astro-ph0204050/ms.tex
1: % SAMPLE1.TEX -- AASTeX sample paper with minimal markup.
2: %
3: %   V1.1 - S.M.   January  2001
4: %   V1.2 - S.M.  30 jan - included Rino's discussion and figure 4
5: %   V1.3 - R.B.  1 feb - ...
6: %   V1.5 - S.M.  4 feb. small corrections
7: %   V1.6 - R.B.  4 feb small corrections
8: %   V1.7 - S.M.  5 feb - abstract
9: %   V1.8 - S.M.  6 feb - small corrections. added Fabrizio's point on spectrum
10: %   V1.9 - S.M.  7 feb - clarified discussion and only 2 tables, new fig.1,
11: %                         new section 2.2 from F.B.
12: %   V2.0 - S.M.  8 feb - editorials after Rino ,  deleted last line of Tab.2
13: %                        and relative part in the conclusion - SUBMITTED VERSION
14: %%  V2.1 - S.M. 11 marzo - after referee comments - nuova figura
15: %%  V2.2 - S.M. 12/13 marzo - dopo commenti Rino e Fabrizio
16: %   V2.3 - S.M. 14 marzo - resubmitted
17: %   V3.0 - S.M. 26 marzo - after new referee's comments: changed abstract, end of
18: %                         section 1, added references on long Po, added line n=3 in fig.4
19: %   V3.1  - S.M. 28 marzo - SUBMITTED
20: 
21: 
22: %\documentstyle[12pt,aasms4]{article}
23: 
24: \documentclass[12pt,preprint]{aastex}
25: \begin{document}
26: 
27: %  \special{!userdict begin /bop-hook{gsave 150 90 translate
28: %  55 rotate /Times-Roman findfont 60 scalefont setfont
29: %  0 0 moveto 0.7 setgray (Version 3.0 - 26/03/2002) show grestore}def end}
30: 
31: 
32: %%%%%%%%%%%%%% MY DEFINITIONS
33: \def\pdot {\dot P}
34: \def\xte{\textit{RossiXTE}}
35: \def\asca{\textit{ASCA}}
36: \def\sax{\textit{BeppoSAX}}
37: \def\psr{PSR~J1846--0258}
38: \def\kes{Kes~75}
39: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
40: 
41: 
42: 
43: \title{\sax\ observations of the young pulsar in the \kes\ Supernova Remnant}
44: \author{S. Mereghetti}
45: \affil{Istituto di Astrofisica Spaziale e Fisica Cosmica -
46: Sezione di Milano ``G.Occhialini'' - CNR \\
47: v.Bassini 15, I-20133 Milano, Italy \\
48: sandro@ifctr.mi.cnr.it}
49: %\authoremail{sandro@ifctr.mi.cnr.it}
50: \author{R. Bandiera}
51: \affil{Osservatorio Astrofisico di Arcetri   \\
52: Largo E. Fermi 5, I-50125 Firenze, Italy \\
53: bandiera@arcetri.astro.it}
54: \author{F. Bocchino}
55: \affil{Osservatorio Astronomico di Palermo    \\
56: Piazza del Parlamento 1, I-90134 Palermo, Italy \\
57: bocchino@astropa.unipa.it}
58: \author{G.L. Israel}
59: \affil{Osservatorio Astronomico di Roma    \\
60: v.Frascati 33, I-00040 Monteporzio  Catone, Italy \\
61: gianluca@mporzio.astro.it}
62: 
63: \begin{abstract}
64: 
65: We present the results of  \sax\  observations of the young
66: X--ray pulsar \psr, recently discovered at the center of the
67: composite supernova remnant \kes. The pulsar (plus nebula) spectrum can be fitted by
68: an absorbed power law with photon index
69: $\alpha_{\rm ph}=2.16\pm0.15$,
70: $N_{\rm H}=(4.7\pm0.8)\times10^{22}$~cm$^{-2}$, and
71: unabsorbed flux $\sim\!3.9\times10^{-11}$~erg~cm$^{-2}$~s$^{-1}$ (2--10~keV).
72: By joining two observations taken at an interval of two weeks we have been able
73: to obtain a precise measurement of the spin period ($P=324.818968\pm0.000006$~ms).
74: This value, when combined with previous measurements, cannot be fitted by
75: a smooth frequency evolution with a canonical braking index $n=3$.
76: % However, the value $n=3$ cannot be excluded if one or
77: % more glitches occurred during the period 1993 -- 2001.
78: With the hypothesis of no glitches and/or significant timing noise,
79: the braking index would be  $n=1.86\pm0.08$
80: and, assuming a short initial period, the pulsar age
81: would be $\sim$1700 years, closer to that    of the   supernova remnant
82: than the simple estimate   $\tau=P/2\pdot=723$~years.
83: Other likely possibilities involve the presence of glitches
84: and lead to a wide range of acceptable ages.
85: For example, we obtain $n$ in the range 1.8--2.5, if a  glitch occurred near MJD 51500
86: while for a glitch between October 1993 and  March 1999 we can only get a lower limit
87: of $n>1.89$.
88: 
89: 
90: \end{abstract}
91: 
92: \keywords{Pulsars: individual (PSR~J1846--0258) -- Stars: neutron --
93: Supernovae: individual (G29.7--0.3)}
94: 
95: \section{Introduction}
96: 
97: 
98: The X--ray pulsar \psr\ ($P=325$~ms) was discovered by Gotthelf et al.\ (2000)
99: at the center of the composite supernova remnant \kes\ (G29.7--0.3).
100: Up to now, no radio detection of \psr\ has been reported, with an
101: upper limit of $\sim\!0.1$~mJy at 1.5~GHz (Kaspi et al.\ 1996).
102: All the information  on the pulsar timing parameters has been determined through
103: observations in the X--ray band. Gotthelf et al.\ (2000) obtained
104: a period derivative of $\pdot=7.1\times10^{-12}$~s~s$^{-1}$, by
105: comparing five period measurements obtained with
106: the \xte\ and \asca\ satellites in the years from 1993 to 2000.
107: Assuming the canonical relation for the spin-down by magnetic dipole
108: radiation, these values lead to an estimate for the magnetic field
109: of $B\!\sim\!5\times10^{13}$~G,
110: above the quantum critical field and in the range
111: of the highest values observed in radio pulsars.
112: Even more remarkably, the  characteristic age of \psr,
113: $\tau=P/2\pdot=723$~years,
114: is the smallest one of any known pulsar.
115: 
116: The association of \psr\ with the supernova remnant \kes\ can be regarded as
117: almost certain: the pulsar is located at the geometrical center
118: of the $3.5'$ diameter shell (Collins, Gotthelf \& Helfand 2002)
119: and is powering a bright radio/X--ray nebula that gives the composite morphology to this
120: supernova remnant.
121: The estimate of the age of \kes\ is subject to a large uncertainty.
122: Assuming that the shell is still in the freely expanding phase with
123: velocity $v$, the age is $1800\,d_{19}\,(v/5000\,{\rm km\,s^{-1}})^{-1}$~years,
124: where $d_{19}$ is the distance normalized to the value of 19~kpc,
125: estimated for \kes\ with 21~cm observations (Becker \& Helfand 1984).
126: Alternatively, if the remnant is already in the Sedov phase, the age
127: can be estimated by the relation between the radius and the shock temperature
128: inferred from the X--ray spectra. In this way Blanton \& Helfand (1996) derived
129: an age of $7000\,d_{19}$~years, based on the temperature of 0.5~keV measured
130: with \asca.
131: It thus appears that   the pulsar characteristic age
132: is smaller than the age of \kes,
133: although the large uncertainties
134: %involved might reconcile the two values.
135: involved do not allow agreement between the two values to be
136: precluded.
137: 
138: 
139: 
140: 
141: \section{Observations and Data Analysis}
142: 
143: 
144: The location of \psr\ was observed twice with the \sax\ satellite in March 2001
145: (see Table~1).
146: The instrument relevant for the observations reported here is
147: the   Medium-Energy Concentrator
148: Spectrometer (MECS), that    operates in  the 1.8--10~keV energy range,
149: providing a good spatial ($\sim\!1'$ full width at half maximum (FWHM))
150: and energy resolution ($\sim\!8.5$\%  FWHM at 6~keV)  over a circular
151: field of view with a diameter of $56'$ (Boella et al.\ 1997).
152: 
153: The main target of the first observation (A)
154: was  the   pulsar AX~J1844--0258, therefore \psr\ was detected at an off-axis
155: angle of $\sim\!23'$  (results on AX~J1844--0258 will be reported elsewhere).
156: The second observation (B) was  pointed on the Kes 75 supernova remnant.
157: Both observations lasted about 2.5~days, but
158: Earth occultations and passage of the satellite in regions
159: of high particle background resulted in net exposure times of 83.7~ks
160: and 105.3~ks for the MECS instrument.
161: 
162: 
163: \subsection{Timing analysis}
164: 
165: 
166: For the timing analysis we used only counts with energy greater than
167: 3~keV in order to reduce the contribution of  the soft emission from the
168: supernova remnant. For observation B  we used a circular extraction
169: region with radius $2'$ (resulting in 15904 counts),
170: while for observation A we used an elliptical
171: region matched to the shape of the off-axis MECS point spread function (3751 counts).
172: The times of arrival were converted to the Solar System barycenter
173: using the source position
174: R.A.$=18^{\rm h}\,46^{\rm m}\,24.5^{\rm s}$,
175: Dec.$=-02^{\circ}\,58'\,28''$ (J2000).
176: 
177: We first analyzed the two observations separately, using a folding program
178: to search over a grid of period ($P$) and period derivative ($\pdot$) values
179: and adopting the method
180: described in Leahy (1987) to determine the best period and its uncertainty.
181: Since the individual observations are relatively short compared to the
182: pulsar spin-down timescale, the period values giving
183: the highest signal do not depend significantly on $\pdot$. We therefore fixed
184: $\pdot$ to the value $7.097\times10^{-12}$~s~s$^{-1}$
185:  (Gotthelf et al.\ 2000)
186: obtaining the period values reported in Table~1.
187: The light curve obtained from observation B is shown, for different
188: energy ranges, in Fig.~1.
189: The difference between the periods measured in the two observations corresponds to
190: $\pdot=(7.127\pm0.096)\times10^{-12}$~s~s$^{-1}$ which is,
191: within the errors, compatible with the value found
192: from the \asca\ and \xte\ data spanning the years 1993--2000
193: (Gotthelf et al.\ 2000).
194: 
195: The accuracy of the \sax\ satellite clock is good enough to
196: combine the two observations (taken about 15 days apart)
197: into a single data set, thus allowing us to derive a smaller error on
198: $P$ and to directly measure $\pdot$.
199: This was again done by folding the data on a grid of  $P$ and $\pdot$ trial values,
200: leading to $P=324.818968\pm0.000006$~ms and
201: $\pdot=(7.095\pm0.086)\times10^{-12}$~s~s$^{-1}$,
202: where the values refer to the epoch MJD=51991.08778.
203: These results are consistent with an extrapolation of the timing solution
204: derived by   Gotthelf et al.\ (2000).
205: 
206: \subsection{Spectral analysis}
207: 
208: 
209: 
210: For the spectral analysis  we used only observation B that provided
211: data of better quality since \psr\ was detected on-axis, where the MECS
212: instrument has the greatest sensitivity and the best spectral and
213: spatial performances.  Moreover, we used also data from the  LECS
214: instrument (Parmar et al. 1997) that covers the soft energy range down
215: to 0.1 keV.
216: 
217: For the source spectrum extraction we used first a circle of radius
218: 4$'$ for the MECS and 8$'$ for the LECS in order to maximize the
219: contribution from both the supernova remnant and the pulsar nebula
220: emission.
221: The background spectrum was estimated from a circular
222: corona between 4.7$'$ and 9.5$'$  for the MECS and between 8$'$ and 10.6$'$ for
223: the LECS. The source spectra were rebinned in order to have at least 30
224: counts per channel.
225: 
226: We   verified that a fit with a power-law only fails
227: to describe the data ($\chi^2$/d.o.f. = 463/404), giving strong residuals
228: corresponding to the presence of the emission lines of Si XIII (1.84
229: keV) and S XV (2.45 keV).
230: We therefore fitted the spectra using a combination of a power law and an optically
231: thin thermal plasma ({\sc mekal} in the XSPEC softare package v.11),
232: plus an interstellar absorption of Morrison and McCammon (1983). This
233: combination of models is known to be appropriate for Kes 75, because
234: it also takes into account the presence of a thermal shell (Blanton \&
235: Helfand 1996).
236: This model produced a  very good fit ($\chi^2$/d.o.f. = 398/402)
237: with the following parameters: interstellar absorption
238: $N_H=4.7(3.9-5.5)\times 10^{22}$ cm$^{-2}$, photon index
239: $\alpha_{\rm ph} = 2.19 (2.07-2.30)$, plasma temperature $kT = 0.40(0.35-0.53)$ keV.
240: This best-fit is shown in Fig.\ 2. The thermal
241: and non-thermal fluxes in the 0.1--2 keV range  are respectively
242: $10.1(3.0-24.4)\times 10^{-10}$
243: and $9.9(7.9-12.0)\times 10^{-11}$ erg cm$^{-2}$ s$^{-1}$,
244: while the corresponding values in the 2--10 keV band
245: are $5.4(1.6-13.0)\times10^{-12}$ and
246: $3.4(2.7-4.1)\times 10^{-11}$ erg cm$^{-2}$ s$^{-1}$.
247: The contribution of the thermal component to the flux density drops below
248: 10\% of the total above 3 keV. To further constrain the spectrum of the
249: nebula, we have repeated the analysis using the data collected from
250: a   smaller circle (2$'$ radius), and fixing the temperature of the
251: thermal component and the interstellar absorption to the values obtained previously.
252: %in the large extraction region.
253: This resulted in  a best-fit with $\alpha_{\rm ph} =
254: 2.16(2.10-2.21)$  ($\chi^2$/d.o.f.=348/317). The non-thermal unabsorbed
255: flux in the 2--10 keV band is $3.9(3.5-4.2)\times 10^{-11}$ erg
256: cm$^{-2}$ s$^{-1}$.
257: These results are consistent with (and have smaller uncertainties than)
258: those obtained by Blanton \& Helfand (1996)
259: with \asca\, that, similarly to our instrument, did not have enough spatial resolution
260: to disentangle the contribution from the shell and the central pulsar/nebula.
261: 
262: 
263: We also examined the pulsar spectrum in two different phase intervals
264: corresponding to the high and low parts of the pulse profile, finding
265: marginal evidence for a hardening  in correspondence of
266: pulse maximum.  This indicates that the surrounding diffuse X--ray
267: nebula, giving a higher relative contribution during the phase of
268: low pulsed emission,  has a  spectrum softer than that of the pulsar.
269: In fact, the preliminary \textit{Chandra} results reported by Collins et al. (2002),
270: yield a power law photon index of $\sim$1.5 for the pulsar and $\sim$1.9 for the
271: nebula.
272: 
273: 
274: 
275: \section{Discussion}
276: 
277: 
278: 
279: The period measurement derived with \sax, together with past
280: measurements (as from Table~1 in Gotthelf et al.\ 2000),
281: allows an analysis of the
282: rotation history of \psr\ in the past decade. All together, there are
283: now 6 different measurements, of which ours is that with the smallest
284: uncertainty (just 12\% of the average of all the others).
285: In the case of a smooth evolution represented
286: with a quadratic function, of the form
287: 
288: $$\nu(t)=\nu_0+\dot\nu_0(t-t_0)+{1\over2}\ddot\nu_0(t-t_0)^2,$$
289: 
290: \noindent
291: a reasonable representation of the data can be achieved
292: with the best fit parameters reported in the first column of
293: Table~2 ($\chi^2=3.52$, for 3
294: d.o.f.; residuals are plotted in Fig.~3.b).
295: The best fit second derivative of the frequency $\ddot\nu_0$ is
296: significantly different from 0.
297: In fact a linear fit ($\ddot\nu_0=0$) can be safely ruled out, as shown
298: by the
299: extremely high $\chi^2$ value ($\sim\!534$, for 4 d.o.f.).
300: We also note that the residuals of a linear fit
301: (Fig.~3.a) are too large to be accounted for by timing noise,
302: even assuming for \psr\ a particularly high activity parameter (see, e.g.,
303: Arzoumanian et al. 1994).
304: 
305: 
306: Under the hypothesis that no glitch has occurred over the time
307: spanned by the period measurements, the available data can be
308: used to determine the braking index $n$, defined by the   law
309: $\dot\nu\propto-\nu^n$ and evaluated, in terms of the frequency and its first
310: two derivatives, as $n=\nu\ddot\nu/\dot\nu^2$.
311: The values of the quadratic fit  correspond to
312: a braking index $n=1.86\pm0.08$,
313: significantly lower than the
314: canonical value ($n=3$) for magnetic dipole losses. For a braking law
315: $\dot\nu\propto-\nu^n$, the pulsar age is:
316: 
317: $$(1)~~~~~~~~~~~~\tau={P\over(n-1)\dot P}\left[1-\left(P_0\over P\right)^{n-1}\right].$$
318: 
319: Assuming an originally fast spinning pulsar ($P_0\ll P$), we derive for \psr\
320: an age $\tau=1675\pm157$~yr.
321: The quoted accuracy on $\tau$ relies on the assumption that irregularities
322: in the spin evolution are negligible in the range of epochs spanned by the
323: observations. In particular, if found that one or more glitches have occurred in
324: between period measurements, the above estimates should be revised.
325: Although population studies indicate small initial periods for radio pulsars
326: (Bhattacharya et al. 1992,  Lorimer et al. 1993) and a value of
327: $P_0\sim17-19$~ms is generally accepted for the Crab pulsar,
328: there is now considerable evidence that fairly long
329: initial spin periods may be common
330: %.  E.g. see Kaspi et al. 2001 ApJ 560, 371 confirming the Torii et al result,
331: % and see Murray et al. 2002 ApJ 568, 226 on a similar result for 3C58
332: % some evidence has
333: % been reported for cases of relatively long initial periods
334: (see, e.g.,  Torii et al. 1999, Kaspi et al. 2001, Murray et al. 2002).
335: We therefore  show in Fig.~4 (solid line), how the age derived from
336: equation (1) depends on the assumed value of $P_0$.
337: 
338: The data can also be interpreted by considering the possibility that glitches
339: occurred between the observations.
340: %Alternative interpretations of the data may in fact include the presence of glitches.
341: For instance, let us first consider the case of a
342: glitch between the first (October 1993) and the second (March 1999) measurements
343: by comparing the October 1993 value with the extrapolation of a fit to the
344: subsequent measurements.
345: A quadratic fit to the   last five points
346: gives a poorly constrained  $\ddot\nu_0$  (corresponding to
347: $n=0.84\pm0.68$; see column (2) in Table~2).
348: The estimated magnitude and even the sign of the   glitch actually
349: depend on the   value of the braking index.
350: The best fit  ($n=0.84$)    corresponds to   a negative jump in
351: frequency, a behavior opposite to what seen in all glitching radio pulsars
352: (Lyne et al.\ 2000). A conventional sign is obtained only for $n>1.89$,
353: which being at only $\sim1.5\sigma$ from the best fit, has a non negligible probability.
354: 
355: Another possibility is the occurrence of a glitch at a later time.
356: In fact the residuals in Fig.~3.b
357: show a pattern suggesting  a glitch
358: in an epoch between the third and fourth data point. We have then chosen
359: 51500.0~MJD as a possible epoch of this glitch: any different choice of the
360: epoch, within the time interval between the third and fourth data point,
361: will have the only effect of slightly changing the best parameters for the
362: glitch, therefore not changing any of the conclusions that we shall draw below.
363: 
364: For different values of the braking index, we have fitted separately the
365: former and latter set of three data points. We find that for values of $n$
366: larger than about 2.5 we would again come out with an unlikely ``anti-glitch''.
367: On the other hand, for $n$ smaller than about 1.8 the
368: (negative) $\dot\nu$ should increase, and also this behavior is in contrast
369: with what seen in radio pulsars (Lyne et al.\ 2000). In the allowed range
370: for $n$ (1.8,2.5), the $\chi^2$ value is monotonically decreasing for
371: increasing $n$, but the trend is so shallow that by itself it cannot be
372: taken as a valid argument for preferring higher values of $n$. The parameters
373: for the two limit cases are shown in columns (3) and (4) of Table~2,
374: and the corresponding pulsar ages by the dashed lines of Fig.~4.
375: 
376: If a glitch at about MJD 51500 actually occurred, how does it compare with
377: known glitches in other pulsars? Fig.~5 shows the positions, in the parameter
378: plane $(\Delta\nu/\nu)$--$(\Delta\dot\nu/\dot\nu)$ of all glitches known in
379: radio pulsars  for which both $\Delta\nu$ and
380: $\Delta\dot\nu$ have been measured (as from Lyne et al.\ 2000).
381: Superposed is the locus of the points
382: representing the MJD 51500 glitch, for different values of $n$. It is
383: apparent that the required glitch has rather normal parameters, with both
384: $(\Delta\nu/\nu)$ and $(\Delta\dot\nu/\dot\nu)$ somehow smaller than the
385: average.
386: 
387: 
388: 
389: 
390: \section{Conclusions}
391: 
392: We have presented a new accurate measurement of the spin frequency of the
393: young pulsar \psr\ associated to the \kes\ supernova remnant.
394: Combined with previous data, and making some hypothesis on the glitch history of
395: \psr\, our new frequency measurement can be interpreted in different ways.
396: 
397: A simple possibility, assuming no glitches
398: in the years 1993--2001, is in terms of a timing solution with
399: $\ddot{\nu}=(2.77\pm0.12)\times10^{-21}$~Hz~s$^{-2}$.
400: This    results in a braking index $n=1.86\pm0.08$ and
401: an estimate for the age of \psr\ more in
402: accord with that of the associated supernova remnant (unless
403: the pulsar initial period was close to the current one).
404: Other likely possibilities involve a single glitch with reasonable parameters
405: and lead to a wide range of acceptable ages.
406: If the glitch occurred near MJD 51500 the resulting braking index  $n$ is
407: in the range 1.8--2.5,
408: while for a glitch between October 1993 and  March 1999 we can only get a lower limit
409: of $n>1.89$.
410: 
411: Braking indices have so far been measured only for five pulsars.
412: Except for PSR J1119--67 which has $n=2.91\pm0.05$ (Camilo et al.\ 2000),
413: all of them  have $n$ significantly smaller than 3, similar to
414: the values suggested by our analysis for \psr. Such  values are
415: $2.51\pm0.01$ (Crab pulsar; Lyne, Pritchard \& Graham-Smith 1993),
416: $2.837\pm0.001$ (PSR B1509--58; Kaspi et al. 1994),
417: $1.4\pm0.2$ (Vela pulsar; Lyne et al. 1996), and
418: $1.81\pm0.07$ (PSR B0540--69; Zhang et al. 2001).
419: 
420: 
421: The pulsar with the highest glitch rate (about one per year) in the
422: sample studied by Lyne et al. (2000), is  PSR B1737--30, which is also the one with
423: the highest magnetic field ($1.7\times10^{13}$~G).
424: We cannot exclude that \psr\ had  more than one glitch in the eight years spanned by the
425: observations, but in this case no information on $n$ can
426: be derived from our data.
427: Interestingly, the glitch parameters we inferred for \psr\ in Fig.~5
428: are similar to those observed in PSR B1737--30.
429: A finer monitoring in the future will establish whether \psr\ is indeed
430: similar to PSR B1737--30.
431: 
432: 
433: As already mentioned by Gotthelf et al.\ (2000), the
434: $P$ and $\pdot$ values of \psr\ are very similar to those of the radio
435: pulsar PSR~J1119--6127
436: ($P=407.6$~ms, $\pdot=4\times10^{-12}$~s~s$^{-1}$, Camilo et al.\ 2000).
437: A remarkable difference between these two young and energetic pulsars
438: ($\dot E_{\rm rot} \gtrsim2\times10^{36}$~erg~s$^{-1}$) is the lack of a bright
439: pulsar wind nebula around PSR J1119--6127 (Crawford et al.\ 2001).
440: These authors suggested that high magnetic field pulsars produce
441: radio nebulae that fade rapidly, an interpretation apparently contradicted by
442: the brightness of the \kes\ core.
443: If the value $n=1.86$ is confirmed for \psr, the difference in the
444: brightness of the wind nebulae of these two pulsars might be related to their
445: different braking index values
446: ($n=2.91\pm0.05$  for PSR J1119--6127; Camilo et al.\ (2000)).
447: While spin-down
448: by magnetic dipole radiation yields $n=3$, a value $n=2$ is
449: expected if  the pulsar braking is driven by the presence of a  strong relativistic
450: wind, which could also have an effect on the emission from a  surrounding nebula.
451: 
452: 
453: 
454: \acknowledgments
455: We thank C. Paizis for useful discussions.
456: This work has been partly supported by the Italian Ministry for University and
457: Research (MIUR) under Grant Cofin2001--02--10.
458: 
459: 
460: 
461: 
462: \begin{references}
463: \reference{} Arzoumanian Z., Nice D.J., Taylor J.H. \& Thorsett S.E. 1994, \apj, 422, 671
464: \reference{} Becker R.H. \& Helfand D.J. 1984, \apj, 283, L93
465: \reference{} Bhattacharya D., Wijers R.A.M.J., Hartman J.W. \& Verbunt F. 1992, A\&A, 254, 198
466: \reference{} Blanton E.L. \& Helfand D.J. 1996, \apj, 470, 961
467: \reference{} Boella G., Chiappetti L., Conti G. et al. 1997, A\&AS, 122, 327
468: \reference{} Camilo F. et al.\ 2000, \apj, 541, 367
469: \reference{} Collins B.F., Gotthef E.V. \& Helfand D.J., 2002,
470: in ''Neutron Stars in Supernova Remnant'', ASP Conf. Proceedings, P.O.Slane \& B.M.Gaensler eds.,
471: in press, astro-ph/0112037
472: \reference{} Crawford F. et al.\ 2001, \apj, 554, 152
473: \reference{} Gotthelf E.V., Vasisht G., Boylan-Kolchin M. \& Torii K. 2000, \apj, 542, L37
474: \reference{} Kaspi V.M., Manchester R.N., Siegman B., Johnston S. \& Lyne A.G. 1994, \apj, 422, L83
475: \reference{} Kaspi V.M., Manchester R.N., Johnston S., Lyne A.G. \& D'Amico N. 1996, \aj, 111, 2028
476: \reference{} Kaspi V.M., Roberts M.E., Vasisht G., Gotthelf E.V., Pivovaroff M. \& Kawai N.   2001, \apj, 560, 371
477: \reference{} Leahy D.A. 1987, A\&A, 180, 275
478: \reference{} Lorimer D.R., Bailes M., Dewey R.J. \& Harrison P.A. 1993, \mnras, 263, 403
479: \reference{} Lyne A.G., Pritchard R.S. \& Graham-Smith F. 1993, \mnras, 265, 1003
480: \reference{} Lyne A.G., Pritchard R.S., Graham-Smith F. \& Camilo F. 1996, \nat, 381, 497
481: \reference{} Lyne A.G., Shemar S.L. \& Graham-Smith F. 2000, \mnras, 315, 534
482: %\reference{} Lyne A.G. \& Graham-Smith F. 1990, Pulsar Astronomy, Cambridge Astrophysics Series
483: \reference{} Morrison D. \& McCammon D. 1983, \apj, 270, 119
484: \reference{} Murray S. et al. 2002, \apj, 568, 226
485: \reference{} Parmar A.N., Martin D.D.E., Bavdaz M. et al. 1997, A\&AS, 122, 309
486: \reference{} Torii K. et al.\ 1999, \apj, 523, L72
487: \reference{} Zhang W., Marshall F.E., Gotthelf E.V., Middleditch J. \& Wang Q.D. 2001, \apj, 554, L117
488: 
489: 
490: %
491: \end{references}
492: 
493: 
494: 
495: 
496: 
497: 
498: 
499: \begin{deluxetable}{cccccc}
500: %\tablewidth{57pc}
501: \tablecaption{Log of \sax\ observations}
502: \tablehead{
503: \colhead{Observation}               &
504: \colhead{Start date} &
505: \colhead{MJD$^a$} &
506: \colhead{Exposure time} &
507: %\colhead{off-axis angle} &
508: \colhead{Period$^b$} \\
509:  (1) & (2) & (3) & (4) & (5)
510: }
511: \startdata
512:  A & March 14, 2001   & 51983.27098   &  83716~s  & 324.81411(12)~ms   \nl
513:  B & March 29, 2001   & 51998.86160   &  105327~s & 324.82371(5)~ms   \nl
514: \hline
515: \enddata
516: 
517: \vspace{0.5cm}
518: 
519: $^a$ Modified Julian Date of the middle of the observation
520: 
521: $^b$ values referred to the Epoch of Column (3) and for fixed
522:      $\pdot=7.097\times10^{-12}$~s~s$^{-1}$ (as from Gotthelf et al.\ 2000)
523: 
524: \end{deluxetable}
525: 
526: 
527: 
528: 
529: %% NEW TABLE 2_______________________
530: 
531:  \begin{deluxetable}{ccccc}
532:  \tablecaption{Derived timing parameters for \psr\ (1$\sigma$ uncertainties)}
533:  \tablehead{
534:   \colhead{ }               &
535:  \colhead{all the points }               &
536:  \colhead{last five points }               &
537:  \colhead{all the points$^a$}               &
538:  \colhead{all the points$^a$} \\
539:  & (1) & (2) & (3) & (4)
540:  }
541:  \startdata
542: %    %\hline
543: %  %       & Quadratic fit &  Fits with n=2.48      &   Fits with n=1.82 \nl
544: %  %       &   (all data)  &  (before/after $t_0$)    & (before/after $t_0$) \nl
545: %  %\hline
546: %
547: $t_o$ (MJD)        & 50000.0 & 50000.0 & 51500.0 & 51500.0 \nl
548:  & & & & \nl
549:  $\nu_0$  (Hz)         & 3.090255(1)   & 3.09024(1)      & 3.0814965(7)   &     3.0814947(7)  \nl
550:                        & &                                &   3.0814965(5)   &    3.0814963(5)  \nl
551:    &   & &  &    \nl
552: $\dot \nu_0$ ($10^{-11}$~Hz~s$^{-1}$)&--6.7768(14)&--6.754(15)&--6.7319(11)  &   --6.7422(11)  \nl
553:                                         & &       &  --6.7448(12)     &   --6.7422(12)  \nl
554:   &  & &   &      \nl
555: $\ddot \nu_0$ ($10^{-21}$~Hz~s$^{-2}$) & $2.77\pm0.12$ & $1.2\pm1.0$  &  [3.647]$^b$       &      [2.685]  \nl
556:                                                                  & &   & [3.661]      &       [2.685]  \nl
557:   &   & &      &      \nl
558:  $\chi^2$/d.o.f.                & 3.52/3  & 1.18/2 &    0.029/1            &  0.086/1  \nl
559:                         & &                & 0.240/1             & 0.508/1  \nl
560:  Total:                 & &             &    0.269/2           &  0.594/2  \nl
561:    &     & &    &     \nl
562:  $n$          & $1.86\pm0.08$  & $0.84\pm0.68$ &     2.48$^c$        &  1.82$^c$   \nl
563:    &        &   & &   \nl
564: $\tau$ (yr)         & $1675\pm157$ &                    & [980.1]       &    [1766.2]  \nl
565:                         & &              &    [978.2]        &   [1766.2]  \nl
566:    &    &    & &   \nl
567: % $\nu_0$ at  MJD 52500.0      & &          &  3.0756826      & 3.0756810  \nl
568: % %\hline
569:  \enddata
570: 
571:   \vspace{0.5cm}
572: 
573:  $^a$  The upper and lower values in each line refer
574:  respectively to the fit to data points before and after MJD 51500.
575: 
576:  $^b$ All the values within square brackets strongly depend on the assumed $n$ value.
577: 
578:  $^c$ Assumed  value
579: 
580:  \end{deluxetable}
581: 
582: 
583: 
584: 
585: 
586: \clearpage
587: 
588: 
589: 
590: 
591: 
592: \figcaption{Folded light curve of \psr\ in three energy ranges,
593: obtained with the MECS instrument in the observation of March 29$^{\rm th}$,
594: 2001.
595: \label{fig1}}
596: 
597: 
598: 
599: \figcaption{Best fit spectrum of \psr\ (upper panel) and residuals (lower panel).
600: The  points correspond to the LECS (0.5-9 keV) and MECS (1.8-10 keV) data.
601: The lines in the upper panel are the best fit model folded through the
602: instrumental response.
603: \label{fig2}}
604: 
605: 
606: 
607: 
608: \figcaption{Residuals of the fits to the spin frequency as a function of
609: time for the cases (a) of a linear   and (b) of a quadratic relation.
610: %The fit parameters are given in Table~2.
611: Note the different scales of the vertical axis.
612: The vertical line in plot (b) indicates the possible location of a glitch, that
613: may account for the pattern of the residuals (see text).
614: \label{fig3}}
615: 
616: 
617: \figcaption{Age of \psr\ as a function of its initial spin period. The solid line
618: refers to $n=1.86$, the dotted line to  $n=3$,
619: and the dashed lines indicate the range allowed by values
620: of $n$ in the interval 1.82 -- 2.48.
621: \label{fig4}}
622: 
623: 
624: 
625: 
626: \figcaption{Parameters of the glitch fitted to the \psr\ data compared with
627: the values of $(\Delta\nu/\nu)$ and $(\Delta\dot\nu/\dot\nu)$ of glitching pulsars
628: (Lyne et al.\ 2000).
629: The line indicates the values obtained for different
630: assumptions on the braking index value; the empty dots represent all braking
631: indices from 1.85 to 2.45, separated by 0.05.
632: The circled dots represent the glitches of the highest field radio pulsar in the
633: sample (PSR~1737--30, with $B=1.7\times10^{13}$~G): they have parameters
634: similar to those expected for the possible MJD~51500 glitch in \psr.
635: \label{fig5}}
636: 
637: 
638: 
639: %\clearpage
640: 
641: 
642: \clearpage
643: \epsscale{.7}
644: \plotone{f1.eps}
645: 
646: 
647: \clearpage
648: \epsscale{.8}
649: \plotone{f2.eps}
650: 
651: \clearpage
652: \epsscale{.8}
653: \plotone{f3.eps}
654: 
655: \clearpage
656: \epsscale{.8}
657: \plotone{f4.eps}
658: 
659: \clearpage
660: \epsscale{.8}
661: \plotone{f5.eps}
662: 
663: 
664: \end{document}
665: