 4πR0 δp dt 2 R03 d 2 δR GM = (4 − 3γ)δR dt 2 R03 We used pV γ = constant or pR 3γ = constant or δp/p0 = −3γδR/R0 . This equation has a harmonic solution δR = A sin ωt if γ > 4/3. s s GM GM √ ω = (3γ − 4) 3 ' ∼ ρ̄. R0 R03 J.M. Lattimer AST 346, Galaxies, Part 5 Pulsational Instability An instability will occur if opacity increases with compression (temperature), which is not normally the case. Normal stellar opacity is of the Kramer’s type κ ∝ ρT −3.5 and ρ ∝ T 3 in adiabatic matter. Thus κ ∝ ρ−1/6 and decreases upon compression. In a partial ionization zone, H↔ H+ + e − , He ↔ He+ + e − , He+ ↔ He++ + e − , the temperature increases less because of the ionization energy sink. This is the key to the kappa mechanism. J.M. Lattimer V I A → B: opacity increases during compression I B → C : heat absorbed ∆SBC = ∆QBC /Thot I C → D: opacity decreases during expansion I D → A: heat released ∆SDA = ∆QDA /Tcold ∆QBC + ∆QDA = ∆SBC (Thot − Tcold )/Thot > 0 AST 346, Galaxies, Part 5 Period-Luminosity Relation I log L = 3.5 log M + a I log L = 2 log R + 4 log Teff I log P = − 12 log M + I 1 log Teff ∼ − 20 log L 3 1 3 log P = ( 4 − 7 + 20 ) log L − 3 log Teff log P = 0.76 log L + b + 7a 2.5 MV = − 0.76 log P + c = −3.3 log P + I I I 3 2 log R + b J.M. Lattimer +b+ a 7 c AST 346, Galaxies, Part 5 Dwarf Spheroidal Galaxies I At least as luminous as globular clusters I Surface brightness 1% of LMC I > 10 in Local Group I No young stars, no gas I Stars of a wide range of ages 3 − 10 Gyr I Low metallicities Z = 0.02Z 7 3 Gyr J.M. Lattimer AST 346, Galaxies, Part 5 15 Galaxy Encounters I Weak or distant encounters I I I I Flyby with associated tides Satellite orbit decay due to dynamical friction Tidal evaporation of orbiting satellite Tidal or gravitational shocks I Strong or close encounters I I I Analytic cases (mergers can only be treated numerically) I I I I J.M. Lattimer Mergers Global gravitational effects become important Dynamical friction (small system moving through a larger one) Tidal evaporation (Jacobi/Roche limit) Slow encounters (adiabatic approximation) Fast (shocking) encounters (impulse approximation) AST 346, Galaxies, Part 5 Distant Weak Encounters and Drag Use the impulse approximation, ignoring the deviation in the stellar paths. The impact parameter is b. The perpindicular pull of star m on star M is GmM/r 2 times b/r , with r 2 = b 2 + V 2 t 2 : dV⊥ GmMb =M F⊥ (t) = 2 dt (b + V 2 t 2 )3/2 Deflection angle: Z +∞ 1 2Gm ∆V⊥ = F⊥ dt = θ= V MV −∞ bV 2 The encounter has symmetry about the vector of closest approach, the line θ/2 backwards from the original perpendicular impact parameter vector. Newton’s 3rd law says m∆v = M∆V . ∆t∆Fdrag = −m∆v|| = − Fdrag 2G 2 M2 m =− . b2 V 3 Z Z = nV ∆Fdrag dt2πbdb =− 4πG 2 M2 nm ln Λ V2 bmax Λ= bmin Drag is caused by the component of the force parallel and backwards to M’s motion. Then J.M. Lattimer 2G Mm θ bV 2 AST 346, Galaxies, Part 5 Distant Weak Encounters and Drag The basic idea is that a moving mass attracts objects to it, but because of its motion, the objects tend to gather behind the mass, pulling it backwards. I Open clusters: ln Λ ≈ 6 I Globular clusters: ln Λ ≈ 11 I Large elliptical galaxy: ln Λ ≈ 22 I Galaxy clusters: ln Λ ≈ 7 Allowing for a range of velocities (Chandrasekhar friction formula): Z V 4πG 2 M2 m ln Λ f (v )d 3 v Fdrag = − V2 0 I V << v , use f (v ) ≈ f (0), resembles Stokes law for motion through a viscous fluid. 16 Fdrag = − π 2 G 2 M2 mf (0)V ln Λ 3 J.M. Lattimer I V >> v , all stars contribute, drag decreases with V : Fdrag = −4πG 2 M2 nmV −2 ln Λ Maxwellian√with dispersion σ, X = V /(σ 2): Fdrag = −4πG 2 M2 nmV −2 ln Λ i h 2 √ × erf (X ) − 2Xe −X / π h i 2 ' −4πG 2 M2 nmV −2 ln Λ 1 − e −0.6X I AST 346, Galaxies, Part 5 Applications of Dynamical Friction I Satellite in circular orbit Take an isothermal galaxy: √ ρ(r ) = Vc2 /(4πGr 2 ), σ = Vc / 2 or X = 1, L = MVc r Fdrag = −0.43G (M/r )2 ln λ dL/dT = Fdrag r = −0.43G (M2 /r ) ln Λ = MVc dr /dt I Massive galaxy encounter M = 1010 M , ri = 20 kpc, V = Vc : tinfall ' 2 × 108 yr ∼ orbit period. I LMC M ≈ 2 × 1010 M , ri = 60 kpc: tinfall ≈ 3 × 109 yr. tinfall = Vc ri2 /(0.86G M ln Λ) Example: globular cluster orbiting Milky Way M = 106 M , Vc = 250 km/s, bmax = ri = 2 kpc: tinfall ' 2.6 × 1010 yr J.M. Lattimer Why is LMC still there? Not on a circular orbit; LMC has been bound to SMC in elliptic (e = 0.2) orbit. Their orbit has decayed by factor 2 but will tidally separate when approaching Milky Way within 30 kpc. They will settle in Galactic center in 1010 yr. AST 346, Galaxies, Part 5 Tidal Limit Satellite and galaxy are fixed in a rotating frame. A star’s energy E = V 2 /2 + Φ is not conserved; but EJ = V 2 /2 + Φeff is. Φeff (r) = Φ(r) − |Ω × r|2 /2 Along a line connecting m and M with origin at m: GM Φeff (x) = − |D−x| − Gm |x| = Ω2 2 (x − D)2 Turning points: xJ ' ±D(m/3M)1/3 is the Jacobi (or tidal, Roche, Hill) radius. For an isothermal potential, xJ ' ±D(m/2M)1/3 . J.M. Lattimer AST 346, Galaxies, Part 5 Satellite Evaporation and Destruction Satellite star with EJ moving away from satellite has decreasing V , and turns around when EJ = Φeff . If EJ > Φeff (rJ ) the star is lost (evaporated). This differs from slow evaporation caused by scattering of stars to V > Ve . Even bound stars (E < 0) can have EJ > Φeff (rJ ). A satellite approaching a galaxy has decreasing rJ and Φeff (rJ ) so it loses more and more stars. Since most stars are only marginally bound (N(E ) peaks near E = 0), a small decrease in Φeff (rJ ) can result in the loss of many stars. J.M. Lattimer AST 346, Galaxies, Part 5 Slow and Fast Encounters Although orbits of many stars significantly affected by a tidal encounter, the tightly bound ones (Porb << tencounter ) are not. The tidal field slowly changes and these stars respond adiabatically. The opposite extreme is a tidal shock: Porb >> tencounter or Vinternal << ∆Vencounter . These stars don’t move during the encounter, so no change in PE. They feel an impulse and ∆ KE. These stars are thus heated. System must expand and cool in response. A two step process (recall Eo,f = - KEo,f ): J.M. Lattimer I shock: KEi = KEo + ∆KE, Ei =Eo + ∆ KE = - KEo + ∆ KE relaxation: Ef = Ei , KEf = KEo − ∆KE Although shock heated system by ∆KE, during relaxation it cooled by −2∆KE and expanded. I Some stars received enough energy to unbind and evaporate. Repeated shocks can disintegrate a cluster. If the encounter is distant, the system is left elongated, long axis pointing to the point of closest approach. AST 346, Galaxies, Part 5 Examples I Open clusters are shocked by passage of dense molecular clouds; most evaporated after 5 × 108 yr. I Globular clusters are shocked by passing through disk. If σ = 5 km/s, r = 10 pc, V⊥ = 170 km/s, R = 3.5 kpc: tdisrupt ≈ 6 × 109 yr. I Galaxies in clusters (galaxy harassment): disks are heated and thicken; spiral arm formation is suppressed; galaxies appear to evolve to earlier Hubble type. Stars and dark matter expand and are lost but join the cluster. Gas loses angular momentum and falls to center, triggering star formation (starburst). J.M. Lattimer I Ring galaxies are formed from tidal shocks. A perturber passes rapidly through center of disk, inducing an inward ∆Vr . Sets up a synchronized epicyclic motion, results in an expanding circular density wave or ring. These trigger star formation. AST 346, Galaxies, Part 5 Andromeda Galaxy I Large than Milky Way: 50% more luminous, hR is double, V (R) ∼ 260 km/s (about 20-30% larger). Twice the number of globular clusters (300). I Bulge is larger than Milky Way’s, with about 30-40% of the luminosity. Stars are older than a few Gyr and relatively rich in heavy elements. I I Has a compact, semi-stellar nucleus, with two concentrations of light 2pc (0.5 arcsec) apart. One is a star cluster, one is a black hole of 4 × 106 M . Nucleus is free of gas and dust. Metal-poor globular clusters have deeply plunging trajectories. J.M. Lattimer I Most stars within a few kpc of disk are not metal-poor, as if the bulge was overflowing. These are about 6 Gyrs old and may be the remnant of a merger with a metal-rich (and therefore massive) galaxy. I A “ring of fire” circles at 10 kpc and contains most of the young stars. Marked by HII regions and CO emission. I Less pronounced, and tightly wound, spiral features makes this an Sb galaxy, compared to Milky Way (SBc). I 4 − 6 × 106 M of HI, 50% more than Milky Way. AST 346, Galaxies, Part 5 M31 J.M. Lattimer AST 346, Galaxies, Part 5 Formation of Local Group Denser clumps collapse, and are already clustered. Tidal torques induce slow rotation J.M. Lattimer AST 346, Galaxies, Part 5 Galactic Chemical Evolution Evolution of total mass M(t) = Mg (t) + M∗ (t) in terms of inflows f (t) and outflows o(t): In the Closed Box Model, f (t) = o(t). The evolution of the gas mass Mg (t) The Initial Mass Function (IMF) is ξ. Integrate over stars dying at time t, i.e., from Mt to the upper limit of the IMF MU . The evolution of species i with mass Mg Zi in the gas is dMg /dt = −Ψ + E + f − o dMg Zi /dt = −ΨZi + Ei + fZi,f − oZi,o dM/dt = f − o in terms of the Star Formation Rate (SFR, Ψ(t)) and the rate of mass ejection by dying stars E (t). A star of mass M is created at time t − τM and dies at time t > τM , leaving a remnant of mass CM . Z MU E (t) = The rate of stellar ejection of species i: Z MU Ei (t) = Yi (M)Ψ(t−τM )ξ(M)dM Mt where Yi (M) is the stellar yield and could depend on t. (M−CM )Ψ(t−τM )ξ(M)dM Mt J.M. Lattimer AST 346, Galaxies, Part 5 Inputs Stellar lifetimes strongly decrease with M. Stellar remnants include white dwarfs, neutron stars and black holes. Their masses depend on the extent of mass loss by winds. Ejected mass fractions increase with M. Stellar yields are better expressed in terms of net yields yi (M) = Yi (M) − Zi (M − CM ), the newly created mass of i minus what was originally there. J.M. Lattimer AST 346, Galaxies, Part 5 Initial Mass Function Two simplified cases: Luminosity function f (L) = dN/dL and M − L relation for MS stars φ(L) = dM/dL. The present-day mass function (PDMF, F (M)) is F (M) = dN/dM = f φ. The IMF can be derived if the SFR is known as it is an integral over time of the star creation rate ΦΨ. If the IMF doesn’t depend on t: Z T ξ(M) = F (M)/ Ψ(t)dt. T −τM Return Mass Fraction R is mass fraction of a stellar generation returned Z to ISM I Short-lived stars ξ(M) = I Eternal stars F (M) hΨiτM R −1 T Ψ(t)dt is where hΨi = T 0 the past average SFR. ξ(M) = Salpeter formula (α = 2.35) ξ(M) = dN/dM = ξo M−α Z MU 1= ξ(M)MdM ML MU (M − CM )ξ(M)dM R= MT ξo = ξ0 Mα−1 R = 0.28 (S), 0.30 (K + S), 0.34 (C + S) J.M. Lattimer F (M) . Ψ0 τM AST 346, Galaxies, Part 5 IMF and Its Evolution J.M. Lattimer AST 346, Galaxies, Part 5 Star Formation Rate Schmidt (1959): Ψ(t) = νMg (t)N N ∼ 2 theoretically if referring to mass density N ∼ 1.4 observationally if referring to surface density, more easily measured. Kennicutt (1998): Ψ ∝ Σ/τdyn where τdyn = R/V (R). J.M. Lattimer AST 346, Galaxies, Part 5 Analytical Chemical Evolution Analytical solutions are possible if Instantaneous Recycling Approximation (IRA) is adopted (Schmidt 1963). Stars are either I “eternal”, M < M , τM ≥ T ∼ 12 Gyr, or I “dead at birth”, M < M , τM ≈ 0. With IRA, Ψ(t − τM ) = Ψ(t) 6= f (M). Define pi to be the amount of i created by a stellar generation per unit mass of “eternal” stars and remnants: Z MU pi = (1 − R)−1 yi (M)ξ(M)dM, MT For a closed box, IRA model dMg Zi dMg = −(1 − R)Ψ(t), = Ψ(t)(1 − R)(pi − Zi ). dt dt Eliminating Ψ and t, and setting Mg (t = 0) = M: Mg dZi = −pi dMg ; Zi = Zi,0 + pi ln M Mg Assuming a Schmidt Law with N = 1: Mg = Me −ν(1−R)t , J.M. Lattimer Zi = Zi,0 + pi ν(1 − R)t. AST 346, Galaxies, Part 5 Models Parameters of the model should be constrained to produce the present-day metallicity. Assuming Zi,0 = 0, R ' 0.3, non-IRA models imply N = 1 and ν = 1.2 Gyr−1 . IRA predicts Mg (T )/M = e −ν(1−R)T ' 4 × 10−5 compared to the observed value Mg (T ) ' 0.2M. Also p = −(Z − Z0 )/ ln(Mg /M) ' 0.6Z ' 0.012 J.M. Lattimer AST 346, Galaxies, Part 5 Metallicity Evolution Nearby F and G stars r v < 80 km/s b v > 80 km/s J.M. Lattimer AST 346, Galaxies, Part 5 Metallicity Distribution Mg = Me −[Z (t)−Z (0)]/p . Where the gas abundance is high relative to stars (outer disk regions) the metallicity is low. Disk gas in M33 J.M. Lattimer AST 346, Galaxies, Part 5 Metal Abundances in Bulge Giant Stars Mass of stars with metallicity between Z and Z + ∆Z is dM∗ ∆Z = Me −[Z (t)−Z (0)]/p ∆Z . dZ Consistency indicates bulge retained all its gas, turning it into stars. p = 0.7Z , G and K giant stars in Galactic bulge. dM∗ /dZ dM∗ /d ln Z J.M. Lattimer AST 346, Galaxies, Part 5 Metal Abundances in Solar-Neighborhood Dwarf Stars If Z (0) = Z0 6= 0, Z0 ∼ 0.15Z : Z − Z0 ' p ln (M/Mg ) and p ' 0.63Z . Closed box model with IRA and Z (0) = 0 suggests for solar neighborhood Z (T ) ' Z ' p ln( M∗ (< 0.25Z ) 1 − e (Z0 −0.25Z )/p = M∗ (< 0.7Z ) 1 − e (Z0 −0.7Z )/p ' 0.25. M ) Mg 50 ), 13 p ' 0.74Z . ≈p ln( Fraction of metal-poor stars with Z < Z /4: M∗ (< 0.25Z ) 1 − e −0.25Z /p = M∗ (< 0.7Z ) 1 − e −0.7Z /p ' 0.52. Observed fraction ∼ 25%. Known as the G-dwarf problem. d[N(Z < Z 0 )/N(Z < ZT )]/d ln Z Pre-enrichment: Z0 = 0.08Z Infall: exponential τ = 7 Gyr J.M. Lattimer AST 346, Galaxies, Part 5 Evidence For Infall There is another inconsistency with the closed box model. Note Ψ ∼ 3 − 5M yr−1 in the solar neighborhood. Present disk gas mass is Mg ∼ 5 − 10 × 109 M . Remaining epoch of star formation should last no longer than Mg /[Ψ(1 − R)] ∼ 1.4 − 4.8 Gyr. Why do we live so near the end of star formation? Additional gas must be accumulating in the Galactic disk. J.M. Lattimer AST 346, Galaxies, Part 5 Models with Infall If it is assumed gas infalls with a metallicity Z0 = Z (t = 0) and an infall rate proportional to star formation rate: f (t) = Λ(1 − R)Ψ(t). The parameter Λ ≤ 1; if Λ = 1 then Mg (t) = Mg (0). Λ−1 −1 Λ/(1−Λ) ! Λ p Mg (t) Mg (t) = 1 − [Z (t) − Z0 ] , Z (t) = 1− . Mg (0) p Λ Mg (0) When Z = Z0 + p/Λ, the maximum metallicity, Mg → 0. Introducing Z M(t) − M(0) 1−R t Mg (t) − M(0) Λ =Λ Ψ(t 0 )dt 0 = , µ(t) = M(0) M(0) 0 M(0) Λ−1 Λ/(1−Λ) Z (t) = Z0 + pΛ−1 1 − 1 − (Λ−1 − 1)µ(t) , Λ 6= 1 h i Z (t) = Z0 + p 1 − e −µ(t) . Λ=1 Z t0 M(0) N(Z < Z 0 ) µ(t 0 ) N(Z < Z 0 ) = Ψ(t)dt = µ(t 0 ), = Λ(1 − R) N(Z < Z (T )) µ(T ) 0 J.M. Lattimer AST 346, Galaxies, Part 5 Evolution of the Solar Neighborhood (the Thin Disk) Assume exponentially decreasing infall with timescale τf = 7 Gyr and Mg (0) = Z (0) = 0. Use surface densities: I Infall normalized to observed value: Z T ΣT = f (t)dt = Σg + Σ∗ ' 12 + 38 = 50M pc−2 0 f (t) = Ae −t/τf =⇒ ΣT = Aτf 1 − e −t/τf −1 =⇒ A = (50/7) 1 − e −12/7 ' 8.7M pc−2 Gyr−1 I Assume Schmidt SFR Ψ(t) = νΣg (t) and Σg (T )/ΣT (T ) = 0.24: dΣg A = −(1 − R)Ψ(t) + Ae −t/τf , Σg (t) = e −t/τf 1 − e −αt , dt α α = (1 − R)ν − τf−1 =⇒ ν ' 0.32 Gyr−1 , Ψ(T ) ' 3.8M pc−2 G Constrain to solar metallicity Z = Z (t = 7.5 Gyr) ' 0.02: 1 − R e αt − αt − 1 Z (t) = pν =⇒ p = 1.17Z α e αt"− 1 # Z t0 0 Aν e −t /τf 0 −(1−R)νt 0 N(Z < Z ) = Ψdt = α+ − (1 − R)νe . α τf 0 J.M. Lattimer AST 346, Galaxies, Part 5 Evolution of the Solar Neighborhood (the Thin Disk) J.M. Lattimer AST 346, Galaxies, Part 5 Oxygen Evolution Oxygen is characteristic of a primary α−nucleus, originating primarily in Type II supernovae. Iron is also primary, produced in a 2:1 proportion by Type I and Type II supernovae, but the former don’t ”turn on” until [Fe/H]> −1.5. thick disk halo thin disk J.M. Lattimer AST 346, Galaxies, Part 5 CNO Evolution Carbon and oxygen are characteristic of primary nuclei, originating primarily in Type II supernovae. Nitrogen thought to originate from Hot Bottom Burning in AGB stars, but the most massive of these (8M ) have lifetimes too long to explain its primary character. It is also produced as a secondary nucleus. Now known to be a primary nucleus produced by rotationally-induced mixing in massive stars, from H-burning of C and O produced inside the star. Requires low-metallicity stars to be rotating muich faster (800 km/s) than high-metallicity stars (300 km/s). J.M. Lattimer AST 346, Galaxies, Part 5 Halo Evolution and Hierarchical Galaxy Formation The metallicity distribution of halo field stars is well fit by a closed box, IRA model dN/d ln Z ∝ (Z /p)e −Z /p . This peaks at Z = p. [Fe/H]= −1.6 implies p ' Z /40. The reduced yield could be explained by an outflow during halo formation, with a rate 8 times the SFR. It could also be interpreted as due to hierarchical galaxy formation. The halo metallicities are lower by more than 3 times than those of nearby galaxies of the same mass as the halo. Those small galaxies show a linear relation between stellar mass and metallicity which results from mass loss: hot supernova ejecta escape more easily from smaller mass galaxies. J.M. Lattimer AST 346, Galaxies, Part 5 The R-Process Elements heavier than Fe are produced by proton capture (p) or by neutron capture, either slow (s) or rapid (s). J.M. Lattimer AST 346, Galaxies, Part 5 Element Evolution Primary α-elements show less scatter and small trends with metallicity. Secondary r-process elements show large scatter. J.M. Lattimer AST 346, Galaxies, Part 5 R-Process and Heirarchical Galaxy Formation J.M. Lattimer AST 346, Galaxies, Part 5 Dwarf Spheroidals and Metallicity Distributions J.M. Lattimer AST 346, Galaxies, Part 5 Neutron Star Mergers and the R-Process Mergers have minimum evolution lifetimes of 10-100 Myrs. But high r-process abundance can be made in small metallicity places. Different places have different [Fe/H]. [Fe/H] is decoupled from time. Ishimaru, Wanajo & Prantzos 2010 J.M. Lattimer AST 346, Galaxies, Part 5 Dwarf Galaxies I Types: I I I I Dwarf Spheroidals - diffuse and devoid of gas and star formation Dwarf Ellipticals - devoid of gas and star formation Dwarf Irregulars -diffuse and abundant gas and star formation I Gas in dwarf irregulars extends beyond stellar disk HI gas on R band image of IC10 Properties: I I I All have horizontal branch stars, i.e., some stars older than 10 Gyr. This is predicted by cold dark matter. M32 is anomalous, with extremely high central surface density. Perhaps the core remnant of a larger galaxy with a massive black hole at its center Metallicities in dwarf galaxies correlates with their masses. This suggests outflow, caused by star formation, which is easier the more shallow is the gravitational potential. J.M. Lattimer AST 346, Galaxies, Part 5 M33 Wolf-Rayet J.M. Lattimer AST 346, Galaxies, Part 5 History of the Local Group Treat the Local Group as the galaxies M31 (M) and the Milky Way (m). They are separated by r = 770 kpc and approaching each other dr /dt = −120 km/s. It can be shown that the equation of motion for two arbitrarily massive objects (m, M) is the same as for a small mass m moving in the field of a much larger mass M + m. In this case, the relative positions of the two masses satisifies d 2r L2 G (M + m) = 3z − 2 dt r r2 r dr 2G (M + m) L2z =± − 2 dt r r Since dr /dt < 0, we use − sign. J.M. Lattimer Pericenter is reached when dr /dt = 0 or rmin = L2z /[2G (M + m)]. Using rmin = 0 gives M + m > (dr /dt)2 r /(2G ) ∼ 1.3 × 1012 M . There is a cyclic solution r = a(1 − e cos η) a3/2 (η − e sin η) t= p G (M + m) where a = L2z /[G (M + m)(1 − e)]. Galaxies began moving apart at t = 0 (η = 0) and are now approaching. η = π(2π) corresponds to apocenter (pericenter). M + m, η, a can be determined from r , dr /dt, t if e is known. We find a minimum 12 M+m > ∼ 5.1 × 10 M when e = 1, 12 times larger than we estimated from circular velocities. p Pericenter time is t2π = 2πa3/2 / G (M + m) = 15 Gyr. AST 346, Galaxies, Part 5