1: %\documentstyle[draft,12pt,psfig]{nature}
2: %\documentstyle[psfig]{nature}
3: \documentstyle[psfig]{nature_local}
4:
5: \title{Nuclear-powered millisecond pulsars and the\\
6: maximum spin frequency of neutron stars}
7:
8: \author{Deepto~Chakrabarty\affiliation{Department of Physics and Center
9: for Space Research, Massachusetts Institute of Technology,
10: Cambridge, Massachusetts 02139, USA}\affiliation{Kavli
11: Institute for Theoretical Physics, University of California,
12: Santa Barbara, California 93106, USA},
13: Edward~H.~Morgan$^*$,
14: Michael~P.~Muno$^*$,\hfill\break
15: Duncan~K.~Galloway$^*$,
16: Rudy~Wijnands\affiliation{School of Physics and Astronomy, University
17: of St. Andrews, North Haugh, St. Andrews, Fife KY16~9SS, UK},
18: Michiel~van~der~Klis\affiliation{Astronomical Institute ``Anton
19: Pannekoek'' and Center for High-Energy Astrophysics, University
20: of Amsterdam, Kruislaan~403, 1098~SJ Amsterdam, The Netherlands},
21: \hfill\break
22: \& Craig~B.~Markwardt\affiliation{Department of Astronomy, University
23: of Maryland, College Park, Maryland 20742,
24: USA}\affiliation{Laboratory for High Energy Astrophysics,
25: NASA Goddard Space Flight Center, Greenbelt, Maryland 20771, USA}
26: }
27:
28: \headertitle{}
29: \mainauthor{}
30:
31: \summary{\noindent
32: Millisecond pulsars are neutron stars that are thought to have been spun-up
33: by mass accretion from a stellar companion\cite{acr+82}. It is
34: unknown whether there is a natural brake for this process, or if it
35: continues until the centrifugal breakup limit is reached at
36: submillisecond periods. Many neutron stars that are accreting mass
37: from a companion star exhibit thermonuclear X-ray bursts that last
38: tens of seconds, caused by unstable nuclear burning on their
39: surfaces\cite{sb03}. Millisecond-period brightness oscillations
40: during bursts from ten neutron stars (as distinct from other rapid
41: X-ray variability that is also observed\cite{van00,wvh+03}) are
42: thought to measure the stellar spin\cite{sb03,szs+96}, but direct
43: proof of a rotational origin has been lacking. Here, we report the
44: detection of burst oscillations at the known spin frequency of an
45: accreting millisecond pulsar, and we show that these oscillations always
46: have the same rotational phase. This firmly establishes burst
47: oscillations as nuclear-powered pulsations tracing the spin of
48: accreting neutron stars, corroborating earlier
49: evidence\cite{szs+96,sm02}. The distribution of spin frequencies
50: of the 11 nuclear-powered pulsars cuts off well below the breakup
51: frequency for most neutron star models, supporting theoretical
52: predictions that gravitational radiation losses can limit accretion
53: torques in spinning up millisecond pulsars\cite{wag84,bil98,aks99}.}
54: \dates{19 March 2003}{21 May 2003}
55:
56: \begin{document}
57: \maketitle
58:
59: The millisecond oscillations observed during X-ray bursts are not
60: perfectly coherent, but usually drift in frequency by several hertz
61: over the course of a burst, generally reaching an asymptotic maximium
62: frequency that is repeatable in a given neutron star\cite{sb03}. This
63: frequency drift has been interepreted as arising from angular momentum
64: conservation in a decoupled surface burning layer that expands and
65: contracts during the burst, so that the asymptotic frequency is the
66: stellar spin frequency\cite{sjg+97,cb00}. A puzzle in this picture is
67: why the oscillation persists late in the burst, well after the nuclear
68: burning has spread over the entire star. Also, in most of these
69: neutron stars, unexplained pairs of kilohertz quasi-periodic
70: oscillations (kHz QPOs) are also observed in the non-burst X-ray
71: emission, with the QPO separation frequency approximately equal to
72: either the burst oscillation frequency or half this value\cite{van00},
73: posing a further puzzle (see ref.~\pcite{wvh+03} for new insight into
74: this problem).
75:
76: We have observed the transient X-ray source
77: SAX~J1808.4$-$3658, which has been detected in four outbursts since
78: its discovery (September 1996, April 1998, January 2000, and October
79: 2002), each lasting several weeks. It was previously established
80: that the source is a weakly magnetized ($<10^{10}$~G), rapidly
81: rotating (401~Hz) accreting neutron star in a 2~hr low-mass X-ray
82: binary\cite{ihm+98,wv98,cm98}. We observed it during its 2002 X-ray
83: outburst for about 700,000 seconds between 15 October and 26 November
84: using the Proportional Counter Array (PCA) on the Rossi X-Ray Timing
85: Explorer (RXTE). As in the two previous outbursts observed by RXTE,
86: persistent 401~Hz accretion-powered X-ray pulsations\cite{wv98}
87: modulated by the 2~hr binary orbit\cite{cm98} were detected throughout
88: our observations, with a fractional r.m.s. amplitude of 3--5\%. The
89: pulsar spin frequency derived from these data was approximately
90: 400.97521~Hz at the start of the outburst, with a mean spin-down rate
91: of about $2\times 10^{-13}$ Hz~s$^{-1}$; a detailed discussion of the
92: pulsar's spin evolution will be presented elsewhere. Four
93: thermonuclear X-ray bursts were also detected on 15, 17, 18, and 19
94: October; these are among the brightest X-ray bursts ever observed by
95: RXTE from any neutron star. Previous analysis of observations from the 1996
96: outburst of SAX J1808.4$-$3658 with the BeppoSAX satellite yielded a
97: marginal detection of a 400$\pm$2~Hz oscillation during a bright X-ray
98: burst\cite{ick+01}.
99:
100: \begin{figure}[t]
101: \centerline{\psfig{file=fig1.ps,width=2.1in}}
102: \caption{Dynamic power spectra of millisecond oscillations in an X-ray
103: burst on 18 October 2002 from SAX J1808.4$-$3658. (a) We analyzed the
104: 2--60~keV PCA data binned at 122~$\mu$s resolution with no energy
105: resolution, with the bin times corrected to the binary
106: center-of-mass frame. We computed overlapping, oversampled Fourier
107: power spectra of 2~s duration spaced at 0.25~s intervals. The
108: contours show Fourier power levels as a function of frequency and
109: time, and the solid histogram shows the X-ray intensity of the burst. The
110: horizontal dotted line shows the (known) pulsar spin frequency. A
111: rapidly drifting oscillation is detected during the burst rise, and
112: a stationary oscillation at nearly the spin frequency is
113: detected in the burst tail. The $n$-th contour level has a
114: single-trial probability of $0.02^n$ of occurring due to noise
115: fluctuations. (b) A close-up view of the burst rise, computed using
116: power spectra of 0.25~s duration at 0.03125~s intervals. The smooth
117: frequency drift is more clearly apparent here.
118: }
119: \end{figure}
120: Strong millisecond oscillations around 401~Hz were clearly detected in
121: all four X-ray bursts in the October 2002 data, with a fractional
122: r.m.s. amplitude of 3--5\% and very similar characteristics in each burst
123: (Fig. 1). First, a rapidly drifting oscillation was detected during
124: the burst rise in the 397--403~Hz range. Second, no oscillations were
125: detected ($<$1.5\% amplitude) during the peak luminosity phase of the
126: burst, when the photosphere is driven away from the star by radiation
127: pressure; this behavior is typical of other burst oscillation sources
128: as well\cite{mcg+02}. Finally, a strong oscillation reappeared during
129: the cooling phase of the burst at a constant frequency nearly equal to
130: the pulsar spin frequency, but exceeding it by an average of $6\pm
131: 1$~mHz in the three best-sampled bursts. The oscillation amplitudes
132: are comparable to those of the persistent (non-burst)
133: accretion-powered pulsations even though the burst emission is
134: considerably brighter than the persistent emission, indicating that
135: the burst flux itself must be pulsed. The burst oscillations are
136: sinusoidal in shape, with a 2$\sigma$ upper limit of $<$0.5\% on the
137: fractional r.m.s. amplitude of an 802~Hz harmonic, compared to a
138: measured mean harmonic amplitude of 0.4\% for the persistent
139: pulsations.
140:
141: The observed frequency drift demonstrates that this is a similar
142: phenomenon as the burst oscillations observed in ten other neutron stars,
143: although the oscillations in this neutron star have some unusual traits.
144: The frequency drift is among the largest observed in any
145: neutron star\cite{wsf01,gcm+01} ($\Delta\nu\approx$5~Hz,
146: $\Delta\nu/\nu\approx$1\%). Also, the drift time scale is an order of
147: magnitude faster than in the other neutron stars\cite{mcg+02}, and the
148: maximum oscillation frequency is reached during the burst rise,
149: inconsistent with angular momentum conservation in a cooling,
150: contracting shell. In fact, the oscillation overshoots the spin
151: frequency during the burst rise. The rapid frequency drift may be an
152: indication that SAX J1808.4$-$3658 has a stronger magnetic field than
153: the other neutron stars, since a sufficiently strong field will suppress
154: rotational shearing in the burning layer\cite{cb00} and may act as a
155: restoring force.
156:
157: This magnetic field argument is particularly appealing given that SAX
158: J1808.4$-$3658 is the only burst oscillation source, and one of only 4
159: neutron stars in low-mass X-ray binaries out of over 70, that shows persistent
160: pulsations in its non-burst emission\cite{mss+02,gcm+02,mss03}. The
161: absence of persistent pulsations in most of these neutron stars suggests
162: that they lack a sufficiently strong field for the accretion flow to be
163: magnetically channeled. Indeed, it has been proposed that the absence
164: of persistent millisecond pulsations from most neutron stars in low-mass X-ray
165: binaries is due to diamagnetic screening of the neutron star magnetic field by
166: freshly accreted material for the typical range of mass accretion
167: rates\cite{czb01} ($\dot M\geq 10^{-10} M_\odot$~yr$^{-1}$, where
168: $M_\odot$ is the solar mass). In this context, we note that all four
169: persistent millisecond X-ray pulsars lie at the low end ($\dot M\sim
170: 10^{-11} M_\odot$~yr$^{-1}$) of the mean $\dot M$ distribution for
171: low-mass X-ray binaries. If this explanation is correct, then most
172: burst oscillation sources are unlikely to show persistent pulsations.
173:
174: \begin{figure}[t]
175: \centerline{\psfig{file=fig2.ps,width=3in}}
176: \caption{Relative phase of the burst tail oscillations in SAX
177: J1808.4$-$3658 and the persistent accretion-powered pulsations just
178: prior to the burst, for all four bursts. The histograms show the
179: profile of the persistent pulsations, and the dashed lines indicate
180: the rotational phase of the burst tail oscillations. The
181: oscillations all have nearly the same rotational phase and slightly
182: lead the persistent pulsations.}
183: \end{figure}
184: We can measure the rotational phase of the oscillations in the burst
185: tails. For each burst, we determined a rotational ephemeris using the
186: persistent (accretion-powered) pulsations in the data for a few
187: thousand seconds prior to the burst and compared this to the phase
188: (epoch of maximum intensity) of the oscillation in the burst tail.
189: The burst tail oscillations all had the same rotational phase within
190: $\pm$6\% and were roughly phase-aligned with the non-burst pulsations,
191: leading them by an average of 11\% (Fig. 2). These measurements
192: will help resolve the puzzle of why oscillations persist in the burst tail,
193: when the nuclear burning has presumably enveloped the entire star.
194: The phase locking indicates that the oscillations are associated with
195: the stellar surface and suggests that the emitting ``hot spot'' has a
196: nearly fixed orientation with respect to the pulsar's magnetic axis.
197: The 11\% offset is consistent with the phase drift accumulated over the
198: duration of the burst tails due to the slight ($<$0.002\%) frequency
199: difference of the burst oscillations and the non-burst pulsations, but
200: this requires an initial phase alignment in the burst tail followed by
201: slight motion of the hot spot\cite{slu02}.
202:
203: Our observations clearly show that brightness oscillations in the
204: tails of thermonuclear X-ray bursts are a nearly exact tracer of
205: stellar spin, establishing these neutron stars as nuclear-powered
206: pulsars. (Other recent observations show that the kHz QPO frequency
207: separation in this neutron star is half the spin frequency\cite{wvh+03},
208: verifying that the burst oscillation traces the spin and not a harmonic.)
209: This provides a powerful tool for studying the most rapidly
210: rotating neutron stars. It is believed that accretion from a binary companion
211: is required in order to spin up old neutron stars to millisecond
212: periods\cite{acr+82}. For sustained accretion in the absence of
213: another angular momentum sink, this process is limited only by the neutron star
214: breakup frequency, up to 3~kHz depending upon the nuclear equation of
215: state\cite{cst94,hlz99}. It has been difficult to determine the
216: actual spin distribution of millisecond pulsars, since there have been
217: strong observational selection effects against detecting very short
218: period millisecond radio pulsars, and persistent millisecond X-ray
219: pulsars are rare. However, RXTE has no significant selection effects
220: against detecting burst oscillations at frequencies well above 1~kHz,
221: so the nuclear-powered pulsars are ideal for probing the fast end of
222: the pulsar spin distribution.
223:
224: The spin frequencies of the 11 nuclear-powered pulsars lie between
225: 270~Hz and 619~Hz, and within that range are marginally consistent
226: with a uniform distribution. However, the absence of
227: spins above 619~Hz is significant. If we make the simple assumption
228: of a uniform spin distribution in the range [$\nu_{\rm low}$,
229: $\nu_{\rm high}$] and consider a sample of $N$ pulsars with spin
230: frequencies $\nu_j$, then a Bayesian analysis yields the probability
231: density for $\nu_{\rm high}$ (within a normalization factor),
232: \begin{equation}
233: p(\nu_{\rm high}) = [\nu_{\rm high}-\min(\nu_j)]^{1-N} - \nu_{\rm
234: high}^{1-N}
235: \end{equation}
236: for $\nu_{\rm high}\geq\max(\nu_j)$, where we have made an {\em a priori}
237: allowance for a $0<\nu<3$~kHz range. For the 11 observed spins, this
238: yields an upper limit of $\nu_{\rm high}<$760~Hz (95\% confidence)
239: on the maximum spin frequency, which is also consistent with the
240: fastest known (641~Hz) millisecond radio pulsar\cite{bkh+82}. This is
241: well below the breakup frequency for nearly all models of rapidly
242: rotating neutron stars (except for those that have an extremely stiff equation
243: of state and contain a pion/kaon condensate in the core, and then only
244: for $M<1.5\,M_\odot$)\cite{cst94b}. Some mechanism clearly acts to
245: halt pulsar spin-up while accretion is still active.
246:
247: Steady magnetic accretion torques should drive an accreting pulsar to
248: an equilibrium spin frequency that depends upon $\dot M$ and the
249: surface dipole magnetic field $B$. Taking the $\dot M$ range observed
250: in the nuclear-powered pulsars and the disk-magnetosphere interaction
251: models relevant for weakly magnetic neutron stars\cite{pc99}, magnetic
252: spin equilibrium can account for the observed spin distribution if all
253: these systems have $B\sim 10^8$~G, consistent with the field inferred
254: in SAX J1808.4$-$3658 (ref. \pcite{pc99}) and in the millisecond radio
255: pulsars\cite{ctk94}. However, fields this strong should be
256: dynamically important for the accretion flow, making it difficult to
257: understand the lack of persistent pulsations in the non-burst emission
258: of all the nuclear-powered pulsars other than SAX J1808.4$-$3658 and
259: instead suggesting a broader range of field strengths. Alternatively,
260: several authors have shown that gravitational radiation can carry away
261: substantial angular momentum from accreting neutron stars, driven
262: either by the excitation of an $r$-mode instability in the neutron
263: star core\cite{wag84,aks99} or by a rotating, accretion-induced
264: crustal quadrupole moment\cite{bil98}. These losses can balance
265: accretion torques for the relevant ranges of $\nu$ and $\dot M$, and
266: the predicted gravitational radiation strengths are near the detection
267: threshold for planned gravitational wave interferometers, especially
268: in the case where an X-ray timing ephemeris is available\cite{bil03}.
269:
270: %\bibliographystyle{nature}
271: %\bibliography{J1808burst}
272: \begin{thebibliography}{10}
273:
274: \bibitem[Alpar {\it et~al.}<1>]{acr+82}
275: Alpar, M.~A., Cheng, A.~F., Ruderman, M.~A. \& Shaham, J. A new class of radio
276: pulsars, {\it Nature} {\bf 300}, 728--730 (1982).
277:
278: \bibitem[Strohmayer \& Bildsten<2>]{sb03}
279: Strohmayer, T. \& Bildsten, L. in {\it Compact Stellar X-Ray Sources} (eds
280: Lewin, W.~H.~G. \& van~der Klis, M.) in press (astro--ph/0301544) (Cambridge
281: U. Press, Cambridge, 2003).
282:
283: \bibitem[van~der Klis<3>]{van00}
284: van~der Klis, M. Millisecond oscillations in X-ray binaries, {\it Ann. Rev.
285: Astr. Astrophys.} {\bf 38}, 717--760 (2000).
286:
287: \bibitem[Wijnands {\it et~al.}<4>]{wvh+03}
288: Wijnands, R. {\it et al.} Kilohertz quasi-periodic X-ray brightness
289: fluctuations from an accreting millisecond pulsar, {\it Nature} {\bf }, in
290: press (2003).
291:
292: \bibitem[Strohmayer {\it et~al.}<5>]{szs+96}
293: Strohmayer, T.~E. {\it et al.} Millisecond variability from an accreting
294: neutron star system, {\it Astrophys. J.} {\bf 469}, L9--L12 (1996).
295:
296: \bibitem[Strohmayer \& Markwardt<6>]{sm02}
297: Strohmayer, T.~E. \& Markwardt, C.~B. Evidence for a millisecond pulsar in 4U
298: 1636$-$53 during a superburst, {\it Astrophys. J.} {\bf 577}, 337--345
299: (2002).
300:
301: \bibitem[Wagoner<7>]{wag84}
302: Wagoner, R.~V. Gravitational radiation from accreting neutron stars, {\it
303: Astrophys. J.} {\bf 278}, 345--348 (1984).
304:
305: \bibitem[Bildsten<8>]{bil98}
306: Bildsten, L. Gravitational radiation and rotation of accreting neutron stars,
307: {\it Astrophys. J.} {\bf 501}, L89--L93 (1998).
308:
309: \bibitem[Andersson, Kokkotas \& Stergioulas<9>]{aks99}
310: Andersson, N., Kokkotas, K.~D. \& Stergioulas, N. On the relevance of the
311: r-mode instability for accreting neutron stars and white dwarfs, {\it
312: Astrophys. J.} {\bf 516}, 307--314 (1999).
313:
314: \bibitem[Strohmayer {\it et~al.}<10>]{sjg+97}
315: Strohmayer, T.~E., Jahoda, K., Giles, A.~B. \& Lee, U. Millisecond pulsations
316: from a low-mass X-ray binary system in the Galactic center region, {\it
317: Astrophys. J.} {\bf 486}, 355--362 (1997).
318:
319: \bibitem[Cumming \& Bildsten<11>]{cb00}
320: Cumming, A. \& Bildsten, L. Rotational evolution during type I X-ray bursts,
321: {\it Astrophys. J.} {\bf 544}, 453--474 (2000).
322:
323: \bibitem[in~'t Zand {\it et~al.}<12>]{ihm+98}
324: in~'t Zand, J.~J. {\it et al.} Discovery of the X-ray transient
325: SAX~J1808.4--3658, a likely low-mass X-ray binary, {\it Astr. Astrophys.}
326: {\bf 331}, L25--L28 (1998).
327:
328: \bibitem[Wijnands \& van~der Klis<13>]{wv98}
329: Wijnands, R. \& van~der Klis, M. A millisecond pulsar in an X-ray binary
330: system, {\it Nature} {\bf 394}, 344--346 (1998).
331:
332: \bibitem[Chakrabarty \& Morgan<14>]{cm98}
333: Chakrabarty, D. \& Morgan, E.~H. The two-hour orbit of a binary millisecond
334: X-ray pulsar, {\it Nature} {\bf 394}, 346--348 (1998).
335:
336: \bibitem[in~'t Zand {\it et~al.}<15>]{ick+01}
337: in~'t Zand, J.~J. {\it et al.} The first outburst of SAX J1808.4--3658
338: revisited, {\it Astr. Astrophys.} {\bf 372}, 916--921 (2001).
339:
340: \bibitem[Muno {\it et~al.}<16>]{mcg+02}
341: Muno, M.~P., Chakrabarty, D., Galloway, D.~K. \& Psaltis, D. The frequency
342: stability of millisecond oscillations in thermonuclear X-ray bursts, {\it
343: Astrophys. J.} {\bf 580}, 1048--1059 (2002).
344:
345: \bibitem[Wijnands, Strohmayer \& Franco<17>]{wsf01}
346: Wijnands, R., Strohmayer, T. \& Franco, L.~M. Discovery of nearly coherent
347: oscillations with a frequency of 567~Hz during type I X-ray bursts of the
348: X-ray transient and eclipsing binary X1658$-$298, {\it Astrophys. J.} {\bf
349: 549}, L71--L75 (2001).
350:
351: \bibitem[Galloway {\it et~al.}<18>]{gcm+01}
352: Galloway, D.~K., Chakrabarty, D., Muno, M.~P. \& Savov, P. Discovery of a 270
353: hertz X-ray burst oscillation in the X-ray dipper 4U 1916$-$053, {\it
354: Astrophys. J.} {\bf 549}, L85--L88 (2001).
355:
356: \bibitem[Markwardt {\it et~al.}<19>]{mss+02}
357: Markwardt, C.~B., Swank, J.~H., Strohmayer, T.~E., in~'t Zand, J.~J.~M. \&
358: Marshall, F.~E. Discovery of a second millisecond accreting pulsar: XTE
359: J1751$-$305, {\it Astrophys. J.} {\bf 575}, L21--L24 (2002).
360:
361: \bibitem[Galloway {\it et~al.}<20>]{gcm+02}
362: Galloway, D.~K., Chakrabarty, D., Morgan, E.~H. \& Remillard, R.~A. Discovery
363: of a high-latitude accreting millisecond pulsar in an ultracompact binary,
364: {\it Astrophys. J.} {\bf 576}, L137--L140 (2002).
365:
366: \bibitem[Markwardt, Smith \& Swank<21>]{mss03}
367: Markwardt, C.~B., Smith, E. \& Swank, J.~H. XTE J1807$-$294, {\it IAU Circ.}
368: {\bf }, No. 8080 (2003).
369:
370: \bibitem[Cumming, Zweibel \& Bildsten<22>]{czb01}
371: Cumming, A., Zweibel, E. \& Bildsten, L. Magnetic screening in accreting
372: neutron stars, {\it Astrophys. J.} {\bf 557}, 958--966 (2001).
373:
374: \bibitem[Spitkovsky, Levin \& Ushomirsky<23>]{slu02}
375: Spitkovsky, A., Levin, Y. \& Ushomirsky, G. Propagation of thermonuclear
376: flames on rapidly rotating neutron stars: extreme weather during type I X-ray
377: bursts, {\it Astrophys. J.} {\bf 566}, 1018--1038 (2002).
378:
379: \bibitem[Cook, Shapiro \& Teukolsky<24>]{cst94}
380: Cook, G.~B., Shapiro, S.~L. \& Teukolsky, S.~A. Recycling pulsars to
381: millisecond periods in general relativity, {\it Astrophys. J.} {\bf 421},
382: L117--L120 (1994).
383:
384: \bibitem[Haensel, Lasota \& Zdunik<25>]{hlz99}
385: Haensel, P., Lasota, J.~P. \& Zdunik, J.~L. On the minimum period of uniformly
386: rotating neutron stars, {\it Astr. Astrophys.} {\bf 344}, 151--153 (1999).
387:
388: \bibitem[Backer {\it et~al.}<26>]{bkh+82}
389: Backer, D.~C., Kulkarni, S.~R., Heiles, C.~E., Davis, M.~M. \& Goss, W.~M. A
390: millisecond pulsar, {\it Nature} {\bf 300}, 615--618 (1982).
391:
392: \bibitem[Cook, Shapiro \& Teukolsky<27>]{cst94b}
393: Cook, G.~B., Shapiro, S.~L. \& Teukolsky, S.~A. Rapidly rotating neutron stars
394: in general relativity: realistic equations of state, {\it Astrophys. J.} {\bf
395: 424}, 823--845 (1994).
396:
397: \bibitem[Psaltis \& Chakrabarty<28>]{pc99}
398: Psaltis, D. \& Chakrabarty, D. The disk-magnetosphere interaction in the
399: accretion-powered millisecond pulsar SAX J1808.4$-$3658, {\it Astrophys. J.}
400: {\bf 521}, 332--340 (1999).
401:
402: \bibitem[Camilo, Thorsett \& Kulkarni<29>]{ctk94}
403: Camilo, F., Thorsett, S.~E. \& Kulkarni, S.~R. The magnetic fields, ages, and
404: original spin periods of millisecond pulsars, {\it Astrophys. J.} {\bf 421},
405: L15--L18 (1994).
406:
407: \bibitem[Bildsten<30>]{bil03}
408: Bildsten, L. in {\it Radio Pulsars} (eds Bailes, M., Nice, D.~J. \& Thorsett,
409: S.~E.) in press (astro--ph/0212004) (Astron. Soc. Pacific, San Francisco,
410: 2003).
411:
412: \end{thebibliography}
413:
414:
415: \smallskip
416: \noindent {\small {\bf Acknowledgements.} We thank L. Bildsten,
417: F. Lamb, D. Psaltis, and S. Thorsett for useful discussions. This
418: work was supported in part by NASA and the Alfred P. Sloan Foundation.
419: }
420:
421: \medskip
422: \noindent {\small {\bf Competing interests statement.} The authors
423: declare that they have no competing financial interests.}
424:
425: \medskip
426: \noindent {\small {\bf Correspondence} and requests for materials should be
427: addressed to D.C. (e-mail: deepto@space.mit.edu).}
428:
429: \end{document}
430: