0709.0925/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: %\documentclass{emulateapj}
3: %% manuscript produces a one-column, double-spaced document:
4: 
5: %\documentclass[manuscript]{aastex}
6: 
7: %% preprint2 produces a double-column, single-spaced document:
8: 
9: %% \documentclass[preprint2]{aastex}
10: 
11: %% Sometimes a paper's abstract is too long to fit on the
12: %% title page in preprint2 mode. When that is the case,
13: %% use the longabstract style option.
14: 
15: %% \documentclass[preprint2,longabstract]{aastex}
16:  
17: %% If you want to create your own macros, you can do so
18: %% using \newcommand. Your macros should appear before
19: %% the \begin{document} command.
20: %%
21: %% If you are submitting to a journal that translates manuscripts
22: %% into SGML, you need to follow certain guidelines when preparing
23: %% your macros. See the AASTeX v5.x Author Guide
24: %% for information.
25: 
26: \newcommand{\vdag}{(v)^\dagger}
27: \newcommand{\myemail}{ehumphreys@cfa.harvard.edu}
28: \newcommand{\velunitns}{km\,s$^{-1}$}
29: \newcommand{\velunits}{km\,s$^{-1}$ }
30: \newcommand{\accunits}{km\,s$^{-1}$\,yr$^{-1}$ }
31: \newcommand{\accunitns}{km\,s$^{-1}$\,yr$^{-1}$}
32: \newcommand{\water}{H$_{2}$O }
33: \newcommand{\about}{$\sim$}
34: \newcommand{\gal}{NGC~4258}
35: \newcommand{\degrs}{\hbox{$^{\circ}$} }
36: \newcommand{\degrns}{\hbox{$^{\circ}$}}
37: \newcommand{\mass}{\mbox{$\rm{\bf{\sf{M}}}$}}
38: 
39: %% You can insert a short comment on the title page using the command below.
40: 
41: \slugcomment{}
42: 
43: %% If you wish, you may supply running head information, although
44: %% this information may be modified by the editorial offices.
45: %% The left head contains a list of authors,
46: %% usually a maximum of three (otherwise use et al.).  The right
47: %% head is a modified title of up to roughly 44 characters.
48: %% Running heads will not print in the manuscript st yle.
49: 
50: \shorttitle{}
51: \shortauthors{Humphreys et al.}
52: 
53: %% This is the end of the preamble.  Indicate the beginning of the
54: %% paper itself with \begin{document}.
55: \usepackage[]{natbib}
56: \usepackage{color}
57: 
58: 
59: 
60: \begin{document}
61: 
62: %% LaTeX will automatically break titles if they run longer than
63: %% one line. However, you may use \\ to force a line break if
64: %% you desire.
65: 
66: \title{Toward a New Distance to the Active Galaxy NGC~4258:\\ 
67: II. Centripetal Accelerations and Investigation of Spiral Structure}
68: 
69: %% Use \author, \affil, and the \and command to format
70: %% author and affiliation information.
71: %% Note that \email has replaced the old \authoremail command
72: %% from AASTeX v4.0. You can use \email to mark an email address
73: %% anywhere in the paper, not just in the front matter.
74: %% As in the title, use \\ to force line breaks.
75: 
76: \author{E. M. L. Humphreys, 
77: {M. J. Reid}, 
78: {L. J. Greenhill},
79: {J. M. Moran}, \&  
80: {A. L. Argon}}
81: \affil{Harvard-Smithsonian Center for Astrophysics, 60 Garden Street,
82:    Cambridge, MA 02138}
83: 
84: 
85: \begin{abstract}
86: We report measurements of centripetal accelerations of maser spectral components of NGC~4258
87: for  51 epochs spanning 1994 to 2004.
88: This is the second paper of a series, in which the goal is determination of a new geometric 
89: maser distance to NGC~4258 accurate to possibly \about3\%. 
90: We measure accelerations using a formal analysis method that involves simultaneous  
91: decomposition of maser spectra for all epochs into multiple, Gaussian components.  
92: Components are coupled between epochs by linear drifts (accelerations) from their  
93: centroid velocities at a reference epoch. For high-velocity emission, accelerations lie in the range 
94: $-$0.7 to $+$0.7 \accunits indicating an origin within 13\degrs of the disk midline (the perpendicular 
95: to the line-of-sight to the black hole). Comparison of high-velocity emission projected positions in VLBI images, 
96: with those derived from acceleration data, provides evidence that masers trace real gas dynamics. 
97: High-velocity emission accelerations do not support a model of trailing shocks associated with spiral arms in the disk. 
98:  However, we find strengthened evidence for spatial periodicity in 
99: high-velocity emission, of wavelength 0.75~mas. This supports suggestions of spiral structure due to 
100: density waves in the nuclear 
101: accretion disk of an active galaxy. Accelerations of low-velocity (systemic) emission lie in the range  
102: 7.7 to 8.9 \accunitns, consistent with emission originating from a concavity where the thin, warped disk 
103: is tangent to the line-of-sight. 
104: A trend in accelerations of low-velocity emission, as a function of Doppler velocity, 
105: may be associated with disk geometry and orientation, or with the presence of spiral structure.
106: %%%%%%%%%%%%%%%%%%%%%%%44 
107: 
108: 
109: 
110: \end{abstract}
111: 
112: %% Keywords should appear after the \end{abstract} command. The uncommented
113: 
114: %% example has been keyed in ApJ style. See the instructions to authors
115: %% for the journal to which you are submitting your paper to determine
116: %% what keyword punctuation is appropriate.
117: 
118: %% Authors who wish to have the most important objects in their paper
119: %% linked in the electronic edition to a data center may do so in the
120: %% subject header.  Objects should be in the appropriate "individual"
121: %% headers (e.g. quasars: individual, stars: individual, etc.) with the
122: %% additional provision that the total number of headers, including each
123: %% individual object, not exceed six.  The \objectname{} macro, and its
124: %% alias \object{}, is used to mark each object.  The macro takes the object
125: %% name as its primary argument.  This name will appear in the paper
126: %% and serve as the link's anchor in the electronic edition if the name
127: %% is recognized by the data centers.  The macro also takes an optional
128: %% argument in parentheses in cases where the data center identification
129: %% differs from what is to be printed in the paper.
130: 
131: \keywords{cosmology: distance scale --- galaxies: active: individual(\objectname[M 106]{NGC~4258})
132: ---  galaxies: nuclei ---  accretion, accretion disks --- masers --- instabilities}
133: 
134: 
135: %% From the front matter, we move on to the body of the paper.
136: %% In the first two sections, notice the use of the natbib \citep
137: %% and \citet commands to identify citations.  The citations are
138: %% tied to the reference list via symbolic KEYs. The KEY corresponds
139: %% to the KEY in the \bibitem in the reference list below. We have
140: %% chosen the first three characters of the first author's name plus
141: %% the last two numeral of the year of publication as our KEY for
142: %% each reference.
143: 
144: \section{Introduction}
145: 
146: The Hubble constant (H$_\circ$) is a cornerstone of the extragalactic distance scale (EDS) 
147: and is a fundamental parameter of any cosmology.  Measured period-luminosity (P-L) relations 
148: for Cepheid variable stars have been used to determine the EDS, based on estimation of the 
149: distances to galaxies within $\sim$30~Mpc and calibration 
150: of distance indicators that are also found well into the Hubble Flow (e.g., the Tully-Fisher relation 
151: and type-1a supernovae).  
152: The best present estimate of H$_\circ$ \citep{Freedman2001} is accurate to 10\%, 
153: 72$\pm3$(random)$\pm7$(systematic) km\,s$^{-1}$\,Mpc$^{-1}$.  Other estimates of H$_\circ$ 
154: are generally consistent, though sometimes model dependent
155: \citep[e.g., microwave background fluctuations;][]{Spergel2006}
156:  or somewhat outside the $\pm$10\% uncertainties
157:  \citep[e.g., H$_\circ$ = 62 km\,s$^{-1}$\,Mpc$^{-1}$;][and references therein]{Sandage2006}.
158: 
159: 
160: Several sources of systematic error affect the accuracy of Cepheid luminosity calibrations.  
161: These include uncertainty in the distance to the Large Magellanic Cloud (LMC), which determines 
162: the zero point of the P-L relation and for which recent estimates differ by up to $\pm$0.25 mag, or $\pm$12\% 
163: \citep{Benedict2002}, and uncertainty in the impact of metallicity on the P-L relation 
164: \citep[e.g.,][]{Udalski2001,Caputo2002,Jensen2003}.
165: This is particularly important because the LMC is metal-poor with respect to other 
166: galaxies in the Freedman et al. study.  For a description of controversy concerning the estimation of 
167: H$_\circ$ see \citet{Macri2006} and \citet[][hereafter Paper I]{Argon2007}.  
168: 
169: For the nearby active galaxy NGC~4258, independent distances may be obtained from analysis of Cepheid brightness 
170: and water maser positions, velocities, and accelerations.  Comparison of the distances would enable refinement of 
171: Cepheid calibrations and the supplementation or replacement of the LMC as anchor, thus reducing uncertainty in H$_\circ$ 
172: and the EDS.  Implications may include a better constraint of cosmological parameters, such as the flatness 
173: of the Universe and the 
174: equation of state for dark energy \citep[e.g.,][]{Hu2005,Spergel2006}, which would discriminate among different 
175: origins, e.g., the cosmological constant and quintessence.
176: 
177: Compact maser emission in NGC~4258, from the 6$_{16}$-5$_{23}$ transition of water (22235.080 MHz) was 
178: first detected near to the systemic velocity (v$_{sys}$) by \citet{Claussen1984}, where 
179: v$_{sys}$ = 472$\pm$4 \velunits 
180: \citep[referenced to the Local Standard of Rest (LSR), and using the radio 
181: definition of Doppler shift;][]{Cecil1992}. The discovery of high-velocity emission at \about v$_{sys}\pm$1000 \velunits provided critical evidence that maser emission probably arises from material in orbit about a 
182: massive central black hole \citep{Nakai1993}. Early Very Long Baseline Interferometry (VLBI) observations 
183: showed that masers delineated an almost edge-on, sub-parsec-scale rotating disk \citep{Greenhill1995a}. 
184: Studies of maser centripetal accelerations, inferred from secular velocity drifts in the peaks of 
185: spectral components, produced independent evidence of a disk geometry 
186: \citep{Haschick1994,Greenhill1995b,Nakai1995,Watson1994}. Low-velocity masers 
187: (near to v$_{sys}$) displayed positive drifts of $\sim$9 \accunitns, which located them 
188: on the front side of the disk, whereas high-velocity masers drifted by $<$ $\pm$1 \accunitns, 
189: confining emission to lie close to the disk midline (the diameter perpendicular to the line of sight). 
190: Very Long Baseline Array \footnotemark[1] (VLBA) studies provided conclusive evidence supporting the disk model, 
191: established a Keplerian rotation curve to better than 1\% accuracy for maser emission, 
192: and traced a warp in the disk structure (\citealt{Miyoshi1995}; \citealt{Herrnstein1996,Herrnstein2005}; 
193:  and early reviews of \citealt{Moran1995,Moran1999}). 
194: 
195: \footnotetext[1]{The Very Long Baseline Array and Very Large Array are operated by the National Radio Astronomy Observatory, a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.}
196: 
197: A geometric distance can be obtained by modeling the 3-D geometry and dynamics traced 
198: by the maser line-of-sight (LOS) velocities, positions, and LOS accelerations or proper motions.  
199: Assuming circular orbits, \citet{Herrnstein1999} obtained the most accurate distance to 
200: NGC~4258 thus far, $7.2\pm0.2$(random)$\pm0.5$(systematic)~Mpc.  The total error is 7\%, and the 
201: systematic component largely reflects an upper limit on the eccentricity of 0.1 for confocal 
202: particle orbits.  This contributes 0.4 Mpc of the systematic uncertainty budget.
203: 
204: This paper is the second of a series in which we revisit estimation of the maser 
205: distance to NGC~4258, using an expanded dataset and more detailed analyses.  Our goal
206: is to reduce the uncertainty by a factor of 2-3, through reduction in both random and systematic errors.  \citet{Herrnstein1999} used four epochs of VLBA data spanning 3 years to estimate distance.  To this, we have added 18 new VLBI epochs over 3.4 years (Paper I).  These combined data will facilitate a more thorough disk modeling, including the incorporation of orbital eccentricity and periapsis angle as parameters.  Here, we address the measurement of radial accelerations for maser emission from a time-series 
207: of spectroscopic measurements, combining VLBI, Very Large Array\footnotemark[1] (VLA), and single-dish spectra.  Future work will present the model of the cumulative data (position, velocity, and acceleration) to obtain a best-fit distance.  
208: 
209: In a parallel series of papers starting with \citet{Macri2006}, analysis of recent 3-color, 
210: wide-field Hubble Space Telescope (HST) Cepheid photometry is used to estimate an improved 
211: ``standard candle'' distance to NGC~4258 \citep[cf.][]{Newman2001}, with supplementary information 
212: from near-infrared and high angular resolution data that place tighter constraints on systematics (crowding and extinction).  
213: 
214: Our approach to the measurement of accelerations is very different from that of previous work, 
215: wherein individual maser spectral Doppler components were identified by local maxima in blended 
216: line profiles at each epoch.  Components were matched up among different epochs on a maximum likelihood 
217: basis or in ``by eye'' analyses, depending on the study, in order to determine drifts in component peak or 
218: fitted centroid velocities \citep{Haschick1994,Greenhill1995b,Nakai1995,Herrnstein1999,Bragg2000}. 
219: For low-velocity emission 
220: in particular, this method is subject to ambiguity in component identification unless the spacing between epochs 
221: is short ($<3$-4 months), because of intensity fluctuations among blended, drifting components.  This informal 
222: fitting technique is also susceptible to biases in the velocities of local maxima due to blending.   Here, we 
223: decompose the spectra into individual Gaussian components for many epochs simultaneously, constrained by a separate 
224: linear velocity drift in time for each feature.  The constraint greatly increases the robustness of the decompositions, 
225:  is less subjective than ``by eye'' techniques, and enables the discernment of weak spectral features with greater confidence. Including the VLBI monitor data reported in Paper I, we applied this fitting to 36 spectra of low-velocity emission spanning $\approx $6 years, and 40 spectra of high-velocity emission spanning $\approx $10 years.
226: 
227: Implicit in the estimation of distance is construction of a disk dynamical model from acceleration, 
228: position, and velocity measurements. These may also be used to detect disk sub-structure.  
229: Analysis of structure on the smallest scales enables identification of maser spots as representative 
230: of physical clumps. Clustering of maser spots enables tests of predicted spiral structure induced by 
231: gravitational instabilities.  \citet{Maoz1995} proposed that spiral density waves form in the NGC~4258 disk, 
232: and the high-velocity masers mark the density maxima along the disk midline, which is where gain paths are greatest.  \citet{Maoz1998} proposed that the spiral pattern 
233: is traced by shocks, with maser emission arising in relatively narrow post-shock regions and visible only 
234: when the spiral arms are tangent to the line-of-sight.  Due to the pitch angle of the spiral pattern, the masers 
235: would arise systematically offset from the midline, and the acceleration signatures predicted by the two models 
236: enable discrimination \citep{Bragg2000}.  
237: 
238: We describe the spectroscopic observations with which accelerations are estimated in Section~\ref{s:data}.  
239: Decomposition of spectra into Gaussian components whose velocities drift in time and the inferred 
240: accelerations are described in Sections~\ref{s:spectrumdecomp} and~\ref{s:results}.  Comparisons with 
241: previous work are in Section~\ref{s:sect5}.  
242: Implications of the new analyses for disk structure follow, with focus on physical identification of 
243: maser clumps, signatures of spiral structure, and strong constraints on orbital eccentricity.
244: The next paper in this series will give the improved distance to NGC~4258 estimated
245: from analyses of maser data presented in Paper I and this paper.
246:   
247: \section{The Dataset}
248: \label{s:data}
249: 
250: The primary purpose of this paper is to report the measurements of maser component accelerations with high accuracy 
251: to be used in determination of a new geometric distance to NGC~4258. We attempted to 
252: reduce statistical uncertainties in the accelerations by inclusion of a larger number of 
253: epochs over a longer time baseline than that done by \citet{Herrnstein1999}. In this study, we analyzed 
254: spectra of water maser emission at 22.235 GHz from NGC~4258 at fifty-one epochs, between 1994 April 19 
255: to 2004 May 21. These data were presented in detail in Paper I. We summarize the observations in 
256: Table~\ref{t:dataset} and display subsets 
257: of the data in Figures~\ref{f:figure1} and~\ref{f:figure2}.
258: 
259: A compilation of the spectra measured from 1994 April 19 to 1997 February 10 was 
260: presented by \citet{Bragg2000}. Observations were made approximately 1 to 2 months using 
261: the VLBA (5 epochs), the VLA (17 epochs) and the Effelsberg 100-m telescope of the 
262: Max-Planck-Institut f\"{u}r Radio Astronomie (5 epochs). The 1$\sigma$ noise in the 
263: spectra range from 15 to 110 mJy for channel spacings of $<$0.35 \velunits (Table~\ref{t:dataset}). 
264: At the majority of epochs using the VLA and VLBA, both low- and high-velocity emission 
265: were measured, however some exceptions are noted in Table~\ref{t:dataset}. Only the red-shifted high-velocity 
266: spectrum was observed at Effelsberg.
267: 
268: The spectra from eighteen VLBI epochs, measured between 1997 March 06 to 2000 August 12, 
269: are displayed in Paper I. Twelve of the epochs used the VLBA only (the ``medium-sensitivity'' epochs) 
270: and had 1$\sigma$ noise in the range 3.6 to 5.8 mJy for 1.3 \velunits wide channels.  At 
271: medium-sensitivity epochs, low-velocity emission plus either red-shifted or blue-shifted 
272: high-velocity emission were observed. The observing set-up was designed to enable exploration 
273: of a continuous range of velocities between known spectral-line complexes, and a limited range to higher velocities.  
274: The remaining six epochs 
275: (the ``high-sensitivity'' epochs) involved the VLBA, the phased VLA and Effelsberg. 
276: The instantaneous bandwidth of the high-sensitivity observations was sufficient to perform 
277: simultaneous imaging of known low- and high-velocity emission, and resulted in 1$\sigma$ noise 
278: of  2.3 to 4.7 mJy per \velunitns. 
279: The average time between the observations was \about2.5 months. We extracted spectra from 
280: VLBI images by fitting 2-D Gaussian brightness distribution functions to all peaks in velocity channel maps. 
281: For a detailed description of the data analysis method, see Paper I. 
282:  Of the remaining epochs, three were obtained using the Green Bank Telescope 
283: (GBT) between 2003 April 10 and 2003 December 8, with 1$\sigma$ noise levels 
284: of \about3 mJy per 0.16 \velunits channel (Modjaz et al.\,2005; Modjaz, priv.\,comm.\,2005; Kondratko, priv.\,comm.\,2006). 
285: We obtained the final epoch of data on 2004 May 21 
286: using the VLA, for limited portions of the red-shifted high-velocity spectrum only, and with 
287: an r.m.s noise of 20 mJy per 0.21 \velunits channel. 
288: 
289: 
290: 
291: \section{Multi-Epoch Spectrum Decomposition}
292: \label{s:spectrumdecomp}
293: 
294: 
295: To measure accelerations of maser components, as well as amplitude and line width variations during
296: the monitoring, 
297: we decomposed portions of spectra from multiple epochs simultaneously into individual Gaussian line profiles. 
298: We identified a component at different epochs by solving 
299: for its constant drift from a centroid velocity 
300: at a reference time. 
301: We could therefore account for each Gaussian at every epoch, 
302: even if the amplitude 
303: had fallen below our detection threshold, which minimized potential 
304: for confusion among components.
305: 
306: We employed a non-linear, multiple Gaussian-component least-squares $\chi^2$ minimization 
307: routine that we dimensioned for a maximum of 84 time-varying Gaussians and 
308: 40 epochs of data in any given fit, which was limited by computational factors. Each maser component was represented by 
309: a Doppler velocity $v_{los}(t_{0})$ at a reference time $t_{0}$, a linear velocity drift $\dot{v}_{los}$, 
310: and a line width and amplitude at every epoch. We could choose to solve 
311: for constant, or time-varying, component line widths. We set {\it a priori}  
312: line widths in order to prevent large deviations from physical values 
313: expected for maser line widths at gas kinetic temperatures of 400 to 1000~K. 
314: The {\it a prioris} were included as extra 
315: data points in the fit, and typically prevented deviations of greater than four times the {\it a priori} uncertainties. 
316: The reference time was chosen to be near the middle of the monitoring period. 
317: All velocities in this paper are quoted for a $t_{0}$ of 1999 October 10, or monitoring day 2000, unless stated otherwise.
318:  
319: 
320: We note that fitting of linear velocity drifts is an approximation for particles on circular orbits 
321: in a disk in which $v_{los}=v_{rot}\cos(\Omega \Delta t+\phi_0) + v_{sys}$, where v$_{rot}$ 
322: is the rotational velocity at any given $r$, $\Omega$ is the angular velocity, 
323: $\Delta t=t - t_{0}$ and $\phi_0$ is the angle from the midline at a reference time t$_{0}$. 
324: From Monte Carlo simulations, we estimate that the maximum error 
325: we introduce by using this approximation is 0.1 \accunitns. We chose the linear approximation in order to determine 
326: velocity drifts without any model assumptions. 
327:  
328: \subsection{High-Velocity Emission}
329: 
330: The high-velocity spectrum consists of isolated blends of small numbers of components, 
331: with low drift rates (Figures~\ref{f:figure1} and~\ref{f:figure2}).  To facilitate the decomposition, we divided 
332: the high-velocity spectra into individual blends, of typical velocity extent 10 to 20 \velunitns. We performed the fitting of each blend 
333: iteratively. First, we identified the number of prominent peaks in the blend at a 
334: high-sensitivity and high-spectral resolution epoch (e.g., 1998 September 5). 
335: We fit this number of Gaussian functions to the data over all epochs and examined the residuals at each epoch. Where 
336: deviations in the residuals exceeded 5$\sigma$, we refit the segment using more Gaussian functions. 
337: We repeated this procedure until there was no systematic structure remaining in the residuals 
338: above the 5$\sigma$ level. We solved for constant line widths as a function of time for high-velocity 
339: emission, because the signal was typically $< $1~Jy,  and line widths at any given epoch were 
340: otherwise not well-constrained. The fits for each high-velocity blend included between four and 
341: forty epochs of data, covering time baselines of 0.5 to 9 years (Tables~\ref{t:table2} and~\ref{t:bluetable}) depending on 
342: component lifetime and time-sampling. Example fits are shown in Figures~\ref{f:redfit} and~\ref{f:bluefit} for red- and 
343: blue-shifted emission, respectively.
344: 
345: 
346: 
347: \subsection{Low-Velocity Emission}
348: 
349: The low-velocity emission spectrum consists of strong (typically 2 to 10~Jy), highly-blended emission 
350: for which previous work has shown components drift at a mean rate of 
351: \about9 \accunits \citep{Haschick1994,Greenhill1995b,Nakai1995,Herrnstein1999,Bragg2000}. 
352: Some studies
353: note a systematic trend in accelerations of low-velocity emission
354: as a function of Doppler velocity, with accelerations of 8.6$\pm$0.5 \accunits at velocities less than 470 \velunitns, and
355:  of 10.3$\pm$0.6 \accunits at velocities greater than 470 \velunits \citep{Haschick1994,Greenhill1995b}. Others 
356: do not draw attention to such an effect \citep{Nakai1995,Herrnstein1999,Bragg2000}.   
357: We therefore approached decomposition of the low-velocity spectra using two methods: (i) assuming
358: that the linear velocity drift approximation holds over the entire 36 epoch monitoring, and including all 36 epochs
359: of data; and, 
360: (ii) assuming that there may be changes in accelerations during the monitoring period, 
361: and dividing the data in four consecutive datasets of nine epochs, for which fitting was performed
362: independently. When weighting the data for fitting, we added 2\% of the channel flux density 
363: in quadrature to the 1$\sigma$ noise, to account for the dynamic range limitations of our observations (see Paper I).
364: 
365: 
366: 
367: To perform the 36-epoch fitting, we first decomposed the 
368: spectrum into a {\it preliminary} set of Gaussian components at one epoch only. We selected 
369: a high-spectral resolution and high-sensitivity epoch for this purpose (1998 September 5). 
370: We performed a fit over a subset 
371: of 6 adjacent epochs for the low-velocity spectrum as a whole, to obtain preliminary values of acceleration. 
372: We increased the number of Gaussian functions used in the fitting with each successive iteration, adding additional Gaussians to the fit, until no 5$\sigma$ systematic deviations 
373: remained (in the residuals).
374: This analysis provided a set of ``seed'' parameters 
375: to perform a new decomposition over 36 epochs. For ease of computation, 
376: we divided the spectrum into segments of \about12 Gaussian components 
377: and performed the decomposition of each 
378: segment iteratively. We used a moving time window 
379: to select the relevant data range for segments at different 
380: epochs in order to track the $\sim$9 \accunits accelerations. At segment edges, we overlapped 
381: fitted regions in each case 
382: by 2 to 3 components to ensure that fits were 
383: consistent across the low-velocity spectrum. 
384: Using this method, we obtained residual deviations of $<$ 5$\sigma$
385: everywhere except at the time extrema of our dataset, where we obtained significantly larger deviations
386: in residuals.
387: 
388: 
389: 
390: For the 9-epoch fits, we also decomposed the 
391: spectrum into a preliminary set of Gaussian components at one epoch only (chosen to be near the middle of each dataset). 
392: We then performed an initial fit over the entire 9 epochs to obtain component accelerations.
393: Unlike the 36-epoch fitting, it was not necessary to sub-divide
394: the spectrum and use a moving time window, and we performed the decomposition
395: across the low-velocity spectrum as a whole in each case.
396: We increased the number of Gaussians in each of the fits iteratively, examining 
397: the residuals after each fit, and adding components where large deviations occurred.
398: We repeated this process until no $>$ 5$\sigma$ systematic deviations remained in the residuals. 
399: In each of the 9-epoch fits, this required \about55 Gaussians of line width 1 to 4 \velunitns.
400: The four, independent fits to the data indicated that there is a persistent trend of component 
401: accelerations as a function of Doppler velocity (or equivalently time), explaining why a good fit was not obtained in the
402: longer time-baseline, 36-epoch method. We adopt results from the 9-epoch fits in the sections
403: that follow.
404: 
405: 
406: \section{Results}
407: \label{s:results}
408: 
409: \subsection{High-Velocity Emission}
410: \label{ss:sect4.1}
411: 
412: 
413: We measured accelerations for 24 red- and 8 blue-shifted high-velocity components 
414: and found them to range between $-$0.40 and $+$0.73 \accunitns, and between 
415: $-$0.72 and $+$0.04 \accunitns, respectively (Tables~\ref{t:table2} and~\ref{t:bluetable}). 
416: We found a weighted 
417: average of red-shifted high-velocity accelerations of 0.02$\pm$0.06  \accunitns, 
418: and a weighted average blue-shifted high-velocity accelerations 
419: of $-$0.21$\pm$0.08 \accunitns. We note that there is a non-normal distribution 
420: with respect to acceleration systematics due presumably to disk structure. We estimated the azimuth angle, $\phi$, 
421: of maser components from the disk midline for a flat disk model in terms of measurable quantities; 
422: the line-of-sight velocity $v_{los}$ with respect to v$_{sys}$ and line-of-sight acceleration $\dot{v}_{los}$, 
423: adopting $\phi \approx \sin \phi / \cos^4 \phi = GM_{BH} \dot{v}_{los} / v^4_{los}$ 
424: \citep{Bragg2000}, where $G$ is the gravitational constant, $M_{BH}$ is the 
425: central mass and where $\phi$ is measured from the midline for red-shifted 
426: high-velocity emission, increasing in the sense of disk rotation. For red-shifted 
427: high-velocity emission, we found that azimuth angles lie in the range -8.9\degrs to 
428: +12.6\degrs and have a mean and standard deviation of 0.2$\pm$3.6$\rm ^o$, i.e., 
429: are centered on the midline. For blue-shifted high-velocity 
430: components we determined that azimuth angles lie in the range 166.8\degrs to +180.5\degrs 
431: and have a mean value of 176.7$\pm$6.9$\rm ^o$. Since for a warped disk $\phi$ $\propto$ $\sin^3 i$, there
432: is a 0.2 to 2\% error in these values \citep{Herrnstein2005}. 
433: Using a flat, edge-on 
434: Keplerian disk model and distance of 7.2~Mpc \citep{Herrnstein1999}, we derived 
435: disk radii of high-velocity emission in the range 0.17 to 0.29~pc. 
436: We measured line widths in the range 1.0 to 5.0 \velunits for the high-velocity emission. 
437: 
438: In Paper I, Argon et al.\,(2007) reported the discovery of new red-shifted 
439: high-velocity emission at 1562 \velunitns, and also detection of emission at 1652 \velunits previously
440: discovered at the GBT by \citet{Modjaz2005}.  We were 
441: unable to estimate the accelerations of these components due to blending and the limited data available. 
442: 
443: \subsection{Low-Velocity Emission} 
444: \label{ss:sect4.2}
445: 
446: We measured accelerations for four, time-consecutive datasets consisting of 9 epochs each. 
447: We found a significant, reproducible trend in acceleration as a function of 
448: component Doppler velocity for each data subset with a systematic variation of  1 \accunits across
449: the low velocity emission range of $\sim$430 to 550 \velunitns.
450: We binned the data from all four fits to yield measured accelerations 
451: in the range 7.7 to 8.9 \accunits for 55 low-velocity components (Figure~\ref{f:systrend}; Table~\ref{t:systable}).
452:  The acceleration trend may be associated with a systematic change in the radius
453: at which maximum velocity coherence is achieved due to the disk warp (see Section~\ref{s:implications_lv}). 
454: We performed a weighted fit to accelerations as a function of velocity and computed a
455: PV diagram that would be consistent from $v_{los}=\sqrt{GM/r^3}b$ where $b$ is the
456: impact parameter and $a_{los}=(GM/r^2)\sin\phi$ for $\phi$=270\degrs and assuming circular rotation. The predicted
457: and observed PV diagrams agree to within 1$\sigma$ for velocities from 390 to 540 \velunitns (Figure~\ref{f:systrendpv}).
458: For the disk shape of \citet{Herrnstein2005}, 
459: we find the maser components
460: lie in a range of deprojected disk radii of 3.9 to 4.2~mas as shown in Figure~\ref{f:velcontours}
461: (0.14 to 0.15~pc at a distance of 7.2 Mpc)
462:  i.e., $<$10\% of the radial extent of observed
463: high-velocity emission.
464: We discuss possible physical origins for the trend in Section~\ref{s:implications_lv}.
465: 
466: \section{Comparison with Previous Work}
467: \label{s:sect5}
468: 
469: Over a monitoring period of ten years, 
470: a high-velocity component orbiting with a period of 700 years travels 5\degrs 
471: in disk azimuth angle, almost entirely along the line-of-sight. In theory, maser components 
472: could therefore cross the midline during our monitoring, or during the time gaps among studies, 
473: and undergo a change in the sign of acceleration. 
474: However, a comparison of the different acceleration studies (Table~5) shows that
475: they are in broad agreement. In particular, we compare our acceleration measurements for 
476: high-velocity emission with those of \citet{Bragg2000} and \citet{Yamauchi2005} (Figure~\ref{f:acccomp}). 
477: The main difference between our results and those of previous studies is that we measure accelerations 
478: for more components due to the greater sensitivity and spectral resolution of our data, and due to the 
479: greater accuracy of the technique we employ to obtain more complete spectrum decomposition. 
480: 
481: High-sensitivity and resolution are of particular importance with respect to blue-shifted high-velocity components. 
482:  The addition of the VLBI data of Paper I enabled measurement of accelerations for 
483: 8 blue-shifted components, whereas \citet{Bragg2000} measured accelerations for 
484:  2 blue-shifted high-velocity components only: 0.043$\pm$0.036 \accunits at -440 \velunits 
485: and -0.406$\pm$0.070 \accunits at -434 \velunitns. \citet{Yamauchi2005} 
486: obtained a single measurement of 0.2$\pm$0.1 \accunits for emission at -287 \velunitns. 
487: We obtain similar accelerations for the components also measured by \citet{Bragg2000}. 
488: However, we spectrally resolve the -287 \velunits emission into 3 components, with 
489: accelerations of  -0.47, -0.12 and -0.38 \accunits (Figure~\ref{f:bluefit}). 
490: Fluctuations in blended features may have led 
491: \citet{Yamauchi2005} to obtain a small positive apparent drift rate. If we fit the 
492: emission at -287 \velunits as a single Gaussian component, we also obtain a 
493: positive acceleration in agreement with \citet{Yamauchi2005} within the uncertainties.
494: 
495: The centripetal accelerations for low-velocity emission lie in the same range as 
496: those measured in previous studies \citep{Haschick1994,Greenhill1995b,Nakai1995,Herrnstein1999,Bragg2000}, 
497: although
498: binned measurements of the present work tend to be lower at any given velocity, see Figure~\ref{f:lowvelacccomp}.
499: This could be due to components at larger radii in the disk, or could be associated with the more robust
500: measurement technique employed here.  
501: A trend in accelerations as a function of component Doppler velocity, with the same sign and 
502: similar magnitude as that measured here, is also reported by \citet{Haschick1994} and by 
503: \citet{Greenhill1995b}, and is evident in the acceleration data of \citet{Nakai1995} and
504: \citet{Herrnstein1999}.
505:  The geometric model of the maser disk by \citet{Herrnstein2005} 
506: indicates that it is warped, both in inclination and position angles, 
507: the result of which is to place low-velocity masers in a concavity on the 
508: front side of the disk (Figures~\ref{f:herrndisk} and~\ref{f:velcontours}). The range of accelerations for 
509: low-velocity emission determined by this, and other 
510: studies, could imply a radial spread in low-velocity masers
511: over the relatively broad bottom of the concavity, which is 
512: consistent with the model \citep{Herrnstein2005} and the PV diagram (Figure~\ref{f:systrendpv}).
513: 
514: \section{Implications for the Accretion Disk from High-Velocity Emission}
515: 
516: \subsection{Disk Geometry}
517: 
518: We can use the accelerations to confirm that the maser dynamic observables 
519: reflect real gas dynamics in the NGC~4258 accretion disk.
520: The small accelerations of the 
521: high-velocity components suggest that they lie close to the disk midline. 
522: For this discussion, it is important to visualize the shape of the disk 
523: in the vicinity of the midline, which is in plane of the sky that includes the 
524: black hole. We show the geometry of the maser disk determined by 
525: \citet{Herrnstein2005} in Figure~\ref{f:herrndisk}. 
526: On the midline, the disk shape is defined by the position angle warp, 
527: as shown in Figures~\ref{f:redsky} and~\ref{f:bluesky} respectively for the red- and blue-side of the disk. 
528: That is, the expected midline curve is $y=r\cos\alpha_{r}$, where $y$ is the projected 
529: vertical position of the warped disk 
530: on the sky relative to the disk dynamical center in units of milli-arcseconds, $\alpha_{r}$ is 
531: the disk position angle measured North of East in the plane of the sky (the x-y plane
532: in Figure~\ref{f:herrndisk}) given by 
533: $\alpha_r=65.6$\hbox{$^{\circ}$}$+5.04$\hbox{$^{\circ}$}$r-0.22$\hbox{$^{\circ}$}$r^2$
534: \citep{Herrnstein2005}, 
535: and where $r$ is the radial position of masers in the disk in mas. We take $r$ \about $b$, 
536: where $b$ is the impact parameter in the plane of the sky measured from the disk dynamical 
537: center.
538: 
539: Over the radial range of emission from 4 to 8~mas on the red-side of the disk, the position 
540: angle varies from 84\degrs to 92\degrns. The ``tilt'' of the disk at the midline 
541: is given by the inclination warp, which is described by $i_r = 107$\hbox{$^{\circ}$}$ - 2.29$\hbox{$^{\circ}$}$ r$  
542: \citep{Herrnstein2005}. Hence, for the red side of the disk 
543: $i_{r}$ varies from 97\degrs at 4~mas, to 89\degrs at 8~mas. The range is the same 
544: for the blue-side of the disk. It is important to note that, at radii of 7.4~mas, we view 
545: the disk exactly edge on. Inside this radius, the near-side of the disk is 
546: tipped down with respect to the observer, while outside it is tipped up. Masers at radii smaller 
547: than 7.4~mas, that lie above the midline in VLBI images, should be behind the midline plane in the disk and have negative 
548: accelerations. 
549:  Similarly masers 
550: beyond 7.4~mas radius, that appear above the midline in Figures~\ref{f:redsky} and~\ref{f:bluesky}, 
551: are expected to be in front of the midline plane and should have positive accelerations. 
552: 
553: We compared the vertical 
554: deviations in the VLBI image caused by the disk projection on the sky (see Figures~\ref{f:redsky} 
555: and~\ref{f:bluesky}), $y_{image}$, 
556: with the vertical deviations expected from 
557: the accelerations, $y_{acc}$. To derive $y_{acc}$ for each maser, we first computed 
558: its distance in the $z$-direction from the midline from its azimuth (which comes from its acceleration), that is 
559: $z = b \tan \phi$ where $\phi$ is the azimuth angle of the maser listed in Tables~\ref{t:redtable}
560: and~\ref{t:bluetable}. 
561: From this azimuth offset, and the local slope of the warped disk at radius $r$,  $y_{acc}$ 
562: was calculated as the projected offset from the midline. The plot of $y_{acc}$ vs.
563: $y_{image}$ is shown in Figure~\ref{f:jim}. The errors 
564: in $y_{image}$ are the quadrature sum of the errors in the VLBI position and 
565: the errors in the midline definition.  The errors in $y_{acc}$ are the result of 
566: acceleration measurement errors, inclination model errors, and errors due 
567: to a finite vertical distribution of masers. We adopt 
568: a disk thickness of 12 $\mu$as, the value determined from
569: low-velocity emission in Paper I.
570: The disk 
571: thickness corresponds to a temperature of about 600 K for hydrostatic equilibrium.
572:  In order to make the reduced $\chi^2$ of the
573: fit equal 1, it was necessary to add a systematic r.m.s. acceleration of 0.1 \accunitns.  
574: Since the magnitude of the maser accelerations is up to \about10 \accunitns, 
575: the systematic errors in the acceleration measurements are at the 1\% level.
576: The errors in Figure~\ref{f:jim} reflect 
577: these contributions. Note that there is one significantly 
578: deviant point at 2.8$\sigma$ (the one at 1271 \velunits and $r=$7.37~mas).
579: 
580: 
581: \subsection{Spiral Structure} 
582: 
583: Accelerations and positions of high-velocity emission can also be used to 
584: investigate theories of spiral structure in the 
585: accretion disk of NGC~4258. Firstly, \citet{Maoz1995} noted
586:  periodic clustering of high-velocity component radii in a visual analysis of
587:  the single VLBI map by \citet{Miyoshi1995}. Maoz inferred a characteristic wavelength 
588: of $\lambda_{char}=$0.75~mas (0.027~pc) for the clustering, and attributed it to spiral density waves 
589: that form by swing amplification.  Spiral density waves can form in a disk that is 
590: unstable to non-axisymmetric perturbation in the 1$< Q \le$2 regime, 
591: where $Q$ is the Toomre stability parameter \citep{Toomre1964}. {\it Swing} refers to the 
592: fact that a non-axisymmetric disturbance is converted to a trailing spiral arm (in which the outer tip points backwards, opposite the direction of disk rotation) due 
593: to shearing by differential rotation. {\it Amplification} occurs during this process, 
594: since the direction of the shear flow of the arm is the same direction as local epicyclic 
595: motions, which enhances the self-gravity of the arms. 
596: To satisfy this stability regime, \citet{Maoz1995} required
597: densities in spiral arms of n(H$_{2}$) 
598: = 10$^{8}$ to 10$^{10}$~cm$^{-3}$ which are sufficient to produce 22~GHz maser emission   \citep[e.g.][]{Cooke1985}, whilst those 
599: in the inter-arm regions are not.
600: In the theory of \citet{Maoz1995}, masers are discrete 
601: physical clumps that can move freely through different spiral arms and can be tracked 
602: between epochs. 
603: 
604: 
605: \citet[][hereafter MM98]{Maoz1998} also propose that maser emission 
606: could arise behind spiral {\it shocks} in the NGC~4258 disk, from density and velocity coherence arguments. The shocks 
607: occur at the interface of spiral arms sweeping through the disk. In 
608: this case masers may not be discrete entities but a wave phenomenon, 
609: with emission arising at the tangents of the spiral shocks along the 
610: LOS, where the path length for amplification is highest. At 
611: different epochs, maser emission would not originate 
612: from the same gas condensations. The accelerations of the masers would no longer 
613: reflect pure Keplerian dynamics, as the action of the shock causes the excitation 
614: points for masers to move to greater radii in the disk. All maser velocities  
615: would shift to lower absolute values. This process would manifest itself observationally as 
616: positive accelerations for blue-shifted high-velocity 
617: components and as negative accelerations for red-shifted high-velocity components. Different spiral theories are schematically represented in Figure~\ref{f:spiralcomp}.
618: 
619: \subsubsection{Spiral Density Wave Model \citep{Maoz1995}}
620: 
621: \citet{Maoz1995} argues that 
622: (i) maser components are periodic in disk radius with a characteristic wavelength 
623: of $\lambda_{char}\sim$0.75~mas; and  (ii) masers are located within 
624: several degrees of the disk midline. Point (ii) is required to avoid a wide range of 
625: accelerations of the masers due to spiral arms, which are not observed. If masers are 
626: near the midline, then the non-circular (peculiar) motion is largely 
627: perpendicular to the arms \citep{Toomre1981} and perpendicular 
628: to the line-of-sight (and therefore undetectable in Doppler shifts). \citet{Maoz1995} 
629: estimates that the component of the peculiar motion along the line-of-sight for high-velocity emission would 
630: not exceed 4.5 \velunitns, comparable to the measured 
631: accuracy of the Keplerian rotation curve \citep{Maoz1995}.
632: 
633: An apparent periodicity in the distribution of maser sky positions is obvious from VLBI 
634: images of the data. For formal quantification, we performed an 
635: analysis using the Lomb-Scargle periodogram for the 18 epochs of VLBI data 
636: described in Paper I.  First, we created a ``binary'' dataset from the VLBI 
637: positions, in which any disk position that had detected emission was assigned a ``flux density'' 
638: of ``1'', to ensure that disparity in flux density between red and blue-shifted emission 
639: did not affect the outcome of the analysis. We gridded the binary dataset 
640: at intervals of 0.005~mas, i.e., on scales much shorter than any expected $\lambda_{char}$, and 
641: placed zeroes at positions at which we did not detect, emission 
642: in the VLBI maps. Using the periodogram, we found a dominant period 
643: in the binary distribution of high-velocity 
644: maser positions of 0.75~mas (Figure~\ref{f:periodogram}), at the same value as 
645: that estimated by \citep{Maoz1995}.  We also predicted that emission 
646: should exist at a velocity of about $-$330 \velunits in the blue-shifted spectrum. 
647: In subsequent observations obtained with the GBT on 2003 October 23, the emission predicted 
648: by the periodogram analysis was detected at $-$329 \velunitns. 
649: At $< $10 mJy, this emission would not have been detected in our VLBI observations.
650:   
651: We note that the model of \citet{Maoz1995} has densities in the spiral arms that can produce
652: maser emission, but not in the inter-arm regions. In the denser spiral arms, gas condensations 
653: naturally form. 
654: In this event, discrete maser-emitting condensations could be
655: tracked as they move across an arm.
656: They are not necessarily persistent in the
657: inter-arm region, i.e. discrete maser condensations may only exist
658: within arms and disperse in the inter-arm regions. 
659: As a result, clumps responsible for low-velocity emission do not survive to beam
660: high-velocity emission toward the observer after one quarter of a rotation, and
661: conversely, high-velocity clumps do not survive to eventually be seen as low-velocity
662: clumps.
663: 
664: 
665: 
666: \subsubsection{Spiral Shock Model (MM98)}
667: 
668: The specific predictions of MM98 for masers, in addition to periodicity 
669: in maser sky positions, are that (i) blue-shifted 
670: components are accelerating and they are located behind the disk midline, and 
671: (ii) red-shifted components are decelerating and should be located in front 
672: of the midline. By contrast, in a disk dominated by Keplerian dynamics, masers 
673: in front of the midline should have a positive sign of acceleration, and masers 
674: located behind the midline should have a negative sign of acceleration.
675: 
676: We do not find that the accelerations of high-velocity components conform to 
677: the MM98 model.  Both the red- and blue-shifted component accelerations are 
678: statistically near zero (see Section~\ref{ss:sect4.1} and Tables~2 and~3), although there is a 2.6$\sigma$ 
679: bias of the latter to negative accelerations which is not in agreement with predictions of MM98. 
680: We note that \citet{Yamauchi2005} 
681: presented evidence in support of MM98. However, \citet{Yamauchi2005} measured the 
682: acceleration of one blue-shifted component only, which had a positive acceleration 
683: at the 2$\sigma$ significance level. We obtain a negative acceleration for the 
684: same component (see Section~\ref{s:sect5}) when it is decomposed into 3 features. 
685: Of the eight blue-shifted components for which 
686: we measured accelerations, only one is positive.                   
687: \citet{Bragg2000} also does not find accelerations that would support the MM98 theory.
688: 
689: \subsubsection{Other Origins for the Periodicity}
690: 
691: We also considered the possibility that orbiting objects (e.g., stars) 
692: create the sub-structure in the NGC~4258 accretion disk, with for example some resonance
693: mechanism causing regularity in the spacings.  If the stellar density is the 
694: same in the central regions of NGC~4258 as in the center of the Milky Way, then integrating 
695: $\rho_* (R) = 1.2 \times 10^6 (R/0.425)^{-\alpha}$
696: between $R=$0.14 and 0.28~pc and assuming $\alpha$=1.4 \citep[from][]{Genzel2003} yields a 
697: stellar mass of  2$\times$10$^{4}$ M$_{\odot}$ in the spherical volume containing the maser disk, 
698: so there are likely at least 
699: several thousand stars. 
700: The cumulative effect of many passages of stars around the black hole might bring them 
701: into a circular orbit co-rotating with the disk, although we note that highly eccentric 
702: stellar orbits are found in our Galactic Center. Each star could then either open up a 
703: gap in the disk (if the disk is thinner than the star's Hill radius and the viscosity 
704: in the disk is sufficiently low) or accrete material from the disk. The vertical thickness 
705: of the maser disk is measured to be 12~$\mu$as (Paper I). We therefore estimate the lower 
706: limit on the mass of a star (or stellar cluster) required to create a gap completely 
707: through the disk to be $>$ 0.05 M$_{\odot}$ ($2r_{Hill} \approx 2 r_{orb}(m_*/3M_{bh})^{1/3} = 0.0003$ pc for
708: $\overline{r_{orb}}$=0.2 pc), such that any star, even a brown dwarf, 
709: could create a gap in the disk equal to its thickness. We note that a number of 
710: massive He I stars (of mass 30 to 100 M$_\odot$) exist between 0.04 to 0.5~pc (1$''$ to 12$''$) of the 
711: Galactic center \citep{Genzel2003}, and that a corresponding \about100 M$_{\odot}$ star 
712: in the NGC~4258 disk would create a gap of $2r_{hill/roche}$=0.006 pc. The stars would be at radii several 
713: orders of magnitude greater distances from the SMBH than their tidal disruption radii $r_{tidal} \simeq r_* (M_{bh}/m_*)^{1/3}$, where $r_{*}$ and m$_{*}$ are the radius and mass of the star respectively \citep[see e.g,][]{Bogdanovic2004}.  
714: 
715: 
716: \section{Implications for the Accretion Disk from Low-Velocity Emission}
717: \label{s:implications_lv}
718: 
719: The range of accelerations measured for low-velocity components is consistent
720: with the warped disk model of \citet{Herrnstein2005} (see our Figures~\ref{f:herrndisk} and~\ref{f:velcontours}). 
721: However, the systematic trend in component acceleration with Doppler velocity 
722: (see Section~\ref{ss:sect4.2}) is not
723: predicted by this model. The trend is likely connected to some aspect of the
724: warped disk geometry that results in a preferred locus in disk radius and azimuth
725: angle within which low-velocity maser action is favored (Figure~\ref{f:threepanel}a) as a result of
726: geometry and orientation to the line of sight, and we
727: investigate this point more in the next paper of this series, in which we 
728: report new modeling of the 3D disk structure and dynamics based on the expanded dataset
729: presented in Paper I and in this paper. In order to 
730: understand other types of ``higher-order'' phenomena that could give rise to the trend
731: however, 
732: we attempted to reproduce the observations by (i) assuming circular orbits in the disk
733: with a spiral arm in the low-velocity maser region (Figure~\ref{f:threepanel}b) and (ii) 
734: allowing for eccentricity in maser orbits (Figure~\ref{f:threepanel}c). 
735: In the preliminary 
736: investigations that follow, we have assumed that the maser disk is flat and is viewed edge-on.
737: 
738: 
739: \subsection{Spiral Structure}
740: 
741: In order to investigate whether spiral structure could give rise to
742: the gradient evident in accelerations of low-velocity components with Doppler velocity, 
743: we performed $N$-body simulations for a maser orbiting a 
744: supermassive black hole \citep[$3.8 \times 10^{7}$ M$_{\odot}$;][]{Herrnstein2005} 
745: in a flat disk, and encountering a spiral arm of various masses (Figure~\ref{f:spiralacc}). 
746: For the simulations, we used a direct $N$-body integration code, NBODY0 \citep{Aarseth1985}, 
747: in which each particle is followed with its own integration step to account 
748: for a range of dynamical rates. See \citet{Aarseth1985} for more details.
749: 
750: We represented a spiral arm at $t= $0 (Figure~\ref{f:spiralacc}a) using 8000 particles 
751: distributed along an arbitrarily-chosen trailing logarithmic spiral arm of 
752: $r=r_0\exp[\,b(\phi-\phi_0)\,]$,  in which $b=$0.42, $\phi_0=$0\degrs 
753: and $r_{0}=$ 1.5~mas or 0.06~pc, with a radial extent of 10$^{-2}$~pc. 
754: The pattern speed of the arm was set to be half that of the Keplerian orbital velocity 
755: (by reducing the force on arm particles by a factor of 4) at any 
756: given radius, and gravitational interaction between arm particles was ignored. We based
757: values for the fractional radial extent and pattern speed of the arm roughly on those of
758: Galactic spiral arms. To avoid potentially infinite acceleration between arm particles
759: and the maser, and to better represent a smooth distribution of mass in the arm, we included a ``softening'' term of $5 \times 10^{-3}$ pc that was added in quadrature
760: to each maser-particle separation.
761: 
762: We varied the mass contained in the arm (bounded by the upper
763: mass limit for the maser disk determined by \citet{Herrnstein2005} of 
764: M$_{disk,upper}=$9$\times$10$^5$ M$_{\odot}$) for different runs of the code. 
765: For any given run, we followed the evolution
766: of a maser cloud through the spiral arm (Figure~\ref{f:spiralacc}a-c) for several hundred years,
767: sampling every 0.55 years.
768: 
769: The computed line-of-sight accelerations for the maser
770: showed significant deviations from that of a mass-less arm, for arm masses greater than a
771: few percent level of M$_{disk,upper}$ (Figure~\ref{f:spiralacc}d). We found the ``amplitude'' 
772: of the acceleration deviation to be directly proportional to arm mass and could 
773: reproduce observed trends in acceleration (see Section~\ref{ss:sect4.2}) for an arm mass of 15\% of the upper limit 
774: maser disk mass. In the simulations, the acceleration trend is reproduced when
775: the maser is ``in'' the spiral arm (between 100 and 200 years in Figure~\ref{f:spiralacc})
776: We note that our calculations include gravitational effects only and
777: do not attempt to incorporate shocks or magneto-hydrodynamic effects of spiral density wave theory.
778: 
779: 
780: 
781: 
782: 
783: \subsection{Eccentric Accretion Flows}
784: 
785: While accretion flows are generally believed to circularize on timescales 
786: much shorter than the viscous timescale, \citet{Statler2001} showed that 
787: elliptical orbits of $e$ $\le$ 0.3 can be long-lived if the disk is thin,   
788: and if the orbits are nested and confocal, with precession 
789: rates that maintain the alignment. 
790: 
791: We describe a system of nested elliptical orbits of identical eccentricity, $e$, of 
792: differing semi-major axes, $a_{semi}$, and with the same periapsis angle, $\omega$, about
793: a supermassive black hole.
794: The periapsis angle is defined here as the angle between the sky plane to the periastron point on each orbit
795: (increasing from the ``red-shifted'' 
796: midline in the sense of disk rotation). In the discussion that follows, it is important 
797: to note that no matter the $e$ and $\omega$ of the system, red-shifted high-velocity 
798: maser emission components must originate from an azimuthal angle of $\phi$ $\approx$ 0\degrns 
799:  (where $\phi-\omega$ is the angle of the component from periastron), and blue-shifted components from 
800: $\phi$ $\approx$ 180\degrns on the basis of velocity coherence and acceleration arguments. 
801: 
802: We first used the symmetry of the position-velocity (P-V) diagram of high-velocity emission to 
803: place limits on eccentricity.  At any given radius in the disk in the plane of the sky (i.e.,
804: where the radial orbital velocity component is perpendicular to the LOS), the Doppler velocity is given by
805: 
806: \begin{equation}
807: v_{los}=\left(\frac{GM}{r}\right)^{1/2}\left[1+e\cos(\phi-\omega)\right]^{1/2}\cos\phi+v_{sys}
808: \end{equation}
809: 
810: \noindent where $r$ is angular radius in the disk and $M=M_{BH}/D$. 
811: We can therefore compute the ratio of $v_{red,\,los}/v_{blue,\,los}$ for 
812: closed orbit ($e$ $<$ 1) values of $e$ and $\omega$ using $\phi$=0\degrs for the red-shifted high-velocity emission 
813: and $\phi$=180\degrs for the blue-shifted high-velocity emission. 
814: Note that this ratio is always unity for circular orbits 
815: ($e=$ 0)  and for orbits of any eccentricity viewed along the semi-major axes i.e., for $\omega$=90\degrs or 270\degrs in our definition.
816: Using values of v$_{sys}$=472$\pm$4 \velunits 
817: \citep{Cecil1992} and a disk dynamical center position of $x_{0}$ = $-$0.1$\pm$0.1 mas 
818: estimated from jet continuum
819: data \citep{Herrnstein1997} within the 1$\sigma$ uncertainties (noting that, by allowing a 1$\sigma$ deviation in
820: each quantity we are actually using a combined uncertainty $> $1$\sigma$), we computed the 
821: {\it observed} ratio of $v_{red,\,los}/v_{blue,\,los}$ from our VLBI data to lie between 0.98 to 1.03. 
822: This constrains maser orbits to have either a low eccentricity of $<$0.05 or to be more highly-eccentric 
823: but viewed (nearly) along  the semi-major axes of the orbits ($\omega$=90\degrs or
824: 270\degrns). The parameter space we {\it eliminate} using the P-V diagram is marked by
825: the white hatching of positive gradient in Figure~\ref{f:stretch}.
826: 
827: To constrain the eccentricity and periapsis angle further, we can also use the low-velocity acceleration 
828: data as a function of Doppler velocity. Instantaneous acceleration is given by
829: $\dot{v}_{los}= - (GM/r^2)\sin\phi$ and $r= a_{semi}(1 - e^2)/(1 + e \cos (\phi - \omega))$ 
830: so we can write
831: 
832: \begin{equation}
833: \frac{d\dot{v}_{los}}{dv_{los}} = \left[\frac{GM}{(1-e^2)^4 a_{semi}^3}\right]^\frac{1}{2}
834: \frac{\gamma^2\cos\phi - 2 \gamma e \sin(\phi-\omega)\sin\gamma}{\sin\phi}
835: \end{equation}
836: 
837: \noindent where $\gamma = 1 + e \cos (\phi - \omega)$ and  where we assume that low-velocity masers occupy the 
838: same orbit such that $a_{semi}$ is
839: a constant, implied by the restriction of low-velocity masers to a narrow portion of the P-V diagram.
840: Using the sign of the gradient, we can further exclude all periapsis angles of  90\degrs $\le$ $\omega$ $\le$ 270\degrs, 
841: marked by the white hatching of negative gradient in Figure~\ref{f:stretch}. Since the $e$ - $\omega$ parameter space
842: remaining after these eliminations is very small (regions with no hatching at all in Figure~\ref{f:stretch}), 
843: it is unlikely that the maser orbits are highly eccentric and observed along a special viewpoint.
844: However, this is a topic we revisit in the next paper of this series, in which we perform 3D accretion disk modeling
845: of the masers including eccentric maser orbits.
846:  
847: 
848: 
849: \section{Conclusions}
850: 
851: In this paper, we measured centripetal accelerations of maser spectral components for data 
852: spanning 1994 to 2004.
853: We found that high-velocity emission accelerations lie in the range %14
854: $-$0.7 to $+$0.7 \accunitns, indicating the emission originates within 13\degrs of the disk midline for
855: material in Keplerian rotation (see Tables 2 and 3). 
856: The good agreement of maser projected vertical positions ($y$-positions) of high-velocity emission derived
857: from the acceleration data, with those of VLBI images, confirms that   
858: masers trace true gas dynamics of the disk.
859: While the high-velocity accelerations do not support the MM98  model of trailing 
860: shocks associated with spiral arms in the disk, we find a spatial periodicity in 
861: high-velocity emission of wavelength 0.75~mas. This supports the model of \citet{Maoz1995} of spiral structure due to 
862: density waves in the disk. 
863: 
864: We measured accelerations of low-velocity emission in the range  %15 
865: 7.7 to 8.9 \accunitns, which is consistent with emission that originates 
866: from a concavity in the front-side of the disk reported by \citet{Herrnstein2005}. 
867: We confirm a systematic trend in accelerations of low-velocity emission as a function of component Doppler velocity, 
868: found by \citet{Haschick1994} and \citet{Greenhill1995b}. 
869: Preliminary investigations into the origin of the trend suggest that eccentricity in maser orbits 
870: is unlikely to be the cause. The trend may be caused either by the effect of a stationary or
871: slowly moving spiral arm, or by a disk feature which causes a stationary pattern of radius versus
872: azimuth angle.
873: 
874: 
875: 
876: We are grateful to Maryam Modjaz and Paul Kondratko for
877: providing their GBT data. We thank Alar Toomre for helpful discussions.
878: 
879: 
880: \bibliographystyle{apj}
881: \bibliography{newrefs2}
882: 
883: 
884: 
885: \clearpage
886: 
887: %% Use the figure environment and \plotone or \plottwo to include
888: %% figures and captions in your electronic submission.
889: 
890: %% To embed the sample graphics in
891: %% the file, uncomment the \plotone, \plottwo, and
892: %% \includegraphics commands
893: %%
894: %% If you need a layout that cannot be achieved with \plotone or
895: %% \plottwo, you can invoke the graphicx package directly with the
896: 
897: %% \includegraphics command or use \plotfiddle. For more information,
898: %% please see the tutorial on "Using Electronic Art with AASTeX" in the
899: %% documentation section at the AASTeX Web site,
900: %% http://www.journals.uchicago.edu/AAS/AASTeX.
901: %%
902: %% The examples below also include sample markup for submission of
903: %% supplemental electronic materials. As always, be sure to check
904: 
905: %% the instructions to authors for the journal you are submitting to
906: %% for specific submissions guidelines as they vary from
907: %% journal to journal.
908: 
909: 
910: %% This example uses \plotone to include an EPS file scaled to
911: %% 80% of its natural size with \epsscale. Its caption
912: %% has been written to indicate that additional figure parts will be
913: %% available in the electronic journal.
914: 
915: \clearpage
916: 
917: \begin{deluxetable}{rlrllccl}
918: \tabletypesize{\scriptsize}
919: %\rotate
920: \tablewidth{0pt}
921: \tablecaption{The Dataset.\label{t:dataset}}
922: \tablehead{
923: \multicolumn{3}{c}{{\bf--------  Epoch --------}} & 
924: \multicolumn{2}{c}{{\bf Observations }}           &
925: \multicolumn{2}{c}{{\bf Observing Details }}      &
926: \multicolumn{1}{l}{{\bf Comments}}               \\
927: %
928: \multicolumn{1}{c}{No.}                           &
929: \multicolumn{1}{c}{Date}                          & 
930: \multicolumn{1}{c}{Day}                           &
931: \multicolumn{1}{c}{Program}                          &
932: \multicolumn{1}{c}{Telescope}                     & 
933: \multicolumn{1}{c}{$\Delta$v\tablenotemark{a}}   &
934: \multicolumn{1}{c}{Sensitivity\tablenotemark{a}}  & 
935: \colhead{}                                       \\
936: %
937: \colhead{}                                        & 
938: \colhead{}                                        &
939: \multicolumn{1}{c}{No.}                           &
940: \multicolumn{1}{c}{Code}                          &
941: \colhead{}                                        & 
942: \multicolumn{1}{c}{(\velunitns)}                    &
943: \multicolumn{1}{c}{(mJy)}                         & 
944: \colhead{}} 
945: \startdata
946:  1  & 1994 Apr 19  &    0 & BM19    & VLBA   & 0.21  & 43    & Systemic, red and  blue-shifted emission \\
947:  2  & 1995 Jan 07  &  263 & AG448 & VLA-CD & 0.33  & 25    & Systemic, red and  blue-shifted emission \\
948:  3  & 1995 Jan 08  &  264 & BM36a   & VLBA   & 0.21  & 80    & Systemic, red and  blue-shifted emission\\
949:  4  & 1995 Feb 23  &  310 & AG448 & VLA-D  & 0.33  & 25    & Systemic, red and  blue-shifted emission \\
950:  5  & 1995 Mar 16  &  331 & \nodata & EFLS   & 0.33  & 85    & Red-Shifted emission only\\
951:  6  & 1995 Mar 24  &  339 & \nodata & EFLS   & 0.33  & 90    & Red-Shifted emission only\\
952:  7  & 1995 Mar 25  &  340 & \nodata & EFLS   & 0.33  & 95    & Red-Shifted emission only\\
953:  8  & 1995 Apr 04  &  350 & \nodata & EFLS   & 0.33  & 85    & Red-Shifted emission only\\
954:  9  & 1995 Apr 20  &  366 & AG448 & VLA-D  & 0.33  & 20    & Systemic, red and  blue-shifted emission \\
955:  10 & 1995 May 29  &  405 & BM36b   & VLBA   & 0.21  & 30    & Systemic, red and  blue-shifted emission \\
956:  11 & 1995 Jun 08  &  415 & AG448 & VLA-AD & 0.33  & 20 & Systemic, red and  blue-shifted emission    \\
957:  12 & 1995 Jun 25  &  432 & \nodata & EFLS   & 0.33  & 110   & Red-Shifted emission only  \\
958:  13 & 1995 Jul 29  &  466 & AG448 & VLA-A  & 0.33  &\nodata& No data\\
959:  14 & 1995 Sep 09  &  508 & AG448 & VLA-AB & 0.33  &\nodata& No data \\
960:  15 & 1995 Nov 09  &  569 & AG448 & VLA-B  & 0.33  & 37-60 & Systemic, red and  blue-shifted emission   \\
961:  16 & 1996 Jan 11  &  632 & AG448 & VLA-BC & 0.33  & 20    & Systemic, red and  blue-shifted emission \\
962:  17 & 1996 Feb 22  &  674 & BM56a   & VLBA   & 0.21  & 25    & Systemic, red and  blue-shifted emission\\    	
963:  18 & 1996 Feb 26  &  678 & AG448 & VLA-C  & 0.33  & 23-30 & Systemic, red and  blue-shifted emission  \\
964:  19 & 1996 Mar 29  &  710 & AG448 & VLA-C  & 0.33  & 20    & Systemic, red and  blue-shifted emission \\           
965:  20 & 1996 May 10  &  752 & AG448 & VLA-CD & 0.33  & 15-24 & Systemic, red and  blue-shifted emission\\
966:  21 & 1996 Jun 27  &  799 & AG448 & VLA-D  & 0.33  &\nodata& No data\\
967:  22 & 1996 Aug 12  &  846 & AG448 & VLA-D  & 0.33  & 23-36 & Systemic, red and  blue-shifted emission\\
968:  23 & 1996 Sep 21  &  886 & BM56b   & VLBA   & 0.21  & 20    & High-velocity emission only\\
969:  24 & 1996 Oct 03  &  898 & AG448 & VLA-AD & 0.33  & 50    & Systemic, red and  blue-shifted emission   \\
970:  25 & 1996 Nov 21  &  947 & AG448 & VLA-A  & 0.33  & 21-26 & Systemic and red-shifted emission \\
971:  26 & 1997 Jan 20  & 1007 & AG448 & VLA-AB & 0.33  & 27-37 & Systemic and red-shifted emission  \\
972:  27 & 1997 Feb 10  & 1028 & AG448 & VLA-AB & 0.33  & 17-23 & Systemic and red-shifted emission\\  
973:  28 & 1997 Mar 06  & 1052 & BM56c   & VLBA   & 0.21  & 4.7   & Systemic, red and  blue-shifted emission  \\
974:  29 & 1997 Oct 01  & 1261 & BM81a   & VLBA   & 0.21  & 4.1   & Systemic, red and  blue-shifted emission  \\
975:  30 & 1998 Jan 27  & 1379 & BM81b   & VLBA   & 0.21  & 5.0   & Systemic, red and  blue-shifted emission   \\
976:  31 & 1998 Sep 05  & 1600 & BM112a  & VLBA   & 0.21  & 4.6   & Systemic, red and  blue-shifted emission\\
977:  32 & 1998 Oct 18  & 1643 & BM112b  & VLBA   & 0.42  & 4.6   & Systemic and red-shifted emission         \\    
978:  33 & 1998 Nov 16  & 1672 & BM112c  & VLBA   & 0.42  & 4.4   & Systemic and blue-shifted emission\\ 
979:  34 & 1998 Dec 24  & 1710     & BM112d  & VLBA   & 0.42  &\nodata& No data \\
980:  35 & 1999 Jan 28  & 1745 & BM112e  & VLBA   & 0.42  & 4.8   & Systemic and blue-shifted emission  \\     
981:  36 & 1999 Mar 19  & 1795 & BM112f  & VLBA   & 0.42  & 4.5   & Systemic and red-shifted emission    \\      
982:  37 & 1999 May 18  & 1855 & BM112g  & VLBA   & 0.42  & 5.2   & Systemic and blue-shifted emission   \\
983:  38 & 1999 May 26  & 1863 & BM112h  & VLBA   & 0.21  & 3.5   & Systemic, red and  blue-shifted emission    \\
984:  39 & 1999 Jul 15  & 1913     & BM112i  & VLBA   & 0.42  &\nodata& No data \\
985:  40 & 1999 Sep 15  & 1975 & BM112j  & VLBA   & 0.42  & 6.3   & Systemic and blue-shifted emission       \\
986:  41 & 1999 Oct 29  & 2019 & BM112k  & VLBA   & 0.42  & 4.7   & Systemic and blue-shifted emission          \\
987:  42 & 2000 Jan 07  & 2089 & BM112l  & VLBA   & 0.42  & 3.9   & Systemic and blue-shifted emission         \\
988:  43 & 2000 Jan 30  & 2112 & BM112m  & VLBA   & 0.42  & 4.7   & Systemic and red-shifted emission    \\ 
989:  44 & 2000 Mar 04  & 2146 & BM112n  & VLBA   & 0.42  & 4.3   & Systemic and blue-shifted emission          \\
990:  45 & 2000 Apr 12  & 2185 & BM112o  & VLBA   & 0.42  & 5.5   & Systemic and red-shifted emission  \\
991:  46 & 2000 May 04  & 2207 & BM112p  & VLBA   & 0.42  & 5.0   & Systemic and blue-shifted emission   \\
992:  47 & 2000 Aug 12  & 2307 & BG107   & VLBA   & 0.42  & 7.2   & Systemic, red and  blue-shifted emission \\
993:  48 & 2003 Apr 10  & 3298 & \nodata & GBT    & 0.21     & 2.9      & Systemic, red and  blue-shifted emission \\ 
994:  49 & 2003 Oct 23  & 3474 & \nodata & GBT    & 0.21  & 2.9   & Systemic, red and  blue-shifted emission \\ 
995:  50 & 2003 Dec 08  & 3520 & \nodata & GBT    & 0.21  & 2.9   & Systemic, red and  blue-shifted emission \\ 
996:  51 & 2004 May 21  & 3685 & AH847 & VLA    & 0.21    &  20     &   Portions of red-shifted spectrum                                      \\ 
997: \enddata
998: \tablenotetext{a}{Channel spacing  and sensitivities (r.m.s)
999: for epochs 1 -- 27 taken from \citet{Bragg2000}; 48 -- 47 from Paper I and
1000: for GBT data (M. Modjaz, P. Kondratko, private communications)}
1001: %% You can append references to a table using the \tablerefs command.
1002: \end{deluxetable}
1003: 
1004: \clearpage
1005: 
1006: 
1007: 
1008: 
1009: 
1010: \begin{deluxetable}{rrrcccc} 
1011: \tabletypesize{\scriptsize}
1012: %\rotate
1013: \tablewidth{0pt}
1014: \tablecaption{Fitted Accelerations For Red-Shifted High-Velocity Doppler Components.
1015: \label{t:redtable}}
1016: \tablehead{
1017: \multicolumn{3}{c}{{\bf -------------- Component --------------}}&
1018: \multicolumn{2}{c}{{\bf --------- Epochs ---------} }&
1019: \multicolumn{1}{c}{{\bf Time-Averaged}}&
1020: \multicolumn{1}{c}{{\bf Component}}\\
1021: %\multicolumn{2}{c}{}\\
1022: \multicolumn{1}{c}{No.} &
1023: \multicolumn{1}{c}{{\bf Velocity\tablenotemark{a}}}&
1024: \multicolumn{1}{c}{{\bf Acceleration\tablenotemark{b}}}& 
1025: \multicolumn{1}{c}{No. in Fit} &
1026: \multicolumn{1}{c}{Time Baseline}&
1027: \multicolumn{1}{c}{{\bf Azimuth Angle}}&
1028: \multicolumn{1}{c}{{\bf Radius}\tablenotemark{c}}\\
1029: %\colhead{} &
1030: \colhead{} &
1031: %\colhead{}                    &
1032: \multicolumn{1}{c}{{\bf (\velunitns)}}&   
1033: \multicolumn{1}{c}{{\bf (\accunitns)}}& 
1034: \colhead{} &
1035: \multicolumn{1}{c}{(yr)}&
1036: \multicolumn{1}{c}{(deg)}&    
1037: \multicolumn{1}{c}{(pc)}} 
1038: \startdata
1039: 1&1248.14$\pm$0.02  &  -0.06 $\pm$0.02 & 24  &   6.0                    & -1.6$\pm$0.6 &  0.28\\
1040: 2&1250.80$\pm$0.03  &  -0.35 $\pm$0.03 & 24  &   6.0                    & -8.9$\pm$0.8 &  0.28\\
1041: 3&1252.24$\pm$0.01  &   0.17 $\pm$0.01 & 24  &   6.0                    &  4.3$\pm$0.3 &  0.28\\
1042: 4&1254.34$\pm$0.02  &  -0.01 $\pm$0.03 & 24  &   6.0                    & -0.3$\pm$0.7 &  0.28\\
1043: 5&1270.89$\pm$0.05  &   0.11 $\pm$0.04 & 24  &   6.0                    &  2.5$\pm$1.0 &  0.27\\
1044: 6&1282.21$\pm$0.06  &  -0.15 $\pm$0.08 &  4  &   0.7                    & -3.1$\pm$1.8 &  0.26\\
1045: 7&1283.87$\pm$0.08  &  -0.05 $\pm$0.14 &  4  &   0.7                    & -1.1$\pm$3.0 &  0.26\\
1046: 8&1309.08$\pm$0.01  &  -0.09 $\pm$0.01 & 30  &   9.0                    & -1.7$\pm$0.1 &  0.24\\
1047: 9&1328.64$\pm$0.04  &   0.73 $\pm$0.02 & 29  &   5.3                    & 12.6$\pm$0.6 &  0.23\\
1048: 10&1330.73$\pm$0.06  &   0.73 $\pm$0.03 & 29  &   5.3                    & 12.4$\pm$0.7 &  0.23\\
1049: 11&1337.36$\pm$0.04  &  -0.16 $\pm$0.03 & 35  &   6.3                    & -2.7$\pm$0.5 &  0.23\\
1050: 12&1339.58$\pm$0.01  &  -0.28 $\pm$0.01 & 35  &   6.3                    & -4.6$\pm$0.3 &  0.22\\
1051: 13&1351.00$\pm$0.03  &   0.40 $\pm$0.03 & 34  &   6.0                    &  6.1$\pm$0.5 &  0.22\\
1052: 14&1353.85$\pm$0.03  &   0.11 $\pm$0.03 & 34  &   6.0                    &  1.8$\pm$0.4 &  0.22\\
1053: 15&1355.56$\pm$0.02  &  -0.10 $\pm$0.02 & 34  &   6.0                    & -1.5$\pm$0.3 &  0.22\\
1054: 16&1395.59$\pm$0.09  &   0.24 $\pm$0.05 & 21  &   4.3                    &  3.1$\pm$0.7 &  0.20\\
1055: 17&1398.09$\pm$0.05  &   0.58 $\pm$0.03 & 21  &   4.3                    &  7.3$\pm$0.4 &  0.20\\
1056: 18&1403.60$\pm$0.01  &  -0.05 $\pm$0.02 & 34  &   6.0                    & -0.6$\pm$0.2 &  0.19\\
1057: 19&1406.74$\pm$0.05  &  -0.06 $\pm$0.04 & 34  &   6.0                    & -0.8$\pm$0.5 &  0.19\\
1058: 20&1449.43$\pm$0.04  &   0.30 $\pm$0.04 & 25  &   2.9                    &  3.0$\pm$0.5 &  0.18\\
1059: 21&1452.35$\pm$0.01  &  -0.40 $\pm$0.02 & 25  &   2.9                    & -4.0$\pm$0.2 &  0.18\\
1060: 22&1452.80$\pm$0.14  &   0.04 $\pm$0.06 & 25  &   2.9                    &  0.4$\pm$0.6 &  0.18\\
1061: 23&1455.26$\pm$0.19  &  -0.33 $\pm$0.09 & 25  &   2.9                    & -3.2$\pm$0.9 &  0.17\\
1062: 24&1468.08$\pm$0.07  &   0.34 $\pm$0.08 &  4  &   2.6                    &  3.2$\pm$0.8 &  0.17\\
1063: \hline
1064: \multicolumn{2}{r}{{{\bf Mean Acceleration:}}\tablenotemark{d}}&\multicolumn{1}{r}{0.02 $\pm$0.06}&
1065: \multicolumn{2}{c}{{{\bf Mean Azimuth Angle:}}\tablenotemark{d}}&\multicolumn{1}{c}{0.2$\pm$3.6}\\
1066: \enddata
1067: \tablenotetext{a}{Component velocities (relativistic definition) for fits that met requirements
1068: of Section~X are quoted for 
1069: 1999 October 10, day 2000 of our monitoring campaign. Uncertainties are the 1-$\sigma$ errors scaled 
1070: by $\chi^2$ per degree of freedom.} 
1071: \tablenotetext{b}{Uncertainties are the 1-$\sigma$ errors scaled 
1072: by $\chi^2$ per degree of freedom.}
1073: \tablenotetext{c}{Calculated using a black hole mass of
1074: 3.8 $\times$ 10$^{7}$ M$_{\odot}$ \citep{Herrnstein2005} for a flat disk.}
1075: \tablenotetext{d}{Quantities are the weighted mean and the weighted deviation 
1076: from the mean.}
1077: \label{t:table2}
1078: \end{deluxetable}
1079: 
1080: %liznote: 1999 October 10 is JD 2451462.5 (day 2000 of monitoring)
1081: % day 0 of monitoring is 2449462.5
1082: % bm81b is on 1998 Jan 27 or JD 2450841.5
1083: \clearpage
1084: 
1085: \begin{deluxetable}{rrrcccc} 
1086: %\label{t:table3}
1087: \tabletypesize{\scriptsize}
1088: %\rotate
1089: \tablewidth{0pt}
1090: \tablecaption{Fitted Accelerations For Blue-Shifted High-Velocity Doppler Components.
1091: \label{t:bluetable}}
1092: \tablehead{
1093: \multicolumn{3}{c}{{\bf -------------- Component --------------}}&
1094: \multicolumn{2}{c}{{\bf --------- Epochs ---------} }&
1095: \multicolumn{1}{c}{{\bf Time-Averaged}}&
1096: \multicolumn{1}{c}{{\bf Component}}\\
1097: %\multicolumn{2}{c}{}\\
1098: \multicolumn{1}{c}{No.} &
1099: \multicolumn{1}{c}{{\bf Velocity\tablenotemark{a,b}}}&
1100: \multicolumn{1}{c}{{\bf Acceleration\tablenotemark{b}}}& 
1101: \multicolumn{1}{c}{No. in Fit} &
1102: \multicolumn{1}{c}{Time Baseline}&
1103: \multicolumn{1}{c}{{\bf Azimuth Angle}}&
1104: \multicolumn{1}{c}{{\bf Radius}\tablenotemark{c}}\\
1105: %\colhead{} &
1106: \colhead{} &
1107: %\colhead{}                    &
1108: \multicolumn{1}{c}{{\bf (\velunitns)}}&   
1109: \multicolumn{1}{c}{{\bf (\accunitns)}}& 
1110: \colhead{} &
1111: \multicolumn{1}{c}{(yr)}&
1112: \multicolumn{1}{c}{(deg)}&    
1113: \multicolumn{1}{c}{(pc)}} 
1114: \startdata
1115: 1& -282.21$\pm$0.07& -0.47 $\pm$0.10 & 6 & 1.5 & -13.2$\pm$2.9 & 0.29 \\
1116: 2& -284.00$\pm$0.06& -0.12 $\pm$0.08 & 6 & 1.5 & -3.3$\pm$2.3&  0.29 \\
1117: 3& -286.05$\pm$0.13& -0.38 $\pm$0.13 & 6 & 1.5 & -10.3$\pm$3.5&  0.29\\
1118: 4& -374.20$\pm$0.05& -0.32 $\pm$0.26 & 5 & 1.1 & -5.6$\pm$4.7& 0.23\\
1119: 5& -375.78$\pm$0.22& -0.72 $\pm$0.49 & 5 & 1.1 & -12.6$\pm$ 8.6& 0.23\\
1120: 6& -435.00$\pm$0.03& -0.29 $\pm$0.05 & 16& 3.5 & -3.9$\pm$ 0.7&  0.20\\
1121: 7& -439.97$\pm$0.02& +0.04 $\pm$0.04 & 15& 2.9 &0.5$\pm$ 0.5&  0.20\\
1122: 8& -514.48$\pm$0.04& -0.24 $\pm$0.16 & 4 & 0.5 &-2.3$\pm$ 1.5&  0.17\\
1123: \hline
1124: \multicolumn{2}{r}{{{\bf Mean Acceleration:}}\tablenotemark{d}}&\multicolumn{1}{r}{-0.21 $\pm$0.08}&
1125: \multicolumn{2}{c}{{{\bf Mean Azimuth Angle:}}\tablenotemark{d}}&\multicolumn{1}{c}{-3.3$\pm$6.9}\\
1126: \enddata
1127: \tablenotetext{a}{Component velocities (relativistic definition) are quoted for 
1128: 1999 October 10, day 2,000 of our monitoring campaign. Uncertainties are the 1-$\sigma$ errors scaled 
1129: by $\chi^2$ per degree of freedom.} 
1130: \tablenotetext{b}{Uncertainties are the 1-$\sigma$ errors scaled 
1131: by $\chi^2$ per degree of freedom.}
1132: \tablenotetext{c}{Calculated using a black hole mass of
1133: 3.8 $\times$ 10$^{7}$ M$_{\odot}$ \citep{Herrnstein2005} for a flat disk.}
1134: \tablenotetext{d}{Quantities are the weighted mean and the weighted deviation 
1135: from the mean.}
1136: \end{deluxetable}
1137: 
1138: 
1139: \clearpage
1140: 
1141: \begin{deluxetable}{ccc} 
1142: %\label{t:table4}
1143: \tabletypesize{\scriptsize}
1144: %\rotate
1145: \tablewidth{0pt}
1146: \tablecaption{Accelerations for Low-Velocity Maser Doppler Components.
1147: \label{t:systable}}
1148: \tablehead{
1149: \multicolumn{1}{c}{{\bf Velocity\tablenotemark{a}}}&
1150: \multicolumn{1}{c}{{\bf Acceleration\tablenotemark{b}}}&
1151: \multicolumn{1}{c}{{\bf Radius in Geometric Model\tablenotemark{c}}}\\
1152: \multicolumn{1}{c}{{\bf (\velunitns)}}&   
1153: \multicolumn{1}{c}{{\bf (\accunitns)}}&
1154: \multicolumn{1}{c}{{\bf (mas)}}}
1155: %colhead{}
1156: %multicolumn{1}{c}{(Yrs)}                  }
1157: \startdata
1158:   438&  7.65$\pm$0.15  &4.20$\pm$0.04\\
1159:   448&  7.81$\pm$0.13  &4.16$\pm$0.03\\
1160:   458&  7.85$\pm$0.13  &4.15$\pm$0.03\\
1161:   468&  7.77$\pm$0.08 & 4.17$\pm$0.02\\
1162:   478&  7.84$\pm$0.17  &4.15$\pm$0.04\\
1163:   488&  8.29$\pm$0.12  &4.04$\pm$0.03\\
1164:   498&  8.16$\pm$0.13  &4.07$\pm$0.03\\
1165:   508&  8.09$\pm$0.09 & 4.09$\pm$0.02\\
1166:   518&  8.75$\pm$0.09 & 3.93$\pm$0.02\\
1167:   528&  8.55$\pm$0.06 & 3.97$\pm$0.02\\
1168:   538&  8.36$\pm$0.10 & 4.02$\pm$0.02\\
1169:   548&  8.87$\pm$0.27 & 3.89$\pm$0.05\\
1170: \enddata
1171: %% Any table notes must follow the \end{tabular} command.
1172: \tablenotetext{a}{Data have been binned into 10 \velunits intervals.}
1173: \tablenotetext{b}{Uncertainties are the standard deviation of the mean for each bin.}
1174: \tablenotetext{c}{For the best-fitting disk geometry of \citet{Herrnstein2005}. See also
1175: Figure~\ref{f:velcontours}.}
1176: \end{deluxetable}
1177: 
1178: 
1179: \clearpage
1180: 
1181: 
1182: 
1183: \begin{deluxetable}{ccccccc}
1184: \label{t:acceleration}
1185: \tabletypesize{\scriptsize}
1186: %\tabletypesize{\small}
1187: %\tabletypesize{\footnotesize}
1188: \rotate
1189: \tablewidth{0pt}
1190: \tablecaption{Studies of Acceleration Among NGC\,4258 Maser Components}
1191: \tablehead{
1192: \colhead{Study\tablenotemark{a}}         &
1193: \colhead{Antennas\tablenotemark{b}}      & 
1194: \colhead{Sensitivity\tablenotemark{c}}   &
1195: \colhead{Dates}                          &
1196: %\colhead{Frequency}                          &
1197: \colhead{Velocity}& 
1198: \colhead{Components}               &
1199: \colhead{Acceleration}         \\
1200: &
1201: &
1202: %&
1203: \colhead{(mJy)}&
1204: &
1205: \colhead{Range\tablenotemark{d}}&
1206: \colhead{Tracked}&
1207: \colhead{Range}\\
1208: &
1209: &
1210: &
1211: &
1212: \colhead{(km s$^{-1}$)}&
1213: &
1214: \colhead{(km s$^{-1}$ yr$^{-1}$)}\\
1215: }
1216: 
1217: \startdata
1218: \citet{Haschick1990} & HAYS & 55$-$70         & 7/1986$-$11/1989 & \ 340 to 590 & 1 (systemic)& 10  \\
1219:                      &      &                 &$\approx$ monthly &              &   &    \\
1220:                      &      &                 &                  &              &   &     \\          
1221: %
1222: \citet{Haschick1994} & HAYS &     30$-$55     & 7/1986$-$11/1993 & \ 340 to 590 & 4 (systemic) & 6.2$-$10.4   \\
1223:                      &      &                 & $\approx$ monthly&              &   &    \\
1224:                      &      &                 &                  &              &   &     \\
1225: %
1226: \citet{Greenhill1995b}&EFLS &  30             &10/1984$-$12/1986 & 335 to 665   & 12& 8.1$-$10.9 \\
1227:                       &     &                 & every few days $-$ months & 1185 to 1515 & 20  &$<$$\pm$1\\
1228:                       &     &                 &02/1993$-$03/1994& $-$570 to $-$240  & 3 &  $<$$\pm$1  \\
1229:                       &     &                 & more frequently   &              &   &     \\
1230:                       &     &                 &                   &              &   &     \\
1231: %
1232: \citet{Nakai1995}     & NRO &  13$-$20        &    1992           & $-$1197 to 2664 & 13 (systemic) & 7.2$-$11.1\\
1233:                       &     &                 &  $\approx$ weekly &   \nodata              & 8 (red) & $<$$\pm$0.7\\
1234:                       &     &                 &                   &   \nodata              & 1 (blue) &\\
1235:                       &     &                 &                   &                 &       &\\
1236: %
1237: \citet{Herrnstein1999}&VLBA+VLA (3 epochs)&  6$-$12 & 04/1994$-$02/1996&430 to 590        & 30 (systemic)  & 6.8$-$11.6\\
1238:                    & VLBA+VLA+GB (1 epoch)&         &                  &1225 to 1490      & 1 (red)   &   0.0  \\
1239:                    &                      &         &                  &$-$450 to $-$350  & 0 (blue   & \\
1240:                    &                      &         &                  &                  &      & \\
1241: %
1242: \citet{Bragg2000}& VLBA (5 epochs)        &  40     &04/1994$-$02/1997 &390 to 660        & 12 (systemic)  & 7.5$-$10.4   \\
1243:                  & EFLS (5 epochs)        &  95     &                  &1235 to 1460      & 17  (red) & -0.77$-$0.38\\
1244:                  & VLA (17 epochs)        &  20     &                  &$-$460 to $-$420,$-$390 to $-$350  & 2  (blue)  & -0.41,$+$0.04\\
1245:                  &                        &         &                  &                                   &      & \\
1246: %
1247: \citet{Yamauchi2005}& NRO                 & 8$-$140 &01/1992$-$04/2005                 & $-$725 to 1720  & 14 (red)  &-0.52$-$0.41    \\
1248:                     &                     &         & 30 irregularly spaced spectra    &    \nodata   & 1 (blue)  & +0.2 \\
1249:                     &                     &         &                                  &    \nodata   &         &   \\
1250: 
1251: This Work           &   Bragg Dataset(27 epochs)   & 20$-$95    & 04/1994$-$05/2004          & -850 to 1850 & 25 (systemic)& 6.1$-$10.8 \\
1252:                     &   VLBA+VLA+EFLS (18 epochs)  & 3.2$-$7.8  & on average every 2.5 months &\nodata & 24 (red) & -0.4$-$0.7 \\              
1253:                     &   GBT(3 epochs)              & a few mJy  &                            &  \nodata& 8 (blue) & -0.7$-$0.04\\
1254:                     &   VLA(1 epoch)               &     5      &                            &              &     & \\
1255: % 
1256: %
1257: \enddata
1258: 
1259: %% Any table notes must follow the \end{tabular} command.
1260: 
1261: \tablenotetext{a}{\citet{Haschick1990} report primarily on a variability study, but 
1262: note that the 465 km s$^{-1}$ component is accelerating.  \citet{Haschick1994}, 
1263: \citet{Greenhill1995b}, \citet{Nakai1995}, \citet{Bragg2000}, and \citet{Yamauchi2005} determine accelerations by 
1264: tracking local spectral maxima and fitting velocity drifts as a linear 
1265: function of time.  
1266: \citet{Bragg2000} performs multiple-Gaussian component decomposition to high-velocity spectra, but
1267: not to systemic spectra.
1268: \citet{Herrnstein1999} determines accelerations by applying a 
1269: Bayesian statistical analysis technique to all possible pairings of components 
1270: among epochs.  Two of the three high velocity studies, i.e., \citet{Greenhill1995b} and 
1271: \citet{Nakai1995}, obtain upper limits only. In this work, we perform multiple-Gaussian
1272: component decomposition to high-velocity {\it and} systemic spectra in multi-epoch
1273: simultaneous fits. 
1274: }
1275: 
1276: \tablenotetext{b}{HAYS: Haystack 37-m antenna; EFLS: Effelsberg 100-m antenna; 
1277: NRO: Nobeyama Radio Observatory 45-m; VLBA: Very Long Baseline Array; VLA: 
1278: 27$\times$25-m Very Large Array; GB: NRAO 140-ft.
1279: }
1280: 
1281: \tablenotetext{c}{Noise level (1$\sigma$) per channel in emission-free portions of the 
1282: spectra (single dish or VLA) or synthesis images (VLBA). See relevant publications
1283: for further details.
1284: }
1285: 
1286: \tablenotetext{d}{Ranges for \citet{Greenhill1995b} and \citet{Herrnstein1999} are 
1287: approximate.  Ranges for \citet{Bragg2000} are for all epochs except the 
1288: first.  The velocity ranges for the first epoch were all shifted by 20 km 
1289: s$^{-1}$ towards higher velocities.  
1290: }
1291: \end{deluxetable}
1292: 
1293: \clearpage
1294: 
1295: 
1296: 
1297: 
1298: 
1299: \begin{figure}
1300: \epsscale{0.5}
1301: %\plotone{fig1_jun19.vps}
1302: \plotone{f1.ps}
1303: \caption{\footnotesize {\it (a)}
1304: Sky positions for 22~GHz maser emission in NGC~4258 obtained using VLBI on 
1305: 18 epochs from 1997 March 06 to 2000 August 12 described in Paper I. Red-shifted 
1306: high-velocity emission occurs from 3 to 8 mas in the disk; low-velocity emission 
1307: occurs in the range -1 to 1 mas and blue-shifted high-velocity emission in the range 
1308: 5 to 9 mas. Note that $(0,0)$ in the plot does not represent the disk dynamical center. 
1309: {\it (b)} GBT spectrum of red-shifted high-velocity emission from \citet{Modjaz2005}, 
1310: obtained on 2003 October 23. The 1$\sigma$ noise in the spectrum is 1.6 mJy. The arrow marks 
1311: weak, newly-detected emission at 1652 \velunits (Paper I).  
1312: {\it (c)} GBT spectrum of low-velocity 
1313: maser emission. The arrows mark emission extrema. The dotted line denotes the galactic 
1314: systemic velocity of 472 \velunits \citep{Cecil1992}. 
1315: {\it (d)} GBT spectrum of blue-shifted maser emission.
1316: \label{f:figure1}}   
1317: \end{figure}
1318: 
1319: \clearpage
1320: 
1321: \begin{figure}
1322: \epsscale{0.7}
1323: %\plotone{systemicsandreds.ps}
1324: \plotone{f2.ps}
1325: \hspace{-1cm}
1326: \caption{\footnotesize {\it (a)} 
1327: Velocity vs. time plot for systemic components using 
1328: data from 1998 January 27 to 2000 August 12 during which sampling 
1329: was every \about3 months. Drifts in Doppler velocities of strong components 
1330: are clearly evident and consistent with centripetal accelerations. The dashed 
1331: line marks the galactic systemic velocity of 472 \velunitns.
1332: {\it (b)} Corresponding plot for red-shifted components. Small magnitudes of 
1333: drifts in component Doppler velocities are consistent with near-zero observed 
1334: accelerations for components near the disk midline. 
1335: \label{f:figure2}}
1336: \end{figure}
1337: 
1338: 
1339: \clearpage
1340: 
1341: \begin{figure}
1342: \epsscale{1.0}
1343: %\includegraphics[angle=-90,scale=0.8]{feature-1450_paper.ps}
1344: \includegraphics[angle=-90,scale=0.8]{f3.ps}
1345: \caption{Example of a simultaneous 4-Gaussian function fit to a blend
1346: of red-shifted components
1347: over 25 epochs (2.9 years). Filled circles mark the data points. Residuals are
1348: plotted with 1$\sigma$ errorbars.
1349: Model Gaussian functions are shown in solid black lines. The epoch day 
1350: number is marked in the upper right of each panel. Table~\ref{t:dataset}
1351:  lists the corresponding dates.
1352: \label{f:redfit}}
1353: \end{figure}
1354: 
1355: \clearpage
1356: 
1357: \begin{figure}
1358: \epsscale{1.0}
1359: %\plotone{feature-280_bw2.vps}
1360: \plotone{f4.ps}
1361: \caption{As for Figure~\ref{f:redfit}, but for a blue-shifted high-velocity emission
1362: blend. 
1363: \label{f:bluefit}}
1364: \end{figure}
1365: 
1366: \clearpage
1367: 
1368: \begin{figure}
1369: %\includegraphics[angle=-90,scale=0.7]{showcrossing4paper.ps}
1370: %\includegraphics[scale=0.8]{showcrossing4paper.ps}
1371: \includegraphics[scale=0.8]{f5.ps}
1372: \caption{Example of Gaussian decomposition for low-velocity emission. 
1373: A velocity-portion of a fit over 9 epochs is
1374: shown, in which dots are data and black symbols are
1375: the residuals including 1$\sigma$ error bars. Individual model Gaussians are coded using 
1376: the same color at each epoch, such that Doppler velocity drifts are clearly evident.
1377: Numbers in the upper right corner of each panel are the number of days since 1994 April 19. 
1378: \label{f:sysfit}}
1379: \end{figure}
1380: 
1381: \clearpage
1382: 
1383: \begin{figure}
1384: %\includegraphics[angle=-90,scale=0.7]{systemic_trend.ps}
1385: \includegraphics[angle=-90,scale=0.7]{f6.ps}
1386: %\includegraphics[scale=0.8]{showcrossing4paper.ps}
1387: %\plotone{systemic_trend.ps}
1388: \caption{Results of Gaussian decomposition for low-velocity emission.
1389: {\it (a)} Measured accelerations from each of the four time-consecutive 9-epoch fits.
1390: Each fit required $\sim$55 Gaussian components. The line-of-best-fit to all the data is shown. 
1391: {\it (b)} Data in {\it (a)} binned into 10 \velunits intervals. The errorbars are the
1392: standard deviation of the mean in each bin.
1393: \label{f:systrend}} 
1394: \end{figure}
1395: 
1396: \clearpage
1397: 
1398: \begin{figure}
1399: %\includegraphics[angle=-90,scale=0.7]{systemic_trend.ps}
1400: \includegraphics[angle=-90,scale=0.7]{f7.ps}
1401: %\includegraphics[scale=0.8]{showcrossing4paper.ps}
1402: %\plotone{systemic_trend.ps}
1403: \caption{Predicted position-velocity diagram for a
1404: linear fit to the low-velocity acceleration data (see Figure~6). 
1405: Symbols mark observed
1406: data from the 18 VLBI epochs of Paper I, binned into 0.05 mas intervals.
1407: The curve shows the diagram predicted from the line-of-best-fit to the
1408: acceleration data. Curve and data are in close correspondence except
1409: at three most positive values of impact parameter.
1410: \label{f:systrendpv}} 
1411: \end{figure}
1412: 
1413: \clearpage
1414: 
1415: \begin{figure}
1416: %\includegraphics[angle=-90,scale=0.7]{vlos_vs_accplus2.cps}
1417: %\includegraphics[angle=-90,scale=0.7]{vlos_vs_accplus2.cps}
1418: \includegraphics[angle=-90,scale=0.7]{f8.ps}
1419: \caption{Comparison of acceleration measurements for red-shifted maser spectral 
1420: components: the current work (black circles); \citet{Bragg2000} (red triangles); 
1421: \citet{Yamauchi2005} 1997 to 2005 (green squares); \citet{Yamauchi2005} 1992 to 2005 
1422: (blue stars). \citet{Yamauchi2005} derive accelerations from their dataset over
1423: two time ranges, one of which is a subset of the other.
1424: All velocities have been converted to a relativistic definition, are referenced to the LSR, and 
1425: have been adjusted to a common reference date (our monitoring day 2000). Measurements from the different 
1426: studies are in broad agreement, however this work tends to identify more components in any given blend 
1427: compared with earlier work.
1428: \label{f:acccomp}}
1429: \end{figure}
1430: 
1431: \clearpage
1432: 
1433: \begin{figure}
1434: %\includegraphics[angle=-90,scale=0.7]{vlos_vs_accplus2.cps}
1435: %\includegraphics[angle=-90,scale=0.7]{vlos_vs_accplus2.cps}
1436: \includegraphics[angle=0,scale=0.7]{f9.ps}
1437: \caption{Comparison of acceleration measurements for low-velocity maser 
1438: spectral components. All studies are plotted
1439: at the mid-date of their monitoring period, in order of increasing
1440: mid-date. Only components for which sufficient velocity drift data
1441: were obtained have been plotted here. The
1442: majority of studies show a trend of increasing acceleration
1443: as a function of Doppler velocity.
1444: We note that the present work is represented by binned data here, whereas
1445: other studies display measurements for individual components. The uncertainties
1446: plotted for the binned values are the standard deviation of the mean of each
1447: bin, as for Figure~\ref{f:systrend}b. For comparison, in the present work we
1448: measure accelerations for $\sim$55 components in the low-velocity spectrum.
1449: \label{f:lowvelacccomp}}
1450: \end{figure}
1451: 
1452: \clearpage
1453: 
1454: \begin{figure}
1455: %\includegraphics[angle=-90,scale=0.2]{newherrndisk.eps}
1456: %\plotone{newherrndisk.eps}
1457: \plotone{f10.ps}
1458: \caption{Best-fitting maser disk from \citet{Herrnstein2005} viewed from [-40,40,-50]. 
1459: The line-of-sight is shown as a line extending beyond the outer edge of the disk in black along the $z$-direction.  
1460: Solid color contours show disk elevation, where red is the maxmium and
1461: dark blue is the minimum. The disk midline for $\phi$=0\degrs and  180\degrs
1462: is shown for red-shifted emission (red line) and blue-shifted emission
1463: (blue line) respectively. Low-velocity (systemic) masers lie in the concavity on the front
1464: side of the disk. 
1465: \label{f:herrndisk}}
1466: \end{figure}
1467: 
1468: 
1469: \clearpage
1470: 
1471: \begin{figure}
1472: \vspace{-4cm}
1473: \epsscale{0.7}
1474: %\plotone{jrh_and_our_masers.ps}
1475: \plotone{f11.ps}
1476: \caption{Location of the low-velocity emission in the concavity \citep{Herrnstein2005}.
1477: Isovelocity contours are marked in color and are in units of \velunits relative
1478: to v$_{sys}$ \velunits \citep{Herrnstein2005}. The LOS is along the $z$-direction, and
1479: therefore a vertical contour would mark a zero velocity gradient in the line-of-sight. The black
1480: line marks the bottom of the concavity. The black symbols mark the position of the low
1481: velocity emission, derived from the acceleration data. Observer views along the $z$-axis.
1482: \label{f:velcontours}}
1483: \end{figure}
1484: 
1485: \clearpage
1486: 
1487: 
1488: 
1489: 
1490: 
1491: \begin{figure}
1492: %\includegraphics[angle=-90,scale=0.8]{reds.cps}
1493: \includegraphics[angle=-90,scale=0.8]{f12.ps}
1494: \caption{Sky positions and accelerations of fitted red-shifted maser components.
1495: {\it (Lower Panel)} LOS accelerations plotted as a function of
1496: disk east-west offset. The solid line marks zero LOS acceleration,
1497: which should occur when masers are on the disk midline.
1498: {\it (Upper Panel)} Sky positions of the corresponding maser components.
1499: The solid line indicates the disk midline predicted by the
1500: geometric model of \citet{Herrnstein2005}. All maser positions
1501: Note that, for this model, masers above the model midline {\it (upper panel)} 
1502: should also have negative accelerations at radii $<$ 7.4~mas, and positive 
1503: accelerations at greater radii.
1504: \label{f:redsky}}
1505: \end{figure}
1506: 
1507: 
1508: 
1509: \clearpage
1510: 
1511: \begin{figure}
1512: %\epsscale{.80}
1513: %\includegraphics[angle=-90,scale=0.8]{bluedisk.cps}
1514: \includegraphics[angle=-90,scale=0.8]{f13.ps}
1515: \caption{As for Figure~\ref{f:redsky}, but for fitted blue-shifted high-velocity 
1516: components. For this model, masers above the model midline {\it (upper panel)}  should have 
1517: negative accelerations at radii $<$ 7.4~mas, and positive accelerations at greater radii.
1518: \label{f:bluesky}}
1519: \end{figure}
1520: 
1521: 
1522: 
1523: \clearpage
1524: 
1525: \begin{figure}
1526: \epsscale{.80}
1527: %\plotone{hv_all.ljgedt.eps}
1528: \plotone{f14.eps}
1529: \caption{Vertical deviations from the warped disk of \citet{Herrnstein2005}
1530: calculated using acceleration data and VLBI data. See Section~6.1 for details.
1531: \label{f:jim}}
1532: \end{figure}
1533: \clearpage
1534: 
1535: \begin{figure}
1536: \hspace{3cm}
1537: %\includegraphics[angle=-90,scale=0.8]{newspiral.ps}
1538: \includegraphics[angle=-90,scale=0.8]{f15.ps}
1539: \caption{
1540: Comparison of spiral structure model predictions in the 
1541: NGC~4258 accretion disk. {\it (Upper)} Schematic of maser disk according 
1542: \citet{Maoz1995}. Masers (filled circles) are periodically located in density peaks 
1543:  of trailing spiral structures (represented here by a 2-arm 
1544: logarithmic spiral). 
1545: Accelerations are negative behind the midline (hatched area) and 
1546: positive in front of the midline. Components should occur within 
1547: several degrees of the midline for maximum velocity coherent maser path lengths. 
1548: {\it (Lower)} Schematic of the maser disk according to MM98. Masers (filled circles) 
1549: are periodically located in the disk, but offset from the midline by 2\degrs (note the 
1550: angle is exaggerated here). Red-shifted high-velocity components have negative accelerations 
1551: (hatched area), whereas blue-shifted component accelerations should be positive 
1552: statistically-speaking. Maser emission occurs at the longest velocity coherent 
1553: path lengths along the line-of-sight through spiral arms.
1554: \label{f:spiralcomp}}
1555: \end{figure}
1556: 
1557: 
1558: \clearpage
1559: 
1560: 
1561: 
1562: \begin{figure}
1563: %\includegraphics[angle=-90,scale=0.7]{paper2_periodogram.ps}
1564: \includegraphics[angle=-90,scale=0.7]{f16.ps}
1565: \caption{
1566: Periodogram of the positions of the high-velocity components in the NGC~4258 disk. The dominant frequency peak at 1.33 mas$^{-1}$ corresponds to a characteristic wavelength of 0.75 mas. The dashed horizontal black line marks the 3$\sigma$ significance level. We confined our analysis to ``wavelengths'' of  $<$ 2 mas, since longer periods would not be well-constrained by our dataset.
1567: \label{f:periodogram}}
1568: \end{figure}
1569: 
1570: 
1571: 
1572: \clearpage
1573: 
1574: \begin{figure}
1575: \hspace{0.7cm}
1576: %\includegraphics[angle=-90,scale=.90]{paper2.ps}
1577: \includegraphics[angle=-90,scale=.90]{f17.ps}
1578: \caption{{\it (Top)} Sky positions of red-shifted high-velocity 
1579: emission in the NGC~4258 disk obtained using VLBI over 18 epochs. 
1580: Dashed lines mark nodes at characteristic wavelengths of 0.75 mas, 
1581: proposed by \citet{Maoz1995}. {\it (Middle)} Blue-shifted components in the 
1582: NGC~4258 disk. {\it (Bottom)} GBT spectrum of the blue-shifted high-velocity 
1583: emission taken on 2003 October 23. Velocity has been represented as approximate 
1584: east-west offset using offset . The arrow marks the velocity of the ``missing'' blue 
1585: component predicted by the Fourier analysis and detected independently in the spectrum.
1586: \label{f:periodicityprediction}
1587: }
1588: \end{figure}
1589: 
1590: \clearpage
1591: 
1592: \begin{figure}
1593: \hspace{0.7cm}
1594: %\vspace{-1.0cm}
1595: %\includegraphics[angle=-90,scale=.75]{nbody_4masses_3epochs.ps}
1596: \includegraphics[angle=-90,scale=.75]{f18.ps}
1597: \caption{{\footnotesize
1598: Scenarios for producing the trend in line-of-sight accelerations
1599: of low-velocity emission as a function
1600: of component Doppler velocity. The line-of-sight direction is along the $z$-axis.
1601: {\it (a)} Masers in circular, Keplerian orbits with locations determined by disk geometry and orientation. 
1602: In this scenario, components (shown as filled, black circles) are in circular
1603: orbits about the black hole (black cross) but only produce visible maser action 
1604: within a favorable locus of disk radius and azimuth angle (shown in gray) determined by
1605: the warping of the disk and tangency of the line of sight to the disk.  The
1606: acceleration trend in this case is due to emission from a number of different
1607: maser components in orbits of different radii.
1608: {\it (b)} Masers in a circular, Keplerian orbit that crosses a spiral arm.
1609: In this scenario, maser components are in a circular
1610: orbit about the black hole and encounter a spiral
1611: arm that perturbs the orbital motions (the arm is represented here by a gray logarithmic spiral, but 
1612: a more realistic scenario would be fragmentary, as in flocculent spiral galaxy disks.) 
1613: The acceleration trend in this case is due
1614: to the passage of maser components through a spiral arm of lower
1615: pattern speed than the local Keplerian velocity (see Figure~\ref{f:spiralacc}). 
1616: {\it (c)} Masers in non-circular, confocal, Keplerian orbits. Maser components in the
1617: same elliptical orbit by definition are at different radii and exhibit different 
1618: accelerations, but they may instead lie on different orbits, with the same effect.}
1619: \label{f:threepanel}}
1620: \end{figure}
1621: 
1622: \clearpage
1623: 
1624: \begin{figure}
1625: \hspace{0.7cm}
1626: %\vspace{-1.0cm}
1627: %\includegraphics[angle=-90,scale=.75]{nbody_4masses_3epochs.ps}
1628: \includegraphics[angle=-90,scale=.75]{f19.ps}
1629: \caption{{\footnotesize
1630: The effect of a spiral arm on accelerations of low-velocity
1631: masers in the NGC~4258 disk.{\it (a)} Initial conditions for an N-body simulation
1632: of a maser clump (black triangle) orbiting a supermassive black hole of $3.8 \times 10^7$ $M_{\odot}$
1633: \citep{Herrnstein2005} (black star). The parameters of the trailing logarithmic spiral arm (grey),
1634: represented by 8000 particles, were arbitrarily chosen to be r$_{0}$ = 1.5 mas, $b$=0.42 and
1635: $\phi_{0}$ = 0\degrs, where  $r = r_0 \exp (b (\phi-\phi_{0}))$. The pattern speed of the spiral arm
1636: is half that of the Keplerian orbital velocity at any given radius.
1637: Note that the true breadth of the arm (10$^{-2}$ pc) is exaggerated for clarity.
1638: {\it (b)} Snapshot at t = 100 years.
1639: {\it (c)} Snapshot at t = 200 years.
1640: {\it (d)} Maser component line-of-sight acceleration as a function
1641: of disk impact parameter due to both the supermassive black hole and a spiral arm of
1642: varying masses. Arm mass is quoted as a percentage of the upper limit maser disk mass,
1643: M$_{disk,upper}$ = 9 $\times$ 10$^5$ M$_{\odot}$ \citep{Herrnstein2005}. Note the significant deviation of the acceleration
1644: profiles at 7, 15 and 30\% from that of the 0\% (black hole only) sinusoid. The curve that best matches
1645: observations is that for an arm mass of 15\%. The grey hatched area indicates the observed projected extent in disk
1646: impact parameter of the low velocity masers.}
1647: \label{f:spiralacc}}
1648: \end{figure}
1649: 
1650: \clearpage
1651: 
1652: \begin{figure}
1653: %\includegraphics[angle=-90,scale=.70]{ecc_analysis_white.ps}
1654: \includegraphics[angle=-90,scale=.70]{f20.ps}
1655: \caption{Contour plot of the ``symmetry'' ratio ($v_{red,\,los}/v_{blue,\,los}$) of model P-V diagrams for
1656: eccentricity ($e$) as a function of periapsis angle ($\omega$). The symmetry of the data 
1657: eliminates values of $e$ and $\omega$ in the positive-gradient hatched regions of the plot.
1658: Using acceleration data for low-velocity emission, we further eliminate all of the negative-gradient
1659: hatched region of 90\degrs $\le$ $\omega$ $\le$ 270\degrs for low-velocity maser components 
1660: on the same orbit. Only unhatched regions of parameter space are permitted by our dataset.
1661: \label{f:stretch}}
1662: \end{figure}
1663: 
1664: 
1665: 
1666: 
1667: 
1668: 
1669: 
1670: 
1671: 
1672: %% If you are not including electonic art with your submission, you may
1673: %% mark up your captions using the \figcaption command. See the
1674: %% User Guide for details.
1675: %%
1676: %% No more than seven \figcaption commands are allowed per page,
1677: %% so if you have more than seven captions, insert a \clearpage
1678: %% after every seventh one.
1679: 
1680: %% Tables should be submitted one per page, so put a \clearpage before
1681: %% each one.
1682: 
1683: %% Two options are available to the author for producing tables:  the
1684: %% deluxetable environment provided by the AASTeX package or the LaTeX
1685: %% table environment.  Use of deluxetable is preferred.
1686: %%
1687: 
1688: %% Three table samples follow, two marked up in the deluxetable environment,
1689: %% one marked up as a LaTeX table.
1690: 
1691: %% In this first example, note that the \tabletypesize{}
1692: %% command has been used to reduce the font size of the table.
1693: %% We also use the \rotate command to rotate the table to
1694: %% landscape orientation since it is very wide even at the
1695: %% reduced font size.
1696: %%
1697: %% Note also that the \label command needs to be placed
1698: %% inside the \tablecaption.
1699: 
1700: %% This table also includes a table comment indicating that the full
1701: %% version will be available in machine-readable format in the electronic
1702: %% edition.
1703: %%
1704: 
1705: \end{document}
1706: %% Tables may also be prepared as separate files. See the accompanying
1707: %% sample file table.tex for an example of an external table file.
1708: %% To include an external file in your main document, use the \input
1709: %% command. Uncomment the line below to include table.tex in this
1710: %% sample file. (Note that you will need to comment out the \documentclass,
1711: %% \begin{document}, and \end{document} commands from table.tex if you want
1712: %% to include it in this document.)
1713: 
1714: %% \input{table}
1715: 
1716: %% The following command ends your manuscript. LaTeX will ignore any text
1717: %% that appears after it.
1718: 
1719: 
1720: %%
1721: %% End of file `sample.tex'.
1722: 
1723: