1: \documentstyle[12pt,aasms]{article}
2:
3: \def\la{\mathrel{\mathpalette\fun <}}
4: \def\ga{\mathrel{\mathpalette\fun >}}
5:
6: %\def\space{\noalign{\medskip}}
7: \def\fun#1#2{\lower0.837ex\vbox{\baselineskip0ex\lineskip0.209ex
8: \ialign{$\mathsurround=0ex#1\hfil##\hfil$\crcr#2\crcr\sim\crcr}}}
9:
10: \def\rh{R_{\rm h}}
11: \def\trh{t_{\rm rh}}
12: \def\pc{{\rm ~pc}}
13: \def\yr{{\rm ~yr}}
14: \def\kms{{\rm ~km~s^{-1}}}
15: \def\sec{{\rm ~sec}}
16: \def\erg{{\rm ~ erg}}
17: \def\cm{{\rm ~ cm}}
18: \def\gr{{\rm ~ g}}
19: \def\Vlasov{collisionless Boltzmann equation}
20: \def\etal{{\it et al. }}
21: \def\MSUN{\rm M_{\odot}}
22:
23: \def\msun{M_\odot}
24: \def\msunyr{M_\odot \ {\rm yr}^{-1}}
25: \def\sles{\lower2pt\hbox{$\buildrel {\scriptstyle <}
26: \over {\scriptstyle\sim}$}}
27:
28: \def\sgreat{\lower2pt\hbox{$\buildrel {\scriptstyle >}
29: \over {\scriptstyle\sim}$}}
30: \def\Rf{\parindent=0pt\medskip\hangindent=3pc\hangafter=1}
31: \def\la{\mathrel{\mathpalette\fun <}}
32: \def\ga{\mathrel{\mathpalette\fun >}}
33: \def\aprop{\mathrel{\mathpalette\fun \propto}}
34: \def\pgf{\noindent\hangindent = 0.5in\hangafter=1}
35:
36: \begin{document}
37:
38:
39: \title{ The October 1985 Long Outburst of U Geminorum:
40: Revealing the Viscous Time Scale
41: in Long Orbital Period Dwarf Novae}
42:
43:
44:
45: \vskip 1truein
46:
47: \author{John K. Cannizzo\footnote{
48: also University of Maryland Baltimore County}}
49: \affil{e-mail: cannizzo@stars.gsfc.nasa.gov}
50: \affil{NASA/GSFC/Laboratory for High Energy Astrophysics,
51: Code 661, Greenbelt, MD 20771}
52: \authoraddr{NASA/GSFC/Laboratory for High Energy Astrophysics,
53: Code 661, Greenbelt, MD 20771}
54:
55: \medskip
56: \medskip
57: \medskip
58:
59:
60: \author{ Neil Gehrels }
61: \affil{e-mail: gehrels@lheapop.gsfc.nasa.gov}
62: \affil{NASA/GSFC/Laboratory for High Energy Astrophysics,
63: Code 661, Greenbelt, MD 20771}
64: \authoraddr{NASA/GSFC/Laboratory for High Energy Astrophysics,
65: Code 661, Greenbelt, MD 20771}
66:
67: \medskip
68: \medskip
69: \medskip
70:
71: \author{ Janet A. Mattei}
72: \affil{e-mail: jmattei@aavso.org }
73: \affil{
74: American Association of Variable Star Observers,
75: 25 Birch Street,
76: Cambridge, MA 02138-1205 }
77: \authoraddr{
78: American Association of Variable Star Observers,
79: 25 Birch Street,
80: Cambridge, MA 02138-1205 }
81:
82: \medskip
83: \medskip
84: \medskip
85:
86:
87: %\vskip 3truein
88: %\vskip 0.25truein
89:
90: %\vskip 0.5truein
91: %\centerline{submitted to the Astrophysical Journal (Letters)}
92: % 1998, February 10, vol. 494}
93:
94: %\received{ 2001 September 7}
95: %\accepted{ 2001 October 1}
96:
97:
98: \begin{abstract}
99:
100: We examine the AAVSO light curve of U Geminorum
101: from 1908 to 2002, with particular
102: focus on
103: the October 1985 outburst.
104: This outburst was longer than any other seen
105: in U Gem by about a factor of 2,
106: and
107: appears to be unique among all
108: dwarf nova outbursts seen in systems
109: with orbital periods longer than 3 hr
110: in that one can measure the decay time
111: scale during the initial slow decay.
