\input{preface} \chapter{Using outflows to track star formation in the W5 HII region complex} \label{ch:w5} \section{Preface} Only a few months after arriving at CU, I was given the opportunity to visit the peak of Mauna Kea to perform observations with the JCMT. I spend about 3 weeks at the telescope over the course of two years primarily mapping the W5 complex. A side-project done during these observations resulted in my Comps II project on IRAS 05358+3543. These data were taken using Jonathan Williams' Hawaii time allocation with the HARP receiver. The data were taken with essentially no plan for how they would be used. The paper may have diminished our group's overall interest in the W5 region: it turns out that star formation is probably at its end here, being quenched by massive-star feedback. However, there is a largely ignored cloud to the northwest of the well-studied W5 bubbles that has significant potential to form new stars. The W5 study was originally intended to include a Bolocam census of cores, but the data in this region turned out to be the most problematic and contained little signal. We acquired additional data in 2009, but never got around to performing a joint analysis of the CO and continuum data. In part, at least, this is because W5 is so faint in the millimeter continuum compared to many Galactic Plane sources. % This work is essentially a very detailed study of a star-forming region with % minimal implications for star forming theories at the moment. \section{Introduction} Galactic-scale shocks such as spiral density waves promote the formation of giant molecular clouds (GMCs) where massive stars, star clusters, and OB associations form. The massive stars in such groups can either disrupt the surrounding medium or promote further star formation. While ionizing and soft UV radiation, stellar winds, and eventually supernova explosions destroy clouds in the immediate vicinity of massive stars, as the resulting bubbles age and decelerate, they can also trigger further star formation. In the ``collect and collapse'' scenario \verb|\citep[e.g.][]{elmegreen:sequential:1977}|, gas swept-up by expanding bubbles can collapse into new star-forming clouds. In the ``radiation-driven implosion'' model \verb|\citep{bertoldi:cometary:1990,klein:implosion:1983}|, pre-existing clouds may be compressed by photo-ablation pressure or by the increased pressure as they are overrun by an expanding shell. In some circumstances, forming stars are simply exposed as low-density gas is removed by winds and radiation from massive stars. These processes may play significant roles in determining the efficiency of star formation in clustered environments \verb|\citep{elmegreen1998}|. % Shortly after igniting, young massive stars may either provide the additional % pressure needed to collapse neighboring gas clumps into stars, or they may % simply disperse and destroy their natal environments. The former phenomenon, % known as triggered star formation, may play a significant role in determining % the efficiency of star formation in clustered environments % \verb|\citet{elmegreen1998}|. In addition, triggering from external processes % asynchronous with star formation by direct collapse (e.g. supernovae, spiral % density waves) are thought to be partly responsible for the evenutal collapse % of parent molecular clouds, but it is not yet known on what scales triggering % is important, nor how strongly each mechanism contributes. Feedback from low mass stars may also control the shape of the stellar initial mass function in clusters \verb|\citep{adams1996,Peters2010}|. Low mass young stars generate high velocity, collimated outflows that contribute to the turbulent support of a gas clump, preventing the clump from forming stars long enough that it is eventually blown away by massive star feedback. It is therefore important to understand the strength of low-mass protostellar feedback relative to other feedback mechanisms. Outflows are a ubiquitous indicator of the presence of ongoing star formation \verb|\citep{reipurth2001}|. CO outflows are an indicator of ongoing embedded star formation at a younger stage than optical outflows because shielding from the interstellar radiation field is required for CO to survive. Although Herbig- Haro shocks and \hh\ knots reveal the locations of the highest-velocity segments of these outflows, CO has typically been thought of as a ``calorimeter'' measuring the majority of the mass and momentum ejected from protostars or swept up by the ejecta \verb|\citep{Bachiller1996}|. %\subsection{W5} The W5 star forming complex in the outer galaxy is a prime location to study massive star formation and triggering. The bright-rimmed clouds in W5 have been recognized as good candidates for ongoing triggering by a number of groups \verb|\citep{lefloch:cometary:1997,thompson:searching:2004,karr:triggered:2003}|. The clustering properties were analyzed by \verb|\citet{koenig:clustered:2008}| using Spitzer infrared data, and a number of significant clusters were discovered. The whole W5 complex may be a product of triggering, as it is located on one side of the W4 chimney thought to be created by multiple supernovae during the last $\sim$10 MYr \citep[][Figure \verb|\ref{fig:color_overview}|] {oey:hierarchical:2005}. Following \verb|\citet{koenig:clustered:2008}|, we adopt a distance to W5 of 2 kpc based on the water-maser parallax distance to the neighboring W3(OH) region \verb|\citep{Hachisuka2006}|. As with W3, the W5 cloud is substantially ($\approx1.5\times$) closer than its kinematic distance would suggest ($v_{LSR}(-40~\kms)\approx3$ kpc). Given this distance, \verb|\citet{koenig:clustered:2008}| derived a total gas mass of 6.5\ee{4} \msun\ from a 2 \um\ extinction map. The W5 complex was mapped in the $^{12}$CO 1-0 emission line by the Five College Radio Astronomy Observatory (FCRAO) using the SEQUOIA receiver array \verb|\citep{heyer:ogs:1998}|. The same array was used to map W5 in the \thirteenco\ 1-0 line (C. Brunt, private communication). Some early work searched for outflows in W5 \verb|\citep{bretherton:unbiased:2002}|, but the low-resolution CO 1-0 data only revealed a few, and only one was published. The higher resolution and sensitivity observations presented here reveal many additional outflows. \begin{figure*} % Generated by GIMP \includegraphics[angle=90,width=5in]{figures_chw5/w345mos_brightened_labeled} \caption{An overview of the W3/4/5 complex (also known as the ``Heart and Soul Nebula'') in false color. Orange shows 8 \um\ emission from the Spitzer and MSX satellites. Purple shows 21 cm continuum emission from the DRAO CGPS \verb|\citep{Taylor2003:CGPS}|; the DSS R image was used to set the display opacity of the 21 cm continuum as displayed (purely for aesthetic purposes). The green shows JCMT \twelveco\ 3-2 along with FCRAO \twelveco\ 1-0 to fill in gaps that were not observed with the JCMT. The image spans $\sim7\arcdeg$ in galactic longitude. This overview image shows the hypothesized interaction between the W4 superbubble and the W3 and W5 star-forming regions \verb|\citep{oey:hierarchical:2005}|.} \label{fig:color_overview} \end{figure*} %\subsection{Neighboring Regions} While W5 is thought to be associated with the W3/4/5 complex, there are other infrared sources in the same part of the sky that are not obviously associated with W5. Some of these have been noted to be in the outer arm (several kpc behind W5) by \verb|\citet{Digel1996}| and \verb|\citet{Snell2002}|. \par \par In section 2, we present the new and archival data used in our study. In section 3, we discuss the outflow detection process and compare outflow detectability in W5 to that in Perseus. In section 4, we discuss the physical properties of the outflows and their implications for star formation in the W5 complex. In section 5, we briefly describe the outer-arm outflows discovered. \section{OBSERVATIONS} \subsection{JCMT HARP CO 3-2} CO J=3-2 345.79599 GHz data were acquired at the 15 m James Clerk Maxwell Telescope (JCMT) using the HARP array on a series of observing runs in 2008. On 2-4 January, 2008, $\sim$ 800 square arcminutes were mapped. During the run, $\tau_{225}$, the zenith opacity at 225 GHz measured using the Caltech Submillimeter Observator tipping radiometer, ranged from 0.1 to 0.4 ($0.420$ \kms) flows exist in the W5 region. Note that the histogram compares quantities that are not directly equivalent: the outflows in \verb|\citet{curtis2010}| and our own data are measured out to the point at which the outflow signal is lost, while the `region' velocities are full-width half-max (FWHM) velocities. Finally, we use the detectability of outflows in Perseus to inform our expectations in W5. Since it appears that we can detect outflows from low-mass protostars with sub-stellar to $\sim30L_\odot$ luminosities at the distance of W5 and these objects should be the most numerous in a standard initial mass function, the distribution of physical properties in W5 outflows should be similar to those in Perseus. However, because W5 is a somewhat more massive cloud ($M_{W5}\approx 5 M_{Perseus}$ \footnote{$M_{W5}$ is estimated from \thirteenco. We also estimate the total molecular mass in W5 using the X-factor and acquire $M_{W5}=5.0\ee{4}$ \msun, in agreement with \verb|\citet{karr:triggered:2003}|, who estimated a molecular mass of 4.4\ee{4} from \twelveco\ using the same X-factor. \verb|\citet{koenig:clustered:2008}| estimated a total gas mass of 6.5\ee{4} from a 2MASS extinction map. The total molecular mass in Perseus is $M_{Perseus} \sim 10^4$ \verb|\citep{bally-perseus2008}|}), we might expect the high-end of the distribution to extend to higher values of outflow mass, momentum, and energy. Since we will likely see clustered outflows confused into a smaller number of distinct lobes, we expect a bias towards higher values of the derived quantities but a lower detection rate. \subsubsection{Velocity, Column Density, and Mass Measurements} \label{sec:measurements} Throughout this section, we assume that the CO lines are optically thin and thermally excited. The measured properties are presented in Table \verb|\ref{tab:outflows}|. These assumptions are likely to be invalid, so we also discuss the consequences of applying `typical' optical depth corrections to the derived quantities. Because we do not measure optical depths and the optical depth correction for CO 3-2 is less well quantified than for CO 1-0 \verb|\citep{curtis2010,Cabrit1990}|\footnote{In \verb|\citet{curtis2010}|, this correction factor ranged from 1.8 to 14.3; \verb|\citet{arce2010}| did not enumerate the optical depth correction they used but it is typically around 7 \verb|\citep{Cabrit1990}|. }, we only present the uncorrected measurements in Table \verb|\ref{tab:outflowsderived}|. The outflow velocity ranges were measured by examining both RA-velocity and Dec-velocity diagrams interactively using the STARLINK GAIA data cube viewing tool. The velocity limits are set to include all outflow emission that is distinguishable from the cloud (i.e. the velocity at which outflow lobes dominate over the gaussian wing of the cloud emission) down to zero emission. An outflow size \citep[or lobe size, following ][]{curtis2010} was determined by integrating over the blue and red velocity ranges and creating an elliptical aperture to include both peaks; the position and size therefore have approximately beam-sized ($\approx18\arcsec$) accuracy. The integrated outflow maps are shown as red and blue contours in Figure \verb|\ref{fig:pv2b}|. The velocity center was computed by fitting a gaussian to the FCRAO \thirteenco\ spectrum averaged over the elliptical aperture. \Figure{figures_chw5/WidthHistogram} {Histogram of the outflow line widths. {\it Black lines}: histogram of the measured outflow widths (half-width zero-intensity, measured from the fitted central velocity of the cloud to the highest velocity with non-zero emission). {\it Blue dashed lines}: outflow half-width zero-intensity (HWZI) for the outer arm (non-W5) sample. {\it Solid red shaded}: The measured widths (HWHM) of the sub-regions as tabulated in Table \verb|\ref{tab:regionspectra}|. {\it Gray dotted}: Outflow $v_{max}$ (HWZI) values for Perseus from \verb|\citet{curtis2010}|. } {fig:widthhist}{0.5}{0} The column density is estimated from \twelveco\ J=3-2 assuming local thermal equilibrium (LTE) and optically thin emission using the equation $ N(\hh) = 5.3\ee{18}\eta_{mb}^{-1} \int T_A^*(v) dv $ for $T_{ex}=20$ K. The derivation is given in the Appendix. The column density in the lobes is likely to be dominated by low-velocity gas and therefore our dominant uncertainty may be missing low-velocity emission rather than poor assumptions about the optical depth. The scalar momentum and energy were computed from \begin{equation} p = M \frac{\sum T_A^*(v) (v-v_{c}) \Delta v}{ \sum T_A^*(v) \Delta v} \end{equation} \begin{equation} E = \frac{M}{2} \frac{\sum T_A^*(v) (v-v_{c})^2 \Delta v}{ \sum T_A^*(v) \Delta v} \end{equation} where $v_c$ is the \thirteenco\ 1-0 centroid velocity. The same assumptions used in determining column density are applied here. We estimate an outflow lifetime by taking half the distance between the red and blue outflow centroids divided by the maximum measured velocity difference ($\Delta v_{max} = (v_{max,red}-v_{max,blue})/2$), $\tau_{flow} = L_{flow} / ( 2 \Delta v_{max})$, where $L_{flow}$ refers to the length of the flow. This method assumes that the outflow inclination is 45\arcdeg; if it is more parallel to the plane of the sky, we overestimate the age, and vice-versa. The momentum flux is then $\dot{P} = p / \tau$. Similarly, we compute a mass loss rate by dividing the total outflow mass by the dynamical age, which yields what is likely a lower limit on the mass loss rate (if the lifetime is underestimated, the mass loss rate is overestimated, but the outflow mass is always a lower limit because of optical depth and confusion effects). The dynamical ages are highly suspect since the red and blue lobes are often unresolved or barely resolved, and diffuse emission averaged with the lobe emission can shift the centroid position. Additionally, it is not clear what portion of the outflow corresponds to the centroid: the bow shock or the jet could both potentially dominate the outflow emission. \verb|\citet{curtis2010}| discuss the many ways in which the dynamical age can be in error. Our mass loss rates are similar to those in Perseus \emph{without} correcting our measurements for optical depth, while our outflow masses are an order of magnitude lower. It therefore appears that our dynamical age estimates must be too low, since we have no reason to expect protostars in W5 to be undergoing mass loss at a greater rate than those in Perseus. However, given more reliable dynamical age estimates from higher resolution observations of shock tracers, the mass loss rates could be corrected and compared to other star-forming regions. Because the emission was assumed to be optically thin, the mass, column, energy, and momentum measurements we present are strictly lower limits. While some authors have computed correction factors to \twelveco\ 1-0 optical depths \verb|\citep[e.g.][]{Cabrit1990}|, the corrections are different for the 3-2 transition \verb|\citep[1.8 to 14.3,][]{curtis2010}|. Additionally, CO 3-2 may require a correction for sub-thermal excitation because of its higher critical density (the CO 3-2 critical density is 27 times higher than CO 1-0; see Appendix \verb|\ref{appendix:dipole}| for modeling of this effect). Additionally, most of the outflow mass is at the lowest distinguishable velocities in typical outflows \verb|\citep[e.g.][]{arce2010}|. It is therefore plausible that in the more turbulent W5 region, a greater fraction of the outflow mass is blended (velocity confused) with the cloud and therefore not included in mass, momentum, and energy measurements. This omission could be greater than the underestimate due to poor opacity assumptions. The total mass of the W5 outflows is $M_{tot}\approx1.5 \msun$, substantially lower, even with an optical depth correction of $10\times$, than the 163 \msun\ reported in Perseus \verb|\citep{arce2010}|. \verb|\citet{arce2010}| also include a correction factor of 2.5 to account for higher temperatures in outflows and a factor of 2 to account for emission blended with the cloud. The temperature correction is inappropriate for CO 3-2 (see Appendix \verb|\ref{appendix:dipole}|, Figure \verb|\ref{fig:approx}|), but the resulting total outflow mass in W5 with an optical depth correction and a factor of 2 confusion correction is about 30 \msun. In order to make our measurements consistent with a mass of 160 \msun\ , a density upper limit in the outflowing gas of $n(\hh) < 10^{3.5} \percc$ is required, since a lower gas density results in greater mass for a given intensity (see Appendix \verb|\ref{appendix:dipole}|, Figure \verb|\ref{fig:coradex}|). However, we expect the total outflow mass in W5 to be greater than in Perseus because of the greater cloud mass, implying that the density in the flows must be even lower, or additional corrections are needed. The total outflow momentum is $p_{tot}\approx10.9 \msun$ \kms, versus a quoted 517 \msun \kms\ in Perseus \verb|\citep{arce2010}|. \verb|\citet{arce2010}| included inclination and dissociative shock corrections for the momentum measurements on top of the correction factors already applied to the mass. If these corrections are removed from the Perseus momentum total (except for optical depth, which is variable in their data and therefore cannot be removed), the uncorrected outflow momentum in Perseus would be about 74 \msun \kms. The W5 outflow momentum, if corrected with a `typical' optical depth in the range 7-14, would match or exceed this value. If an additional CO 3-2 excitation correction (in the range 1-20) is applied, the W5 outflow momentum would significantly exceed that in Perseus. Assuming a turbulent line width $\Delta v \sim 3$ \kms\ (approximately the smallest FWHM line-width observed), the total turbulent momentum in the ambient cloud is $p = M_{tot} \Delta v = 1.3\ee{5} \msun$ \kms, which is $\sim10^5$ times the measured outflow momentum - the outflows detected in our survey cannot be the sole source of the observed turbulent line widths, even if corrected for optical depth and missing mass. Table \verb|\ref{tab:regionspectra}| presents the turbulent momentum for each sub-region computed by multiplying the measured velocity width by the integrated \thirteenco\ mass. Even if the outflow measurements are orders of magnitude low because of optical depth, cloud blending, sub-thermal excitation, and other missing-mass considerations, outflows contribute negligibly to the total momentum of high velocity gas in W5. This result is unsurprising, as there are many other likely sources of energy in the region such as stellar wind bubbles and shock fronts between the ionized and molecular gas. Additionally, in regions unaffected by feedback from the HII region (e.g. W5NW), cloud-cloud collisions are a possible source of energy. Figure \verb|\ref{fig:outflowhist}| displays the distribution of measured properties and compares them to those derived in the COMPLETE \verb|\citep{arce2010}| and \verb|\citet{curtis2010}| HARP CO 3-2 surveys of Perseus. Our derived masses are substantially lower than those in \verb|\citet{arce2010}| even if corrected for optical depth, but our momenta are similar to the CPOC (COMPLETE Perseus Outflow Candidate) sample and our energies are higher, indicating a bias towards detecting mass at high velocities. The bias is more heavily towards high velocities than the CO 1-0 used in \verb|\citet{arce2010}|. The discrepancy between our values and those of \verb|\citet{arce2010}| and \verb|\citet{curtis2010}| can be partly accounted for by the optical depth correction applied in those works: \thirteenco\ was used to correct for opacity at low velocities, where most of the outflow mass is expected. Those works may also have been less affected by blending because of the smaller cloud line widths in Perseus. %\Figure{OutflowHistograms}{Histograms of outflow properties.}{fig:outflowhist}{1.0} \FigureFour{figures_chw5/OutflowMassHistogram_legend}{figures_chw5/OutflowEnergyHistogram}{figures_chw5/OutflowMomentumHistogram}{figures_chw5/OutflowColumnHistogram} {Histograms of outflow physical properties. %All display a peak at low values in addition %to a high excess. None are obviously well-fit by power-law or gaussian distributions. The solid unfilled lines are the W5 outflows (this paper), the forward-slash hashed lines show \verb|\citet{arce2010}| CPOCs , the dark gray shaded region shows \verb|\citet{arce2010}| values for known outflows in Perseus, and the light gray, backslash-hashed regions show \verb|\citet{curtis2010}| CO 3-2 outflow properties. The outflow masses measured in Perseus are systematically higher partly because both surveys corrected for line optical depth using \thirteenco. The medians of the distributions are 0.017, 0.044, 0.33, and 0.14 \msun\ for W5, Curtis, Arce Known, and Arce CPOCs respectively, which implies that an optical depth and excitation correction factor of 2.5-20 would be required to make the distributions agree (although W5, being a more massive region, might be expected to have more massive and powerful outflows). It is likely that CO 3-2 is sub-thermally excited in outflows, and CO outflows may be destroyed by UV radiation in the W5 complex while they easily survive in the lower-mass Perseus region, which are other factors that could push the W5 mass distribution lower. } {fig:outflowhist} The momentum flux and mass loss rate are compared to the values derived in Perseus by \verb|\citet{hatchell2007}| and \verb|\citet{curtis2010}| in Figures \verb|\ref{fig:outflowPflux}| and \verb|\ref{fig:outflowmdot}|. Both of our values are computed using the dynamical timescale $\tau_d$ measured from outflow lobe separation, while the \verb|\citet{hatchell2007}| values are derived using a more direct momentum-flux measurement in which the momentum flux contribution of each pixel in the resolved outflow map is considered. The derived momentum fluxes (Figure \verb|\ref{fig:outflowPflux}|) are approximately consistent with the \verb|\citet{curtis2010}| Perseus momentum fluxes; \verb|\citet{curtis2010}| measure momentum fluxes in a range $1\ee{-6} -55$ \kms) and could be associated with different clouds within the same spiral arm. The other 7 have central velocities $v_{LSR} < -55$ \kms\ and are associated with the outer arm identified in previous surveys \verb|\citep[e.g.][]{Digel1996}|. The properties of these outflows are given in Tables \verb|\ref{tab:outeroutflows}| and \verb|\ref{tab:outeroutflowsderived}|; the distances listed are kinematic distances assuming $R_0=8.4$ kpc and $v_0=254$~\kms\ \verb|\citep{Reid2009}|. Of these outflows, all but one are within 2\arcmin\ of an IRAS point source. Outflow 54 is the most distant in our survey at a kinematic distance $d=7.5$ kpc ($v_{lsr}=-75.6$ \kms) and galactocentric distance $D_G = 14.7$ kpc. It has no known associations in the literature. Outflows 41 - 44 are associated with a cloud at $v_{LSR}\sim -62$ \kms\ known in the literature as LDN 1375 and associated with IRAS 02413+6037. Outflows 53 and 55 are at a similar velocity and associated with IRAS 02598+6008 and IRAS 02425+6000 respectively. All of these sources lie roughly on the periphery of the W5 complex. Outflows 45 - 52 are associated with a string of IRAS sources and HII regions to the north of W5 and have velocities in the range $-55 < v_{LSR} < -45$. They therefore could be in the Perseus arm but are clearly unassociated with the W5 complex. Outflows 45 and 46 are associated with IRAS 02435+6144 and they may also be associated with the Sharpless HII region Sh 2-194. Outflows 47 and 48 are associated with IRAS 02461+6147, also known as AFGL 5085. Outflows 49 and 50 are nearby but not necessarily associated with IRAS 02475+6156, and may be associated with Sh 2-196. Outflows 51 and 52 are associated with IRAS 02541+6208. % \subsection{Comparison to other studies} % % XXXX: Use otherpapers compare.txt. Discuss what the most powerful outflows in each region % would look like at 2kpc % \Table{ccccc}{Comparison to other outflow surveys} % {{Paper} & {Region} & {Tracer} & {Distance (pc)} & {Area (arcmin$^2$)} & {Area (pc$^2$)} % & {Resolution (\arcsec)} & {Resolution (pc)} & {Outflows detected} \\ } % {tab:surveys} % { % \verb|\citet{arce2010}| & CO 1-0 & Perseus & 250 & 57600 & 305 & 46 & 0.056 & 96 \\ % \verb|\citet{dionatos2010}| & CO 3-2 & Serpens & 310 & 29 & 0.24 & 14 & 0.021 & 20 \\ % \verb|\citet{davis:jcmt:2010}| & CO 3-2 & Taurus & 140 & 2705 & 4.5 & 14 & 0.0095 & 16 \\ % \verb|\citet{hatchell2007}| \tablenotemark{a} & CO 3-2 & Perseus & 250 & 204 & 1.1 & 14 & 0.017 & 37 \\ % This paper & CO 3-2 & W5 & 2000 & 7200 & 2440 & 14 & 0.14 & 38 \\ % }{ % \tablenotetext{a}{This survey targeted 51 mm cores, hence its much higher detection rate per unit area.} % } \section{Conclusions} We have identified \nwfive\ molecular outflow candidates in the W5 star forming region and an additional \nouter\ outflows spatially coincident but located in the outer arm of the Galaxy. \begin{itemize} % \item The majority of the millimeter sources are associated with outflows, % though some of the brightest millimeter sources lack outflows. These % sources may consist of warmer dust that has not yet become Jeans % unstable. Millimeter emission is a good tracer of active embedded star % formation, while far-infrared brightness is not. % \item The majority of the gas seen in \thirteenco\ and 1.1 mm emission is % associated with star formation. There is therefore only a small amount % of \item The majority of the CO clouds in the W5 complex are forming stars. Star formation is not limited to cloud edges around the HII region. Because star formation activity is observed outside of the region of influence of the W5 O-stars, it is apparent that direct triggering by massive star feedback is not responsible for all of the star formation in W5. \item The W5 complex is seen nearly face-on as evidenced by a strict upper limit on the CO column through the center of the HII-region bubbles. It is therefore an excellent region to study massive star feedback and revealed and triggered star formation. \item Outflows contribute negligibly to the turbulent energy of molecular clouds in the W5 complex. This result is unsurprising near an HII region, but supports the idea that massive star forming regions are qualitatively different from low-mass star-forming regions in which the observed turbulence could be driven by outflow feedback. Even in regions far separated from the O-stars, there is more turbulence and less energy injection from outflows than in, e.g., Perseus. \item Despite detecting a significant number of powerful outflows, the total outflowing mass detected in this survey ($\sim 1.5$ \msun\ without optical depth correction, perhaps $10-20$ \msun\ when optical depth is considered) was somewhat smaller than in Perseus, a low to intermediate mass star forming region with $\sim 1/6$ the molecular mass of W5. \item The low mass measured may be partly because the CO 3-2 line is sub-thermally excited in outflows. Therefore, while CO 3-2 is an excellent tracer of outflows for detection, it does not serve as a `calorimeter' in the same capacity as CO 1-0. \item Even considering excitation and optical depth corrections, it is likely that the mass of outflows in W5 is less than would be expected from a simple extrapolation from Perseus based on cloud mass. CO is likely to be photodissociated in the outflows when they reach the HII region, accounting for the deficiency around the HII region edges. However, in areas unaffected by the W5 O-stars such as W5NW, the deficiency may be because the greater turbulence in the W5 clouds suppresses star formation or hides outflows. \item Velocity gradients across the tails of many cometary clouds have been observed, hinting at their geometry and confirming that the outflows seen from their heads must be generated by protostars within. %The data presented contain tens of cometary clouds with %precise kinematic information about their molecular gas. \item Outflows have been detected in the Outer Arm at galactocentric distances $\gtrsim12$ kpc. These represent some of the highest galactocentric distance star forming regions discovered to date. %\item Feedback in W5 is primarily destroying rather than % triggering SF. \end{itemize} % \section{Acknowledgements} % We thank the two anonymous referees for their assistance in refining this % document. We thank Devin Silvia for a careful proofread of the text. This work % has made use of the APLpy plotting package % (\url{http://aplpy.sourceforge.net}), the pyregion package % (\url{http://leejjoon.github.com/pyregion/}), the agpy code package % (\url{http://code.google.com/p/agpy/}) , IPAC's Montage % (\url{http://montage.ipac.caltech.edu/}), the DS9 visualization tool % (\url{http://hea-www.harvard.edu/RD/ds9/}), the pyspeckit spectrosopic analysis % toolkit (\url{http://pyspeckit.bitbucket.org}), and the STARLINK package % (\url{http://starlink.jach.hawaii.edu/}). IRAS data was acquired through IRSA % at IPAC (\url{http://irsa.ipac.caltech.edu/}). DRAO 21 cm data was acquired % from the Canadian Astronomical Data Center % (\url{http://cadcwww.hia.nrc.ca/cgps/}). The authors are supported by the % National Science Foundation through NSF grant AST-0708403. This research has % made use of the SIMBAD database, operated at CDS, Strasbourg, France %{\it Facilities:} JCMT, VLA %\bibliography{w5outflows} {\tiny \clearpage \include{outflowtable_mnras} % tab:outflows \include{outflowsumtable_mnras} % tab:outflowsums \clearpage \include{outerarm_outflowtable_mnras} } \include{individual_outflows_preface} %\include{individual_outflows} \include{w5_appendix} %\end{document} \input{solobib} \end{document}