1: \documentclass[preprint,12pt]{aastex}
2: \usepackage{natbib}
3: \bibliographystyle{apj}
4: \newcommand{\kms}{km s$^{-1}$}
5: \newcommand{\vsys}{systematic velocity}
6: \newcommand{\pa}{position angle}
7: \newcommand{\inc}{inclination}
8: \newcommand{\vrad}{$v_R$}
9: \newcommand{\vtan}{$v_\theta$}
10: \newcommand{\vx}{$v_x$}
11: \newcommand{\vy}{$v_y$}
12: \newcommand{\ps}{$\Omega_p$}
13: \newcommand{\halpha}{H$\alpha$}
14: \shorttitle{Spiral Streaming in M51}
15: \shortauthors{Shetty et al}
16:
17: \begin{document}
18: \title{Kinematics of Spiral Arm Streaming in M51}
19: \author{Rahul Shetty, Stuart N. Vogel, Eve C. Ostriker, and Peter
20: J. Teuben}
21: \affil{Department of Astronomy, University of Maryland, College Park,
22: MD 20742-2421}
23: \email{shetty@astro.umd.edu, vogel@astro.umd.edu,
24: ostriker@astro.umd.edu, teuben@astro.umd.edu}
25:
26: \begin{abstract}
27:
28: We use CO and \halpha\ velocity fields to study the gas kinematics
29: in the spiral arms and interarms of M51 (NGC 5194), and fit the 2D
30: velocity field to estimate the radial and tangential velocity
31: components as a function of spiral phase (arm distance). We find
32: large radial and tangential streaming velocities, which are
33: qualitatively consistent with the predictions of density wave theory
34: and support the existence of shocks. The streaming motions are
35: complex, varying significantly across the galaxy as well as along
36: and between arms. Aberrations in the velocity field indicate that
37: the disk is not coplanar, perhaps as far in as 20\arcsec\ (800 pc)
38: from the center. Velocity profile fits from CO and \halpha\ are
39: typically similar, suggesting that most of the \halpha\ emission
40: originates from regions of recent star formation. We also explore
41: vortensity and mass conservation conditions. Vortensity
42: conservation, which does not require a steady state, is empirically
43: verified. The velocity and density profiles show large and varying
44: mass fluxes, which are inconsistent with a steady flow for a single
45: dominant global spiral mode. We thus conclude that the spiral arms
46: cannot be in a quasi-steady state in any rotating frame, and/or that
47: out of plane motions may be significant.
48:
49: \end{abstract}
50:
51: \keywords{galaxies: individual (M51) -- galaxies: spiral structure --
52: galaxies: kinematics and dynamics}
53:
54: \section{Introduction}
55: Spiral arms are the dominant morphological features of most disk
56: galaxies. From a theoretical perspective, two frameworks have been
57: proposed to describe the nature of the spiral arms: one is that the
58: spiral arms are generally long-lasting, or slowly evolving, and the
59: other is that the arms are transient features
60: \citep[e.g.][]{ToomreToomre72}. Observational studies have yet to
61: show definitively whether the arms are evolving or long lived, though
62: it has been over 40 years since the landmark paper by \citet{LinShu64}
63: suggesting that spiral structure in galaxies is a long lived
64: phenomenon --- the Quasi-Stationary Spiral Structure (QSSS) hypothesis
65: \citep{Lindblad63}. In the QSSS depiction, though material passes in
66: and out of the arms, the slowly evolving global pattern rotates with a
67: single angular speed that results from the excitation of global modes.
68: The spiral arms are formed from self-excited and self-regulated
69: standing density waves \citep{Bertin89a, Bertin89b, BertinLin96}.
70:
71: However, interaction between a disk galaxy and a companion is another
72: explanation for the presence of spiral arms. In such a framework, the
73: arms are transient features that are generated by the tidal
74: interaction \citep[e.g.,][]{ToomreToomre72}. Any spiral arms existing
75: before the encounter are overwhelmed by the tidal driving
76: \citep{SaloLaur00I}.
77:
78: Regardless of the origin of the stellar arms, gas in the disk will
79: respond strongly to the gravitational perturbations those arms impose.
80: Numerical studies have indicated that shocks can develop if the
81: relative speed between the spiral perturbation and the gas is large
82: \citep{Roberts69, ShuMilioneRoberts73, Woodward75}. The presence of
83: dust lanes in the spiral arms and the enhancement in ionized emission
84: downstream, indicating regions of star formation, is attributed to this
85: shock scenario. Such shocks are also thought to be the cause of the
86: well defined molecular arms seen in many grand design galaxies,
87: including M51. Numerical and analytical studies have provided
88: predictions for the velocity and density profiles of the matter
89: affected by the spiral gravitational perturbation
90: \citep[e.g][]{LubowBalbusCowie86, KO02, GittinsClarke04}.
91:
92: There have been numerous observational studies addressing the nature
93: of spiral structure that have focused on the gaseous components.
94: \citet{Visser80a} showed that steady state density wave models fit the
95: HI kinematics of M81 quite well. \citet{Lowe94} used the modal theory
96: of density waves to describe the spiral pattern in M81. Both
97: \citet{Rand93} and \citet{Aalto99} used observed molecular velocities
98: along 1D cuts on the major and minor axes of the grand design spiral
99: M51, and found qualitative agreement with the density wave models of
100: \citet{RobertsStew87}. \citet{KunoNakai97} fitted observed CO
101: velocities from single dish observations to obtain gas streamlines.
102: The smooth shape of the velocity profiles led them to conclude that
103: galactic shocks do not exist in M51. However, the study by
104: \citet{Aalto99}, using higher resolution interferometric data, found
105: steeper velocity gradients, supporting the presence of shocks.
106:
107: Yet, other observational studies have suggested that the arms are not
108: long lived. In fact, the classic kinematic study of M51, that of
109: \citet{Tully74sp}, found evidence for a transient pattern in the outer
110: disk, due to the interaction with its companion, but that a steady
111: state is probably appropriate for the inner arms.
112: \citet{ElmSeidenElm89} and \citet{Vogel93} also suggested the presence
113: of multiple pattern speeds. \citet{HenryQG03} argued that the spiral
114: pattern may be a superposition of an $m$=2 mode and a weaker $m$=3
115: mode, suggesting a transient pattern for the arms of M51.
116:
117: This paper presents a detailed study of the gaseous velocity structure
118: associated with the spiral pattern in M51. In a future paper, we will
119: discuss and compare the spiral pattern in different tracers. Here, we
120: use the CO and \halpha\ velocities to map the 2D velocity field in
121: M51.
122:
123: Our study makes use of the full 2D velocity field in M51 from
124: interferometric CO and Fabry-Perot \halpha\ observations, rather than
125: just major and minor axis cuts. Noting that variations in the
126: observed velocity field are mainly associated with the spiral arms, we
127: fit the observed velocity field to obtain the radial and tangential
128: components as a function of arm phase (i.e. distance perpendicular to
129: the arm). We then analyze whether the fitted velocity field and
130: density maps are consistent with the predictions of steady state
131: theory.
132:
133: In the next section we briefly describe our CO and \halpha\
134: observations. In $\S$\ref{fitting} we describe the method we employ
135: to estimate the radial and tangential velocity components throughout
136: the disk. Since our method is sensitive to the assumed values of the
137: \vsys, major axis \pa, and disk \inc\ with respect to the sky, in
138: $\S$\ref{methodtest} we present results from our effort to constrain
139: these parameters, and describe how errors could affect the fitting
140: results. In $\S$\ref{results} we present and discuss the fitted
141: profiles of radial and tangential velocities for a range of radii. We
142: then use the velocity and density profiles to empirically test
143: conservation of vortensity in $\S$\ref{vort_sec}. Next, in
144: $\S$\ref{consmass}, we examine whether (quasi) steady state mass
145: conservation is applicable, as would be necessary for a QSSS
146: description. Finally, in $\S$\ref{summary}, we summarize our
147: conclusions.
148:
149: \section{Observations}
150:
151: The CO and \halpha\ intensity and velocity maps are shown in Figures
152: \ref{COarm} and \ref{Haarm}, respectively. The CO J=1-0 data for M51
153: were obtained in part from BIMA SONG (Survey of Nearby Galaxies). The
154: observations and data reduction are described in \citet{Regan01} and
155: \citet{Helfer03}. The SONG map is based on 26 pointings and has an
156: angular resolution of 5.8\arcsec\ $\times$ 5.1\arcsec. Later, we
157: obtained data for 34 additional pointings, so that the spiral arms
158: were mapped as far as the companion galaxy to the north and to a
159: similar distance along the spiral arm to the south. Additionally,
160: inner fields were mapped in a higher resolution array (B array),
161: yielding higher angular resolution. The newer data were reduced using
162: the same procedures as described in \citet{Helfer03} for BIMA SONG.
163: Together the data sets cover 60 pointings. The maps used for this
164: paper have variable resolution, reaching as high as 4\arcsec\ in the
165: inner spiral arms but degrading to 6\arcsec\ -- 13\arcsec\ in the
166: interarms and in the outer arms.
167:
168: \halpha\ data were obtained with the Maryland-Caltech Palomar
169: Fabry-Perot, which covered the optical disk at an angular resolution
170: of 2\arcsec\ and a velocity resolution of 25 \kms. The observations
171: and reduction are described in \citet{Gruendl96} and also
172: \citet{Vogel93}. Both CO and \halpha\ intensity maps are obtained by
173: fitting Gaussian profiles to the spectrum at each location. The
174: velocity maps indicate the velocity of the peak of the fit Gaussian
175: intensity.
176:
177: Also shown in Figures \ref{COarm} and \ref{Haarm} are two lines
178: tracing logarithmic spirals. The bright CO arm is well represented by
179: a logarithmic spiral with a pitch angle of 21.1\degr. The weaker arm
180: also generally follows a logarithmic spiral, although, as will be
181: discussed, a number of its arm segments either lead or lag the
182: depicted line. The logarithmic spirals will be discussed extensively
183: in the following sections.
184:
185:
186: \section{Estimation of Spiral Streaming Velocities \label {fitting}}
187:
188: The observed line of sight velocity $V_{obs}$ can be decomposed as a
189: sum of terms involving the \vsys\ $V_{sys}$, the radial velocity
190: \vrad\, and the tangential velocity \vtan:
191: \begin{equation}
192: {V_{obs}(R,\theta)=V_{sys}+[v_{R}(R,\theta)\sin(\theta-\theta_{MA})+v_{\theta}(R,\theta)\cos(\theta-\theta_{MA})]\sin i} ,
193: \label{vobs}
194: \end{equation}
195: where $R$ and $\theta$ are the galactocentric radius and azimuthal
196: angle, and $\theta_{MA}$ and $i$ are the \pa\ of the major axis and
197: \inc\ of the galaxy, respectively. This equation does not include a
198: velocity component perpendicular to the disk. The exclusion of the
199: vertical velocity component is reasonable since studies of face-on
200: grand design spirals indicate that the $z$-component of velocity is
201: less than 5 \kms \citep{KruitShostak82}, provided the disk has no
202: significant warp (we return to this issue in $\S$\ref{vsyssec}).
203:
204: Inspection of the velocity maps indicates that the isovelocity
205: contours near the spiral arms tend to run parallel to the arms. For a
206: disk in pure circular rotation and with a flat rotation curve, on the
207: other hand, the isovelocity contours of the projected velocity field
208: are purely radial. It is evident that the velocity field of M51 is
209: significantly different from this sort of simple ``spider diagram,''
210: due to the non-axisymmetric perturbations associated with spiral
211: streaming. Clearly, \vrad\ and \vtan\ vary with azimuth.
212:
213: Previous estimates of streaming velocities have used observed
214: velocities near the major axis (where the projections of \vrad\
215: vanishes) to estimate \vtan, and velocities near the minor axis (where
216: the projections of \vtan\ vanishes) to estimate \vrad\
217: \citep[e.g.,][]{Rand93}. However, much of the CO gas is organized
218: into GMCs and larger complexes known as GMAs \citep{Vogel88}.
219: Further, \citet{Aalto99} have found that the streaming velocities of
220: M51 GMAs in the same spiral arm have a significant dispersion.
221: Observations along a single cut, e.g. the major axis, sample discrete
222: GMCs and therefore may give a misleading estimate of the streaming
223: velocities. As an alternative, an approach that fits a streaming
224: profile to all the observed velocities in an annulus as a function of
225: distance from the arm peak may better characterize the streaming
226: velocities. Also, the gas surface density distribution varies
227: significantly at any distance from the peak (i.e. the gas is clumpy),
228: and so averaging parallel to the arm may better characterize the
229: variation in gas surface density as a function of arm distance.
230:
231: Typically, 2D fits to a galaxy velocity field assume that \vrad\ and
232: \vtan\ are constant along rings (e.g. the tilted ring analysis
233: described in \citet{Begeman89}). By contrast, as mentioned earlier,
234: \vrad\ and \vtan\ do vary with azimuth, and indeed inspection of
235: Figures \ref{COarm} and \ref{Haarm} indicates that the primary
236: variations are due to the flow through the spiral arms rather than
237: variations with galactocentric radius. As in most galaxies, the
238: rotation curve of M51 is relatively flat; the radial variations that
239: do occur are associated with spiral arm streaming. Thus, we are
240: motivated to assume that radial variations of azimuthally-averaged
241: quantities are negligible (at least over relatively limited radial
242: ranges) and that \vrad\ and \vtan\ vary primarily with spiral arm
243: phase $\psi$. The left panel of Figure \ref{armphase} shows the
244: relevant geometry depicting the spiral arm phase. Our assumption is
245: that \vrad\ and \vtan\ are constant along narrow spiral arcs, such as
246: the segments in Figure \ref{armphase}, that are congruent to the
247: spiral arms. Thus, we rewrite equation (\ref{vobs}) (for a limited
248: range of radii) as
249: \begin{equation}
250: {V_{obs}=V_{sys}+[v_R(\psi)\sin(\theta-\theta_{MA})+v_\theta(\psi)\cos(\theta-\theta_{MA})]\sin i}.
251: \label{newvobs}
252: \end{equation}
253:
254: In order to simplify the process of identifying regions of constant
255: arm phase, we adopt a coordinate system in which the spiral arms are
256: straight. \citet{ElmSeidenElm89} show that the spiral arms of M51
257: appear as straight line segments in a $(\theta, log(R))$, or
258: logarithmic polar, coordinate system. The right panel of Figure
259: \ref{armphase} shows the logarithmic polar diagram corresponding to
260: the features in the left panel. Figure \ref{pgCO} shows the CO
261: intensity and velocity maps of M51 in log-polar coordinates, and
262: Figure \ref{pgHa} shows the corresponding H$\alpha$ maps.
263:
264: The sky images in Figures \ref{COarm} and \ref{Haarm} are first
265: deprojected before being transformed into a $(\theta, log(R))$
266: coordinate system. In order to deproject the sky view of a galaxy,
267: the center position, \pa, and \inc\ are required. We initially use
268: the canonical values for these parameters, which are listed in Table
269: \ref{paramtab}. We discuss the estimation of these parameters in the
270: next section.
271:
272: The two straight lines overlaid on Figures \ref{pgCO} and \ref{pgHa}
273: indicate the adopted pitch angle of 21.1\degr\ and also correspond to
274: the spiral loci shown in Figure \ref{COarm}. It is clear from the
275: overlaid lines, which are separated by 180\degr, that the weaker arm
276: is not symmetric with the brighter one, as discussed by
277: \citet{HenryQG03}. Yet, both CO arms wrap around approximately
278: 360\degr\ of the galaxy, even though they appear to jump in phase at
279: one or more positions. The arms in H$\alpha$ show more jumps in phase
280: and variations in the pitch angle. In spite of their asymmetries, the
281: CO spiral arms are particularly well described as logarithmic spirals,
282: better even than H$\alpha$, or the optical arms shown by
283: \citet{ElmSeidenElm89}.
284:
285: We will refer to overlaid logarithmic spiral arcs (or lines) as
286: ``slits,'' for we will extract observed CO and H$\alpha$ velocities as
287: a function of position along the arc (or line), similar to obtaining
288: long-slit spectra. Each slit marks a region of constant arm phase
289: $\psi$. Thus, while the observed velocity varies along the slit due
290: to projection, \vrad\ and \vtan\ are assumed constant. We arbitrarily
291: define the arm phase marked by the leftmost slit in Figures \ref{pgCO}
292: and \ref{pgHa} as $\psi=0$\degr. The other CO arm appears at an arm
293: phase of approximately $\psi$ = 180\degr; other features such as the
294: stellar arms and the gravitational potential minimum may of course be
295: offset from the CO arms.
296:
297: As noted previously, our fit will assume that the intrinsic \vrad\ and
298: \vtan\ are constant at a given arm phase, i.e. along a given slit, but
299: that \vrad\ and \vtan\ vary with $\psi$ as the slit is translated in
300: azimuth. Translating the slit amounts to shifting a straight line to
301: the right in the logarithmic polar diagram; this direction of
302: increasing azimuth is the same as the direction of rotation for M51.
303: We then fit equation (\ref{newvobs}) to the observed velocities
304: extracted at each arm phase $\psi$, thereby obtaining \vrad$(\psi)$
305: and \vtan$(\psi)$.
306:
307: Although \vrad\ and \vtan\ vary primarily with arm phase, they may of
308: course also vary with radius. Therefore, we limit the radial range of
309: an annulus (or equivalently the length of a slit) as much as possible
310: while still fitting a sufficiently extended azimuth range to obtain
311: good leverage on both \vrad\ and \vtan. In other words, an annulus
312: should be sufficiently broad to cover the spiral arm both near the
313: major axis and the minor axis; the width of the annulus thus
314: depends on the pitch angle and galactocentric radius of the arm.
315:
316: We first test our method by applying it to a model spiral galaxy with
317: known radial and tangential velocities. The solid lines in Figure
318: \ref{modvfit} shows the averaged density, \vrad, and \vtan\ profiles
319: in an annulus from a snapshot of a hydrodynamical simulation of a disk
320: responding to a spiral perturbation. The model spiral galaxy is a 2D
321: version of a 3D model described in detail in \citet{GC02}, and the
322: annulus used here extends from 8.38 - 8.92 kpc. The direction of gas
323: flow is in the direction of increasing phase. As the gas approaches
324: the arm, (marked by density maxima) the radial velocity, \vrad,
325: decreases by $\sim$40 \kms. The sign reversal of \vrad\ indicates
326: that the gas is moving away from the nucleus before the shock and
327: towards the nucleus after the shock. As the gas emerges from the arm,
328: the radial velocity increases again. The tangential velocity \vtan\
329: gradually decreases as the gas approaches the arm, then receives a
330: strong boost and reaches a maximum just downstream from the arm.
331:
332: In order to test the fitting algorithm, the model \vrad\ and \vtan\ at
333: all locations are used in equation (\ref{vobs}) to create a model
334: observed velocity field. This velocity field, along with the model
335: density map, are transformed into logarithmic polar projections.
336: Equation (\ref{newvobs}) is then fit to the model observed velocities
337: at each arm phase in an annulus, using slits parallel to the spiral
338: arms; the dashed lines in Figure \ref{modvfit} are the results of the
339: velocity fits, in the same annulus (8.38 - 8.92 kpc). The results
340: reproduce the overall shape of the velocity profiles quite well,
341: although with slight phase shifts and offsets. The offsets and the
342: shallower minimum in \vrad\ are likely due to the variation of the
343: pitch angle of the arms with radius; i.e. the spiral arms are not
344: perfectly logarithmic, whereas the ``slit'' used to extract velocities
345: at constant arm phase is. Despite these offsets, we were able to
346: reproduce the major features of the velocity profiles of the model
347: spiral galaxy, indicating that our method of fitting observed
348: velocities at each arm phase recovers a 2D velocity field reasonably
349: accurately.
350:
351: We now apply our fitting method to the M51 data, adopting systematic
352: parameters listed in Table \ref{paramtab}. As an example, Figure
353: \ref{COvrvt} shows the \vrad\ (top) and \vtan\ (bottom) fits to the
354: observed CO and \halpha\ velocity field for one annulus between the
355: galactocentric radii of 21\arcsec\ and 36\arcsec. The CO intensity
356: averaged along a slit as a function of phase angle $\psi$ is also
357: shown as dashed lines, indicating the distribution of molecular gas.
358: As mentioned, $\psi = 0$\degr\ is arbitrary and is marked by the
359: leftmost line in Figures \ref{pgCO} and \ref{pgHa}, corresponding to
360: the brighter arm, which we will refer to as Arm 1. We show a phase
361: range greater than 360\degr\ so that both upstream and downstream
362: velocities can easily be seen for both arms. The direction of gas
363: flow through the arms (assuming we are inside corotation) is from left
364: to right, so that the right sides of the CO peaks correspond to the
365: downstream side of the arm. In most cases, as the gas flows through
366: the arm, the radial velocity decreases and then increases, and the
367: tangential velocity receives a boost, as predicted qualitatively by
368: density wave theory.
369:
370: In a conventional tilted-ring velocity fitting analysis, galaxy
371: parameters such as the \inc, \pa, dynamical center, and \vsys\ can be
372: directly fit. However, even though the \inc\ and \pa\ appear
373: explicitly in equation (\ref{newvobs}), for our fits all but the
374: \vsys\ must be assumed prior to deprojecting a galaxy velocity field
375: and therefore before the fit. For our initial fits we employed the
376: standard assumed values for these parameters for M51, shown in Table
377: \ref{paramtab}. In the next section we explore the effects of errors
378: in these assumed global parameters on the estimation of \vrad\ and
379: \vtan.
380:
381: \section{Method Testing and Parameter Constraints \label
382: {methodtest}}
383:
384: To test the sensitivity of the \vrad\ and \vtan\ fits to errors in the
385: global (fixed) parameters, we generated test velocity fields and
386: created sky projections with known parameters. We then applied our
387: fitting technique to estimate \vrad\ and \vtan\ for the model
388: galaxies, assuming incorrect values for the fixed parameters, and
389: compared the fitting results to the actual model values of \vrad\ and
390: \vtan. This enables us to quantify the sensitivity of the fits to the
391: fixed parameters. In addition to constraints obtained from fitting
392: our kinematic data, we also use standard methods to constrain the
393: values of $V_{sys}$, $\theta_{MA}$, and $i$. As we shall show, one of
394: our conclusions is that some of the basic parameters for M51, many of
395: which date to \citet{Tully74sp}, may in fact be poorly constrained due
396: to the morphological and kinematic perturbations induced by the tidal
397: interaction with its companion.
398:
399: As an initial test, we generated a simple model with \vtan\ = 240
400: \kms\ and \vrad\ = $-$35, i.e. an axisymmetric disk with a flat
401: rotation curve and uniform radial inflow. We refer to this model as
402: the ``constant velocity'' model. We apply our general method to fit
403: \vrad\ and \vtan\ as a function of arm phase using ``observed''
404: velocities. If we assume the input values of $V_{sys}$ = 464 \kms,
405: $\theta_{MA}$ = 170\degr, and $i$ = 20\degr, we indeed recover the
406: input values for \vrad\ and \vtan\ as independent of phase. We now
407: consider the effects of assuming incorrect values for the parameters.
408:
409: \subsection{Position of Dynamical Center \label{censec}}
410:
411: In testing the sensitivity of the fits to the assumed center position,
412: we applied the fitting algorithm to a model for which the center
413: position was shifted by 1\arcsec\ in both RA and DEC. We found that a
414: 1\arcsec\ error in the assumed center has a negligible effect on the
415: fit velocities. BIMA observations have an astrometric accuracy of
416: $\sim$10\% of the synthesized beam. The highest resolution of our CO
417: observations is 4.5\arcsec, so the error in position will likely not
418: be greater than $\sim$0.5\arcsec. Thus, observational errors will
419: likely not affect the results of our fits. For all the analysis that
420: follows, we will adopt the center position listed in Table
421: \ref{paramtab}. This choice assumes that the dynamical center
422: coincides with the location of a weak AGN known to exist in the
423: nucleus of M51 \citep{Hoetal87, NakaiKasuga88}. We use the position
424: of the radio continuum source observed with the VLA, which has an
425: accuracy of $\pm$0$\farcs$01 \citep{Hagiwara01}.
426:
427: \subsection{Systematic Velocity \label{vsyssec}}
428:
429: Before discussing methods for determining $V_{sys}$, we first explore
430: the effect that an error in an assumed value of $V_{sys}$ would have
431: on fits for \vrad\ and \vtan\ in which $V_{sys}$ is held fixed, using
432: the constant velocity model. As expected, an error $\Delta V_{sys}$
433: in the assumed $V_{sys}$ produces a sinusoidal variation in both
434: fitted velocity components, with an amplitude of $\Delta V_{sys}/\sin
435: i$, and a period of 360\degr (see eqn. \ref{vobs}). Clearly,
436: $V_{sys}$ needs to be well determined.
437:
438: One approach to obtaining $V_{sys}$ is to fit the data for its value
439: using equation (\ref{newvobs}). Figure \ref{M51vsys} shows the
440: results of fits to the M51 data in which $V_{sys}$ was fit, along with
441: \vrad\ and \vtan, as a function of arm phase. Although $V_{sys}$
442: should be constant, it can be seen that the fit value of $V_{sys}$
443: varies with arm phase. Similar variations in fitted $V_{sys}$ result
444: regardless of what values of the \pa\ and \inc\ are assumed.
445:
446: One possible explanation for the apparent variation of $V_{sys}$ is
447: that the galactic disk of M51 is twisted and/or warped, i.e. the \pa\
448: and/or \inc\ may vary with radius. We therefore compare the results
449: of fitting for $V_{sys}$ from a model galaxy with no warp to a model
450: with a warp. We again make use of the constant velocity model, for
451: this model also represents an unwarped disk. Instead of keeping
452: $V_{sys}$ fixed, we allowed this parameter to be free in the fit. If
453: we use the true \pa\ and \inc, we correctly recover the adopted values
454: for all three free parameters, $V_{sys}$, \vrad, and \vtan.
455:
456: To generate a warp model, we increase $i$ monotonically from 25\degr\
457: at the inner radius (100\arcsec) to 35\degr\ at the outer radius
458: (200\arcsec); the \pa\ is kept fixed. If we allow $V_{sys}$ to be
459: free in fitting the model data, Figure \ref{warp} shows that the
460: $V_{sys}$ varies almost sinusoidally about the true model value 400
461: \kms. The mean fitted \vsys\ is equal to this value. This is the
462: case regardless of what values of \inc\ or \pa\ we use (within
463: reasonable limits), and regardless of the radius range of the annulus
464: used for fitting. Thus, regardless of the assumed fixed parameters
465: and of the limits in radius, for a simple warp the mean value of the
466: fits gives the correct \vsys.
467:
468: Motivated by our finding that even with a warp the average fit value
469: of $V_{sys}$ gives the true value, we calculated the mean of the
470: $V_{sys}$ values shown in Figure \ref{M51vsys}, obtaining $V_{sys}$ =
471: 470.6 \kms. Comparison of the $V_{sys}$ fits of the actual M51 data
472: (Figure \ref{M51vsys}) with $V_{sys}$ fits to the simple warp model
473: (Figure \ref{warp}) shows that the warp model has a slower variation.
474: Hence, if a warp is responsible for producing variations in the fitted
475: \vsys\, it must be more complex than our simple model; we return to
476: this question in $\S$\ref{consmass}.
477:
478: We therefore apply two additional methods to estimate the value of
479: $V_{sys}$. The first method is based on a standard tilted-ring
480: analysis \citep{Begeman89}, in which the galactic disk is represented
481: as a series of nested tilted rings. In its most general form, each
482: tilted ring may have a different center, \vsys, \pa, \inc, and
483: rotational velocity. We use 10\arcsec\ rings from an inner radius of
484: 20\arcsec\ to an outer radius of 120\arcsec, fixing the center
485: position, \inc, and \pa\ to the values in Table \ref{paramtab}. We
486: obtain mean \vsys\ of 471.4 $\pm$ 0.5 \kms. When we allow the \pa\
487: to vary as well, we obtain a mean of 471.3 $\pm$ 0.3 \kms.
488:
489: A third method we use to constrain $V_{sys}$ is to assume a functional
490: form for the rotation curve, using the {\tt NEMO} program {\tt
491: rotcurshape} \citep{Teuben95}. In contrast to the tilted-ring
492: method, which fits each ring independently, {\tt rotcurshape} fits
493: $V_{sys}$, $\theta_{MA}$, $i$, center position $\alpha$ and $\delta$,
494: and the coefficients of the function used to describe the rotation
495: curve simultaneously to the entire velocity field. Therefore, it can
496: yield a single $V_{sys}$ that best fits the entire velocity field. It
497: is particularly useful for finding $V_{sys}$ if the kinematic center
498: position can be fixed. For this fit, we limit the {\tt rotcurshape}
499: fit to the inner 20\arcsec\ in radius. This is inside the main spiral
500: arms, in the region where the rotation curve is rising and the
501: isovelocity contours are relatively straight. We assume the center,
502: $\theta_{MA}$, and $i$ listed in Table \ref{paramtab} and a rotation
503: curve of the form $v = V_{sys} + v_o x/(1+x)$ where $x$ is the ratio
504: of the radius to the core radius, and fit for $V_{sys}$, $v_o$, and
505: the core radius. We obtain $V_{sys}$ = 473 $\pm$ 0.5 \kms. If we
506: allow the \pa\ to vary as well, we obtain 473.2 $\pm$ 0.3 \kms. It is
507: encouraging that this is within 2 \kms of $V_{sys}$ determined from
508: the other two methods even though this fit uses a different method and
509: fits an entirely different region (i.e. the inner 20\arcsec, inside the
510: main CO arms, as opposed to outside 20\arcsec).
511:
512: The canonical value of $V_{sys}$ for M51 is 472 $\pm$ 3 \kms\
513: \citep[and references therein]{Tully74sp}. Table \ref{vsystab} lists
514: Tully's value for $V_{sys}$ as well as the results from applying the
515: three techniques described above. Our different methods give a mean
516: \vsys\ of 471.7 $\pm$ 0.3 \kms. Henceforth, we will fix $V_{sys}$ to
517: be 472 \kms\ (LSR, corresponding to a heliocentric velocity of 464
518: \kms) in fitting the velocity field to estimate $v_R(\psi)$ and
519: $v_\theta(\psi)$.
520:
521: \subsection{Position Angle \label{pasec}}
522:
523: To investigate the effect of errors in the assumed galaxy \pa\ on the
524: fitted values of \vrad\ and \vtan, we first use the aforementioned
525: constant velocity model. Note that the \pa\ is required to deproject
526: the galaxy image, as well as in equation (\ref{newvobs}). Using
527: incorrect \pa s, but correct model values of \vsys\ and \inc, yields a
528: greater effect on \vrad\ than on \vtan. This is because an error
529: $\Delta\theta_{MA} \ll 1 $ in $\theta_{MA}$ results in an error
530: $\approx -v_{\theta}\sin\Delta\theta_{MA}$ in the fitted \vrad,
531: whereas the corresponding error in \vtan\ is $\approx
532: +v_R\sin\Delta\theta_{MA}$. Since \vtan\ is large compared to \vrad,
533: the shift in \vrad\ is larger than the shift in \vtan. Thus , a
534: $\pm$10\degr\ error in \pa\ in the ``constant velocity'' model which
535: has \vtan\ = 240 \kms\ produces approximately a $\mp$40 \kms\ shift in
536: the fitted radial velocity. Position angle errors also produce small
537: perturbations in both velocity components. We conclude that using an
538: accurate value of $\theta_{MA}$ is very important to obtain accurate
539: \vtan\ and especially \vrad\ fits.
540:
541: Unfortunately, the \pa\ of the major axis of M51 is particularly
542: difficult to determine. The strong spiral arms and the tidal
543: interaction with NGC 5195 distort the stellar disk, making it
544: effectively impossible to determine a \pa\ from the orientation of the
545: isophotes. Thus, it is necessary to go beyond morphology in
546: determining $\theta_{MA}$. We therefore revisit the determination of
547: the galaxy's position angle. We apply the method of \citet{Tully74sp}
548: to our velocity data from H$\alpha$ and CO observations. We also
549: study the effect of streaming motions on this method by using model
550: galaxies with known position angles and streaming velocities. In
551: addition, we also apply two alternate methods to derive the position
552: angle.
553:
554: The widely used value for the position angle of M51, 170\degr , was
555: determined by \citet{Tully74sp} using kinematic information. Tully
556: assumed that the observed velocity should reach its extreme value at
557: the \pa\ of the galaxy major axis, $\theta_{MA}$. To determine
558: $\theta_{MA}$, he averaged the observed velocities in wedges extending
559: over 5\degr\ in azimuth, and then for each radius took the \pa\ of the
560: wedge with the extreme velocity as the estimated major axis at that
561: radius. Tully excluded radii at which he was not confident that the
562: true major axis had measured velocities (e.g. the faint interarm
563: regions near the major axis).
564:
565: Figure \ref{wedges} shows the results of applying Tully's position
566: angle determination method to H$\alpha$ and CO observations. For each
567: annulus of radial extent 5\arcsec, the position of the wedge with the
568: extreme velocity is marked. Due to the lack of data in the outer
569: regions of the CO observations, only the wedges in the inner
570: 70\arcsec\ provide reliable measures of the \pa\ of the extreme
571: velocity. Similar to Tully, we did not attempt to estimate the \pa\
572: of the extreme velocity at radii for which data are sparse in the
573: range of plausible \pa s of the major axis. From the location of
574: these extreme velocity wedges, we found $\theta_{MA}$ to be 172\degr\
575: (from an error weighted average) from both CO and \halpha\
576: observations.
577:
578: However, streaming motions can shift the velocities, resulting in the
579: extreme velocity occurring at position angles not corresponding to the
580: true major axis. Indeed, inspection of positions of the velocity
581: extremum wedges overlaid on the intensity maps shows that the \pa s of
582: the wedges in the interarms are clearly shifted counter-clockwise from
583: those in the arms. This is most evident in the H$\alpha$ maps, for
584: which emission is detected from almost everywhere in the disk. We
585: further explore streaming effects on the Tully method using a model
586: with known streaming motions, generated using one of our \vrad\ and
587: \vtan\ fits to the M51 CO data, with $\theta_{MA}$ = 170\degr. (Since
588: this test is designed simply to reveal the twists in the apparent \pa\
589: due to streaming, the particular value assumed for the true \pa\ and
590: the particular \vrad\ and \vtan\ fits used is not significant.)
591: Figure \ref{modwedges} shows the results of applying Tully's method to
592: this streaming model galaxy. In any given annulus, the extreme
593: velocity averaged in the 5\degr\ wedges occurs in the interarm
594: regions. In Tully's analysis, however, only spiral arm regions (near
595: the apparent major axis) were considered, due to observational
596: limitations. Therefore, the major axis position angle he found is
597: likely biased clockwise from the true major axis. As shown in Figure
598: \ref{modwedges}, even if the interarm regions are considered, the \pa
599: s of the locations of the extreme velocities do not necessarily
600: correspond to the major axis. Thus, such an extrema method can be
601: biased due to the inherent streaming in M51, regardless of whether the
602: arms or interarms are considered.
603:
604: We employ two alternate position angle determination methods that make
605: use of the full observed velocity field. In the first method, we
606: average the observed velocity at each position angle in a wedge for
607: both the northern and southern sides of the galaxy; then we fit a
608: cosine curve to these averaged velocities as a function of azimuthal
609: angle. We will refer to this method as the ``radial-averaged''
610: method. This is most easily accomplished in the polar projection,
611: where we can average along a column to perform the radial average.
612: The radial-averaged velocity as a function of \pa\ is shown in Figure
613: \ref{radslits}, along with the corresponding cosine fitted curves. We
614: assume that the galaxy major axis should be at the \pa\ of the extrema
615: of such curves. The mean position angle determined from the H$\alpha$
616: and CO fits is $\sim$177\degr, larger than the position angle
617: determined by the Tully method. Again, the position angle determined
618: in such a way is sensitive to streaming. Earlier we showed that
619: streaming tends to cause the position angle of extreme interarm
620: velocities to be biased counter-clockwise from the true value. Since
621: the interarms occupy a greater fraction of the galaxy compared to the
622: arms, streaming will introduce a counter-clockwise bias to the apparent
623: position angle of the major axis. The effect of streaming on this
624: method is further discussed below, following a discussion of our
625: second position angle determination method.
626:
627: Our second method to determine the \pa, the ``azimuthal fit'' method,
628: is similar to the one described in the previous paragraph, but instead
629: the cosine curve is fit to the observed velocities along a projected
630: circle with constant (projected) radius. As in the previous method,
631: the polar projection of the velocity field is useful; in this case we
632: simply fit a cosine curve to the velocities along a row of constant
633: projected radius. The results from applying this method are shown in
634: Figure \ref{circslits}. Note that the \pa\ of the velocity extrema
635: varies systematically as a function of radius; it is approximately
636: 180\degr\ in the inner region 30\arcsec\ from the center, declining to
637: 165\degr\ 120\arcsec\ from the center; this trend, including the rise
638: near Log($R$) = 1.7, is also evident from simple inspection of Figures
639: \ref{pgCO} and \ref{pgHa}. Averaging over the radius range displayed
640: in Figure \ref{circslits}, we obtain the same \pa\ of $\sim$177\degr\
641: as in the previous method. Again, the velocities in the interarms
642: bias this determination of position angle, for the same reasons stated
643: in the previous paragraph.
644:
645: In order to understand the effect of streaming motions on the \pa\ of
646: the major axis derived using the radial averaged and azimuthally fit
647: methods, we apply these methods to streaming models with known \pa s
648: and streaming velocities. The model velocity fields are produced from
649: our \vrad\ and \vtan\ profiles obtained by assuming fixed values of
650: $\theta_{MA}$; we then apply the radial averaged and azimuthally fit
651: methods to these model velocity fields. Both methods recover a
652: $\theta_{MA}$ of $\sim$176\degr, for all models, similar to the actual
653: M51 velocity field, even though the \pa s assumed in generating the
654: streaming models can be very different. This is because for different
655: \pa s, the streaming velocities were derived from fits designed to
656: best match the observed velocities. So in fact all the streaming
657: models give virtually identical observed velocity fields regardless of
658: assumed \pa.
659:
660: However, if we recreate the models setting the radial component to be
661: zero everywhere, then we correctly recover the assumed position
662: angles. This is clear evidence that radial streaming affects methods
663: to determine $\theta_{MA}$, not only near the minor axis, as recognized
664: by Tully, but also elsewhere including even the major axis.
665:
666: In order to quantify the effect of the non-zero radial velocities on
667: the apparent position angle, we apply the position angle determination
668: methods to models with known constant radial velocities. In these
669: artificial models tangential streaming velocities are assumed to vary
670: with arm phase, but radial streaming is assumed constant. We found
671: that for every $\pm$10 \kms\ in radial velocity, the derived position
672: angle differs from the actual position angle by $\pm$ 3\degr. This
673: degeneracy between the position angle and radial velocity renders it
674: difficult to accurately identify the true position angle, or to map
675: the radial velocity. In order to accurately determine the position
676: angle, we need to know the radial streaming. But in our effort to map
677: the radial and tangential velocities of M51, we need to know the
678: position angle. Thus, as we carry out our investigation, we shall use
679: a range of position angles in deriving the two dimensional velocity
680: components of M51.
681:
682: \subsection{Inclination \label{incsec}}
683:
684: Estimating the inclination based on the orientation of the isophotes
685: is unreliable, as discussed in $\S$\ref{pasec}, due to the strong
686: perturbations from the spiral arms and the tidal interaction. In
687: principle, the inclination can be determined from a fit to the
688: velocity field, as we did to obtain the systematic velocity in
689: $\S$\ref{vsyssec}. However, the fit inclination is not well
690: determined by the available data, presumably due to the streaming. To
691: test the sensitivity of the \vrad\ and \vtan\ fits to the inclination,
692: $i$, in equation (\ref{newvobs}), we assume incorrect values of the
693: \inc\ which differ from the true value by $\Delta i$ for the constant
694: velocity model (where \vrad\ and \vtan\ are constant). We then fit
695: for \vrad\ and \vtan. As expected, an error $\Delta i$ results in
696: errors in the fit value of \vrad\ and \vtan\ with magnitudes $\propto
697: \sin\Delta i$, along with small perturbations about this offset.
698:
699: Since we find that error introduced in the velocity components due to
700: an incorrect \inc\ can be large (although just a simple scaling), we
701: sought other constraints on the inclination. In particular, the
702: Tully-Fisher \citep{TullyFisher77} relation can be used to estimate
703: the inclination. The well known Tully-Fisher relation is a
704: correlation between galaxy luminosity and maximum rotation speed. The
705: inclination can be estimated by comparing the rotational velocity
706: predicted by the Tully-Fisher relation with the observed velocity of
707: the flat part of the rotation curve. We use the baryonic form of the
708: Tully-Fisher relation discussed by \citet{McGaugh05} \citep[see
709: also][]{McGaugh00}:
710: \begin{equation}
711: {M_b=50\,V_c^4},
712: \label{Stacy}
713: \end{equation}
714: where $M_b$ is the baryonic mass (in $M_\odot$), and $V_c$ is the
715: circular rotational velocity (in \kms). Since the dispersion in the
716: Tully-Fisher relation, $L \propto V_c^4$, is relatively small, given
717: the luminosity the uncertainty in $V_c$ is small.
718:
719: In order to determine the baryonic mass $M_b$, we require the stellar
720: mass $M_*$, which is related to the $B$-band luminosity $L_B$ and the
721: $B$-band mass to light ratio $(M/L_B)$,
722: \begin{equation}
723: {M_{*}=L_B\cdot(M/L_B)}.
724: \label{Lb}
725: \end{equation}
726:
727: We use the correlation of the galaxy color with $M/L$ discussed by
728: \citet{BelldeJong01} applicable to the \citet{CharlotBruzual91}
729: population synthesis models, to obtain $(M/L_B)$:
730: \begin{equation}
731: {(M/L_B)=10^{[-0.63+1.54(B-V)]}}.
732: \label{BC}
733: \end{equation}
734: The RC3 catalog \citep{RC3} gives $(B-V)=0.53$ for M51, so
735: $(M/L_B)$=1.54.
736:
737: The last quantity required to determine $M_*$ is the luminosity $L_B$,
738: which can be derived if we know the distance. Two independent studies
739: have given similar M51 distance estimates: observation of planetary
740: nebulae gives a distance modulus of $m-M$=29.62 $\pm$ 0.15
741: \citep{Feldmeier97}, and a study of surface brightness fluctuations in
742: the companion NGC 5195 gives $m-M$=29.59 $\pm$ 0.15 \citep{Jensen96}.
743: We thus employ a distance modulus of $m-M$= 29.6, corresponding to a
744: distance of $8.4 \pm\ 0.6$ Mpc. Using the RC3 catalog value of
745: $B=8.67$, corrected for extinction,
746: \begin{equation}
747: {L_B=10^{-0.4(8.67-29.6-5.48)}L_\odot = 3.66\times10^{10} L_\odot}.
748: \label{lum}
749: \end{equation}
750: Using this value and the $(M/L_B)$ value of 1.54 (equation [\ref{BC}])
751: in equation (\ref{Lb}), we obtain
752: \begin{equation}
753: {M_*=5.64\times10^{10} M_\odot}.
754: \end{equation}
755:
756: We can now apply the \citet{McGaugh05} relation in equation
757: (\ref{Stacy}) to obtain the circular velocity:
758: \begin{equation}
759: {V_c=[(M_* + M_{gas})/50]^{1/4}=188\,\,{\rm km\, s}^{-1}},
760: \label{Vc}
761: \end{equation}
762: where $M_{gas}$ is the total gas mass; in the case of M51 the gas is
763: predominantly molecular. From our CO observations, we compute
764: $M_{gas} = 5.4\times10^{9} M_\odot$, using an X-factor of of $2\times
765: 10^{20}$ cm$^{-2}$ [K \kms]$^{-1}$ \citep[e.g.][]{Strong88}. Due to
766: the small value of the exponent in equation \ref{Vc}, errors in the
767: mass, due to variations in the X-factor, for example, will not
768: significantly affect the resulting rotational velocity.
769:
770: The observed velocity is related to the circular velocity by
771: \begin{equation}
772: {V_{c,obs}=V_c\,\sin i}.
773: \label{sin}
774: \end{equation}
775: Adopting the center, $\theta_{MA}$, and $V_{sys}$ described in this
776: section, we apply a tilted-ring analysis to determine the flat part of
777: the rotation curve. We obtain an observed circular velocity between
778: 70 and 80 km s$^{-1}$, implying
779: \begin{equation}
780: {22^\circ\,\,\lesssim\,\,i\,\,\lesssim\,\,25^\circ}
781: \label{ninc}
782: \end{equation}
783: Therefore, for our subsequent fits, we adopt an inclination of
784: 24$^\circ$.\footnote{Employing the standard Tully-Fisher relation
785: instead, we obtain a mean inclination of $\sim$23\degr, using the slope
786: and zero-point fits from \citet{Verheijen01}.}
787:
788: \subsection{Summary: System Parameter Values}
789:
790: In summary, we have shown that the fit values of \vrad\ and \vtan\ are
791: sensitive to the assumed values for the fixed parameters in equation
792: (\ref{newvobs}), $V_{sys}$, $\theta_{MA}$, and $i$. Uncertainties in
793: the assumed position of the dynamical center are too small to
794: significantly affect the derived streaming velocities. We have used
795: three different methods to determine $V_{sys}$, which resulted in a
796: value similar to the $V_{sys}$ found by Tully. We have found it to be
797: extremely difficult to constrain the value of the \pa\ of the major
798: axis using the velocity field, due to the significant streaming that
799: shifts the position angle of the extreme velocities. As a result, in
800: fitting for \vrad\ and \vtan, we allow for a range of plausible \pa s.
801: Lastly, we will adopt an \inc\ of 24\degr, which is determined by
802: using the baryonic Tully-Fisher relation \citep{McGaugh00} between the
803: baryonic mass and rotational velocity. To estimate \vtan\ and \vrad,
804: we thus use the center position and systematic velocity listed in
805: Table \ref{paramtab}, but use a range of \pa s and an \inc\ of
806: 24\degr.
807:
808: \section{Results: Velocity Profile Fits \label{results}}
809:
810: With our improved estimates of the global parameters, we apply the
811: fitting algorithm to the observed velocity field in different annuli
812: to determine the radial and tangential velocities \vrad\ and \vtan\ as
813: a function of arm phase $\psi$. We initially adopt a \pa\
814: $\theta_{MA}$ of 170\degr. We address the issue of a varying
815: $\theta_{MA}$ in $\S$\ref{consmass}. Figures \ref{COvans} and
816: \ref{Havans} show the CO and H$\alpha$ \vrad\ and \vtan\ fits in 6
817: overlapping annuli between a galactocentric radii of 21\arcsec\ and
818: 105\arcsec, and Figure \ref{anreg} shows the corresponding overlapping
819: annular regions.
820:
821: An initial inspection of the streaming profiles indicates that the
822: velocity structure is rather complex. Models of density wave
823: streaming qualitatively predict that as gas encounters the arm, \vrad,
824: which was positive (i.e. outward) in the interarm, becomes negative,
825: and as the gas exits the arm again becomes positive. The azimuthal
826: velocity \vtan\ is predicted to increase rapidly as gas flows through
827: the arm, and then decline more gradually in the interarm
828: \citep[e.g.][and $\S$\ref{fitting}]{RobertsStew87}. First we
829: concentrate on Arm 1 (the brighter arm, shown at $\psi$ = 0\degr\ and
830: more fully at 360\degr). For \vrad\ there is a pronounced minimum
831: close to the arm position, seen in both CO and H$\alpha$. There is a
832: boost in \vtan\ through the arm, again seen in both CO and H$\alpha$.
833: For this arm, the streaming is qualitatively as expected from
834: steady-state spiral shock models. The velocities associated with Arm
835: 2 (located at $\psi$ $\sim$ 200\degr) however do not agree with simple
836: predictions. For \vrad, a clear minimum is only apparent in the outer
837: annuli, and the boost in \vtan\ is weak or nonexistent. In the
838: interarms, the structure appears somewhat more complex than the simple
839: model expectation of a relatively constant or slowly rising \vrad\ and
840: a slowly declining \vtan. We suggest that Arm 1 matches simple theory
841: because its structure is simple, i.e. well described as a log spiral
842: of constant phase. By contrast, for Arm 2 the CO distribution is not
843: as well described by a single log spiral segment. Instead, it has
844: several segments with different pitch angles and jumps in phase; thus
845: the velocities associated with this arm is complex.
846:
847: One explanation for the differences in the two arms, as discussed by
848: \citet{RixRieke93}, is that the spiral pattern in M51 is actually a
849: superposition of a strong spiral mode with a $m$=2 Fourier component
850: with weaker $m$=1 and $m$=3 components. \citet{HenryQG03}, using the
851: spatial distribution of CO emission obtained from the BIMA CO map,
852: found such a scenario to be feasible by explaining the bright arm as
853: the result of constructive interference between the $m$=2 and $m$=3
854: components, and the weak arm the result of a destructive interference
855: between the two components. There is evidence for interarm structure
856: possibly supporting such a multiple density wave component description
857: of the spiral arms, which would be expected to manifest itself in the
858: kinematics. Recent Spitzer observations of M51 clearly show spiral
859: structure between the main CO arms; the Spitzer image and interarm
860: features are discussed in the next section.
861:
862: There are also clear differences between different annuli. For
863: example, in the 36\arcsec\ - 61\arcsec\ annuli, the \vrad\ increase
864: downstream from the weaker arm is much more pronounced than in the
865: 21\arcsec\ - 36\arcsec\ annuli. In addition, there is a \vrad\
866: decrease to as low as $< -$50 \kms\ in the arms of the outer regions,
867: which perhaps can be attributed to an incorrect choice of a fixed \pa\
868: for the disk (see $\S$\ref{pasec}). There are also clear differences
869: in the \vtan\ gradients between different annuli.
870:
871: By and large, similar velocity structure is apparent in both CO and
872: \halpha. For example, in the 47\arcsec\ to 80\arcsec\ annulus, the
873: gradual rise in \vrad\ from -50 \kms\ at $\psi$=180\degr\ to 70 \kms\
874: at $\psi$=300\degr\ is shown in both tracers. Further, there is a
875: strong \vrad\ peak at $\psi$=120\degr\ in the 27\arcsec\ to 47\arcsec\
876: annulus in both CO and \halpha; however, such pronounced local extrema
877: in the interarms are not expected in the theory for a single spiral
878: mode. In general, the overall amplitude of the streaming and the
879: location of most features coincide, and regions in which the velocity
880: structure is somewhat different tend to be interarm regions where
881: little CO is detected.
882:
883: Such similarities are not unexpected due to the dynamical coupling
884: between the different components. CO, which traces the molecular
885: component, is dynamically cold, with a velocity dispersion of only 4 -
886: 8 \kms. Thus, the molecular component of the disk reacts strongly to
887: any perturbation, as evident in the strong \vrad\ and \vtan\ gradients
888: associated with the spiral arms. The spiral arms compress the gas,
889: triggering star formation. The newly formed hot O and B stars
890: subsequently ionize the surrounding gas, resulting in \halpha\
891: emission. Due to the fact that much of the \halpha\ emission comes
892: from gas {\it near} the region of birth, observed \halpha\ velocities
893: will be similar to observed CO velocities. There may also be a
894: diffuse ionized medium not closely associated with the O and B stars,
895: and this medium is likely not dynamically coupled with the molecular
896: gas. However, as can be seen in Figures \ref{pgCO} and \ref{pgHa},
897: the brightest regions of \halpha\ emission occur just downstream from
898: the molecular spiral arms. Further, the generally good agreement
899: between the velocity measurements from CO and \halpha\ observations
900: suggest that most of the ionized emission originates in gas associated
901: with star forming regions. This similarity in velocity structure
902: derived from independent observations also gives confidence that the
903: fitted velocities are reliable and that the deviations from simple
904: theory, including interarm features, are real.
905:
906: The profiles in Figures \ref{COvans} and \ref{Havans} qualitatively
907: agree with previous studies of streaming in M51 involving 1D cuts
908: along the major and minor axes \citep[e.g.][]{Rand93, Aalto99}. The
909: gradients of the velocity profiles through the arms in different
910: annuli is in accordance with the conclusion of \citet{Aalto99}
911: supporting the presence of shocks in the arms from a qualitative
912: comparison of velocities along 1D cuts to streaming models of
913: \citet{RobertsStew87}. In $\S$\ref{analys} we analyze the feasibility
914: of a steady or quasi-steady spiral pattern in M51, which has been a
915: working hypothesis for many analyses of the spiral arms of this
916: galaxy.
917:
918: \subsection{Interarm Structure}
919:
920: In estimating the radial and tangential velocity components, we fit
921: observed velocities along log-spiral segments. The slope of the slit
922: is determined by the slope of the main CO arms on the logarithmic
923: polar projection, i.e. the pitch angle of the arms. Though the slope
924: of the CO arms, or at least the bright arm, is well defined, that
925: slope may not be appropriate for the interarms. In other words,
926: velocity may not be constant along the interarm log-spiral segments
927: congruent to the main CO arms.
928:
929: The recent Spitzer 8 $\mu$m image of M51 \citep{Calzetti05,
930: Kennicutt03}, shows clear interarm features not seen in the CO map
931: due to the lower resolution of the CO observations. Many of these
932: features are spurs (or feathers) which have been found to be
933: ubiquitous in grand design spirals \citep{LaVigneVogelOstriker06}.
934: These interarm features will also cause kinematic perturbations. In
935: fact, close inspection reveals that interarm perturbations in the
936: velocity field of M51 coincide with strong interarm features apparent
937: in the 8 $\mu$m image. Since the features have different pitch angles
938: from the main CO arms, we are likely smearing out these finer interarm
939: velocity perturbations. As a result, the interarm velocity profiles we
940: have derived do not reveal the details of the velocity perturbations
941: associated with interarm substructure; in a detailed study of the 2D
942: velocity field between the main arms, the interarm structure would
943: need to be considered.
944:
945: In the next section, we use the \vrad\ and \vtan\ fits to assess the
946: feasibility of the hypothesis of a quasi-steady pattern. Again, our
947: fitting method is designed to reveal streaming solely associated with
948: the spiral arms, and does not capture smaller scale perturbations,
949: such as those associated with interarm features. Both observations
950: \citep{Elm80,LaVigneVogelOstriker06} and numerical simulations
951: \citep{KO02,ShettyOstriker06} have shown that spurs and feathers are
952: associated with star formation, indicating that these features are not
953: long lasting. The modal theory, hypothesizing quasi-stationary grand
954: design spiral structure, acknowledges that such smaller scale features
955: can be transient \citep[e.g.][]{BertinLin96}. In this study, smearing
956: out the interarm perturbations likely does not affect the overall
957: conclusions we draw from the fitted velocity profiles.
958:
959:
960: \section{Tests of Conservation Laws \label{analys}}
961:
962: \subsection{Conservation of Vortensity \label{vort_sec}}
963:
964: For a flattened system, the conservation of mass and angular momentum
965: can be combined to yield
966: \begin{equation} {\frac{\partial }{\partial t}\left(\frac{\nabla
967: \times {\bf v}_{inertial}}{\Sigma}\right) +
968: {\bf v}_{inertial}\cdot \nabla\left(\frac{\nabla \times
969: {\bf v}_{inertial}}{\Sigma}\right)=0},
970: \label{fllvort}
971: \end{equation}
972: where $\Sigma$ is the surface density, and ${\bf v}_{inertial}$ is the
973: velocity in the inertial frame. Equation (\ref{fllvort}) states that
974: the vorticity per unit surface density, known as vortensity, is
975: conserved along streamlines. For steady systems, the conservation of
976: vortensity can be simplified:
977: \begin{equation} {\frac{1}{\Sigma}\cdot \left( \frac{v_\theta}{R} +
978: \frac{\partial v_\theta}{\partial R} - \frac{1}{R}\frac{\partial
979: v_R}{\partial \theta} \right) } = constant,
980: \label{vorteqn}
981: \end{equation}
982: because the temporal term in equation (\ref{fllvort}) vanishes. Even
983: if the flow is not steady, portions of the galaxy that originated in a
984: region of constant vortensity will still satisfy equation
985: \ref{vorteqn}.
986:
987: In order to test whether equation (\ref{vorteqn}) is satisfied for the
988: gas in M51, we need the surface density $\Sigma$, which we can
989: estimate using the observed CO brightness to derive the corresponding
990: H$_2$ column density.\footnote{As we will describe in
991: $\S$\ref{contsec}, HI can be neglected since the gas in M51 is
992: mostly molecular in the region studied.} Most studies suggest that
993: the relationship between CO and H$_2$ is reasonably linear, though the
994: conversion factor, known as the X-factor, is controversial. In our
995: analysis, we will assume that CO is indeed a linear molecular tracer,
996: and employ an X-factor of $2\times 10^{20}$ cm$^{-2}$ [K \kms]$^{-1}$
997: \citep[e.g.][]{Strong88}.
998:
999: We first test the vortensity condition from the velocity profiles
1000: derived in the 47\arcsec-80\arcsec\ annulus, (see Figure \ref{COvans}
1001: - \ref{anreg}). We choose this particular annulus because \vtan\
1002: variations are relatively smooth in both arm and interarm regions, and
1003: are likely due primarily to spiral streaming. This annulus clearly
1004: shows the characteristic \vtan\ boost in the arm, and the more gradual
1005: interarm decrease in \vtan. We consider the fits derived from this
1006: annulus assuming a $\theta_{MA}$ = 170\degr (see Fig.
1007: \ref{circslits}). As shown in $\S$\ref{pasec}, changes in
1008: $\theta_{MA}$ affect \vtan\ only modestly.
1009:
1010: For the first term in equation (\ref{vorteqn}), $v_\theta / R$, we use
1011: the mean value of the tangential velocities fit in the given region.
1012: To measure the radial gradient of the tangential velocity, which
1013: appears in the second term, we use
1014: \begin{equation}
1015: \left.\frac {\partial{}}{\partial R} \right|_\theta = \frac{1}{R \tan i_p}\frac{d}{d\psi},
1016: \label{rphase}
1017: \end{equation}
1018: where $i_p$ is the pitch angle of the spiral arms, and we assume
1019: negligible variation parallel to the arm. We adopt a pitch angle of
1020: 21.1$\degr$ (which is also the slope of the ``slit''). We fit
1021: straight lines to the velocity profiles in order to approximate the
1022: last two terms in equation (\ref{vorteqn}). For the surface
1023: densities, we use the peak value for the arm, and for the interarm we
1024: use the value of $\Sigma$ at phase separated by 90$\degr$ from the
1025: arm. Again, we are assuming that CO directly traces the molecular
1026: abundance. Table \ref{vorttab} shows the vortensity values for the
1027: 47\arcsec-80\arcsec\ annulus, including the values of each of the
1028: terms in equation (\ref{vorteqn}).
1029:
1030: Table \ref{vorttab} shows that, within the errors, the arms and
1031: Interarm 2 have consistent vortensity values. The value for Interarm
1032: 1, however, is lower than in the other regions. The lower value for
1033: Interarm 1 can be inferred directly from the profile itself (Fig.
1034: \ref{Havans}). The tangential velocities in both arms are clearly
1035: rising, and \vtan\ in Interarm 2 (downstream from Arm 2 at $\psi = 200
1036: \degr$) is predominantly decreasing, suggesting that the spiral arms
1037: have the most significant influence on the velocities in these
1038: regions. In Interarm 1 (downstream from Arm 1 at $\psi = 0\degr$),
1039: however, there is more structure to the velocities, suggesting that
1040: there are other sources of perturbations in addition to the spiral
1041: arms. We find that for most of the velocity profiles in Figure
1042: \ref{COvans} and \ref{Havans} the vortensity values are consistent
1043: between the two arms, within the errors. However, in the interarms,
1044: the vortensity values differ. We find varying vortensity values in
1045: all interarm regions except for Interarm 2 indicated in Table
1046: \ref{vorttab}. Overall, the agreement between vortensity in arm
1047: regions in each annulus indicates either that a steady state depiction
1048: of the vortensity is valid and there is very little radial migration
1049: of gas, or else that in a given annulus much of the gas originated in
1050: a region of constant vortensity and has been conserved along
1051: streamlines as gas in a whole annulus flows inward or outward.
1052:
1053: \subsection{Conservation of Mass \label{consmass}}
1054:
1055: \subsubsection{Flux Weighted Average \vrad \label{fluxvrave}}
1056:
1057: In the QSSS scenario, the spiral pattern --- as defined by its
1058: amplitude, phase, and rotation rate --- would not change significantly
1059: over the course of a few revolutions in a frame rotating along with
1060: the spiral pattern \citep{Lindblad63, BertinLin96}. Such a framework suggests that
1061: on average any accretion of material into the arms should be balanced
1062: by the same amount of material exiting downstream. This condition
1063: corresponds to conservation of mass for a steady state system; if this
1064: condition holds it should be apparent in the variation of observed
1065: velocities with spiral arm phase.
1066:
1067: As can be seen in Figures \ref{COvans} and \ref{Havans}, the spiral
1068: arms clearly perturb \vtan\, with deviations of $\ge 100$ \kms\, but
1069: \vtan\ always remains positive, indicating that the orbital flow is in
1070: one direction only. However, the radial velocities do change sign,
1071: indicating both inflow and outflow. If the spiral arms are indeed a
1072: quasi-stationary pattern, then large amounts of matter should not be
1073: undergoing net inflow or outflow, i.e. the mass-weighted average
1074: radial velocity cannot be too large. A large or spatially strongly
1075: variable mass-weighted average \vrad\ would imply a very dynamic
1076: system. In particular, if the sign of this quantity changes, then
1077: there would be a buildup or depletion of mass in one or more radial
1078: locations.
1079:
1080: As discussed in $\S$\ref{pasec}, the fit for \vrad\ is very sensitive
1081: to the assumed value of the \pa. Thus, in investigating the
1082: mass-weighted average radial velocity, we consider a range of \pa s.
1083: Figure \ref{pavr} shows the \vrad\ fits for \halpha\ in annuli with
1084: radii of 47\arcsec\ - 80\arcsec\ and 61\arcsec\ - 105\arcsec\ for
1085: three different \pa s, 170\degr, 175\degr, and 180\degr. The CO fits
1086: are similar, but noisier and have larger error bars (see Figs.
1087: \ref{COvans} - \ref{Havans}). A striking aspect of the fits in Figure
1088: \ref{pavr} is the large magnitude of inflow in the arms for all three
1089: \pa s; the radial velocity drops to as low as $-$75 \kms, suggesting
1090: significant inflow for gas in the spiral arms. In the upstream
1091: regions \vrad\ is positive, approaching 70 \kms\ for some parameter
1092: choices. For a region farther in, in the 27\arcsec\ - 47\arcsec\
1093: annulus shown in Figures \ref{COvans} and \ref{Havans}, the fitted
1094: radial velocity (assuming $\theta_{MA}$ = 170$\degr$) reaches values
1095: greater than 100 \kms, indicating tremendous outflow in the inner
1096: regions; assuming a \pa\ of 180$\degr$ for this annulus only reduces
1097: the peak velocity from $\approx$100 \kms\ to $\approx$75 \kms.
1098:
1099: Figure \ref{flxvr} shows the flux-weighted average radial velocity,
1100: $\langle v_R \rangle$, for different \pa s in the different annuli
1101: used in the fitting process. With the canonical \pa\ of 170\degr,
1102: there is significant outflow in the inner regions of M51. On the
1103: other hand, a \pa\ of 180\degr\ seems appropriate for the innermost
1104: regions of M51, since this yields a lower value of $\langle v_R
1105: \rangle$. However, with such a \pa\ we find significant inflow in the
1106: outer region. If we adopt an intermediate \pa\ of 175\degr, there is
1107: outflow in the inner regions and inflow in the outer regions. For a
1108: $\theta_{MA}$ of 175\degr, $\langle v_R \rangle$ = 0 for the
1109: 36\arcsec\ - 61\arcsec\ annulus, with mean radius $\langle R \rangle$
1110: = 1.98 kpc, while adjacent annuli have $\langle v_R \rangle$ = 10 and
1111: $-$20 \kms, for $\langle R \rangle$ = 1.5 kpc and 2.6 kpc respectively
1112: (1\arcsec\ = 40.7 pc at a distance of 8.4 Mpc). If this were true,
1113: then the gas would all collect near $R \approx$ 2 kpc in less than one
1114: orbital time-scale ($\sim$200 Myr), which is not consistent with a
1115: steady state.
1116:
1117: This analysis leads us to conclude that if the spiral pattern is
1118: long-lived, the large variations in the radial velocity shown in
1119: Figure \ref{flxvr} suggests that the position angle must vary with
1120: radius, indicating a disk which is not coplanar. This trend
1121: suggesting a larger \pa\ in the inner regions and a smaller \pa\ in the
1122: outer regions is also in accordance with the \pa\ tests described in
1123: $\S$\ref{pasec} (see Figure \ref{circslits}).
1124:
1125: We schematically show a disk with a varying position in Figure
1126: \ref{epic}; the \pa s of the ellipses are arranged as indicated by
1127: Figure \ref{circslits}. As discussed, one effect of a variation of
1128: \pa\ is a disk that is not coplanar. The inclination in this
1129: schematic is exaggerated; the observed morphology, including the
1130: apparent spiral structure, depends on the viewing angle, among other
1131: factors.
1132:
1133: \subsubsection{Continuity and Spiral Pattern Speed \label{contsec}}
1134: In this section we explore the plausibility of QSSS using the gas
1135: continuity equation. The continuity equation for gas flow in a
1136: two-dimensional system is
1137: \begin{equation}
1138: \frac{\partial\Sigma}{\partial t} +
1139: \nabla\cdot(\Sigma{\bf v})=0;
1140: \label{cont}
1141: \end{equation}
1142: this holds in any frame, e.g. whether the velocity is measured in an
1143: inertial frame or one rotating at a constant pattern speed. The first
1144: term, $\partial \Sigma / \partial t$, represents the temporal growth
1145: or decay of the surface density $\Sigma$ at any given radius $R$ and
1146: azimuthal angle $\theta$ in the plane of the galaxy, where those
1147: coordinates are with respect to the frame in which the velocity is
1148: being measured.
1149:
1150: If the flow is in a steady state, then the temporal term vanishes,
1151: leaving only the mass flux term $\Sigma{\bf v}$. If the gas is
1152: responding primarily to a single dominant spiral perturbation, as would
1153: be required for a fixed spiral pattern, and when ${\bf v}$ is measured in the frame
1154: rotating at the pattern angular velocity $\Omega_p$,
1155: \begin{equation}
1156: \nabla\cdot[\Sigma({\bf v}_{inertial}-\Omega_p R \hat{\theta})]=0.
1157: \label{cont_anal}
1158: \end{equation}
1159: Thus, for an exact steady state the mass flux must be constant (in the
1160: frame rotating with the same angular velocity as the spiral mode).
1161: For a quasi-steady state, the temporal variations in $\Sigma$ will
1162: only be small, and thus variations in mass flux would also be small.
1163: This condition can be further simplified using a reference frame
1164: aligned locally with the spiral arms. Figure \ref{armgeom} shows this
1165: reference frame; the $x$ and $y$ coordinates are the directions
1166: perpendicular and parallel to the local spiral arm, respectively. The
1167: transformation between cylindrical coordinates and this arm frame is
1168: achieved using
1169: \begin{equation}
1170: \hat{x}=\cos i_p \hat{R} + \sin i_p \hat{\theta},
1171: \label{xtran}
1172: \end{equation}
1173: \begin{equation}
1174: \hat{y}=-\sin i_p \hat{R} + \cos i_p \hat{\theta},
1175: \label{ytran}
1176: \end{equation}
1177: where $i_p$ is the pitch angle of the arms.
1178:
1179: The velocities in the arm frame are given by
1180: \begin{equation}
1181: v_x=v_R \cos i_p + (v_\theta -\Omega_p R) \sin i_p ,
1182: \label{vxeqn}
1183: \end{equation}
1184: and
1185: \begin{equation}
1186: v_y=-v_R \sin i_p + (v_\theta -\Omega_p R) \cos i_p .
1187: \label{vyeqn}
1188: \end{equation}
1189:
1190: From the maps shown in Figures \ref{COarm} and \ref{pgCO}, it is
1191: apparent that the intensity and velocity vary significantly more
1192: across the arms, in the $x$-direction, than along them, in the
1193: $y$-direction. Thus, the variation in the product of $\Sigma v_y$
1194: along $\hat{y}$ is much smaller than the variation in the product of
1195: $\Sigma v_x$ along $\hat{x}$, reducing equation (\ref{cont_anal}) to
1196: \begin{equation}
1197: {\Sigma v_x\approx constant}.
1198: \label{sigvx}
1199: \end{equation}
1200: Namely, for a steady pattern, as the gas decelerates (in the $x$
1201: direction, perpendicular to the arm), mass accumulates and the surface
1202: density increases; as the gas velocity increases, the surface density
1203: decreases.
1204:
1205: One difficulty in testing whether equation (\ref{sigvx}) holds is that
1206: neither $\theta_{MA}$ nor $\Omega_p$ is well constrained. Errors in
1207: $\theta_{MA}$ yield errors in the fitted value of \vtan\ of $\approx
1208: v_R \sin \Delta \theta_{MA}$ and in the fitted value of \vrad\ of
1209: $\approx -v_\theta \sin \Delta \theta_{MA}$. If $\Delta\Omega_p$ is
1210: the error in the pattern speed, then the fitted value of $v_x$ will be
1211: approximately given by $v_x + (-v_\theta \sin \Delta \theta_{MA} \cos
1212: i_p + v_R \sin \Delta \theta_{MA}\sin i_p - \Delta \Omega_p R \sin
1213: i_p)$. Since the \vrad\ term has two factors of the $\sin$ of small
1214: angles, that term will be much smaller compared with the other two
1215: terms. The true value of $v_x$ will therefore differ from the fitted
1216: value by $C_x \approx v_\theta \sin \Delta \theta_{MA} \cos i_p +
1217: \Delta \Omega_p R \sin i_p$.
1218:
1219: In order to assess whether steady state continuity as expressed by
1220: equation (\ref{sigvx}) holds in the case of M51, we therefore consider
1221: the quantity
1222: \begin{equation}
1223: \Sigma \tilde{v}_x=\Sigma (v_R \cos i_p + (v_\theta - \Omega_p R) \sin
1224: i_p + C),
1225: \label{modvxeqn}
1226: \end{equation}
1227: where \vrad\ and \vtan\ are fitted values and $C \equiv \langle C_x
1228: \rangle$, i.e. the (unknown) azimuthally-averaged correction due to
1229: the errors in $\theta_{MA}$ and $\Omega_p$. We apply equation
1230: (\ref{modvxeqn}) by solving for the value of $C$ using the values of
1231: \vrad, \vtan, and $\Sigma$ in the two arm segments of an annulus:
1232: \begin{equation}
1233: C = \frac{[v_{R,arm1} \cos i_p + (v_{\theta,arm1} - \Omega_p R) \sin i_p]
1234: \Sigma_{arm1} - [v_{R,arm2} \cos i_p + (v_{\theta,arm2} - \Omega_p R)
1235: \sin i_p] \Sigma_{arm2}}{\Sigma_{arm2} - \Sigma_{arm1}}.
1236: \label{modvxC}
1237: \end{equation}
1238: We then test whether the value of $C$ obtained using equation
1239: (\ref{modvxC}) also satisfies equation (\ref{modvxeqn}) in the
1240: interarm regions. If equation (\ref{modvxeqn}) is satisfied for both
1241: interarm and arm regions, it would suggest an approximate steady
1242: state.
1243:
1244: We again focus on the 47\arcsec-80\arcsec\ annulus, where the \vrad\
1245: (and \vtan\ ) are relatively ``smooth.'' The \vrad\ profiles for this
1246: annulus is shown in Figure \ref{pavr}, assuming three different values
1247: of the position angle. Table \ref{mctab} shows the relevant values
1248: associated with equation (\ref{modvxeqn}) for the arms and interarm
1249: regions. We employ the pattern speed of 38 $\pm$ 7 \kms\ kpc$^{-1}$,
1250: calculated by \citet{ZimmerRandMcG04} by applying the
1251: Tremaine-Weinburg method CO observations of M51. Corotation
1252: corresponding to this pattern speed is marked on Figure \ref{anreg}.
1253: After solving for $C$ using quantities from the arms, it is clear that
1254: the flow in the interarm region is not consistent with a steady state
1255: description. A \pa\ of 170$\degr$ produces large negative mass flux
1256: in the arms, and positive flux in one interarm region. Even
1257: variations in the X-factor cannot resolve this discrepancy. Assuming
1258: larger values of the \pa\ still produces mass fluxes with vastly
1259: different magnitudes, and even different signs. Increasing the error
1260: in $C$ up to an order of magnitude still cannot result in consistent
1261: mass fluxes between the arm and interarm. This suggests that any
1262: reasonable changes to the values of $\theta_{MA}$ or $\Omega_p$ will
1263: still result in varying mass fluxes. We have checked the mass flux in
1264: other annuli using the same method as for the 47\arcsec-80\arcsec\
1265: annulus, as well as in other localized regions not presented here, and
1266: found similar discrepancies in the mass flux.
1267:
1268: In our analysis of continuity so far, we have not taken into account
1269: the contribution from the atomic component of the disk. In fact, in
1270: most galaxies the majority of the gas exists in the form of HI. In
1271: M51, \citet{TilanusAllen89} showed that the downstream offset of HI
1272: relative to the dust lanes is likely due to dissociation of molecular
1273: gas by recently formed massive stars. However, the inner disk of M51
1274: has an unusually large fraction of molecular gas, so even at peaks of
1275: the HI photodissociation arms, the contribution of atomic gas to the
1276: total gas surface density is negligible. Using the HI maps of
1277: \citet{Rotsetal90}, we find that the atomic column density N(HI) is
1278: significantly less than the molecular column density N(H$_2$) in the
1279: vast majority of locations in the inner disk (21\arcsec\ $\le R \le
1280: 105$\arcsec); N(HI) exceeds N(H$_2$) in only $\sim$ 7\% of the inner
1281: disk. The mean value of N(H$_2$)/N(HI) throughout the inner region is
1282: $\sim$10. Though we used a constant X-factor to obtain the molecular
1283: surface density, moderate variations in the X-factor \citep[for M51,
1284: see][]{NakaiKuno95} will not be sufficient to account for the
1285: discrepancy. Nevertheless, no change in the X-factor, or in the
1286: contributions of the molecular or atomic matter to the total mass, can
1287: account for the change in sign of the mass flux; the varying sign of
1288: $\Sigma v_x$ can only be due to a sign change in $v_x$, not $\Sigma$.
1289:
1290: Our conclusion, after analyzing the mass flux, is that the kinematics
1291: are not consistent with a quasi-steady spiral pattern in a flat disk.
1292: We find that no single pattern speed can satisfy quasi-steady state
1293: continuity, suggesting that the QSSS hypothesis is not applicable to
1294: M51. It is essentially the tremendous variations of the radial
1295: velocity within a given annulus --- amounting to $\sim 100$ \kms\ ---
1296: that lead the QSSS hypothesis into difficulty. One explanation for
1297: the transient nature of the spiral arms in M51, perhaps due to the
1298: interaction with its companion, is a spiral perturbation with a
1299: constant pattern speed, but with time-varying amplitude. Or, there
1300: may be multiple modes at work in the disk of M51, which may be
1301: construed as a mode with a radially varying pattern speed
1302: \citep[e.g.][]{MerrRandMeidt06}. Multiple patterns speeds in M51 have
1303: been previously suggested by \citet{Vogel93} and
1304: \citet{ElmSeidenElm89}. However, the extreme variations in the radial
1305: velocity cannot be explained by multiple patterns alone. Possible
1306: causes for the large observed \vrad\ gradients are large out-of-plane
1307: motions or a variation in inclination; since the inclination of M51 is
1308: small, a variation in $i$ due to a warped or twisted disk will produce
1309: large variations in the observed velocity due to projection effects.
1310:
1311:
1312: \subsection{Discussion}
1313:
1314: We have shown that the density and velocity structure in M51 does not
1315: support a quasi-steady state depiction for the spiral pattern, using
1316: measurements of the mass flux. Further evidence that the observed
1317: structure is inconsistent with steady state can be obtained by
1318: adopting the fitted 2D velocity field, and demonstrating that the
1319: density structure is then non-steady. We have carried out this
1320: exercise using a modified version of the {\tt NEMO} task {\tt
1321: FLOWCODE} \citep{Teuben95}. In this exercise, a disk is populated
1322: with gas tracer particles using the intensity profiles averaged along
1323: spiral segments, reproducing the spiral density pattern of M51. Each
1324: location in the disk has an associated \vrad\ and \vtan, given by the
1325: fitted velocity profiles (e.g. Figures \ref{COvans} and \ref{Havans}),
1326: and an assumed value of the pattern speed. The motion of the
1327: particles is then integrated using {\tt FLOWCODE}: after a suitably
1328: small timestep, the particles take on new velocities depending on
1329: their location in the disk. In essence, this simulation is a purely
1330: kinematic test to determine whether the steady state continuity
1331: equation (eqn. [\ref{cont_anal}]) is satisfied or not, using the
1332: density and fitted velocity profiles of M51 (Figs.
1333: \ref{COvans}-\ref{Havans}). We find that the input spiral pattern
1334: vanishes in less than one orbital time scale ($\sim$200 Myr),
1335: regardless of what values of the \pa\ and pattern speed we assume.
1336:
1337: The precise nature of the velocities is one of a number of issues that
1338: need to be considered in further studying the the global spiral
1339: pattern in M51. For example, our result suggests the role of a warp
1340: certainly needs to be taken into account. There are strong
1341: indications that the outer disk of M51 is warped; our finding suggests
1342: that the disk is not coplanar even further inward. The non-coplanar
1343: attribute may be the result of the tidal interaction between M51 and
1344: its companion.
1345:
1346: The possible warp and/or twist in the disk of M51 would of course
1347: affect the projected velocities, and would present itself as gradients
1348: in the velocity components, as discussed in the previous section. If
1349: this were indeed the case, then the single or multiple in-plane modes
1350: would have to be in phase with the vertical mode in order to sustain a
1351: spiral pattern. The inherent uncertainty in deriving three velocity
1352: components from the single observed component leads to difficulty in
1353: estimating and analyzing both the vertical and in-plane modes.
1354:
1355: \section{Summary\label{summary}}
1356: We have analyzed the velocity field of M51, using CO and \halpha\
1357: observations, to investigate the nature of the spiral structure. We
1358: summarize the main results here:
1359:
1360: 1) The velocity field is quite complex. Observed velocities show
1361: significant azimuthal streaming associated with the spiral arms, as
1362: well as strong gradients in the radial velocities.
1363:
1364: 2) The aberrations in the velocity field strongly suggest that the
1365: disk is not coplanar, perhaps as far in as 20\arcsec\ ($\sim$800 pc)
1366: from the center.
1367:
1368: 3) We obtain fitted radial and tangential velocity profiles by
1369: assuming that velocities in any annulus vary only with arm phase.
1370: Strong gradients in the radial and tangential velocities are found in
1371: the profile fits. In general, the shape of both the \vrad\ and \vtan\
1372: profiles are in qualitative agreement with theory of nonlinear density
1373: waves, and support the presence of shocks.
1374:
1375: 4) In detail, the velocity profiles from different radial regions of
1376: M51 differ significantly. In addition, velocity profiles associated
1377: with the two arms also show differences in a given annulus. For the
1378: arm that is well described by a logarithmic spiral (bright arm), the
1379: associated velocities are in good agreement with simple theoretical
1380: spiral shock profiles. For the other arm, which is not as well
1381: described by a logarithmic spiral, the velocities are more complex.
1382:
1383: 5) The velocity profile fits from CO and \halpha\ emission are rather
1384: similar, suggesting that most of the \halpha\ emission originates from
1385: gas associated with star forming regions.
1386:
1387: 6) When we assume a single value for the position angle of the major
1388: axis of M51 and inclination, we find that large amounts of material
1389: flows toward an annulus of intermediate radius, due to the large
1390: gradients and change of sign in the flux weighted average radial
1391: velocity. As a result, either the position angle of the major axis or
1392: the inclination must vary with radius, suggesting that the disk of M51
1393: is warped and twisted.
1394:
1395: 7) We analyze conservation of vortensity, using the radial and
1396: tangential velocity profile fits. We find that vortensity is fairly
1397: consistent within a given annulus, indicating that the gas there all
1398: originated in a region of uniform vortensity.
1399:
1400: 8) Using the equation of continuity, we find that the density and
1401: fitted velocity profiles are inconsistent with quasi-steady state mass
1402: conservation in any frame rotating at a constant angular speed, at
1403: least for a planar system. Variations in the pattern speed, \pa, and
1404: X-factor alone cannot account for the differences in the mass flux,
1405: suggesting that spiral arms are quite dynamic, and possibly that
1406: out-of-plane motions are significant.
1407:
1408: \acknowledgements We are grateful to S. McGaugh for his help in using
1409: the baryonic Tully-Fisher relation to estimate the inclination of M51.
1410: We thank W.T. Kim for useful discussions, and G. Gomez for providing
1411: velocity profiles from his hydrodynamic models. We also thank the
1412: anonymous referee for raising many interesting questions and useful
1413: suggestions. This research was supported in part by grants AST
1414: 02-28974 and AST 05-07315 from the National Science Foundation.
1415:
1416: \bibliography{ref}
1417:
1418: \clearpage
1419:
1420: \begin{deluxetable}{lcc}
1421: \tablecolumns{3}
1422: \tablewidth{0pt}
1423: \tablecaption{Initially Adopted Parameters for M51 \label{paramtab}}
1424: \tablehead{
1425: \colhead{Parameter} & \colhead{Value} & \colhead{Reference}}
1426: \startdata
1427: Center RA ~($\alpha$) (J2000) & $13^h29^m52^s.71$ & \citet{Hagiwara01} \\
1428: Center DEC ~($\delta$) (J2000) & $47^\circ11'42\farcs80$ & '' \\
1429: Systematic Velocity ~($V_{sys}$) & 472 (LSR) & \citet{Tully74sp} \\
1430: Position Angle of Major Axis ~($\theta_{MA}$) & 170\degr & '' \\
1431: Inclination ~($i$) & 20\degr & '' \\
1432: \enddata
1433: \end{deluxetable}
1434:
1435: \begin{deluxetable}{lcc}
1436: \tablecolumns{3} \tablewidth{0pt} \tablecaption{Estimation of the
1437: Systematic Velocity of M51 \label{vsystab}} \tablehead{
1438: \colhead{Method} & \colhead{$V_{sys}$ (\kms)\tablenotemark{a}} &
1439: \colhead{Error (\kms)}} \startdata
1440: \citet{Tully74sp} & 472 & 3 \\
1441: Freeing $V_{sys}$ in fitting & 470.6 & 0.4 \\
1442: for \vrad$(\psi)$ and \vtan$(\psi)$ & & \\
1443: Tilted Rings Analysis & 471.4 & 0.5 \\
1444: Rotation Curve Fitting & 473.3 & 0.5 \\
1445: \tableline
1446: Weighted Mean & 471.7 & 0.3 \\
1447: \enddata \tablenotetext{a}{Velocity in LSR frame}
1448: \end{deluxetable}
1449:
1450: \begin{deluxetable}{cccccc}
1451: \tablewidth{0pt} \tablecaption{Vortensity in the 47\arcsec\ -
1452: 80\arcsec\ Annulus} \tablehead{ \colhead{Region} & \colhead
1453: {$\overline \Sigma$\tablenotemark{a}} & \colhead {$\left(
1454: \frac{v_\theta}{R} \right)$\tablenotemark{b}} & \colhead
1455: {$\left(\frac{\partial v_\theta}{\partial
1456: R}\right)$\tablenotemark{b}} &
1457: \colhead{$\left(\frac{1}{R}\frac{\partial v_R}{\partial
1458: \theta}\right)$\tablenotemark{b}} & \colhead{Vortensity
1459: Value\tablenotemark{c}}} \startdata
1460:
1461: Arm 1 ($\psi \approx 360\degr$) & 244 & 70 $\pm$ 1 & 103 $\pm$ 2 &
1462: -30
1463: $\pm$ 3 & 0.8 $\pm$ 0.2 \\
1464: Interarm 1 ($\psi \approx 90\degr$) & 16 & 67 $\pm$ 1 & -52 $\pm$ 3
1465: & 6 $\pm$ 1 & 0.5 $\pm$ 0.2
1466: \\
1467: Arm 2 ($\psi \approx 190\degr$)& 128 & 63 $\pm$ 0.5 & 59 $\pm$ 3 &
1468: -18
1469: $\pm$ 3 & 1 $\pm$ 0.2 \\
1470: Interarm 2 ($\psi \approx 275\degr$) & 19 & 59 $\pm$ 1 & -31 $\pm$ 2
1471: &
1472: 11 $\pm$ 1 & 0.9 $\pm$ 0.2 \\
1473: \enddata {\singlespace \tablenotetext{a}{$\rm \left[M_\odot \,\,
1474: pc^{-2}\right]$; error of $\sim$20\%}
1475: \tablenotetext{b}{$\left[\rm km \,\, s^{-1} \,\,kpc^{-1}\right]$}
1476: \tablenotetext{c}{$\rm \left[ km \,\, s^{-1} \,\,kpc^{-1} \,
1477: (M_\odot\,\, pc^{-2})^{-1}\right]$} }
1478: \label{vorttab}
1479: \end{deluxetable}
1480:
1481: \begin{deluxetable}{cccccc}
1482: \tablewidth{0pt}
1483: \tablecaption{Mass Flux in the 47\arcsec\ - 80\arcsec\ Annulus}
1484: \tablehead{
1485: \colhead{Region} & \colhead {$\Sigma$\tablenotemark{a}} &
1486: \colhead {$ v_R \cos i_p$\tablenotemark{b}} &
1487: \colhead {$(v_\theta-\Omega_pR) \sin i_p$\tablenotemark{b}} &
1488: \colhead {$C$\tablenotemark{c}} &
1489: \colhead{$\Sigma\tilde{v_x}$\tablenotemark{d}}}
1490: \startdata
1491: $\theta_{MA} = 170$\degr: \\
1492: Arm 1 ($\psi \approx 360\degr$) & 244 & -24 & 41 & -58 $\pm$ 30 & -10113 \\
1493: Arm 2 ($\psi \approx 190\degr$)& 128 & -53 & 32 & -58 $\pm$ 30 & -10113 \\
1494: Interarm 1 ($\psi \approx 90\degr$) & 16 & 23 & 18 & - & -259 \\
1495: Interarm 2 ($\psi \approx 275\degr$) & 19 & 49 & 21 & - & 237 \\
1496: \tableline
1497: $\theta_{MA} = 175$\degr: \\
1498: Arm 1 ($\psi \approx 360\degr$) & 244 & -41 & 41 & -32 $\pm$ 18 & -8103 \\
1499: Arm 2 ($\psi \approx 190\degr$)& 128 & -63 & 32 & -32 $\pm$ 18 & -8103 \\
1500: Interarm 1 ($\psi \approx 90\degr$) & 16 & 12 & 18 & - & -38 \\
1501: Interarm 2 ($\psi \approx 275\degr$) & 19 & 36 & 21 & - & 464 \\
1502: \tableline
1503: $\theta_{MA} = 180$\degr: \\
1504: Arm 1 ($\psi \approx 360\degr$) & 244 & -59 & 41 & -8 $\pm$ 17 & -6345 \\
1505: Arm 2 ($\psi \approx 190\degr$)& 128 & -71 & 32 & -8 $\pm$ 17 & -6345 \\
1506: Interarm 1 ($\psi \approx 90\degr$) & 16 & 1 & 18 & - & 182 \\
1507: Interarm 2 ($\psi \approx 275\degr$) & 19 & 27 & 21 & - & 761 \\
1508: \enddata
1509: {\singlespace
1510: \tablenotetext{a}{$\left[\rm M_\odot \,\,pc^{-2}\right]$; error of $\sim$20\%}
1511: \tablenotetext{b}{$\left[\rm km \,\, s^{-1} \right]$}
1512: \tablenotetext{c}{$\left[\rm km \,\, s^{-1} \right]$; error largely
1513: due to errors in $\Sigma$ and $\Omega_p$}
1514: \tablenotetext{d}{$\left[\rm M_\odot \,\, pc^{-2} \,\, km \,\, s^{-1} \right]$}
1515: }
1516: \label{mctab}
1517: \end{deluxetable}
1518:
1519: \begin{figure}
1520: \epsscale{1.7}
1521: \includegraphics*[scale=0.95]{f1-s.eps}
1522: \caption{\footnotesize CO (1-0) velocity-integrated intensity (bottom)
1523: and velocity (top) maps of M51. Velocity contours increment by 10
1524: \kms, between 360 and 560 \kms. Overlaid lines are logarithmic
1525: spirals with a pitch angle of 21.1\degr, separated by 180\degr.}
1526: \label{COarm}
1527: \end{figure}
1528:
1529: \begin{figure}
1530: \epsscale{1.7}
1531: \includegraphics*[scale=0.95]{f2-s.eps}
1532: \caption{\footnotesize \halpha\ velocity-integrated intensity (bottom)
1533: and velocity (top) map of M51. The overlaid spirals, as well as the
1534: velocity contours, are as described in Figure \ref{COarm}.}
1535: \label{Haarm}
1536: \end{figure}
1537:
1538:
1539: \begin{figure}
1540: \epsscale{0.9}
1541: \plotone{f3.eps}
1542: \caption{\footnotesize Geometry depicting the spiral arm phase $\psi$.
1543: The diagram on the left is the geometry in the plane of the galaxy.
1544: The phase, $\psi$, represents the angular displacement between two
1545: locations with equal galactocentric radius $R_o$ for two congruent
1546: spiral segments. The diagram on the right is the logarithmic polar
1547: projection of the geometry on the left, showing the corresponding
1548: spiral segments.}
1549: \label{armphase}
1550: \end{figure}
1551:
1552: \begin{figure}
1553: \includegraphics*[angle=-90,scale=0.70]{f4-s.eps}
1554: \caption{\footnotesize Logarithmic polar projections of the CO
1555: intensity and velocity maps. Though the origin of the abscissa
1556: (azimuthal angle) is arbitrary, in this case it is aligned with due
1557: North. The direction of rotation is to the right (counter-clockwise
1558: as seen on the sky). Also shown are the two logarithmic spiral
1559: lines positioned along the two spiral arms, which correspond to the
1560: lines overlaid on the maps of Figure \ref{COarm}.}
1561: \label{pgCO}
1562: \end{figure}
1563:
1564: \begin{figure}
1565: \includegraphics*[angle=-90,scale=0.70]{f5-s.eps}
1566: \caption{\footnotesize Logarithmic polar projections of the H$\alpha$
1567: intensity and velocity maps. Coordinate system and log spiral
1568: overlays are as in Figure \ref{pgCO}.}
1569: \label{pgHa}
1570: \end{figure}
1571:
1572: \begin{figure}
1573: \epsscale{.8}
1574: \plotone{f6.eps}
1575: \caption{\footnotesize Gas profiles as a function arm phase $\psi$ for
1576: a model spiral galaxy (see text). Solid lines: density (top),
1577: \vrad\ (middle), and \vtan\ (bottom) profiles averaged at each arm
1578: phase, in an annulus extending from 8.38 - 8.92 kpc. The \vrad\ and
1579: \vtan\ dashed lines in lower panels are obtained by fitting equation
1580: (\ref{newvobs}) to the ``observed'' velocities in the 8.38 - 8.92
1581: kpc annulus.}
1582: \label{modvfit}
1583: \end{figure}
1584:
1585: \begin{figure}
1586: \epsscale{1.2}
1587: \plottwo{f7a.eps}{f7b.eps}
1588: \caption{\footnotesize CO (left) and \halpha\ (right) \vrad\ and
1589: \vtan\ fits as a function of arm phase for an annulus with an inner
1590: radius of 21\arcsec\ and an outer radius of 36\arcsec. The one
1591: sided 3$\sigma$ error-bars are also shown on the bottom of each
1592: panel. Dashed lines are the corresponding mean CO velocity
1593: integrated intensities, with the scale depicted on the right
1594: ordinate. Table \ref{paramtab} shows the fixed (canonical)
1595: parameters used in obtaining these fits.}
1596: \label{COvrvt}
1597: \end{figure}
1598:
1599: \begin{figure}
1600: \includegraphics*[angle=-90,scale=0.50]{f8.eps}
1601: \caption{\footnotesize Result of fit to M51 CO velocity data in which
1602: $V_{sys}$, \vrad, and \vtan\ were allowed to vary. The radial range
1603: of the annulus is 14\arcsec\ - 136\arcsec. The fact that the fit
1604: value of $V_{sys}$ varies with phase $\psi$ shows that other
1605: parameters (e.g. $i$, $\theta_{MA}$) vary with radius within the
1606: annulus. The mean of the fits is 470.6 \kms\ (LSR), shown by the
1607: dashed line.}
1608: \label{M51vsys}
1609: \end{figure}
1610:
1611: \begin{figure}
1612: \includegraphics*[angle=-90,scale=0.50]{f9.eps}
1613: \caption{\footnotesize Fit $V_{sys}$ as a function of arm phase,
1614: similar to Fig. \ref{M51vsys}, but for a model galaxy with a warp.
1615: The \vsys\ adopted for the model is 400 \kms\ (dashed line), equal
1616: to the mean of the fits (solid line).}
1617: \label{warp}
1618: \end{figure}
1619:
1620: \clearpage
1621:
1622: \begin{figure}
1623: \epsscale{0.77}
1624: \plotone{f10-s.eps}
1625: \caption{\footnotesize Tully ``wedge'' method for estimating galaxy
1626: \pa. The extreme velocity for each 5\arcsec\ annulus, averaged in
1627: 5\degr\ wedges, is marked. The upper panels show the \halpha\
1628: velocity (left) and velocity-integrated intensity (right); lower
1629: panels show the same for CO. For CO, emission was too weak to apply
1630: the method at some radii, especially in the outer galaxy.}
1631: \label{wedges}
1632: \end{figure}
1633:
1634: \begin{figure}
1635: \plotone{f11-s.eps}
1636: \caption{\footnotesize Application of Tully wedge method to a model
1637: with streaming. The velocity field is shown (left) along with the
1638: corresponding intensity map (right). The position angle assumed in
1639: the model is 170\degr, shown by the solid line. The 5\degr\ wedges
1640: with the extreme velocity for each annulus is marked. It can be
1641: seen that streaming shifts the estimated \pa\ from the true \pa.
1642: Note that the extreme velocities do not occur in the arm.}
1643: \label{modwedges}
1644: \end{figure}
1645:
1646: \begin{figure}
1647: \includegraphics*[angle=-90,scale=0.60]{f12.eps}
1648: \caption{\footnotesize Mean observed velocity plotted vs azimuthal
1649: angle. All observed velocities were averaged over radius at each
1650: azimuth. Velocities are fit by cosine functions (solid line);
1651: extremum of the cosine curve indicates the best fit \pa\ of the
1652: major axis.}
1653: \label{radslits}
1654: \end{figure}
1655:
1656:
1657: \begin{figure}
1658: \includegraphics*[angle=-90,scale=0.55]{f13.eps}
1659: \caption{\footnotesize Fit \pa\ of the major axis $\theta_{MA}$ as a
1660: function of galactocentric radius from \halpha\ (upper) and CO
1661: (lower) velocity fields. The \pa s were obtained by fitting a
1662: cosine function to the distribution of observed velocity vs
1663: azimuthal angle at each radius.}
1664: \label{circslits}
1665: \end{figure}
1666:
1667: \begin{figure}
1668: \epsscale{0.93}
1669: \plotone{f14.eps}
1670: \caption{\footnotesize \singlespace CO \vrad\ (left panels) and \vtan\
1671: (right panels) fits as a function of arm phase $\psi$ in different
1672: annuli (with radii labeled in the upper right of each panel). The
1673: thickness of the line shows a range of $\pm3\sigma$. Only \vrad\
1674: and \vtan\ fits with 3$\sigma$ $\leq$ 20 \kms\ and $\leq$ 60 \kms,
1675: respectively, are shown. Dashed lines are the corresponding mean CO
1676: intensities, with the scale shown on the right ordinate. We assume
1677: a \pa\ of 170\degr, an \inc\ of 24\degr\ and the center position and
1678: systematic velocity listed in Table \ref{paramtab}. Figure
1679: \ref{anreg} shows the annular regions of M51 considered for these
1680: fits.}
1681: \label{COvans}
1682: \end{figure}
1683:
1684: \begin{figure}
1685: \epsscale{0.93}
1686: \plotone{f15.eps}
1687: \caption{\footnotesize H$\alpha$ \vrad\ and \vtan\ fits as a function
1688: of $\psi$ in different annuli, as in Figure \ref{COvans}.}
1689: \label{Havans}
1690: \end{figure}
1691:
1692: \begin{figure}
1693: \epsscale{.6}
1694: \plotone{f16-s.eps}
1695: \caption{\footnotesize Deprojected CO map of M51 showing the
1696: overlapping annuli for which \vrad\ and \vtan\ are fitted as a
1697: function of arm phase (shown in Figures \ref{COvans} and
1698: \ref{Havans}). The radii of (solid) circles, from the inner to the
1699: outer, are: 21\arcsec, 27\arcsec, 36\arcsec, 47\arcsec, 61\arcsec,
1700: 80\arcsec, and 105\arcsec. The annulus marked by dashed circles
1701: (4.2 kpc $\le R \le$ 6.1 kpc) spans possible corotation
1702: radii corresponding to an adopted spiral pattern speed $\Omega_p$ =
1703: 38 $\pm$ 7 \kms\ kpc$^{-1}$ (see $\S$\ref{contsec}).}
1704: \label{anreg}
1705: \end{figure}
1706:
1707: \begin{figure}
1708: \includegraphics*[scale=0.80]{f17.eps}
1709: \caption{\footnotesize H$\alpha$ \vrad\ fits as a function of arm
1710: phase for 3 different \pa s $\theta_{MA}$, 170\degr, 175\degr, and
1711: 180\degr, for two annuli (47\arcsec\ - 80 \arcsec\ and 61\arcsec\ -
1712: 105\arcsec). We fix the inclination at 24\degr, and other
1713: parameters used in the fitting are shown in Table \ref{paramtab}.
1714: The dashed line is the mean CO intensity along the arm for a
1715: $\theta_{MA}$ of 170\degr, which varies only slightly with
1716: $\theta_{MA}$. The error bars are not shown because they are
1717: similar to those shown in Figure \ref{Havans}.}
1718: \label{pavr}
1719: \end{figure}
1720:
1721: \begin{figure}
1722: \includegraphics*[angle=-90, scale=0.70]{f18.eps}
1723: \caption{\footnotesize Mass-weighted average radial velocities
1724: $\langle v_R \rangle$ in the different annuli, two of which are
1725: shown in Figure \ref{pavr}. The abscissa indicates the mean radius
1726: $\langle R \rangle$, in arcsecs, of each annulus. The three panels
1727: show the mass-weighted average \vrad\ assuming three different
1728: values for the \pa\ $\theta_{MA}$. The error bars include both
1729: fitted errors in \vrad\ (see Fig. \ref{COvans}) and an estimated
1730: error of 20\% in $\Sigma$.}
1731: \label{flxvr}
1732: \end{figure}
1733:
1734: \begin{figure}
1735: \includegraphics*[scale=0.70]{f19.eps}
1736: \caption{\footnotesize Model disk showing the variation of the \pa\
1737: with radius. The \pa\ profile is taken from Fig. \ref{circslits}.
1738: The inclination is exaggerated to show a more edge-on view.}
1739: \label{epic}
1740: \end{figure}
1741:
1742: \begin{figure}
1743: \includegraphics*[angle=-90, scale=0.60]{f20.eps}
1744: \caption{\footnotesize Coordinate transformation geometry, from
1745: $(R,\theta)$ galactocentric coordinates to the $(x,y)$ spiral arm
1746: frame.}
1747: \label{armgeom}
1748: \end{figure}
1749:
1750: \end{document}
1751: