0706.1811/ms.TEX
1: 
2: %\documentstyle[12pt,aas2pp4]{article}
3: %\documentclass[12pt,preprint]{aastex}
4: \documentclass[preprint2]{aastex}
5: 
6: \begin{document}
7: 
8: \title{Direct determination  of the epicycle frequency in the galactic disk,
9:  and the derived rotation velocity $V_0$}
10: \author{ J.R.D. L\'epine\footnote{ Instituto de Astronomia, Geof\'isica e 
11: Ci\^encias Atmosf\'ericas, Universidade
12:  de S\~ao Paulo, Cidade Universit\'aria, S\~ao Paulo, SP, Brazil; 
13: E-mail: jacques@astro.iag.usp.br}
14: \and
15: Wilton S. Dias \footnote{UNIFEI, Instituto de Ci\^encias Exatas, Universidade
16: Federal de Itajub\'a, Itajub\'a, MG, Brazil 
17: }
18: \and Yuri Mishurov \footnote {South Federal University (Rostov State
19: University), Rostov-on-Don,  Russia}
20:  }
21: 
22: \begin {abstract} 
23: 
24: We present a method which allows a direct measurement of the epicycle frequency
25: $\kappa$ in the galactic disk, using the large database on open clusters completed by our
26: group. The observed velocity vector (amplitude and direction) of the clusters in
27: the galactic plane is derived from the catalog data.   In the epicycle approximation,
28: this velocity is the sum of the circular velocity, described by the galactic rotation
29: curve, and of a residual velocity, which has a direction that  rotates with the
30: frequency $\kappa$.  If for some reason the clusters are formed with non-random initial
31: perturbation velocity direction (measured for instance with respect to the direction of
32: circular rotation), then a plot of the orientation angle of the residual velocity as
33: a function of age reveals the epicycle frequency.  The data analysis confirms that
34: this is the case; due to the non-random initial velocities, it is possible to measure 
35: $\kappa$ for different galactic radii. Our analysis considers that the effect of the
36: arms on the stellar orbits is small (the Galactic potential is mainly axis-symmetric); 
37: in this sense our results do not depend on any specific model of the spiral structure,  
38: like the existence of a given number of spiral arms, or on a particular choice of the
39: radius of corotation.  The values of $\kappa$ provide constraints on the rotation 
40: velocity of the disk and on its minimum beyond the solar radius; in particular, 
41:  $V_0$ is found to be 226 $\pm$ 15 kms$^{-1}$ even if the short scale (R$_0$ = 7.5 kpc)
42:  of the galaxy is adopted. The mesured $\kappa$ at the solar radius is 42$\pm$4 
43:  kms$^{-1}$kpc$^{-1}$.
44: 
45: \end{abstract}
46: 
47: \keywords{Galaxy: spiral arms: Galaxy - rotation curve; Open  Clusters}
48: 
49: 
50: % ------------------------ Section 1 ---------------------
51: 
52: \section{Introduction} 
53: 
54: 
55: Open clusters are ideal objects to study the dynamics of the spiral structure of 
56: the Galaxy. A large number of them have known distances and space velocities 
57: (proper motions and radial velocities), that are more precisely determined than
58: those of individual stars. Another important parameter is their age, which is
59: obtained by means of the HR diagrams. In a previous work, Dias \& L\'epine (2005,
60: hereafter DL) integrated backwards the orbits of a sample of open clusters to 
61: find their birthplaces as a function of time, and obtained the velocity of the
62: spiral arms. In a similar manner, by integrating the orbits, it is possible to
63: determine the velocity (direction and amplitude) of the clusters at the time
64: of their birth. Since the gas of the disk is known to present systematic 
65: perturbations with respect to circular rotation, or "streaming motions" related
66: to the spiral structure, it would not be surprising if the new born clusters also
67: present some systematic deviations from circular motion. The study of initial 
68: velocities allows us to verify if the dominant process of star formation produces
69: clusters with random velocities or, on the contrary, with velocities oriented in 
70: some preferential direction(s). The answer to this question will certainly be 
71: able to put restrictions to the mechanisms that trigger star formation.
72: 
73: One result of our investigation is the discovery that part of the open
74: clusters have initial velocities in a constant narrow angle for long period of
75: times. This property allowed us to develop a method for a direct determination
76: of the epicycle frequency. According to the perturbation 
77: theory, if an open cluster was born with an initial velocity different from that
78: of the circular velocity given by the rotation curve, it will follow an orbit that
79: can be described by the epicycle approximation.  The motion can be seen as the sum
80: of the circular motion around the galactic center, plus a small amplitude rotation
81: around the equilibrium point in the circular orbit. The epicycle frequency $\kappa$
82: is a function of the galactic rotation velocity $\Omega$ and of its derivative
83:   $d\Omega/dR$ only:
84:  $$\kappa^2 = 4 \Omega^2 (1 + \frac{1}{2} \frac{R}{\Omega} \frac{d\Omega}{dR}) ~~~~   (1)$$
85: Based on this equation, the  epicycle frequency is easily derived from any observed
86: rotation curve. However, since there are uncertainties  concerning the exact shape of the
87: rotation curve, as several different curves are proposed in the literature, a direct
88: measurement of the epicycle frequency, without passing through equation (1), is 
89: of great interest. Such a measurement, in turn, can help to select the best galactic
90: parameters and rotation curve. We discuss in this work, in addition to the statistics
91: of initial velocities of open clusters,  the restrictions on the
92: galactic parameters that can be derived from the measured epicycle frequency. 
93: 
94: \section{The Open Clusters Catalog}
95: 
96: 
97: We make use of the {\it New Catalogue of Optically visible Open Clusters and Candidates} 
98: published by Dias et al. (2002) and updated by Dias et al. (2006) \footnote{Available
99: at the web page http://astro.iag.usp.br/$_{\symbol{126}}$wilton}. This catalog updates the previous
100: catalogs of Lyng\"a (1987) and of Mermilliod (1995). The present 2.7 version of the catalog
101: contains 1756 objects, of which 850 have published ages and distances, 889 have
102: published proper motions (most of them determined by our group, Dias et al. 2006 and 
103: references therein) and 359 have radial velocities. Recently, Pauzen and Nepotil (2006)
104: compared statistically our catalog with averaged data from the literature 
105: and reached a result which is in favor of the data from the catalog.
106: 
107: \section{The method}
108: 
109: The main result  of  the perturbation theory of stellar orbits in a central potential,
110: applied to the galactic disk, is that the motion of a star (or of an open cluster)
111: can be described as the sum of a circular rotation around the galactic center, 
112: with angular velocity $\Omega(R)$,  plus an epicycle perturbation, as illustrated in 
113: Figure 1 ( eg. Binney \& Tremaine, 1987, hereafter BT87).
114: The epicycle perturbation is an oscillation with frequency $\kappa$ around the
115: unperturbed circular orbit, in both the radial direction and the direction of circular
116: rotation: 
117: $$r(t) = r_0 + \xi(t), ~~~         with ~ \xi(t)  = b ~sin (\kappa t + \phi)~~~(2)$$
118: $$\theta(t) = \theta_0  + \Omega t   +  \frac{\eta (t)}{r_0}, ~~~   with ~ \eta(t) =  
119: a ~ cos(\kappa t + \phi)~~~(3)$$
120: ($\xi$ and $\eta$ represent the displacements in the two directions,  $b$ and $a$
121: the amplitudes in units of length, and $\phi$ the phase). In the frame of reference 
122: of the guiding center of the epicycle motion, 
123: rotating with angular velocity $\Omega$, the perturbed displacement describes an 
124: ellipse, with the  ratio of the amplitudes $ a/b  =   2 \Omega/\kappa$  (BT87), 
125: which is approximately 1.5. Like in the case of a circular motion, the perturbed
126: velocity rotates with the same frequency of the perturbed  displacement.  The 
127: velocity components are:
128: $$\xi'=\frac{d\xi}{dt} = b~\kappa~cos (\kappa t + \phi)~~~(4) $$
129: $$\eta'= \frac{d\eta}{dt} =  -a ~\kappa~sin(\kappa t + \phi)~~~(5)$$
130: The ratio of the velocity amplitudes is the same of the displacement amplitudes. 
131: The angle $\gamma$ between the perturbed velocity vector and the direction of circular 
132: rotation, taken as a reference (see Figure 2), is: 
133: $$ \gamma(t) = tan^{-1} (\frac{\xi'}{\eta'}) = tan^{-1} [(b/a) tan(\kappa t + \varphi)]~(6)$$ 
134: We introduced a new symbol, $\varphi$, for the phase, because we want to adopt 
135: the direction of circular rotation as the origin for $\gamma$; the relation between 
136: the two phases is  $\varphi = \phi + \pi/2$. In the figures, we represent the galactic
137: rotation in the clockwise direction, while $\gamma$ increases in the counterclockwise
138: direction. The situation $\gamma = 0^o$ occurs when the cluster is at its minimum galactic
139: radius,  along the epicycle trajectory ($\phi = -\pi/2$, in Equation 2).  
140: 
141: \begin{figure}
142: \plotone{figure1.eps}
143: \caption{Epicycle orbit around an equilibrium point rotating with circular velocity
144: V around the Galactic Center. V$_p$ represents the perturbed velocity.}
145: \label{figure1}
146: \end{figure}
147: 
148: \begin{figure}
149: \plotone{figure2.eps}
150: \caption{The geometry of the velocity vectors. We adopt a usual convention with the Y axis 
151: oriented along the circular rotation at the solar position, and the X axis in the
152: galactic radial direction, oriented towards external regions. One can see that
153: $\gamma = \theta + tan^{-1} (V_x/V_y)$, with $\theta = sin^{-1}(y/R)$}
154: \label{figure2}
155: \end{figure}
156: 
157: In the hypothetic case $a=b$, $\gamma$  would increase linearly with time, with a slope
158: $\kappa$. In the case $a/b$ = 1.5, $\gamma$ is a smooth increasing function of time,
159: with a slope that on the average is equal to $\kappa$, since in a time $T= 2\pi/\kappa$,
160: $\gamma$ varies $360^o$. If all the clusters had a same starting angle, 
161: in a plot of $\gamma$ versus age of the clusters the slope of the fitted line would
162: give directly the value of $\kappa$. 
163: 
164: The present day velocity vector of an open cluster with respect to the Sun, projected  on
165: the galactic plane, is obtained from the data given in the catalog: it is the sum of the
166: radial velocity projected on the plane, and of the velocity in the plane  of the sky,
167: derived from  proper motion and distance,  also projected on the galactic plane. We
168: first convert the velocities to the Local Standard of Rest (LSR). Different sets of
169: solar velocity components ($u,v,w$) with respect to the LSR are recommended in the
170: literature, like the Standard Solar Motion, with ($u,v,w$) = (-10.4, 14.8, 7.3)
171: (in kms$^{-1}$, Mihalas \& Binney, 1981),  (-10.4, 5.3, 7.3) from the analysis of 
172: the Hipparcos data by Dehnen et al. (1987), (-9.7, 5.2, 6.7) also based on Hipparcos
173: data by Bienaym\'e (1999), or (-9.9, 12.1, 7.5) according to Abad et 
174: al.(2003), among others. Since differences of a few kms$^{-1}$ between the 
175: diverse systems only occur in the $v$ component, we varied it to verify how it
176: affects our results. We finally adopted  $v$ = 9.0 kms$^{-1}$, since this value 
177: corresponds to a minimum of the rms deviation of the adjusted points in the 
178: determination of the epicycle frequency. However, this choice is not critical for
179: our results; almost the same values of epicycle frequencies were 
180: obtained with $v$ in the range 6 to 12 kms$^{-1}$. Interestingly, we have 
181: another argument  based on the new ideas of the present paper in favor of this
182: intermediate value of $v$ between the extreme ones, 5 and 15 kms$^{-1}$. The
183: perturbation velocities are rotating, and since the clusters span a range of ages,
184: even if their initial velocities had some preferential direction, the present direction
185: must be almost random (or cover almost uniformly the range -180$^\circ$
186: to 180$^\circ$). However, if we adopt an artificially incorrect value for   $v$,
187: like 0 kms$^{-1}$ or 20 kms$^{-1}$, the present day distribution becomes
188: either very concentrated around $\pm$ 180$^\circ$ or around 0$^\circ$. This 
189: experiment tells us that the correct $v$ must be near the center of the range
190:  0 kms$^{-1}$- 20 kms$^{-1}$.
191: 
192: A further step in the treatment of the velocity vectors is their transformation to
193: the local reference frame of each cluster, as illustrated in Figure 2. The new frame 
194: has an axis in the local direction of rotation and the other in the local radial 
195: direction. This step includes the corrections for differential rotation, described 
196: below. The local frame of the clusters at present time is the most convenient one to 
197: start numerical integration of orbits. However, for the other use of the 
198: velocity vectors, which is the direct determination of the epicycle frequency, a
199: small correction is still required to place the velocity in the frame of reference 
200: of the guiding center, as later discussed. 
201: 
202: In this work, we do not take into account the velocities perpendicular to the 
203: plane; only the velocity components projected on the galactic plane are investigated.
204: 
205:  
206: \subsection{Corrections for differential rotation and local frame of reference}
207: 
208: After electing a rotation curve, we subtract from the observed velocity of a cluster
209: the velocity expected from pure circular rotation at its position; what is left
210: is the perturbation velocity. This is not different from what DL called correction
211: for differential rotation. The corrections can be made in different ways; the one that
212: we adopted is to compute, for the longitude and distance of each cluster, 
213: the expected  U and V velocity components, assuming that the cluster
214: presents pure circular rotation, and subtracting the  velocity components
215: (0, V$_0$) of the LSR. Of course, the perturbation velocity vectors that we 
216: derive (observed - expected velocities) depend slightly on the choice of the galactic parameters 
217: $R_0$, $V_0$, $(dV/dR)_0$ (the radius of the solar orbit, the rotation velocity of 
218: the LSR around the galactic center, and the slope of the rotation curve, respectively).
219:  However, we are dealing with second order (or differential) effects; if we adopt a
220:  rotation curve that is too low or too high, the LSR velocity and the expected
221: cluster velocity are affected in a similar way, and the computed perturbation velocity 
222: does not change too much. As already discussed, once a perturbation velocity vector 
223: is established, its angle $\gamma$ will be measured with respect to the local  
224: rotation direction.
225: 
226: Since the epicycle frequency depends strongly on the galactic radius (equation 1), we must
227: select the clusters in a narrow range of radius, to be able to observe a well defined
228: rotation of the perturbation velocity. We must make a compromise, since too narrow 
229: galactic radius ranges will contain only  a few clusters. After a number of experiments, 
230: we found that a radial extent of the order of 0.6 - 1 kpc is convenient. A special radius
231: range is the one that contains the Sun (let us say, the range  $R_0 \pm $ 0.4 kpc),
232: because it is not affected by differential rotation; whatever the rotation
233: curve we choose, the expected circular rotation velocity of the clusters is the same
234: of LSR rotation around the galactic center.
235: 
236: \subsubsection{Rotation curves}
237: 
238: For radius ranges (or rings) relatively distant from $R_0$, the choice of the rotation
239: curve is relevant.  Our task is risky, since we are willing to observe perturbations of
240: the order of 10 kms$^{-1}$ over a line-of-sight projection of rotation velocity 
241: of the order of 130 kms$^{-1}$.
242: For instance, if the true rotation velocity of the Galaxy in a ring is
243: larger than the one that we  use for the corrections, the perturbation velocities
244: are under-corrected in the direction of $\eta'$ and the observed $\gamma$s will tend
245: to be closer to zero than they should be. The effect is similar to the one that we
246: described previously, concerning the $v$ component of the solar velocity. In principle,
247: the fact that under-corrections or over-corrections tend to destroy the "normal" 
248: distribution of present day angles of the velocity vectors could be used as a guide
249: to select the correct rotation curve. But this is not the purpose of the present work. 
250: 
251: Anyway, in order to check the effect of the differential rotation corrections on
252: the derived epicycle frequency, we performed the calculations over a grid of rotation
253: curves. We varied $R_0$ from 7.0 to 8.5 kpc, in steps of  0.5 kpc, and $V_0$ from
254: 170 to 250 kms$^{-1}$, in steps of 10 kms$^{-1}$. The method is very similar to that
255: of DL for the determination of the pattern rotation speed; for each point of the
256: grid we make use of a rotation curve that passes at the corresponding ($R_0$, $V_0$)
257: and  fits the observed CO data taken from Clemens (1985);  the same set of rotation
258: curves of DL was used here. These are curves directly derived from observations, 
259: with the rotation velocities recomputed for each adopted ($R_0$, $V_0$). A simple
260: and smooth analytical expression was fitted to each of these curves, in order to be 
261: able to compute the corresponding "theoretical" epicycle frequency. We refer to 
262: these curves as CO-based curves or smooth curves. We also performed a number of
263: experiments with constant velocity curves, or pure "flat"curves. 
264: 
265: Not all the points of this grid of rotation curves and radial ranges produce a valid 
266: measurement of the epicycle frequency. We later discuss in more detail an  example
267: with a specific choice of $R_0$, in order to clarify what we consider a valid
268: measurement. 
269: 
270: \subsubsection{The reference frame of the guiding center}
271: 
272: The equations 2 to 5 are valid in the reference frame of the guiding center of
273: the epicycle motion, which does not coincide with the local reference frame at present
274: epoch. A small correction to the angle $\gamma$, corresponding to a 
275: change of the velocity component in the $\eta$ direction, is required for this last
276: frame conversion. The velocity shift only affects the direction of circular rotation,
277: since  the local frame and the the frame of the guiding center have the same 
278: (zero) radial velocity.
279: 
280: The expression for the correction is derived in Appendix A, based on the equations
281: of motion  derived by  Makarov et al.(2004, hereafter MOT). It should be noted that the
282:  MOT's equations are mainly intended to describe the positions of stars as a
283:  function of time. In the analysis of positions, a secular drift appears and is
284:  quite important, this being the reason for the stretching of open clusters. In
285: the present work, we are not dealing with positions, but only with velocities 
286: (our analysis is based on plots of velocity directions as a function of age). Velocities
287: are not affected by secular terms like positions are; this is why young associations
288: that are considerably stretched (like Sco-Cen, and the stellar streams) are
289: easily recognized through their common spatial velocities. In our particular case
290: the present day position of the particles (clusters) coincide with the origin of
291: the individual local frame. As we move towards the past or  the future, 
292: the local frame rotates with a constant circular velocity and the particles move out
293: from the origin.
294: 
295: It is shown in Appendix A that the amplitude of the correction $\delta \gamma$ to
296: the present day angles, that we apply to each point on the graphs like figure 4, has 
297: an rms value about 15$^\circ$, and is on the average equal to zero. Since the tolerance
298: adopted in our analysis (section 3.3) for a point to be considered as belonging 
299: to a fitted line is 20$^\circ$, and the fitted parameters ($\kappa$ and 
300: $\varphi$) depend on many points, the errors that would be introduced by not
301: performing the $\delta \gamma$ correction would be  small. For such a small correction,
302: a first order estimation of it is largely sufficient. For this reason,
303: the fact that the correction is based on a theory that makes use of the
304: Oort's constants, of which we do not know a priory the exact values, is not relevant.
305: The use of different sets of Oort's constants produces minor differences in the
306: corrections (as illustrated in Appendix A), that do not affect the fitted parameters. 
307: 
308: \subsubsection{The modulus of the velocity perturbation}
309: The histogram of the modulus of the velocity vectors, or residual velocities after
310: correction for differential rotation,  is presented in Figure 3. The peak of the 
311: distribution is at about 10 kms$^{-1}$. In parts of the following study we remove
312: from the sample the clusters with modulus larger than 20 kms$^{-1}$. The  reason
313: is that if the amplitude of the perturbation is large, the simple epicycle
314: approximation is not valid. However, when we are dealing with numerical integration
315: of the orbits, which do not rely on the epicycle approximation, this restriction is 
316: not needed. We also remove from the sample the too small velocities 
317: ($ \mid V \mid <$ 3 kms$^{-1}$ ), since the velocity components would be of the
318: same order of the measurement errors, and the error on the angle $\gamma$ could be large.
319: 
320:  
321: \begin{figure}
322: \plotone{figure3.eps}
323: \caption{Histogram of the modulus of the residual velocity vectors of the open clusters in
324: the galactic plane, after subtracting the normal rotation velocity at that radius.  
325: }
326: \label{figure3}
327: \end{figure}
328: 
329: 
330: \subsection{The case $R_0$=7.5 kpc, and clusters near galactic radius R$_0$}
331: 
332: The measurement of the epicycle frequency gives us a link between $R_0$ and $V_0$,
333: like the well known link between the same parameters given by the Oort's 
334: constants A and B. If we fix one of these parameters, we can derive the other. 
335: Recent work often adopts R$_0$ = 7.5 kpc (Racine \& Harris, 1989, Reid, 1993,
336: and many others). This shorter galactic scale, compared to the IAU recommended 
337: 8.5 kpc scale, is supported by VLBI observations of H$_2$O masers associated 
338: with the Galactic center. The distance to the Galactic center derived from 
339: infrared photometry of bulge red clump stars (Nishiyama et al. 2006) is 
340: R$_0$=7.52 $\pm$0.10 kpc, while astrometric and spectroscopic observations 
341: of the star S2 orbiting the massive black hole in the Galactic center 
342: taken at the ESO VLT (Eisenhauer et al., 2005) gives 7.94 $\pm$ 0.42 kpc.
343: 
344: Figure 4 (a,b,c) are plots of the angle of the perturbed velocity vector
345: with the direction of circular rotation, as a function of age. In these
346: examples, the galactic positions of the clusters were computed considering that
347: R$_0$ = 7.5 kpc, and the plots a, b, c, show clusters selected in different
348: ranges of galactic radius. In particular, the range shown in Figure 4b is
349:  7.1 $< R <$ 7.9 kpc. As we already mentioned, this is a special radius range,
350: since the choice of the rotation curve has no influence on the  fitted $\kappa$.
351: The positions and velocities in the galactic plane of the same sample of clusters
352: are shown in Figure 5. In Figure 4a and 4c, the two lines   represent the
353: function (6), for a same value of $\kappa$, but  different values of $\varphi$. 
354: When a line reaches the maximum angle that can be measured  ($180^o$), 
355:  it starts again at $-180^o$. In Figure 4b only one line was plotted, as it seems 
356:  to be a dominant one, fitting alone 12 clusters, considering that a cluster 
357:  is fitted if it is situated within $\pm 20^o$, in the vertical direction, from
358:  the line. The points situated around this line are reproduced in Figure
359:  6 without the folding at multiples of 180$^\circ$; 2 points slightly off the 
360:  $\pm 20^o$ limit were included. The rms deviation of the points from the line
361:  is 14$^\circ$. It can be noted from Figure 4 that for ages larger than
362:  50 Myr, no points considered as not belonging to the line are closer than about
363:  130$^\circ$(or 9$\sigma$) from it. This justifies our method of "isolating" a 
364:  line to fit the theoretical function. In the present case the fitted value of $\kappa$
365:  is 42.0 $\pm $ 4 kms$^{-1}$kpc$^{-1} $; this is the epicycle frequency at the
366:  solar radius, one of the main results reported here. The initial angle is 
367:  61$^\circ$, in the frame of the guiding center, or a little more, about
368:  80$^\circ$ in the local frame of the clusters. This initial angle is in good
369:  agreement with what is obtained from a totally different method (Figure 4e). 
370:  The probability of the apparent alignment of points being a chance coincidence
371:  is discussed in the next section. 
372: 
373: \begin{figure*}[t]
374: \plotone{figure4.eps}
375: \caption{Left-side (a,b,c): angle $\gamma$ between the perturbed velocity and the
376:  direction of circular rotation as a function of the age of the clusters. The 
377:  lines represent the function (6) for different values of $\varphi$. The clusters
378:  were selected in different ranges of galactic radius, indicated in each frame. 
379:  The fitted values of $\kappa$ were 46, 42 and 45 kms$^{-1}$kpc$^{-1}$ 
380:  respectively. The right side frames are the histograms of initial velocities for
381:  the corresponding radius ranges, obtained by numerical integration of the orbits
382:  (section 3.4)} 
383: \label{figure4}
384: \end{figure*}
385: 
386: 
387: \begin{figure}
388: \plotone{figure5.eps}
389: \caption{Observed position and perturbation velocity vectors of the clusters in 
390: the range of galactic radius 7.1$< R <$ 7.9 kpc. The  radius range is indicated
391: by full lines and the solar radius by a dotted line. In the text we refer to this
392: region as the R = 7.5 kpc ring. The galactic center is not contained in the figure.}
393: \label{figure5}
394: \end{figure}
395: 
396: \begin{figure}
397: \plotone{figure6.eps}
398: \caption{Angle $\gamma$ between the perturbed velocity and the direction of circular
399: rotation as a function of the age of the clusters, for the clusters situated around
400: the line shown in Figure 4b, with the 180$^\circ$ folding removed. The adjusted line
401: gives $\kappa$ = 42.0 $\pm $ 4 kms$^{-1}$kpc$^{-1} $ and $\varphi$ = 70$^\circ$, which 
402: corresponds to an initial angle 61$^\circ$ (Equation 6).}
403: \label{figure6}
404: \end{figure}
405: 
406: 
407: The fact that a reasonable number of points fall on a same line (for instance, 
408: 12 clusters within $\pm$ 20$^o$ of the line that begins at 61$^o$) reveals that some
409: process confines the direction of the initial velocity of the clusters to a same 
410: angle with respect to the direction of circular rotation, for periods of time of 
411: more than 100 Myr. We postpone for the moment the discussion on the nature of such
412: process.
413: 
414: \subsection{Probabilities of alignment in a random process}
415: 
416: In the case illustrated in Figures 4b, the sample contains 52 clusters.
417: The probability of 12 clusters being fitted by a theoretical line, as it happens 
418: in this case, is small. The probability of a point falling on a given line which
419: has a "thickness" of 40$^o$ is 40/360 = 1/9. The probability of 12 points being 
420: on a given line, out of 52 experiments, or 12 successes and 40 failures, is 
421: [52!/(12! 40!)](1/9)$^{12}$(8/9)$^{40}$, $\approx$  7$\times 10^{-3}$.
422: 
423: In the example of Figure 4a, with a sample of 40 points, 2 lines have been fitted,
424: with a total of 16 points on them. In a random process, the probability of
425: falling on one of the two lines is 2/9. If  40 points are randomly placed on
426: the diagram, the probability of 16 points  falling on any one of 2 given lines
427: (15 successes and 21 failures) is 
428:  [40!/ (16! 24!)](2/9)$^{16}$(7/9)$^{24}$ $\approx 5\times 10^{-3}$.
429:  
430:  In the first example, one could argue that the problem cannot be reduced to that
431:  of throwing 52 points on graph and computing the probability of 12 of them being within
432:  a given distance of a line, since the line was not previously present;
433:  we fitted their parameters so that they  pass through a group of points that
434:  resulted to be aligned. This is equivalent to reduce the number of "successes", 
435:  for instance, we can choose 2 points to define the phase and slope of the  line.
436:  In this case we require only 12  successes and 40 failures; the probability
437:  is still only about 3$\%$. In reality, the probability  of the observed 
438:  alignments of points being spurious can be discarded, since the fitted
439:  value of $\kappa$ are always very close to the ones expected from theory,
440:  and the initial phases are confirmed by another method. Furthermore, there is 
441:  a continuous variation of $\kappa$ from one ring to the other; for instance,
442:  the clusters in  figures 4a and 4b are not the same, since the galactic rings do
443:  not overlap, but the fitted slope changes smoothly from one ring to the other. 
444:  We note that the choice of the number of fitted lines  is arbitrary; similar
445:  conclusions can be obtained with 1, 2, or 3  lines. We also verified that
446:  our conclusions do not depend strongly on the  "thickness" of the lines; if we
447:  decide that points at distances up to 15$^o$ or 25$^o$ from a line are fitted
448:  points, the number of fitted points varies, but the change in the probability 
449:  of being on a line by chance tends to compensate the variation, in the calculation
450:  of probability.
451:  
452:  The simple analysis of probabilities tells us that we are not dealing with random 
453:  alignments, but also warns us that we should not attempt to draw conclusions
454:  from cases with a small number of aligned points, like for instance less
455:  than half of the total number of points being on 3 candidate lines, or less than
456:  one third of the points being on two lines, within a tolerance of 20$^o$. 
457:  When the number of clusters decrease, or when the galactic parameters
458:  or rotation curves are different from the correct ones, we are
459:  sometimes confronted with situations in which  very different values of $\kappa$ 
460:  seem to fit the points with about the same quality of the fit. This limits the
461:  number of galactic rings and the range of parameters  that can be explored. 
462:  
463:  \subsection{Numerical integration of the orbits and histograms of initial directions}
464:  
465: The approximation described in the previous sections is  useful to provide
466: an estimate of the epicycle frequency without depending on a precise 
467: knowledge of the rotation curve. A rough idea of the distribution of initial 
468: velocity directions is obtained simply by looking at the starting angles of
469: the fitted lines (at age=0). 
470: 
471: To obtain the distribution of initial angles, a second method is 
472: to perform a numerical integration of the stellar orbits. For this integration,
473: a first step is to obtain the axis-symmetric potential of the Galaxy, by integrating the
474: force $V(R)^2/R$ based on the adopted rotation curve. In this way, the potential
475: does not depend on any assumption on the different components of the Galaxy
476: (disk, bulge, etc.). The present day position and velocity of a cluster define a 
477: quadrivector q(t) containing the polar coordinates, the radial velocity
478: and the angular momentum, which are the initial conditions of the integration. 
479: In the integration, at each time step, $q(t+dt)$ is deduced from
480: $q(t)$ using a Runge-Kutta IV method. The value of $dt$ is chosen so as to 
481: have about  360 points along a circle around the Galactic center. The integration
482: is performed backward in time, for a total time equal to the age.
483: 
484: If the adopted rotation curve is close to the true one, the two methods
485: should produce similar initial velocity distributions. Therefore, the comparison 
486: between the results of the two methods is a  tool to make a choice among
487: different rotation curves. As  examples of  histogram of initial velocity 
488: directions, those obtained with the same samples used in Figure 4 (a,b,c), are shown in 
489: Figure 4 (d,e,f, respectively).  Note that we expect to obtain similar, but not
490: precisely the same distribution of initial velocities with the two methods, since 
491: the methods rely on different simplifying assumptions. For instance, the numerical
492: integration takes into account the present day galactic radius of individual clusters,
493: while the slope-fitting method considers intervals of galactic radius, and supposes that 
494: the equilibrium radius of the circular orbit is approximately the center of 
495: the radius range. It should be remembered that the initial angle in the slope-fitting
496: method is in the frame of reference of the guiding centers (Section 3.1.2). 
497: 
498: The agreement between the two methods is found to be quite satisfactory when rotation
499: curves with high values of $V_0$ (220, 230 kms$^{-1}$kpc$^{-1} $) are used, but become
500: poor with small $V_0$, like 180 kms$^{-1}$kpc$^{-1} $.
501:  
502:  \subsection{Possible sources of errors}
503:  
504:  A source of uncertainty in the value of $\kappa$ that we obtain, in the examples
505:  that we discussed,  illustrated in Figure 4, is that $\kappa$  depends on the 
506:  definition of the LSR. It may happen that we are not using the best set of solar
507:  velocity components for the LSR corrections, in the choice described in section 3.
508:  The "correct" LSR would be the reference frame that rotates around the
509:  Galactic center with the velocity of the rotation curve at R$_0$. It can be seen
510:  from Figure 2 that if we add a correction to $V_y$ (in the direction of rotation),
511:  the computed value of $\gamma$ will change. A small change in the LSR
512:  definition produces a same $\delta V_y$ for all the clusters, but different angles
513:  $\gamma$ are affected in a different way, so that the slope of the fitted 
514:  lines in Figure 4 may change. The arguments which led us to adopt $v$ = 9 kms$^{-1}$
515:   are explained in section 3.0. We verified that typically, if we add 1 kms$^{-1}$ 
516:   to the $V_y$ component of each cluster, to mimic the effect of an error on
517:   the $v$ component of the solar motion,  the fitted $\gamma$ is changed about
518:   1 kms$^{-1}$kpc$^{-1}$. The error on $\kappa$ introduced by the uncertainty in $v$
519:   is about $\pm$3 kms$^{-1}$kpc$^{-1}$.
520:  
521:  It was mentioned in Section 3 that the ratio a/b is approximately 1.5, but the precise 
522:  value depends on the adopted $\Omega$ and $\kappa$. A number of tests showed that
523:  a small variation of this ratio (like for instance taking a/b = 1.4) does not produce
524:  any change in the fitted values of $\kappa$, so that it was not even necessary to use 
525:  an iterative procedure to determine $\kappa$. Similarly, in section 3.1.2 we commented
526:  that the correction $\delta \gamma$ for the frame of reference of the guiding centers
527:  depend on the Oort's constants, which in turn depends on the adopted rotation curve.
528:  In this case too, these are second order effects, and an iterative procedure is not 
529:  required. In principle, when a line is fitted to the observed points like in Figure 6,
530:  the scattering of the points due to the errors in radial velocities and proper motions
531:  contained in the catalog, as well as the small effects that we just mentioned,
532:  are taken into account in the error on the slope given by the least-square fitting.
533:  We consider that our final errors on $\kappa$ are $\pm$4 kms$^{-1}$kpc$^{-1}$.
534:  
535:  Other effects like  the possible  perturbation of the orbits by the spiral structure
536:  (see eg L\'epine et al, 2003), or the effects of the local wiggles of the rotation
537:  curve (section 3.8) are not considered here as errors of measurements, but the cause
538:  of discrepancies between the measured  $\kappa$ and theoretical values expected
539:  from too smooth rotation curves.
540:  
541:  \subsection{Detailed analysis and the derived value of $V_0$}
542: 
543: We expect that if a rotation curve is a good approximation of the true one,
544: it will produce a match between the theoretical and observed $\kappa$, not
545: only near $R_0$, but at other galactic radii as well. In Figure 7 we present 
546: $\kappa$ as a function of radius, obtained from the same slope-fitting method
547: illustrated in Figures 4(a,b,c). We remark that all the determinations of 
548: $\kappa$ rely on this method, the numerical integration is used only in the 
549: statistics of initial velocities. The method does not give reliable results between
550: 8.0 and 8.9 kpc, because the number of clusters becomes too small, and in addition, the 
551: number of initial phases seem to be larger. Only at 8.9 kpc we found again a
552: reliable result. Although the observed epicycle frequencies do not depart much 
553: from those predicted (based on Eq. 1) by a smooth rotation curve with $V_0$=
554: 220 kms$^{-1}$, the slope of $\kappa$ as a function of radius is steeper than that
555: of  smooth rotation curves. The data in Figure 7 was also fitted by a smooth $V_0$ =
556: 230 kms$^{-1}$ rotation curve modified by the addition of a Gaussian  minimum
557: centered at 7.8 kpc and peak value 6 kms$^{-1}$ , the corresponding value of $V_0$ 
558: being about 225 kms$^{-1}$. The  existence of a minimum is further discussed in a
559:  next section. We do not intend to present a precise fitting of it, but only to
560:  illustrate that it could be an explanation for the anomalous slope of  $\kappa$.
561:  Such a minimum is almost not perceptible in the rotation curve (Figure 10).
562:   
563:  
564: \begin{figure}
565: \plotone{figure7.eps}
566: \caption{Epicycle frequency measured at different galactic radii. The dashed
567: lines represent the theoretical $\kappa$ corresponding to smooth rotation 
568: curves with $V_0$ =  230 kms$^{-1}$(upper curve) and 220 kms$^{-1}$ (lower curve)
569: both with $R_0$ = 7.5 kpc. The dotted line correspond to the theoretical $\kappa$ 
570: obtained with a smooth rotation curve with $V_0$ = 232 kms$^{-1}$, to which 
571: a Gaussian minimum centered at 7.8 kpc has been added.} 
572: \label{figure7}
573: \end{figure}  
574: 
575:  Interesting results are obtained by varying the rotation curves with a fixed $R_0$.
576:  Figure 8 presents the theoretical and observed epicycle frequency as a function
577:  of the adopted $V_0$, for a fixed $R_0$= 7.5 kpc. What we call here the 
578:  theoretical $\kappa$ is the value obtained from direct use of equation (1),
579:  with $\Omega(R)$ given by the adopted rotation curve; the observed $\kappa$
580:  is obtained from the slopes of lines in plots similar to those shown in Figure 4.
581:  For each $V_0$, the same rotation curve is used to determine the theoretical
582:  frequency and for the differential rotation corrections. Figure 8 turns clear
583:  how the best value of $V_0$ can be derived, and how the uncertainty on $\kappa$
584:  affects this $V_0$ value.
585:  
586:  Let us first analyse the case of pure flat rotation curves (V($R$)=V($R_0$)= 
587:  constant). For such curves, it can be derived from equation (1) that  $\kappa$ = 
588:  $\sqrt2 \Omega$. At $R_0$, $\Omega$ = $V_0 /R_0$; the dashed line in Figure 8
589:  gives $\kappa$  as a function of $V_0$, for flat curves with R$_0$ = 7.5 kpc.
590:  This line cuts the line of observed $\kappa$ at 42 kms$^{-1}$kpc$^{-1}$ 
591:  and V$_0$ = 226 kms$^{-1}$. In other words, if the rotation curve were a flat
592:  one, our measurement of $\kappa$ = 42 $\pm$ 4 kms$^{-1}$kpc$^{-1}$ would imply
593:  that V$_0$ is 226 $\pm$ 15 kms$^{-1}$. 
594:  
595:   
596:  \begin{figure}
597: \plotone{figure8.eps}
598: \caption{The epicycle frequency as a function of $V_0$ for different rotation
599: curves, all based on $R_0$ = 7.5 kpc. The full line with positive slope
600: is the theoretical prediction for the series of CO-based rotation curves  
601: with different values  of $V_0$ (see section 3.1.1). The dashed line
602: is the theoretical  $\kappa$ for pure flat curves with different $V_0$. The
603: full line with negative slope shows the observed values of $\kappa$, based on 
604: angle versus age fitting, like in Figures 4(a,b,c). The theoretical and observed
605: $\kappa$  intersect at at about 226 kms$^{-1}$.}
606: \label{figure8}
607: \end{figure} 
608: 
609:   
610: Besides the pure flat (slope zero) curves, another family of curves that we 
611: consider is that of smooth curves, obtained by fitting the CO data 
612: with different values of $V_0$ (section 3.1.1). The line representing the
613: theoretical $\kappa$ for this family of curves  also cuts the observed 
614: line at 226 kms$^{-1}$. The coincidence of the two "theoretical" lines
615: intersecting at 226 kms$^{-1}$ is not surprising, since the smooth curves 
616: for $V_0$ about 220-230  kms$^{-1}$ are  very close to flat ones
617: (see Figure 10).
618: 
619: \subsection{Rotation curves with $R_0$ = 7.5 kpc at different galactic radii}
620: 
621: 
622: We  present in Figure 9  the observed values of $\kappa$
623: (based on figures similar to Figure 4) and theoretical values of $\kappa$
624: (derived from the family of CO-based curves) for two distinct ranges of radius,
625: one smaller than $R_0$, 5.9 $< R < 6.5$ kpc (dashed lines), and the other 
626: larger than $R_0$, 8.2 $< R < 9.8$ kpc (dotted-dashed lines). The lines with positive
627: slope are the theoretical ones. The two lines corresponding to $R$ = 6.2 kpc 
628: intersect at about $V_0$ = 220 kms$^{-1}$; while the two lines corresponding to 
629: $R$ = 9.0 kpc intersect at about 235 kms$^{-1}$; both values have errors $\pm$ 10 kms$^{-1}$
630: and are in agreement   with $V_0$ = 226 kms$^{-1}$, previously obtained from the 
631: clusters situated near $R$ = 7.5 kpc, in Figure 6. 
632: 
633: \begin{figure}
634: \plotone{figure9.eps}
635: \caption{Theoretical and observed values of $\kappa$ as a function of $V_0$
636: of the CO-based rotation curves (all with $R_0$ = 7.5 kpc), for two distinct
637: galactic radii: 6.2 kpc (dashed lines) and 9.0 kpc (dashed-dotted lines). The 
638: positive slope lines are the theoretical ones, and the negative slope are 
639: the observed ones, like in Figure 8.}
640: \label{figure9}
641: \end{figure} 
642: 
643: \subsection{Rotation curves with $R_0$= 7.0 , 8.0 and 8.5 kpc }
644:  
645:  An analysis similar to that of the previous subsections has been performed, 
646:  adopting $R_0$ = 7.0 kpc, $R_0$= 8.0 kpc and $R_0$= 8.5 kpc. For each adopted
647:  $R_0$, families of smooth rotation curves  with different values of $V_0$
648:  (as explained in section 3.1.1) were used to derive the theoretical $\kappa$
649:  at the radius $R_0$. On the other hand, the angle of the observed velocity
650:  perturbation with respect to the direction of circular rotation have been 
651:  plotted,  using the samples of clusters in the radius range $R_0~ \pm$ 0.4 kpc.
652:  The study was also made for rings that do not coincide with $R_0$; the width
653:  of the rings was always about $\pm$ 0.4 kpc. In all the cases the figures 
654:  look similar to Figure 9, with a rising theoretical line and a decreasing
655:  line of "observed"  $\kappa$. The intersection of the two lines is considered
656:  as the best estimate of $V_0$ for the given $R_0$. 
657:   The results are summarized in Table 1.
658:  
659:  
660: \begin{table}[t1]
661: \centering \caption[]{\label{tab1}Values of $V_0$ for which the observed 
662: and theoretical $\kappa$ are equal, for different adopted $R_0$ and different
663: radii of the galactic ring.}
664: \begin{flushleft}
665: \begin{tabular}{ccccccccc}
666: \hline $R_0$& ring radius & best $V_0$ & $\kappa$   \\
667:   kpc    & kpc     &  kms$^{-1}$ & kms$^{-1}$kpc${-1}$ \\
668: \hline \noalign{\smallskip}
669: 7.0 & 5.7 & 195   & 50 \\
670: 7.0 & 7.0  & 230 & 47   \\
671: \hline \noalign{\smallskip}
672: 7.5 & 6.2   & 220  & 48  \\
673: 7.5 & 7.5   & 225  & 42  \\
674: 7.5 & 9.0   & 230  & 35  \\
675: \hline \noalign{\smallskip}
676: 8.0 & 6.7   & 225  & 48 \\
677: 8.0 & 8.0   & 230  & 40  \\
678: \hline \noalign{\smallskip}
679: 8.5 & 7.3   & 210   & 42 \\
680: 8.5 & 8.5   & 245   & 42 \\
681: \noalign{\smallskip} \hline
682: \end{tabular}
683: \end{flushleft}
684: \end{table}
685: 
686:  It can be seen that the observed $\kappa$ at $R$(ring)= $R_0$ is almost the same for
687:  the different adopted $R_0$ (the average of the $R$= $R_0$ rings in Table 1
688:  is $\kappa$=43 ). This is an expected result since the samples of 
689:  clusters are  the same in these cases, as the positions of the clusters are given
690:  relative  to the Sun. And, as we already mentioned, when we study a ring at $R_0$,
691:  the rotation curve has only a second order effect in the determination of $\kappa$.
692:  The fact that a same  best estimate of $V_0$ is obtained for $R_0$= 7.0, 7.5 and 8.0
693:  kpc is an indication that this value is a robust result.
694:  Furthermore, the fact that for $R_0$ = 7.5 and 8.0 kpc almost the same value of $V_0$ is
695:  obtained from two different rings is an indication that the smooth rotation
696:  curves that we are using in these cases are close to the real ones. 
697:  
698:  \subsection{A wiggled rotation curve and the Oort's constants}
699:  
700:   It is well accepted that the rotation curve presents irregularities in the 
701:   solar neighborhood. According to Olling and Merrifield (1998), there is a minimum 
702:   at about $R_0$ + 3 kpc, and a step to higher rotation velocities at about $R_0$ + 4 kpc. 
703:   Honma and Sofue 1997) used the same method of Merrifield (1992), and both groups obtain a
704:  step-like increase of rotation velocity, but at different radius: about 1.2 $R_0$ 
705:  (HS) or about 1.5 $R_0$ (M). Brand \& Blitz (1993) found the deepest minimum at
706:  about 1.15 $R_0$, and the start of a step-like increase at about 1.3 $R_0$. Amaral
707:  et al. (1996), based on an analysis of OH/IR stars, found the minimum at 1.07 $R_0$.
708:  We propose the curve presented in Figure 10, which contains a minor wiggle
709:  represented by a Gaussian minimum centered at 7.8 kpc with a half-width of 1kpc,
710:  and depth 8 kms$_{-1}$. Such a wiggle produces a reasonable fit of the epicycle
711:  frequency (see Figure 7), and the general curve  fits the Clemens (1985) data,
712:  excepting for the minimum at about 8.5 kpc. It should be noted that Clemens did 
713:  not use the same VLSR transformation that we adopted in the present work, and that 
714:  the we applied corrections for the different $R_0$ that are only strictly valid 
715:  for the data inside the solar circle.
716:  
717:  In most of the descriptions that we mentioned, the minimum is at 1.1 $R_0$ or beyond.
718:  So up to  $R_0$ the curve is still quite flat, and we believe that our analysis of
719:  the sample of clusters near $R_0$ is not seriously affected by the wiggles.
720:  Indeed, when we plot the angle $\gamma$ as a function of age, like in Figure 4, but
721:  using our  curve with aminor wiggle to perform the differential rotation correction, 
722:  the result looks very similar to Figure 4, except that the best fit of the slope of the lines
723:  gives us 41 kms$^{-1}$kpc$^{-1}$, instead of 42 kms$^{-1}$kpc$^{-1}$. This difference 
724:  is well within the error bars.
725:   
726:    
727: \begin{figure}
728: \plotone{figure10.eps}
729: \caption{The rotation curve of the Galaxy for $R_0$ = 7.5 kpc. 
730: The line is a fit to the CO data, using the empirical expression 
731: V= 240 exp(-(3.6/r)$^2$) + 350 exp(-r/3.5 -0.12/r); the units are
732:  km$^{-1}$ and kpc. A very small wiggle was added, represented by
733: a  Gaussian minimum -8 exp(-[(r-7.8)/1.0]$^2$), in order to better
734: fit the epicycle frequecy data (Figure 7). Including this term,
735: $V_0$ = 226 kms$^{-1}$ precisely at $R_0$, but outside the minimum
736: $V$ =230  kms$^{-1}$.}
737: \label{figure10}
738: \end{figure}
739: 
740: An important aspect of the minimum in the curve is that it could explain the apparent
741: contradiction between our results and previous results based on the Oort's constants.
742: The value of $V_0$ that we obtain may seem too high, since it corresponds to 
743: $V_0 /R_0$ = 30.1 (the units are  kms$^{-1}$kpc$^{-1}$ in what follows). In a first
744:  order approximation, the parameters V$_0$ and R$_0$ are linked together through 
745:  the Oort's constants A and B, which are determined by observations (V$_0$/R$_0$ 
746:   = A - B,  (dV/dR)$_{R0}$ = -A-B ). A value of A-B  which seemed  to be well 
747:  established is 26, confirmed by recent observations of a different nature 
748:  (Kalirai et al., 2004) which give 25.3. Feast \& Whitelock (1997) obtained 27.2 based
749:  on Hipparcos proper motions of Cepheids. However, Olling and Merrifield (1998)
750:  verified that the Oort's constants A and B differ significantly from the general
751:  $V_0 /R_0$ dependence, in the solar neighborhood. They attributed this effect to
752:  an anomaly in the local gas distribution. Olling \& Dehnen (2004) showed that 
753:  the most reliable tracers of the "true" Oort's constants are the red giants, and derived
754:  A-B = 33. Another previous determination with high value of A-B is that of 
755:  Backer \& Sramek (1999), who measured the proper motion of Sgr A, at the Galactic
756:  center, equal to 6.18 $\pm$ 0.19 mas yr$^{-1}$, equivalent to 29.2 $\pm$ 0.9. This
757:  is an interesting result since it does not depend on local wiggles. Branham (2002) 
758:  obtains $\Omega$= 30.3,  Fern\'andez et al. (2001),$\Omega$ =30, and Miyamoto and
759:  Zu (1998), $\Omega$ = 31.5,  all of them  from the kinematics of OB stars. Meztger et
760:  al. (1998) obtained 31  from Cepheid kinematics, and Mendez
761:  et al. (1999) obtained 31.7 from the Southern Proper Motion Program. We conclude 
762:  from this discussion that the value of $V_0$ that we obtain, and our choice
763:  $R_0$ = 7.5 kpc, are not in contradiction with the constraints of the Oort's constants.
764:  
765:   As we already mentioned, one of the main properties of the wiggles is a minimum 
766:   at about 1.1-1.2 $R_0$, observed by different authors and using different tracers.
767:   We only insist here on the existence of the minimum because it has not always been
768:   recognized, possibly because there seems to be no obvious reason for it. 
769:  The existence of this sharp (half-width about 1.0 kpc) minimum could  possibly
770:  be related to corotation. This resonance is a frontier between two
771:  regions of the galactic disk, one in which the matter flows towards the center, and
772:  one in which the matter flows towards external regions. This tends to form a "vacuum" 
773:  around corotation, as confirmed by hydrodynamic simulations (eg. L\'epine et al. 2001). 
774:  If less gas is present, less stars form, and after a few billion year a "gap" ring
775:  with a smaller stellar density forms. 
776:  
777: \subsection{Statistics of initial directions}
778: 
779: Typically, a ring with a width about 0.8 kpc contains 3 or 4 main directions 
780: of initial velocity perturbation, seen as peaks in the histograms. A preliminary
781: analysis of the position on the galactic plane where open clusters were born,
782: indicates that at least two different directions can originate from a same
783: position in the galactic plane, at a same epoch. It was shown by DL that the
784: clusters were born in spiral arms. We can now say that a same spiral arm can produce star
785: clusters going in different directions. However, this kind of analysis is difficult
786: to perform, since the samples become very small when narrow bins of ages and initial
787: positions are considered.
788: 
789: Different rings may have different main directions, but some general tendencies 
790: can be observed. Figures 11 to 13 present the distribution of initial velocities
791: in the local (U,V) frame of each cluster, for different ranges of radius, obtained
792: by numerical integration. One is the local region, from 7.1 to 7.9 kpc (Figure 11),
793: which has already been studied in detail, and is supposed to include the corotation
794: radius. Another region, from 5.0 to 7.1 kpc (Figure 12) is inside the  corotation
795: circle. We consider a large galactic radius interval to obtain better statistics.
796:  Finally, in Figure 13 we show the distribution for clusters in the radius range 
797: 8.1 to 12 kpc, external to the corotation circle. One can see some preferential 
798: or non-random directions, which are the basis of the present work, since they 
799: offer the possibility of measuring the epicycle frequencies. Although the statistics
800: are poor, it does not seem that a description in terms of pure ellipsoids
801: would be an adequate one.
802: 
803:  Clear differences appear between the three regions, with dominant positive U velocities at
804: large galactic radii, and dominant negative U velocities in the internal region.
805: There seems to be a change of about 180$^\circ$ in the main velocity direction in the
806: comparison of internal and external regions (from -50$^\circ$ to + 130$^\circ$). 
807: This is in general agreement with the idea that outside corotation, the spiral arms 
808: tend to pull the galactic material towards the exterior, while inside corotation 
809: the arms pull the material towards the center.
810: 
811: In principle, we might expect that the clusters start their life traveling along the
812: spiral arms, so that they would not leave the arms before the brightest stars
813: reach the end of their life. This would contribute to the well defined visible
814: aspect of the arms. DL showed that if we select clusters with ages smaller than
815: 20 Myr, the spiral structure appears clearly, while for ages larger than 30 Myr, 
816: it almost disappears. This time scale is comparable to the time required for a 
817: velocity vector to change its direction by 90$^\circ$, which is 1/4 of the epicycle
818: period. Interestingly, the dominant initial directions for the internal regions
819: of the Galaxy satisfy the condition above. If we consider the vector sum of the 
820: circular velocity and of the perturbation velocity, for instance a vector with modulus
821: about 230 kms$^{-1}$ in the V direction plus a perturbation with modulus 15 kms$^{-1}$
822: in the direction -70$^o$ or in the direction -140$^o$ (main directions seen in
823: Figure 12), the resulting velocity is approximately along the spiral arms,
824: considering a typical pitch angle of about 14$^o$. 
825: 
826: \begin{figure}
827: \plotone{figure11.eps}
828: \caption{Distribution of initial velocities of clusters in the galactic radius range 7.1 to
829: 7.9 kpc}
830: \label{figure11}
831: \end{figure}
832: 
833: 
834: \begin{figure}
835: \plotone{figure12.eps}
836: \caption{Distribution of initial velocities of clusters in the galactic radius
837: range 5.0  to 7.1 kpc}
838: \label{figure12}
839: \end{figure}
840: 
841: 
842: \begin{figure}
843: \plotone{figure13.eps}
844: \caption{Distribution of initial velocities of clusters in the galactic radius
845: range 8.1  to 12.0 kpc}
846: \label{figure13}
847: \end{figure}
848: 
849: 
850: \section{Conclusions}
851: 
852: The study of samples of open clusters situated in  small intervals of galactic
853: radius or rings with width $<$ 1 kpc, reveals the presence of several sequences
854: of 4 to 12 clusters that have been formed with a same initial direction of the perturbation
855: velocity, as measured with respect to the direction of circular rotation. The expected
856: rotation of the perturbation velocity vector from the epicycle theory can be observed
857: in plots of the angle versus age of the clusters.  In such plots, the sequences of 
858: clusters along lines that correspond to a same initial angle span ranges of ages
859: of about 40 to 150 Myr. In a given ring, usually 2, 3 or more such preferential
860: directions are observed. Probably, the initial angles are dictated by the local
861: hydrodynamics in the spiral arms. The observations suggest that portions of the 
862: shock waves associated with the arms are coherent, or keep the same characteristics,
863: during  period of times of more than 100 Myr. Therefore, the observations
864: provide an argument in favor of star formation  triggered directly by the 
865: spiral shocks, and not by secondary processes like the explosion of supernovae. 
866: 
867: If we average the initial direction of the perturbation velocity over a relatively 
868: large range of galactic radius (a few kpc), the individual coherent
869: groups become less prominent, or average out, but a broad general distribution
870: of initial directions appears, with a peak at about -70$^o$, for galactic radii
871: smaller than R$_0$ (or smaller than the corotation radius). This particular initial 
872: direction helps the arms to keep their sharply defined aspect, since the clusters do
873: not move out from the arms before the bright short-lived stars reach the end of their life. 
874: 
875: The property of the cluster formation mechanism  of producing initial perturbation
876: velocities in a given direction for groups of clusters spanning a range of ages
877: allows us to perform a direct determination of the epicycle frequency $\kappa$ in
878: the galactic disk. This measurement shows only a minor dependence on the rotation
879: curve which is used to perform the corrections for differential rotation, to find
880: the perturbation velocity. For a given rotation curve, there are two ways of
881: obtaining the epicycle frequency at a radius $R$. One is to make direct use of the 
882: expression for $\kappa$ as a function of $\Omega$ and its derivative, and the other
883: uses the plots of the angle  versus age of the open clusters.  We compared the
884: results obtained using different rotation curves, based on different values of
885: $R_0$ and of $V_0$. If we fix for instance $R_0$ and vary $V_0$, we can  find a
886:  best value of $V_0$, for which the "observed" and the "theoretical" values of 
887: $\kappa$ coincide.
888: 
889: If $R_0$ = 7.5 kpc is adopted, using the sample of clusters in the range 
890: $ 7.1 < R < 7.9$ kpc, we find $V_0$= 226$\pm$15 kms$^{-1}$. The same value
891:  of $V_0$ is obtained when we adopt $R_0$ = 7.0 kpc or $R_0$ = 8.0 kpc. 
892:  Therefore,  $V_0$= 226$\pm$15 kms$^{-1}$ is a robust result, which does not
893:  depend  on the precise choice of $R_0$, in the 7-8 kpc range.
894: 
895: We examined samples situated in rings with different galactic radii; for instance with
896: mean radius between 6 and 8  kpc. The rotation curve $R_0$ = 7.5 kpc , 
897:  $V_0$= 226 kms$^{-1}$, with a minimum at 7.8 kpc is found to be a
898: satisfactory one, since it predicts values of $\kappa$ that are consistent with
899: the observed ones, not only at $R_0$, but also at these different galactic radii.
900: The ratio $V_0 / R_0$ that we recommend is 30.1 kms$^{-1}$kpc$^{-1}$, which is
901: larger than that "classical" value obtained from the Oor's constants (A-B),
902: 25 kms$^{-1}$kpc$^{-1}$. However, as we discussed in a previous section, there are
903:  many recent measurements that result in $V_0 / R_0$ close to 30 kms$^{-1}$kpc$^{-1}$,
904: including a method that does not depend on local irregularities of the rotation 
905: curve. The method to determine $V_0$ that we  proposed is a new one, which is something
906: much needed in view of the conflicting results that have been published in the 
907: last decade, almost always based on analyses of Oort's constants. 
908: 
909: \acknowledgments{The work was supported in part by the Sao Paulo State
910: agency FAPESP (fellowship 03/12813-4)by the Conselho Nacional de 
911: Desenvolvimento Cientifico e Tecnol\'ogico.} 
912:  
913: \section{Appendix A}
914: 
915: The present paper deals with open clusters that are spread over distances
916: distances of several kpc, so that it is convenient to use an individual
917: frame of reference for each cluster, to describe their motion.
918: The local frame is chosen to rotate around the galactic center with the
919: velocity of the rotation curve at the radius where the cluster is at present epoch.
920: In such a reference frame, in which $\xi_0$ = 0 and $\eta_0$ =0, the MOT's equations
921: of motion (see section 3.1.2) reduce to: 
922: 
923: $$ \xi = \frac{\eta_0'}{2B}  + \frac{\eta_0'}{2B} cos(\kappa t)
924:  + \frac{\xi_0'}{\kappa}sin(\kappa t)$$
925: 
926: $$ \eta = \frac{\xi_0'}{2B}  - \frac{\xi_0'}{2B} cos(\kappa t) + \eta_0'\frac{A}{B}t$$
927:  $$  + \eta_0'\frac{A-B}{\kappa B}sin(\kappa t) $$
928:   
929:   where A and B are the Oort's constants. Taking the derivative:
930:   
931:   $$ \xi' = \frac{\eta_0'\kappa}{2B} sin(\kappa t)
932:  + \xi_0' cos(\kappa t)$$
933:  
934:  $$ \eta' = \frac{\xi_0' \kappa}{2B} sin(\kappa t)
935:   + \eta_0'\frac{A}{B} - \eta_0'\frac{A-B}{ B}cos(\kappa t) $$
936:  
937:   
938:   The only non-harmonic term is $\eta_0'\frac{A}{B}$, a term that does not
939:   increase with time and represents a  shift in velocity, which corresponds to the 
940:   change from the local frame to the frame of the guiding center (the frame in 
941:   which the motion is described by pure harmonic functions in both axes). At the
942:   instant t=0 the velocity in the local frame is ($\xi_0', \eta_0'$) and in the 
943:   frame of the guiding center ($\xi_0', \eta_0'[1-\frac{A}{B}]$).
944:   
945:   The initial angle $\gamma$ in the frame of reference of the guiding center
946:   is:
947:   
948:   $$\gamma_{0 GC} = atan(\frac{\xi_0'}{\eta_0'[1-\frac{A}{B}]})  $$ 
949:   so that:
950:   $$\delta \gamma = atan (C\times tan [\gamma_0]) - \gamma_0$$
951:   
952:   where $\delta \gamma$ is the correction to $\gamma_0$ needed to convert it from the
953:   local reference frame to that of the guiding center. The constant $C$ obtained
954:   using the classical values of the constants A =15 kms$^{-1}$kpc$^{-1} $ and B=
955:   -11 kms$^{-1}$kpc$^{-1} $ (Mihalas \& Binney, 1981, chapter 8) is 0.42; 
956:   using the Oort's constants adopted by MOT, A= 12 kms$^{-1}$kpc$^{-1} $ and B=
957:    -15 kms$^{-1}$kpc$^{-1} $, $C$= 0.55. since only an order of magnitude is needed,
958:    we can adopt $C$= 0.5. The function $\delta \gamma$ is illustrated in Figure A1;
959:   its average is zero, its maximum about 24$^o$ and rms value 15$^o$ in the worst case
960:   ($C$= 0.42).
961: 
962: 
963: \begin{figure}
964: \plotone{figA1.eps}
965: \caption{Correction to the angle $\gamma$ to move it to the reference frame of
966: the guiding center of the epicycle motion, as a funtion of $\gamma$. The dashed line 
967: is with $C$=0.55, full line with $C$=0.42 (see Appendix).}
968: \label{figA1}
969: \end{figure}
970: 
971: 
972: 
973: 
974: 
975: 
976: 
977: % -------------------------- Bibliography --------------------------------
978: \begin{references}
979:  
980: \reference{AOLM} Amaral, L.H., Ortiz, R., L\'epine, J.R.D., Maciel, M., 1996, MNRAS 281, 339
981: 
982: \reference{BS}   Backer, D.C., Sramek, R.A., 1999, ApJ 524, 805
983: 
984: \reference {B}  Bienaym\'e, 1999,  A\&A 341, 86 
985: 
986: \reference {BT} Binney, J., Tremaine, S., 1987, {\it Galactic Dynamics}, Princeton
987:  University Press, Princeton, New Jersey. 
988:  
989: \reference {BB} Brand, J., Blitz, L., 1993, Atron. Astrophys. 275, 67
990:  
991: \reference {B} Branham Jr., R.L., 2002, ApJ 570, 190
992: 
993: \reference{Cl} Clemens D. P., 1985, ApJ 295, 422
994: 
995: \reference{Di4} Dias, W. S., Assafin, M., L\'epine, J.R.D., Alessi, 
996: B. S., 2002, A\&A 388, 168
997: 
998: \reference {Di5} Dias, W. S., Assafin, M., Florio, V., Alessi, B. S., Libero, V., 
999:  2006, A\&A 446, 949
1000:  
1001: \reference{Di2} Dias, W. S., L\'epine, J.R.D., 2005, ApJ 629, 825
1002: 
1003: \reference{Di3b} Dias, W. S., L\'epine, J.R.D., Alessi, B. S., 2001, A\&A 376, 441
1004: 
1005: \reference{Di3} Dias, W. S., L\'epine, J.R.D., Alessi, B. S., 2002, A\&A 388, 168
1006: 
1007: \reference {E++} Eisenhauer,F., Genzel, R., Alexander, T, + 18 authors, 2005, ApJ, 628, 246
1008: 
1009: \reference {FW} Feast, M., Whitelock, P., 1997, MNRAS 291, 683
1010: 
1011: \reference{FFT+} Fern\'andez, D., Figueras, F., \& Torra, J., 2001, A\&A 372, 833.
1012: 
1013: \reference {HS} Honma, M.,\&  Sofue, Y., 1997, Publ. Astron. Soc. Japan 49, 453
1014: 
1015: \reference{kal+} Kalirai, J.S., et al., 2004, ApJ 601, 277
1016: 
1017: \reference{LAM} L\'epine, J.R.D., Acharova, I.A. , Mishurov, Yu.N., 2003, ApJ 589, L210
1018: 
1019: \reference{LMD+} L\'epine, J.R.D., Mishurov, Yu.N., \& Dedikov, S.Yu., 2001, ApJ 546, 234.
1020: 
1021: \reference{Ly}  Lynga, G., 1987, Computer Based Catalogue of Open Cluster Data, 5th ed., 
1022: (Strasbourg; CDS)
1023: 
1024: \reference{MOT} Makarov, V.V., Olling, R.P.,  Teuben, P.J., 2004, MNRAS 352, 1199
1025: 
1026: \reference{MPT+} M\'endez, R.A., Platais, I., Girard, T.M., Kozhurina-Platais, V., 
1027: van Altena, W.F.,1999, ApJ 524, L39
1028: 
1029: \reference{Me} Mermilliod, J. C., 1995, in Information and On-Line  Data in Astronomy, 
1030: ed. D. Egret \& M.A. Albrecht (Dordrecht: Kluwer), 127 
1031: 
1032: \reference{M}  Merrifield, M.R., 1992, AJ 103 (5), 1552 
1033: 
1034: \reference {MCS} Metzger, M. R.,  Cadwell, J.A.R., Schechter, P.L., 1998, ApJ 115, 635 
1035: 
1036: \reference {MB}  Mihalas, D., Binney, J.J., 1981, {\it Galactic Astronomy}, 2nd ed.
1037: San Francisco: Freeman
1038: 
1039: \reference {MZhu} Miyamoto, M., Zhu, Z., 1998, ApJ 115, 1483
1040: 
1041: \reference {N++}  Nishiyama, S., Nagata, T., Sato, S., et al., 2006, ApJ 647, 1093
1042: 
1043: \reference{OM} Olling, R.P., Merrifield, M.R., 1998, MNRAS 297, 943
1044: 
1045: \reference {OD} Olling, R.P., Dehnen, W., 2003, ApJ 599, 275
1046: 
1047: \reference {PN} Paunzen, E.; Netopil, M., 2006, MNRAS 371, 1641
1048: 
1049: \reference{RH}	Racine, R.; Harris, W. E., 1993, AJ 98, 1609
1050: 
1051: \reference {R}  Reid, M.J., 1993, ARA\&A, 31, 345
1052: 
1053: \end{references}
1054: 
1055: \end{document}
1056: 
1057: 27.	Abad, C., Vieira, K., Bongiovanni, A., Romero, L.and Vicente, B., 
1058: An extension of Herschel's method for dense and extensive catalogues.
1059:  application to the determination of solar motion., A&A 397, 345-351 (2003)
1060: