astro-ph0410456/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: \usepackage{natbib,psfig}
3: %% If you want to create your own macros, you can do so
4: %% using \newcommand. Your macros should appear before
5: %% the \begin{document} command.
6: %%
7: %% If you are submitting to a journal that translates manuscripts
8: %% into SGML, you need to follow certain guidelines when preparing
9: %% your macros. See the AASTeX v5.0 Author Guide
10: %% for information.
11: \renewcommand{\floatpagefraction}{0.9}
12: \renewcommand{\textfraction}{0.00}
13: \newcommand{\vdag}{(v)^\dagger}
14: \newcommand{\myemail}{kmb@astro.ox.ac.uk}
15: \newcommand{\gtsim}{\mbox
16: {{\raisebox{-0.4ex}{$\stackrel{>}{{\scriptstyle\sim}}$}}}}
17: \newcommand{\ltsim}{\mbox
18: {{\raisebox{-0.4ex}{$\stackrel{<}{{\scriptstyle\sim}}$}}}}
19: \newcommand{\mc}{\multicolumn}
20: %% You can insert a short comment on the title page using the command below.
21: 
22: \slugcomment{to appear in ApJ Letters}
23: 
24: \shorttitle{Symmetry in the changing jets of SS\,433}
25: \shortauthors{Blundell \& Bowler}
26: 
27: \begin{document}
28: 
29: \title{Symmetry in the changing jets of SS\,433\\ and its true distance from us} 
30: 
31: \author{Katherine M.\ Blundell\altaffilmark{1} and Michael G.\ Bowler
32: \altaffilmark{1}}
33: 
34: \altaffiltext{1}{University of Oxford, Department of Physics, Keble
35:   Road, Oxford, OX1 3RH, U.K.}
36: 
37: \begin{abstract}
38: We present the deepest yet radio image of the Galactic jet source,
39: SS\,433, which reveals over two full precession cycles ($> 2 \times
40: 163$\,days) of the jet axis.  Systematic and identifiable deviations
41: from the traditional kinematical model for the jets are found:
42: variations in jet speed, lasting for as long as tens of days, are
43: necessary to match the detailed structure of each jet.  It is
44: remarkable that these variations are equal and opposite, matching the
45: two jets simultaneously.  This explains certain features of the
46: correlated redshift residuals found in fits to the kinematic model of
47: SS\,433 reported in the literature.  Asymmetries in the image caused
48: by light travel time enabled us to measure the jet speeds of
49: particular points to be within a range from $0.24\,c$ to $0.28\,c$,
50: consistent with, yet determined independently from, the speeds derived
51: from the famous moving optical emission lines.  Taken together with
52: the angular periodicity of the zigzag/corkscrew structure projected on
53: the plane of the sky (produced by the precession of the jet axis),
54: these measurements determine beyond all reasonable doubt the distance
55: to SS\,433 to be $5.5 \pm 0.2$\,kpc, significantly different from the
56: distance most recently inferred using neutral hydrogen measurements
57: together with the current rotation model for the Galaxy.
58: 
59:  
60: \end{abstract}
61: 
62: \keywords{Stars: Binaries: Close, Radio Continuum: Stars, Stars:
63:   Individual: SS\,433, W\,50} 
64: 
65: \section{Introduction}
66: \label{sec:intro}
67: 
68: SS\,433 is famous as the first known relativistic jet source in the
69: Galaxy.  Red- and blue-shifted optical lines, indicating velocities of
70: $0.26c$, were discovered by \citet{Mar79a,Mar79b} and interpreted as
71: being from gas accelerated in oppositely-directed jets
72: \citep{Fab79,Mil79}.  \citet{Mar84} successfully fitted a kinematic
73: precessing jet model, finding an intrinsic jet speed of
74: $\approx0.26c$, a precession period of $\sim 163\rm\,days$, a cone
75: opening angle of $\sim20^\circ$, and an inclination to the
76: line-of-sight of the precession axis of $\sim80^\circ$.  Subsequent
77: radio imaging was consistent with this model
78: \citep[e.g.,][]{Hje81,Ver87,Ver93}, with the images of \cite{Hje81}
79: resolving ambiguities in the parameters from optical studies.
80: 
81: While optical spectroscopic data observed over many precession cycles
82: \citep[e.g.][]{Mar84,Eik01} have confirmed the basic parameter values
83: of what has come to be known as the kinematic model, a detailed
84: analysis of any deviations of the data from this model is hindered by
85: inherent degeneracies in the parameters to be fitted, since the
86: measurement is that of line-of-sight Doppler shifts.  Detailed
87: analyses from quasi-daily milli-arcsec scale monitoring are similarly
88: afflicted by such degeneracies since, for example, a variation in
89: observed proper motion can in principle arise from changes in jet
90: velocity projected on the plane of the sky, or changes in the angle
91: the jet axis makes with our line-of-sight, perhaps caused by a change
92: in precession rate.  Moreover, in milli-arcsec scale monitoring only a
93: limited fraction of the precession period of 162 days has been sampled
94: to date.  However, the extended and spatially resolved jet output
95: observed over several arcseconds, while not time-resolved in the
96: conventional sense, is an historical record of the geometry of the jet
97: ejection over two complete precession periods.
98: 
99: \section{The deepest yet image of SS\,433}
100: \label{sec:obs}
101: On 2003 July 10, SS\,433 was observed for 10\,hrs with the VLA in its
102: most extended A-configuration.  The primary flux calibrator was
103: 3C\,286 and the secondary calibrator was 1922+155.  The data were
104: reduced using standard techniques within the AIPS software and the rms
105: background noise was 29\,$\mu$Jy/beam.  The resulting total intensity
106: image is shown in Fig\,\ref{fig:stdmodel}a and reveals two oppositely
107: directed jets, each of which can be followed over more than 2 complete
108: precession periods.  The compression to a zigzag on the sky of the
109: eastern (mainly approaching) jet and corkscrewing of the western
110: (mainly receding) jet, both due to the finite speed of light, are
111: remarkably clear.  Superimposed on the image is the path traced on the
112: sky in the simple kinematic model of \cite{Hje81}, assuming a jet
113: speed of $0.26\,c$, period 162 days and a distance of 5.5\,kpc.  The
114: match in Fig\,\ref{fig:stdmodel}a is sufficiently good for it to be
115: obvious that if the mean speed is $0.26\,c$ and the period 162 days
116: then the distance to SS\,433 is 5.5\,kpc with an uncertainty of
117: perhaps $\pm 0.1\,$kpc.  If the true distance were 3.1\,kpc
118: \citep{Dub98} the path traced on the sky would, for a speed of
119: $0.26\,c$, have almost twice the angular periodicity.  The image could
120: only be (approximately) matched for a mean speed of $\sim
121: 0.15\,c$. Then the bolides would have had to decelerate from $0.26\,c$
122: \citep{Eik01} and in doing so have lost $\sim 20\,$MeV per hydrogen
123: atom, two thirds of the original bulk energy, in a distance less than
124: $\sim 10$ light days. We regard this as so implausible that one might
125: conclude without further analysis that the distance to SS\,433 is
126: $\sim 5.5\,$kpc.  However, we establish this distance below {\em
127: without} assuming the mean speed.
128: 
129: 
130: \section{The true distance to SS\,433}
131: \label{sec:distance}
132: Close inspection of Fig\,\ref{fig:stdmodel}a shows that the real trace
133: departs from the simple kinematic model significantly.  This is
134: clearer in Fig\,\ref{fig:stdmodel}b.  The image shown is that of
135: Fig\,\ref{fig:stdmodel}a but with a Sobel operator having been
136: applied to delineate the ridgeline.  The Sobel operator \citep{Sob90}
137: is a simple edge detect filter which is widely used in magnetic
138: resonance imaging of brains.  The algorithm generating the filter
139: computes the square root of the sum of the squares of the directional
140: derivatives in two orthogonal directions.  The ridgeline shows up as
141: dark since the gradient changes from positive to negative across the
142: jet profile.
143: 
144: Distortions in the image introduced by the finite speed of light are
145: independent of distance so in the absence of such detailed departures
146: as are evidence in Fig\,\ref{fig:stdmodel}, the jet speed and hence
147: the distance would be extremely accurately determined.  Nonetheless,
148: the speed can be determined quite accurately under the assumptions
149: that the jets are symmetric and that special relativity is correct:
150: for a pair of bolides launched simultaneously but ejected in opposite
151: directions, each having velocity of magnitude $\beta$ in units of $c$
152: and making an angle $\theta$ to our line-of-sight, the ratio of their
153: separations from the central core is given by
154: \begin{equation}
155:   \frac{S_{\rm app}}{S_{\rm rec}} = \frac{1 + \beta \cos\theta}{1 -
156:   \beta \cos\theta},
157: \end{equation}
158: where $S_{\rm app}$ is the separation of the bolide which is
159: approaching us from the central core projected on the plane of the sky
160: and $S_{\rm rec}$ is the same for the receding bolide.  The validity
161: of the assumption of symmetry is established in
162: \S\,\ref{sec:deviations}.
163: 
164: 
165: A comparison of the separations from the central core to opposite
166: pairs of well-defined points where the ridgeline crosses the mean jet
167: axis (shown as the yellow line in Fig\,\ref{fig:stdmodel}b) yielded
168: speeds of $0.24 \pm 0.015\,c$, $0.25 \pm 0.03\,c$, $0.27 \pm 0.04\,c$
169: and hence a distance of $5.5 \pm 0.2$\,kpc.  Note that this is {\em
170: independent} of assumptions from optical data, in contrast with
171: previous estimates.  This value of the distance is rather larger than
172: that derived from quasi-daily milli-arcsec monitoring \citep{Sti02},
173: which is vulnerable to temporary deviations from the kinematic model,
174: which we explore in \S\ref{sec:deviations}, during observation,
175: together with the degeneracies mentioned in \S\ref{sec:intro}.
176: 
177: A distance of $5.5 \pm 0.2$\,kpc is somewhat above the weak upper
178: limit of 3.8\,kpc for the distance to SS\,433 which \cite{van80} found
179: from HI absorption measurements (together with a strong lower limit of
180: 3.0\,kpc).  $5.5 \pm 0.2$\,kpc is substantially greater than the
181: distance of 3.1\,kpc inferred from neutral hydrogen measurements by
182: \cite{Dub98}.  They suggested that a gas cloud, seen in HI emission,
183: is interacting with W50: the inferred velocity of this cloud implies a
184: distance of 3.1\,kpc if the rotation model of the Galaxy is correct
185: (SS\,433's Galactic co-ordinates are [39.69, $-2.24$]).  However, this
186: model assumes purely circular motion and takes no account, for
187: example, of the presence of a bar in the Galaxy on the motion of gas
188: near SS\,433 \citep{Bin91}.  A distance of 5.5\,kpc for SS\,433
189: implies the same distance to the W50 nebula.
190: 
191: \section{Deviations from the kinematic model}
192: \label{sec:deviations}
193: 
194: To verify the assumption of symmetry and investigate deviations from
195: the simple kinematic model, the distance was set to 5.5\,kpc and
196: simulated bolides were launched every ten days with equal speeds in
197: each jet, but the speeds were chosen to match the ridgeline of the
198: east jet.  This is illustrated in Fig\,\ref{fig:sobel5p5}a, where the
199: beads are colour coded such that matching colours on opposite sides
200: correspond to bolides launched simultaneously.  The beads shown for
201: the west jet are the symmetric counterparts of the eastern beads.  The
202: assumption of symmetry is very well justified over the entire image,
203: and outside the innermost precession period the test has precision.
204: (Note that although the FWHM of the point spread function of these
205: images is $0.35^{\prime\prime}$, the accuracy with which the centroid
206: of the peak of the jet cross-section may be found (and hence the jet
207: ridgeline known) is $\sim 0.035^{\prime\prime}$.)  The astonishing
208: accuracy with which the western jet is reproduced for a distance of
209: 5.5\,kpc validates the assumption of symmetry, our delineation of the
210: ridgeline and our distance to SS\,433.  Fig\,\ref{fig:sobel5p5}b shows
211: the best fit that can be found for the eastern jet if the distance to
212: SS\,433 is 3.1\,kpc.  The beads shown for the west jet are the mirrors
213: of those shown for the east jet; inspection of
214: Fig\,\ref{fig:sobel5p5}b reveals that this does a very poor job of
215: fitting the west jet (quantified in Table\,\ref{tab:chisq}) and is
216: wholly inconsistent with jet symmetry.  As a consistency check on the
217: measurements of Table\,\ref{tab:chisq}, we calculated $\chi^2$ for the
218: east-jet beads at an assumed distance of 3.1\,kpc against the east-jet
219: beads at an assumed distance of 5.5\,kpc (this was 19.9) and for the
220: west-jet beads at an assumed distance of 3.1\,kpc against the west-jet
221: beads at an assumed distance of 5.5\,kpc (this was 110.3).
222: 
223: \begin{deluxetable}{lrr}
224: \tablecaption{\label{tab:chisq} To quantify the discrepancy between the fit
225: of the west jet to the ridgeline taking the fitted east jet values and
226: assuming symmetry, an independent smooth line was drawn along the
227: ridgeline of the jet, which was determined in most places to
228: $0.035^{\prime\prime}$.  With this uncertainty we then calculated the
229: $\chi^2$ as tabulated below for the sensitive outer regions. }
230: \tablehead{\colhead{Measure  } & \multicolumn{2}{c}{Distance}  \\
231:                    & \colhead{5.5\,kpc (Fig.\,\ref{fig:sobel5p5}a)}
232:                    & \colhead{3.1\,kpc (Fig.\,\ref{fig:sobel5p5}b)}\\ }
233: %% All data must appear between the \startdata and \enddata commands
234: \startdata
235: $\chi^2$ for East jet (20 beads) & 16.2\phantom{xxxxxxxx} &  17.3\phantom{xxxxxxxx}  \\
236: $\chi^2$ for West jet (22 beads) & 17.3\phantom{xxxxxxxx} &  140.1\phantom{xxxxxxxx}  \\
237: \enddata
238: \end{deluxetable}
239: 
240: Fig\,\ref{fig:finalfit} is a different rendering of the total
241: intensity image shown in Fig\,\ref{fig:stdmodel}a, with the beads
242: from Fig\,\ref{fig:sobel5p5}a (5.5\,kpc distance) superimposed, to
243: clearly demonstrate the consistency between the total intensity image
244: and that which had the Sobel operator applied.  The jet speeds vary
245: between $0.243\,c$ and $0.275\,c$.  These are shown in
246: Fig\,\ref{fig:redshifts}a, with the corresponding Doppler redshifts
247: for (hypothetical) optical spectroscopy shown in
248: Fig\,\ref{fig:redshifts}c plotted as residuals with respect to the
249: kinematic model.  These residuals are very modest compared with those
250: presented in \cite{Eik01}.
251: 
252: There is nothing in the accumulated data on optical Doppler shifts
253: \citep[e.g.,][]{Mar84,Eik01} which contradicts our attribution of the
254: observed departures from the simple kinematic model to these small
255: variations in velocity. On the contrary, there is in fact independent
256: evidence of speed variations in the spectroscopic data presented by
257: \cite{Eik01}. The principal features of the correlation between their
258: $z1$ and $z2$ are readily explained by (symmetric) velocity variations
259: in addition to symmetric pointing angle variations (as is the lack of
260: marked correlation with phase); we will discuss this in more detail in
261: a forthcoming paper (Blundell \& Bowler, submitted, see astro-ph).
262: 
263: The same symmetric match to the radio image may be obtained by keeping
264: the jet speed constant at $0.26\,c$ and varying the phase of the
265: precession.  The image on the sky is less sensitive to the phase than
266: is optical spectroscopy; the necessary phases are shown as a function
267: of time in Fig\,\ref{fig:redshifts}b, with the corresponding Doppler
268: residuals shown in Fig\,\ref{fig:redshifts}d.  The phase (in the
269: outer regions) varies approximately linearly with time, corresponding
270: to a rotation period of 276 days, rather different from the mean
271: period of 162 days.  There is also a phase jump of $\sim 60^{\circ}$
272: in less than 10 days.  The corresponding Doppler residuals are much
273: larger than those reported by \cite{Eik01}, precluding phase-only
274: variations being responsible for the observed deviations from the
275: kinematic model (unless our radio image captured a highly anomalous
276: episode).
277: 
278: We suggest that the  systematic deviations from the kinematic
279: model observed in optical spectroscopy arise from a combination of
280: both strictly symmetric oscillations in jet speed and strictly
281: symmetric variations in pointing angle.  
282: 
283: \section{Concluding remarks}
284: \label{sec:conclusions}
285: 
286: We have presented the deepest yet radio image of SS\,433 on arcsecond
287: scales, from which we have derived its distance from us to be
288: 5.5\,kpc, independently of any assumptions from optical data.  We have
289: identified deviations from the simple kinematic model in our image,
290: which last over timescales of 10s of days.  These may be formally
291: fitted in either of two extreme scenarios: by small oscillations
292: purely in jet velocity about a mean of $0.26\,c$ (which are perfectly
293: symmetric in both jets) or by large oscillations purely in the rate of
294: precession about a mean of 162 days (which are perfectly symmetric in
295: both jets).  Additional information from redshift residual data
296: \citep{Eik01} strongly suggests that variations in velocity and in
297: phase are both occurring.
298: 
299: We speculate that the phase (precession rate) variations might arise
300: because of a varying effective moment of inertia of the nozzle (which
301: in turn might arise because of variations in the mass transfer rate).
302: We suggest that these same changes in mass distribution might cause
303: variation in the inner radius of the accretion disc which may determine
304: the speed of jet bolide ejection \citep{Mei01}.
305: 
306: \acknowledgments
307: 
308: K.M.B.\ thanks the Royal Society for a University Research Fellowship.
309: We warmly thank Rob Fender and Simone Migliari for suggesting deep
310: radio imaging of this source as part of a collaborative programme to
311: investigate the related X-ray and radio emission.  The VLA is a
312: facility of the NRAO operated by AUI, under co-operative agreement
313: with the NSF.  It is a pleasure to thank James Binney and Philipp
314: Podsiadlowski for useful discussions.  
315: 
316: \begin{thebibliography}{}
317: \bibitem[Binney et al(1991)]{Bin91}
318: Binney, J., Gerhard, O.E., Stark, A.A., Bally, J.\ \& Uchida, K.I.,
319: 1991, \mnras, 252, 210
320: 
321: \bibitem[Dubner et al(1998)]{Dub98}
322: Dubner, G.M., Holdaway, M., Goss, W.M., \& Mirabel, I.F., 1998, \aj, 116,
323: 1842 
324: %A high-resolution study of the W50--SS\,433 system and the surrounding
325: %medium 
326: 
327: \bibitem[Eikenberry et al(2001)]{Eik01} Eikenberry, S.S., 
328: Cameron, P.B., Fierce, B.W., Kull, D.M., Dror, D.H., Houck, J.R., \& 
329: Margon, B.\ 2001, \apj, 561, 1027 
330: 
331: \bibitem[Fabian \& Rees(1979)]{Fab79}
332: Fabian, A.C.\ \& Rees, M.J., 1979, \mnras, 187, 13{\sc p}
333: %SS\,433: a double jet in action?
334: 
335: \bibitem[Hjellming \& Johnston(1981)]{Hje81}
336: Hjellming, R.M.\ \& Johnston, K.J.\ 1981, \nat, 290, 100
337: 
338: \bibitem[Margon(1984)]{Mar84}
339: Margon, B.\ 1984, \araa, 22, 507
340: %Observations of SS\,433
341: 
342: \bibitem[Margon et al(1979a)]{Mar79a}
343: Margon, B., Ford, H.C., Katz, J.I., Kwitter, K.B., Ulrich, R.B.,
344: Stone, R.P.S., \& Klemola, A., 1979a, \apj, 230, {\sc l}41
345: %Enormous periodic Doppler shifts in SS\,433
346: 
347: \bibitem[Margon et al(1979b)]{Mar79b}
348: Margon, B., Ford, H.C., Grandi, S.A., \&
349: Stone, R.P.S., 1979b, \apj, 233, {\sc l}63
350: %Discovery paper
351: %The bizarre spectrum of SS\,433
352: 
353: \bibitem[Meier, Koide, \& Uchida(2001)]{Mei01} 
354: Meier, D.L., Koide, S., \& Uchida, Y.\ 2001, Science, 291, 84 
355: %Magnetohydrodynamic Production of Relativistic Jets
356: 
357: \bibitem[Milgrom(1979)]{Mil79}
358: Milgrom, M., 1979, \aap, 76, {\sc l}3
359: 
360: \bibitem[Sobel(1990)]{Sob90} Sobel, I., 1990, in ``Machine Vision for
361: Three-Dimensional Scenes'' eds H.\ Freeman, Academic Press, p376
362: %An isotropic image gradient operator
363: 
364: \bibitem[Stirling et al(2002)]{Sti02} 
365: Stirling, A.M., Jowett, F.H., Spencer, R.E., Paragi, Z., Ogley,
366: R.N., \& Cawthorne, T.V.\ 2002, \mnras, 337, 657
367: 
368: \bibitem[van Gorkom, Goss \& Shaver(1980)]{van80}
369: van Gorkom, J.~H., Goss, W.~M., \& Shaver, P.~A.\ 1980, \aap, 82, L1  
370: %H I absorption in the direction of SS 433
371: 
372: \bibitem[Vermeulen et al(1987)]{Ver87}
373: Vermeulen, R.C., Schilizzi, R.T., Icke, V., Fejes, I., \& Spencer,
374: R.E., 1987, \nat, 328, 309
375: %Evolving radio structure of the binary star SS\,433 at a resolution of
376: %15 milli-arcsec
377: 
378: \bibitem[Vermeulen et al(1993)]{Ver93}
379: Vermeulen, R.C., Schilizzi, R.T., Spencer, R.E., Romney, J.D., \&
380: Fejes, I., 1993, \aap, 270, 177
381: %A series of VLBI images of SS\,433 during the outbursts in May/June
382: %1987 
383: 
384: \end{thebibliography}
385: 
386: \newpage
387: 
388: \begin{figure}
389: \epsscale{0.7}
390: \plotone{f1.eps}
391: %\psfig{figure=f1.eps,width=0.7\textwidth,angle=0}
392: \figcaption{\label{fig:stdmodel} {\bf (a)} A total intensity image at
393:   4.85\,GHz observed with the VLA in A-configuration.  Overlaid is the
394:   projection on the sky of two oppositely directed jets (composed of
395:   individual bolides ejected one per day in each direction) a distance
396:   5.5\,kpc from us, having constant ejection speed $0.26\,c$, whose
397:   axis precesses every 162 days and is at an angle of 78$^{\circ}$ to
398:   our line-of-sight (making an angle of 10$^{\circ}$ with respect to
399:   East-West) and tracing out a cone of semi-angle 19$^{\circ}$.  The
400:   blue regions of the line, indicate the bolides (whose motion is
401:   assumed to be ballistic) which are travelling towards us and the red
402:   regions indicate the bolides moving away from us.  To first order,
403:   the kinematic model seemingly shows a reasonable average fit to the
404:   data. {\bf (b)} A Sobel-filtered version of the image shown in (a)
405:   reveals more clearly the inadequacies of the standard kinematic
406:   model with respect to the data.  Implementation of the nodding
407:   parameters quoted by \cite{Sti02} makes only a modest difference to
408:   the appearance of the trace from the simple kinematic model.  The
409:   white circle in the lower right corner is the size of the
410:   $0.35^{\prime\prime} \times$ $0.35^{\prime\prime}$ synthesized beam.
411: }
412: \end{figure}
413: 
414: \newpage
415: 
416: \begin{figure}
417: \epsscale{0.8}
418: \plotone{f2.eps}
419: %\psfig{figure=f2.eps,width=0.8\textwidth,angle=0}
420: \figcaption{\label{fig:sobel5p5} As Fig~\ref{fig:stdmodel}b, with the
421: image having been passed through a Sobel filter.  The coloured beads
422: (now representing one bolide emitted every 10 days in each direction
423: along the ambient jet axis) are the best by eye fits if jet speed is
424: the only varying parameter in the kinematic model and the distance of
425: SS\,433 from us is {\bf (a)} 5.5\,kpc requiring speeds which vary
426: around $0.26\,c$ and {\bf (b)} 3.1\,kpc requiring speeds which vary
427: around $0.15\,c$.  White crosses represent the positions of bolides
428: launched 235 and 245 days previously.  Details in main text.  }
429: \end{figure}
430: 
431: \newpage
432: \begin{figure}
433: \epsscale{0.9}
434: \plotone{f3.eps}
435: %\psfig{figure=f3.eps,width=0.96\textwidth,angle=0}
436: \figcaption{\label{fig:finalfit} The same total intensity image as
437: Fig\,\ref{fig:stdmodel}a but shown with a transfer function which is
438: not monotonic, in order to emphasise local gradients in intensity over
439: a range of intensity levels and hence the good fit of the beads, which
440: are those of Fig\,\ref{fig:sobel5p5}a with the addition of a few
441: outside $\pm 3^{\prime\prime}$ relative Right Ascension. }
442: \end{figure}
443: \newpage
444: 
445: \begin{figure}
446: \epsscale{0.9}
447: \plotone{f4.eps}
448: %\psfig{figure=f4.eps,width=0.96\textwidth,angle=0}
449: %% This figure was made with /home/kmb/projects/ss433/mallard:
450: %% (a) speed_v_time.gle (b) phase_v_time.gle (c and d)
451: %% lineofsight_fig5.gle and all four with fig5_for_paper.gle
452: \figcaption{\label{fig:redshifts} {\bf (a)} The velocities of
453: individual bolides versus the number of days since those bolides would
454: have been observed to be at the core, if our structural deviations are
455: purely attributable to variations in jet speed.  The colour coding is
456: the same as in Fig\,\ref{fig:sobel5p5}a.  {\bf (b)} The phases of the
457: jet axis as individual bolides were launched versus the number of days
458: since those bolides would have been observed to be at the core, if our
459: structural deviations are purely attributable to phase variations in
460: the precession of the jet axis.  Panels (c) and (d) pertain to the
461: lower panels of figures 2 and 3 of \cite{Eik01}: {\bf (c)} The
462: difference in redshifts which would have been observed had the
463: kinematic model been obeyed exactly and the redshifts which would have
464: been observed as the bolides were launched as in the scenario of (a).
465: {\bf (d)} As (c), but for the scenario of (b).  }
466: \end{figure}
467: 
468: \end{document}
469: