M82. This is due to the limited sensitivity and spatial resolution of currentinstruments, but this is changing. New mm-wave arrays (e.g., CARMA, andesp ecially ALMA) will map the molecular gas in a large sample of nearbygalaxies with excellent resolution (< 1!!). As described in §3.3.5, Spitzer hasrecently helped constrain the amount and location of dust in the winds.3. Zone of influence and escape efficiency. Current estimates of these two quan-tities are limited by the sensitivity of the instrument used for the measure-ments. Deeper emission-line, X-r ay, and radio maps of wind galaxies willprovide better constraints on the extent of the wind and the probabilitythat the outflowing material escapes from t he pot ential well of the host.4. Thermalization efficiency. Observational constraints on the thermalizationefficiency of GWs are rare because o f an incomplete account ing of the varioussources of thermal energy and kinetic energy in the wind. A mult iwavelengthapproach that considers all gas phases is needed.5. Wind / ISM interface and influence of magnetic fields. Constraints on mi-crophysics at the interface between the wind and ga laxy ISM are availablein only a handful of gala xies. High-resolution (≤ parsec scale) imaging andspectra of the entrained disk material in a sizable sample of local objects arerequired. The lar ge-scale morphology of the magnetic field lines has beenmapped in a few winds, but the strength of the field on pc scale is unknown.This information is crucial in estimating the conductivity between the hotand cold fluids.6. Positive feedback. Star-forming radio jet/gas interactions have been foundin a few nearby systems and are suspected to be responsible for the “align-ment effect” between the radio and UV continua in distant r adio galaxies.The same physics may also provide positive feedback in wind galaxies. Con-vincing evidence for positive feedback has been found in the disk o f M82(Matsushida et al. 2004), but the frequency of this phenomenon is com-pletely unknown.7. Galactic winds in the distant universe. Absorption-line studies of high-zgalaxies and QSOs will remain a powerful tool to search for distant galacticwinds and to constrain their environmental impact. Future large ground andspace telescopes will extend such studies to the r eionization epoch. Thesegalaxies are very faint, but gravitational lensing by foreground clusters canmake them detectable and even spatially resolved.4 Elemental Abundances as Tracers of Star Formation4.1 Introduction : Basics of Chemical EvolutionHydrogen, helium, and tr aces o f lithium, boron, a nd beryllium were producedearly on in the Big Bang. All other elements (i.e. a ll other “metals”) were24produced through nucleosynthesis in stars. The abundances of these elementsare t herefore a direct tracer of past star formation in a galaxy.Gas is transformed into stars. Each star burns hydrogen and helium in itsnucleus and produces heavy elements. These elements are partially returnedinto the interstellar gas at the end of the star’s life via stellar winds or super-novae explosions. Some fraction of the metals are locked into the remnant ofthe star. If there is no gas infall f r om the outside or selective loss of metals tothe outside, the metal abundance of the gas, and of subsequent generationsof stars, should increase with time. So in principle the evolution of chemi-cal element abundances in a galaxy provides a clock for galactic aging. Oneshould expect a relation between metal abundances and stellar ages. On aver-age, younger stars should contain more iron than older stars. This is partia llythe case for the solar neigborhood, where an age-metallicity relation is seenfor nearby disk stars, but a lot of scatter is seen at old a ges (> 3 Gyr; e.g.,Nordstrom, Andersen, & Mayor 2005). Clearly, our Galaxy is not as simpleas described here and we need to add a few more ingredients t o better matchthe observations.In §4.2, I describe a few simple models to account for the complexity of galax-ies. An extensive literature exists on this topic. I refer the readers to the sem-inal paper by Tinsley (1980) as well as Binney & Tremaine (1987; §§9.2 and9.3) a nd Binney & Merrifield (1998; §5.3) . In §4.3, I compare the predictions ofthese models with the observations in local star-forming and starburst galax-ies and in distant star-forming galaxies and quasars. These comparisons helpus understand the integrated star formation history and chemical evolution ofthese objects.4.2 Simple ModelsAll models discussed here a ssume that the galaxy’s gas is well-mixed i.e. uni-form metal abundance, and that the (high-mass) stars return their nucleosyn-thetic products rapidly, much faster than the time to form a sig nificant fractionof the stars (this is called the “instantaneous recycling approximation”).4.2.1 Closed Bo xThe closed-box model further assumes that no infall or outflow is taking place.In that case, the total baryonic mass of the galaxy, Mbaryons= Mg(as)+ Ms(tar)= constant. If Z is the fraction by mass of heavy elements (the Sun’s abun-dance is Z"∼ 0.02 and the most metal-poor stars in the Milky Way haveZ ≤ 10−4Z"), t he mass of heavy elements in the gas Mh= ZMg.25If the total mass made into stars is dM!sand the amount of mass instanta-neously returned to the ISM (from supernovae and stellar winds, enriched withmetals) is dM!!s, then the net matter turned into stars is dMs= dM!s− dM!!s.The mass of heavy elements returned to t he ISM is y dMs, where y is theyield of heavy elements (made instantaneously). As a rule-of-thumb, only starsmore massive than ∼ 8M"make heavies (supernovae). The fraction of massreturned t o the ISM dM!!s/dMs∼ 0.20, the yield y ∼ 0.01 (dependent onstellar evolution and the Initial Mass Function ≡ IMF), and the metallicityof the shed gas Z(shed gas) = (heavies shed) / (mass shed) = y dMs/dM!!s∼0.01/0.2 = 0.05 (i.e. about 2.5 × Z").In the closed-box model, mass conservation impliesdMg+ dMs=0 (6)The net change in metal content of the gas isdMh= ydMs− ZdMs=(y − Z)dMs(7)Since dMg= −dMsand Z = Mh/Mg, the change in Z isdZ = dMh/Mg− MhdMg/M2g=(y − Z)dMs/Mg+(Mh/Mg)(dMs/Mg)=ydMs/MgdZ/ dt = −y(dMg/dt)/MgAssuming y = constant (i.e. independent of time and Z):Z(t)=Z(0) − y ln [Mg(t)/Mg(0)] (8)= Z(0) − y ln µ(t),where µ = gas (mass) fraction ≡ Mg(t)/Mg(0) = Mg(t)/Mt. The metallicityof the gas grows with time, as new stars are f ormed and the gas is