112: This rate
113: is $\sim26\pm6$ d mag$^{-1}$.
114: Using estimates of
115: the rate of accretion during outburst
116: taken from Froning et al.,
117: one can show that
118: $\sim10^{24}$ g
119: of gas was accreted onto the white dwarf
120: during the outburst
121: which
122: constrains the surface density
123: in the outer accretion disk
124: to be $\sim600$ g cm$^{-2}$.
125: The observed time scale
126: for decay is consistent
127: with that expected in U Gem,
128: given its orbital period and disk mass
129: at the time the outburst began.
130: The data are not of
131: sufficient quality to be able to
132: ascertain a
133: deviation from exponentiality
134: in the decay light curve (as in the
135: SU Ursa Majoris stars' superoutbursts).
136: % as
137: %is expected qualitatively during a
138: %``viscous decay''
139: % in which the dominant mode of
140: % depletion
141: % of the gas stored
142: % in the accretion disk
143: %is accretion onto the central object.
144: When coupled with the viscous time
145: inferred from the (short orbital period) SU UMa stars,
146: the U Gem viscous time scale lends support
147: to the standard model in which the decays
148: in dwarf novae can either be viscous or
149: thermal, with the ratio between them
150: being roughly $h/r$ where $h$ is the vertical
151: pressure scale height in the disk.
152: Indeed, dwarf novae are the only systems
153: for which one can be reasonably certain of
154: the identification of ``viscous'' and ``thermal''
155: decays.
156:
157:
158:
159:
160:
161: \end{abstract}
162:
163:
164: \medskip
165: \medskip
166:
167:
168:
169: {\it Subject headings:}
170: accretion, accretion disks $-$ binaries: close $-$
171: stars: individual (U Gem)
172:
173:
174: \section{ Introduction }
175:
176:
177: Cataclysmic variables (CVs)
178: are
179: interacting binaries in which a Roche lobe filling
180: K or M dwarf secondary transfers matter
181: onto a white dwarf (WD) primary (Warner 1995).
182: Dwarf novae (DNe) make up
183: a subclass of CVs which is characterized
184: by semiperiodic outbursts
185: caused by episodic accretion of stored
186: gas onto the WD (Meyer \& Meyer-Hofmeister 1981,
187: see Cannizzo 1993a for a review).
188: U Geminorum is one of the two prototypical
189: dwarf novae, the other one being SS Cygni.
190: U Gem was the first CV to be discovered
191: (in 1855 by John Russell Hind of Bishop's
192: Observatory, London), and is among the better studied.
193: It has an orbital period of 4.25 hr,
194: a primary mass $\sim1.1\msun$, secondary
195: mass $M_{\rm WD}\sim0.4\msun$, and inclination $\sim67^{\deg}$
196: (Krzeminski 1965, Smak 1971, 1976,
197: Warner \& Nather 1971,
198: Sion et al. 1998, Long \& Gilliland 1999).
199: Krzeminski (1965) showed U Gem to be an eclipsing binary
200: system, and Warner \& Nather (1971) and independently Smak (1971)
201: developed the now standard model for nonmagnetic CVs,
202: with U Gem as the prototype.
203: (The inference on orbital inclination comes from the fact that
204: the disk is eclipsed, but not the WD.)
205: U Gem also has a reliable distance
206: $96.4\pm4.6$ pc
207: determined from {\it HST}
208: astrometric parallax (Harrison et al. 1999).
209: % In terms of its outburst characteristics,
210: The mean recurrence time for outbursts
211: in U Gem
212: is $\sim120$ d,
213: during which time it brightens
214: from $m_V\simeq14$ to $m_V\simeq 9$ (Szkody \& Mattei 1984).
215:
216:
217: The outburst durations in U Gem follow a bimodal
218: distribution as with other DNe.
219: The alternation of long and short outbursts
220: has been shown to be a natural consequence
221: of the accretion disk limit cycle model
222: (Buat-Menard et al. 2001).
223: The outbursts last typically
224: either $\sim 3-8$ d or $\sim12-16$ d (measured
225: through $m_V =12$ on the rising and decaying branches).
226: There was, however, one notable exception to this:
227: in October 1985 a long outburst occurred
228: which lasted $\sim40$ d. This long duration
229: is interesting from the standpoint of
230: the accretion disk limit cycle theory which
231: provides the theoretical underpinnings for our
232: current understanding of dwarf nova outbursts.
233: According to theory, a decrease in the light
234: output from the disk can occur in one of two ways:
235: (1) One can have the entire disk completely in a
236: hot, ionized state, and have the mass and concomitant
237: mass accretion rate decrease by virtue of loss of
238: disk material onto the WD $-$ a ``viscous'' decay, or
239: (2) A cooling front can sweep through the disk from
240: outside to inside and revert the disk back to a
241: neutral gas, with very low viscosity, that does
242: not accrete appreciably $-$ a ``thermal'' decay.
243: The ratio of the times between these two
244: time scale is roughly $h/r \simeq 0.03$, the ratio
245: of a local pressure scale height to radius.
246: For some reason (which would be difficult to uniquely
247: identify),
248: the cooling front in the Oct 1985 outburst
249: was delayed for much longer than usual, thereby forcing
250: the entire disk to remain in the hot, ionized state
251: and to continue accreting slowly onto the WD.
252:
253:
254: Although the ``fast'' decays during which the
255: cooling front traverses the disk have been extensively
256: studied in dwarf novae, and in fact form the basis
257: for the ``Bailey'' relation which provides our only direct
258: measure of the Shakura \& Sunyaev (1973) $\alpha$
259: parameter for ionized gas,
260: the long outbursts seen in the well-studied
261: long orbital systems such as U Gem and SS Cyg
262: do not exhibit enough dynamic range in flux
263: to be able to determine
264: a time scale for the slowly decaying
265: portion of the light curve.
266: The term ``viscous plateau''
267: has been used for the slow decays
268: to convey the fact that the flux
269: is roughly constant (Cannizzo 1993b).
270: For instance, an analysis of the long term AAVSO
271: light curve of SS Cyg reveals no ``outlier'' outburst
272: of anomalously long duration (see Fig. 2 of Cannizzo \& Mattei
273: 1992) like the Oct 1985 U Gem outburst.
274: This is significant given that the
275: Cannizzo \& Mattei study
276: covers $\sim10$ times more outbursts than
277: in the current U Gem study.
278: Only in the ``superoutbursts''
279: of the SU UMa subclass of DNe
280: can the slow decay time scale be reliably determined
281: (e.g., Cannizzo 2001b presents a study of the 2001
282: superoutburst of WZ Sge).
283: The October 1985 outburst of U Gem
284: appears to be the only non-SU UMa-type event ever observed
285: for which one can also make this determination
286: and therefore ``test'' (in one limited way)
287: the prevailing paradigm for
288: viscous and thermal decays in accretion disks.
289:
290:
291: \section { Background }
292:
293:
294: Froning et al. (2001)
295: present {\it FUSE} observations
296: of U Gem at three times near
297: outburst maximum and one near quiescence.
298: From their spectral fitting they derive a
299: rate of accretion in the disk
300: near maximum of $\sim5\times 10^{-9}\msunyr$.
301: At this rate, a total mass of $\sim 1\times 10^{24}$ g
302: would be accreted in 35 d, the approximate duration of
303: the viscous plateau for the outburst of current interest.
304: If roughly half of the mass in the disk
305: at the start of the outburst
306: is accreted onto the WD during the outburst,
307: this implies an initial disk mass
308: of $\sim2\times 10^{24}$ g.
309: For an outer disk radius
310: (determined by the orbital
311: period and WD mass) of $4.5\times 10^{10}$ cm,
312: this gives a surface density at the
313: outer disk edge of $\sim600$ g cm$^{-2}$
314: at the start of an outburst,
315: for a quasi-steady disk in which $\Sigma\propto r^{-3/4}$.
316:
317: The controlling (i.e., slowest) time scale
318: in a quasi-steady disk without transition fronts
319: is the viscous time scale at the outer
320: edge $\tau_{\nu}(r_{\rm outer})$.
321: One can obtain a simple estimate of
322: $\tau_{\nu}(r_{\rm outer})$
323: by using
324: the ``vertically-averaged''
325: equations which give the
326: radial disk structure (Shakura \& Sunyaev 1973).
327: In qualitative terms, the viscous time
328: scale depends on the dimensions of the disk,
329: on the amount of stored gas, and on the viscosity
330: parameter $\alpha$.
331: Eqn. [A6] of Cannizzo \& Reiff (1992)
332: gives the midplane disk temperature $T=T(\Omega, \Sigma, \alpha_{\rm hot})$,
333: where the Keplerian frequency $\Omega(r) = (GM_{\rm WD} r^{-3})^{1/2}$
334: and $\alpha_{\rm hot}$ is the value of alpha in the outbursting
335: disk ($\simeq 0.1$, Smak 1984, 1998).
336: If we adopt an opacity law for the
337: ionized state $\kappa = 2.8\times 10^{24}$
338: g cm$^{-2}$ $\rho T^{-3.5}$
339: and use the condition of local hydrostatic equilibrium
340: $\Omega^2 h^2 = c_S^2 = R T \mu^{-1} $ (where
341: $c_S$ is the local sound speed, $h$ is the local
342: disk semi-thickness, $R$ the ideal gas constant, and $\mu$
343: the mean molecular weight $=0.67$),
344: we then get for the viscous time
345:
346:
347: $\tau_{\nu} \equiv (\alpha\Omega)^{-1} (r/h)^2 = 26$ d
348: $m_1^{5/14}$ $(r_{10}/4.5)^{13/14}$ $(\alpha_{\rm hot}/0.1)^{-8/7}$
349: $(\Sigma/6\times 10^2$ g cm$^{-2})^{-3/7}$,
350:
351:
352: \noindent where $m_1=M_{\rm WD}/\msun$
353: and $r_{10} = r_{\rm outer}/(10^{10}$ cm).
354: The viscous time scale $\tau_{\nu} $
355: represents an $e-$folding time over which
356: local surface density
357: perturbations at a given radius
358: are smoothed out in a dynamically
359: evolving disk.
360: This scaling predicts that if the depletion
361: of disk mass were large enough,
362: one should see a flattening
363: in the decay light curve, given that
364: $\tau_{\nu} \propto \Sigma^{-3/7}$.
365:
366:
367:
368: \section { Data and Modeling }
369: \subsection { Data }
370:
371:
372: We examined the long term
373: AAVSO light curve of U Gem from 1908
374: to 2002 which consists of over 51,000
375: observations in the AAVSO
376: International Database
377: from observers worldwide.
378: Figure 1 presents the frequency histogram
379: distribution for the outburst durations,
380: measured from $m_V=12$
381: on the rising and declining portions of outburst.
382: U Gem lies in the plane of the ecliptic
383: so that there are yearly sun gaps in the data.
384: We omit from our analysis outbursts
385: that straddle a sun gap, or that contain
386: any gap of more than 4 d.
387: As with other DNe, one
388: can see the characteristic bimodal distribution of
389: outburst durations.
390: Figure 2 shows the $m_V$ data from the AAVSO
391: of U Gem during a 140 d span beginning at JD 2446320
392: (i.e., 12.5 Sep 1985).
393: This interval covers the $\sim35$ d of the outburst,
394: as well as the following short outburst.
395: The duration of the Oct 1985 outburst exceeded
396: by about a factor of 2 the second longest one.
397: The two straight lines have slopes of 26 d mag$^{-1}$
398: and 1.4 d mag$^{-1}$, respectively.
399: According to theory these
400: should represent the viscous and thermal
401: time scales, which differ by $\sim h/r$.
402: The combination of the finite scatter
403: in the data points and the limited dynamic
404: range of variation in $m_V$ on the slow part of
405: the decay preclude one from being able to assess
406: a possible deviation from exponentiality
407: in the slow decay light curve
408: as the disk drains onto the WD.
409: This effect can be seen in the SU UMa superoutbursts
410: (e.g., Cannizzo 2001b for WZ Sge).
411:
412:
413: \subsection { Time Dependent Modeling}
414:
415:
416:
417:
418: As a final check, we utilize
419: our time dependent accretion disk code
420: to calculate a detailed light curve
421: (for descriptions of the code see
422: Cannizzo 1993b, 2001a).
423: One recent upgrade to the code has been
424: the provision for a variable outer disk radius,
425: which has been implemented as described in
426: Buat-Menard et al. (2001).
427: One sees a rapid expansion of the
428: outer disk in outburst due to the outward
429: movement of hot, ionized gas in the heating
430: front, followed by a slow contraction in
431: quiescence due to the addition of low
432: angular momentum material at the outer edge.
433: In addition, the disk mass in outburst is
434: generally larger, which leads to a slightly
435: brighter outburst.
436: We assume $M_{\rm WD} = 1.1\msun$,
437: $r_{\rm inner} = 5\times 10^8$ cm, and
438: $r_{\rm outer} = 4.5\times 10^{10}$ cm.
439: To calculate $m_V$ we assume $D=96$ pc.
440: We perform a long run taking a mass transfer
441: rate ${\dot M}_T = 6.3\times 10^{18}$ g s$^{-1}$,
442: and wait until the initial transients
443: associated with the initial surface density profile
444: have disappeared.
445: We then extract a long outburst from this run.
446: Figure 3 shows the light curve, disk mass, and
447: accretion rate onto the WD for this model.
448: We have not performed any special tuning
449: to enhance the duration of this long outburst,
450: therefore its duration is somewhat shorter
451: than the (unusually long) outburst of current interest ($\sim20$
452: d versus $\sim35$ d).
453: The two parallel dotted lines in the first panel indicate
454: a decay time of 26 d mag$^{-1}$. This slope matches well with
455: that of the computed light curve.
456: Also, the variation in disk mass indicated in the second
457: panel is in accord with the previous estimate of $\sim10^{24}$ g
458: for the accreted mass.
459:
460:
461:
462: \section {Discussion and Conclusion }
463:
464:
465:
466:
467: Our result supports the standard model
468: for dwarf nova outbursts in which
469: the outbursting disk decays primarily
470: via simple accretion onto the WD when
471: in the ``viscous plateau'' (Cannizzo 1993b)
472: stage, and $\alpha_{\rm hot}\simeq 0.1$.
473: Of all outbursts in U Gem
474: observed by the AAVSO
475: over the past century, only the Oct 1985 outburst
476: had a long enough duration to allow a
477: fairly precise determination of its viscous
478: plateau decay time scale.
479: The decay time $\sim26\pm6$ d mag$^{-1}$
480: is about $\sim2-5$ times longer than observed
481: in, for instance, the 2001 superoutburst
482: of WZ Sge (Cannizzo 2001b),
483: and is consistent with what one would predict based
484: on the orbital period and outburst energetics.
485: The more normal state of affairs for long period DNe
486: in which
487: one frequently sees flat-topped outbursts
488: is that
489: the short duration ($\sim7-10$ d)
490: of the viscous plateau,
491: combined
492: with the
493: slow viscous time in the outer
494: disk,
495: conspire to produce a much smaller
496: dynamic variation in $m_V$.
497: In fact the variations that are
498: usually seen may be
499: due almost entirely
500: to sloshing action of the gas in
501: the disk as it responds
502: to the matter redistribution accompanying the outburst,
503: therefore
504: one cannot reliably determine a slope for the
505: viscous plateau.
506: DNe
507: appear to be unique in that the identification
508: of viscous and thermal decays is reasonably firm.
509: Physical effects extrinsic to the underlying accretion
510: disk limit cycle theory such as strong irradiation
511: and evaporation do not alter the basic picture in
512: a fundamental way.
513: In contrast, the identification of viscous decays
514: with outbursts seen in the X-ray novae - interacting
515: binaries in which the accretor is a black hole or neutron
516: star - is much more problematic
517: (Ertan \& Alpar 1998,
518: King \& Ritter 1998,
519: Shahbaz, Charles, \& King 1998,
520: Cannizzo 2000,
521: Wood et al. 2001).
522:
523:
524: We sincerely thank variable star
525: observers worldwide who, for more
526: than 90 y, have
527: contributed observations
528: on U Gem to the AAVSO International
529: Database.
530: It is their observations that make this study possible.
531:
532:
533:
534: \begin{references}
535:
536: \reference Buat-Menard, V., Hameury, J.-M., \& Lasota, J.-P.
537: 2001, \aap, 366, 612
538:
539: \reference {} Cannizzo, J. K., 1993a, in Accretion Disks in Compact
540: Stellar Systems, ed. J. C. Wheeler (Singapore: World Scientific), 6
541:
542: \reference {} Cannizzo, J. K. 1993b, \apj, 419, 318
543:
544: \reference {} Cannizzo, J. K. 2000, \apj, 534, L35
545:
546: \reference {} Cannizzo, J. K. 2001a, \apj, 556, 847
547:
548: \reference {} Cannizzo, J. K. 2001b, \apj, 561, L175
549:
550: \reference {} Cannizzo, J. K., \& Mattei, J. A. 1992, \apj, 401, 642
551:
552: \reference {} Cannizzo, J. K., \& Reiff, C. M. 1992, \apj, 385, 87
553:
554: \reference {} Ertan, U., \& Alpar, M. A. 1998, \aap, 336, 220
555:
556: \reference {} Froning, C. S., Long, K. S., Drew, J. E.,
557: Knigge, C., Proga, D. 2001, \apj, 562, 963
558:
559: \reference {} Harrison, T. E., McNamara, B. J., Szkody, P., McArthur, B. E.,
560: Benedict, G. F., Klemona, A. R., \& Gilliland, R. L. 1999, ApJ, 515, L93
561:
562: \reference {} King, A. R., \& Ritter, H. 1998,
563: MNRAS, 293, L42
564:
565: \reference {} Krzeminski, W. 1965, \apj, 142, 1051
566:
567: \reference {} Long, K. S., \& Gilliland, R. L. 1999, ApJ, 511, 916
568:
569: \reference {} Meyer, F., \& Meyer-Hofmeister, E. 1981, \aap, 104, L10
570:
571: \reference {} Shahbaz, T., Charles, P. A., \& King, A. R. 1998, \mnras, 301, 382
572:
573: \reference {} Shakura, N. I., \& Sunyaev, R. A. 1973, \aap, 24, 337
574:
575: \reference {} Sion, E. M., Cheng, F. H., Szkody, P., Sparks, W.,
576: Gaensicke, B., Huang, M., \& Mattei, J. 1998, ApJ, 496, 449
577:
578: \reference {} Smak, J. 1971, Acta Astr., 21, 15
579:
580: \reference {} Smak, J. 1976, Acta Astr., 26, 277
581:
582: \reference {} Smak, J. 1984, Acta Astr., 34, 161
583:
584: \reference {} Smak, J. 1998, Acta Astr., 48, 677
585:
586: \reference {} Szkody, P., \& Mattei, J. A. 1984, PASP, 96, 988
587:
588: \reference {} Warner, B. 1995, Cataclysmic Variable Stars, (Cambridge:
589: Cambridge University Press)
590:
591: \reference {} Warner, B., \& Nather, R. E. 1971, \mnras, 152, 219
592:
593:
594: \reference {} Wood, K. S., Titarchuk, L., Ray, P. S.,
595: Wolff, M. T., Lovellette, M. N., Bandyopadhyay, R. M.
596: 2001 ApJ, 563, 246
597:
598:
599:
600: \end{references}
601:
602: \vfil\eject
603: \centerline{ FIGURE CAPTIONS }
604: %\medskip
605:
606:
607: Figure 1. Frequency
608: histogram distribution of all outburst
609: durations in U Gem, compiled from the long term
610: AAVSO light curve.
611: We utilize 1 d bins, and measure the times
612: from $m_V=12$ on the rising and declining portions
613: of the outbursts.
614: The total number of outbursts
615: in the distribution is 78,
616: with 31 below $t_{\rm outburst}=9$ d,
617: and 47 above.
618: As with SS Cyg,
619: the distribution is strongly bimodal, with
620: peaks at $\sim5$ d and $\sim12$ d.
621: The 39 d duration of the Oct 1985 outburst is
622: indicated by the arrow.
623:
624: Figure 2.
625: AAVSO data of U Gem spanning 140 d, starting from
626: JD 2446320.
627: The two straight lines have slopes of 26 d mag$^{-1}$
628: and 1.4 d mag$^{-1}$, respectively.
629:
630: Figure 3. Time variation during a computed
631: ``long'' outburst of the apparent visual
632: magnitude ({\it top panel}), the disk mass ({\it middle panel}),
633: and the rate of accretion onto the WD ({\it bottom panel}).
634: The two dotted lines in the first panel indicate a rate
635: of decay of 26 d mag$^{-1}$.
636:
637:
638: %2445619 12.97996
639: %2445718 5.977156
640: %2446383 39.15527
641: %2446438 3.850909
642: %2446568 4.193646
643:
644:
645:
646:
647:
648: \end{document}
649: