1: %\documentstyle[12pt]{article}

2: %\documentstyle[12pt,aastex,psfig]{article}

3: %\documentclass[12pt,preprint]{aastex}

4: %\documentclass{emulateapj5}

5: %\documentclass[psfig]{aastex}

6: %\documentstyle[preprint,psfig]{aastex}

7: %\documentclass[usenatbib,bm,psfig]{mn2e}

8: %\documentclass[usenatbib,bm,psfig]{mn2e}

9: \documentclass[12pt,preprint,url]{aastex}

10: %\usepackage{emulateapj5}

11: %\usepackage{longtable,lscape}

12: \usepackage{epsfig}

13: \usepackage{graphicx} % include figs

14: \usepackage{amsbsy}

15: \usepackage{amsmath}

16: \newcommand{\be}{\begin{equation}}

17: \newcommand{\ee}{\end{equation}}

18: \newcommand{\ere}{{\bf r}}

19: \newcommand{\vn}{{\bf v}_{n}}

20: \newcommand{\vns}{{\bf v}_{ns}}

21: \newcommand{\vs}{{\bf v}_{s}}

22: \newcommand{\vl}{{\bf v}_{L}}

23: \newcommand{\vv}{{\bf v}}

24: \newcommand{\VV}{{\bf V}}

25: \newcommand{\vse}{{\bf v_{se}}}

26: \newcommand{\vsb}{{\bf v_{s,b}}}

27: \newcommand{\vsi}{\dot{{\bf s}}}

28: \newcommand{\vsl}{{\bf v_{sl}}}

29: \newcommand{\pa}{\partial}

30: \newcommand{\om}{\boldsymbol{\omega}}

31: \newcommand{\ep}{\boldsymbol{\epsilon}}

32: \newcommand{\nna}{\nabla}

33: \newcommand{\si}{\boldsymbol{\sigma}}

34: \newcommand{\tri}{\xi}

35: \newcommand{\Om}{\boldsymbol{\Omega}}

36: \newcommand{\rna}{\rho_{n}}

37: \newcommand{\rsa}{\rho_{s}}

38: \newcommand{\stok}{{\bf u}_{stokes}}

39: \newcommand{\Rey}{{\it Re}}

40: \newcommand{\f}{\frac}

41: \newcommand{\de}{\text{d}}

42: \newcommand{\bs}{\boldsymbol}

43: \newcommand{\D}{\displaystyle}

44: 

45: \def\ltsima{$\; \buildrel < \over \sim \;$}

46: 

47: 

48: \def\lsim{\lower.5ex\hbox{\ltsima}}

49: % Define "greater than or approximately" symbol.

50: \def\gtsima{$\; \buildrel > \over \sim \;$}

51: \def\gsim{\lower.5ex\hbox{\gtsima}}

52: % e.g. $a \simlt b$ gives a~< b

53: % e.g. $a \simgt b$ gives a~> b

54: 

55: \begin{document}

56: \title{Avalanche dynamics of radio pulsar glitches}

57: 

58: %\maketitle

59: 

60: \author{A. Melatos\altaffilmark{1}, C. Peralta\altaffilmark{1,}\altaffilmark{2}

61: and J. S. B. Wyithe\altaffilmark{1} }

62: 

63: \email{a.melatos@unimelb.edu.au}

64: 

65: \altaffiltext{1}{School of Physics, University of Melbourne,

66: Parkville, VIC 3010, Australia}

67: 

68: \altaffiltext{2}{Max-Planck-Institut f\"ur

69: Gravitationsphysik, Albert-Einstein-Institut, Am M\"uhlenberg 1,

70: D-14476 Golm, Germany}

71: 

72: \begin{abstract}

73: \noindent

74: We test statistically the hypothesis that radio pulsar glitches result from

75: an avalanche process, in which angular momentum is transferred erratically

76: from the flywheel-like superfluid in the star to the slowly decelerating, solid crust

77: via spatially connected chains of local, impulsive, threshold-activated events,

78: so that the system fluctuates around a self-organised critical state.

79: Analysis of the glitch population (currently 285 events from 101 pulsars)

80: demonstrates that the size distribution in individual pulsars

81: is consistent with being scale invariant, as expected for an avalanche process.

82: The measured power-law exponents fall in the range $-0.13\leq a \leq 2.4$,

83: with $a\approx 1.2$ for the youngest pulsars.

84: The waiting-time distribution is consistent with being 

85: exponential in seven out of nine pulsars

86: where it can be measured reliably,

87: after adjusting for observational limits on the minimum waiting time,

88: as for a constant-rate Poisson process.

89: PSR J0537$-$6910 and PSR J0835$-$4510 are the exceptions;

90: their waiting-time distributions show evidence of quasiperiodicity.

91: In each object, stationarity requires that the rate $\lambda$ equals

92: $- \epsilon \dot{\nu} / \langle\Delta\nu\rangle$,

93: where $\dot{\nu}$ is the angular acceleration of the crust,

94: $\langle\Delta\nu\rangle$ is the mean glitch size,

95: and $\epsilon\dot{\nu}$ is the relative angular acceleration of the crust and superfluid.

96: Measurements yield $\epsilon \leq 7 \times 10^{-5}$

97: for PSR J0358$+$5413 and $\epsilon \leq 1$ (trivially)

98: for the other eight objects, which have $a < 2$.

99: There is no evidence that $\lambda$ changes monotonically

100: with spin-down age.

101: The rate distribution itself is fitted reasonably well by an exponential

102: for $\lambda \geq 0.25\,{\rm yr^{-1}}$,

103: with $\langle \lambda \rangle = 1.3^{+0.7}_{-0.6}\,{\rm yr^{-1}}$.

104: For $\lambda < 0.25\,{\rm yr^{-1}}$, its exact form is unknown;

105: the exponential overestimates the number of glitching pulsars

106: observed at low $\lambda$, where the limited total observation time

107: exercises a selection bias.

108: In order to reproduce the aggregate waiting-time distribution of the

109: glitch population as a whole,

110: the fraction of pulsars with $\lambda > 0.25\,{\rm yr^{-1}}$

111: must exceed $\sim 70$ per cent.

112: \end{abstract}

113: 

114: %\begin{keywords}

115: \keywords{dense matter --- pulsars: general --- stars: interior --- stars: neutron --- stars: rotation}

116: %\end{keywords}

117: 

118: \section{Introduction 

119:  \label{sec:gli1}}

120: Glitches are tiny, impulsive, randomly timed increases in the spin frequency $\nu$ 

121: of a rotation-powered pulsar,

122: sometimes accompanied by an impulsive change in the frequency derivative $\dot{\nu}$.

123: They are to be distinguished from timing noise,

124: a type of rotational irregularity

125: where pulse arrival times wander continuously,

126: although there is evidence that timing noise is the cumulative result of 

127: frequent microglitches in certain pulsars

128: \citep{cor85,dal95}.

129: 

130: At the time of writing,

131: 285 glitches in total have been detected in 101 objects

132: ($\sim 6\%$ of the known radio pulsar population),

133: the majority in the last four years, facilitated by the

134: Parkes Multibeam Survey,

135: refined multifrequency ephemerides,

136: and better interference rejection algorithms

137: \citep{hob02,kra03,kra05,lew05,jan06}.

138: Efforts to analyse the data statistically have focused on

139: the correlation of glitch activity with age

140: \citep{mck90,she96,ura99,lyn00,wan00}

141: and Reynolds number

142: \citep{peralta06,mel07},

143: the post-glitch relaxation time-scale

144: \citep{wan00,won01},

145: the size distribution

146: \citep{mor93a,mor93b,peralta06},

147: and the correlation between glitch sizes and waiting times

148: \citep{wan00,won01,mid06,peralta06}.

149: \citet{hob02} reviewed the role of observational selection effects.

150: 

151: Most glitching pulsars ($65\%$) have been seen to glitch once,

152: but a minority glitch repeatedly;

153: the current record holder is PSR J1740$-$3015, with 33 glitches.

154: Of those objects which glitch repeatedly,

155: most do so at unpredictable intervals, 

156: but two (PSR J0537$-$6910 and Vela) are quasiperiodic;

157: Vela, in particular, has been likened to a relaxation oscillator

158: \citep{lyn96}.

159: The fractional increase in $\nu$ spans seven decades 

160: ($3\times 10^{-11} \leq \Delta\nu \leq 2\times 10^{-4}$) 

161: across the glitch population 

162: and as many as four decades in a single object

163: (e.g.\ $7\times 10^{-10} \leq \Delta\nu \leq 2\times 10^{-6}$

164: in PSR J1740$-$3015).

165: The spin-down age $\tau_{\rm c}= -\nu/(2\dot{\nu})$

166: of glitching pulsars spans four decades,

167: from $1\times 10^3\,{\rm yr}$ to $3\times 10^7\,{\rm yr}$.

168: In many respects, therefore, the glitch phenomenon is {\em scale invariant}.

169: This striking property invites physical interpretation.

170: 

171: Theories of pulsar glitches have focused mainly on the local microphysics

172: of the superfluid in the stellar interior and its coupling to the solid crust,

173: for example the strength of vortex pinning

174: \citep{and75,jon98}, 

175: the rate of vortex creep

176: \citep{lin96},

177: or the conditions for exciting superfluid turbulence

178: \citep{per05,per06,mel07,and07}.

179: Ultimately, however, the local microphysics must be synthesized with

180: the global, {\em collective} dynamics in order to make full contact 

181: with observational data.

182: (Likewise, a practical model of earthquakes must synthesize

183: the microphysics of rock fracture with the

184: macrodynamics of interacting tectonic plates.)

185: For example, if approximately $10^{16} (\Delta\nu/1\,{\rm Hz})$ 

186: vortices unpin from crustal lattice sites in sympathy during a glitch,

187: they must communicate rapidly across distances much greater than 

188: their separation. How?

189: And why does the number that unpin fluctuate so dramatically

190: (by up to four orders of magnitude)

191: from glitch to glitch in a single pulsar,

192: while always amounting to a small fraction ($\Delta\nu/\nu$) 

193: of the total?

194: 

195: Such collective, scale invariant behavior is a generic feature of 

196: a class of natural and synthetic far-from-equilibrium systems, 

197: called self-organized critical systems, 

198: that are discrete, interaction dominated, and slowly driven, 

199: and that adjust internally via erratic, spatially connected {\em avalanches}

200: of local, impulsive, threshold-activated, relaxation events

201: \citep{jen98}.

202: Such systems fluctuate around a stationary state

203: towards which they evolve spontaneously,

204: in which global driving balances local relaxation on average over the long term.

205: The archetype of a self-organized critical system is the sandpile

206: \citep{bak87}.

207: 

208: In this paper, we study pulsar glitches as an avalanche process, 

209: as first proposed by \citet{mor93a}.

210: After reviewing self-organised criticality in \S\ref{sec:gli2},

211: we define the statistical sample on which our study is based

212: (\S\ref{sec:gli3})

213: and analyze the observed distribution of glitch sizes

214: (\S\ref{sec:gli4})

215: and waiting times

216: (\S\ref{sec:gli5}).

217: Some implications for glitch physics are explored in \S\ref{sec:gli6}.

218: We only include radio pulsars in the sample, to preserve its homogeneity,

219: even though glitches have now been observed in anomalous X-ray pulsars

220: (magnetars) as well

221: \citep{dal03,kas03}.

222: 

223: \section{Avalanche dynamics

224:  \label{sec:gli2}}

225: A system in a self-organised critical state exhibits the following

226: distinguishing features

227: \citep{jen98}.

228: \begin{enumerate}

229: \item

230: It is composed of many discrete, mutually interacting elements,

231: whose motions are dominated by local (e.g.\ nearest neighbor) rather than

232: global (e.g.\ mean field) forces.

233: \footnote{Tectonic plates, or grains of sand in a pile,

234: are terrestrial examples of interacting elements.}

235: \item

236: Each element moves when the local force exceeds a threshold

237: (stick-slip motion).

238: Hence stress accumulates sustainedly at certain random locations

239: while relaxing quickly elsewhere;

240: at any instant, the system houses numerous metastable 

241: stress reservoirs, separated by relaxed zones.

242: \item

243: An external force drives the system slowly, in the sense that

244: elements adjust to local forces rapidly compared to the driver time-scale.

245: Combined with local thresholds, this ensures that the system

246: evolves quasistatically through a history-dependent sequence of metastable states

247: (a huge number of which are available).

248: \item

249: Transitions from one metastable state to the next occur

250: via avalanches:

251: spatially connected chains of local equilibration events,

252: in which one element relaxes and

253: redistributes some local stress to its neighbors,

254: which in turn can exceed their thresholds and relax

255: (knock-on effect).

256: The duration of even the largest avalanches is short compared to

257: the driving time-scale (see previous point).

258: \item

259: Avalanches have no preferred scale: they can involve a few

260: (commonly) or all (rarely) of the elements in the system.

261: Their sizes and lifetimes follow power-law distributions,

262: whose exponents are related.

263: The numerical values of the exponents depend on the spatial dimensionality 

264: of the system,

265: the spatial symmetries of the local forces and redistributive channels,

266: the strength of the local forces 

267: \citep{fie95}, 

268: and the level of conservation 

269: \citep{ola92}.

270: \footnote{In this respect, far-from-equilibrium critical systems

271: differ from equilibrium critical systems 

272: (e.g.\ second-order phase transition in a ferromagnet),

273: whose exponents depend only on the dimensionality of the system

274: and its order parameter(s).}

275: \item

276: Over the long term, the system tends to a critical state,

277: which is stationary on average but not instantaneously.

278: For example, on average, the power input by the external driver equals the 

279: energy per unit time released by avalanches.

280: But there are fluctuations, because, at any instant, 

281: a random amount of energy is stored in metastable local reservoirs.

282: \end{enumerate}

283: 

284: Avalanche dynamics are generically observed in nature

285: when conditions (i)--(iii) are met,

286: and properties (iv)--(vi) emerge irrespective of the detailed microphysics

287: \citep{jen98}.

288: Likewise, in this paper, we remain agnostic about the microphysics

289: of pulsar glitches;

290: the statistical analysis presented below makes no assumptions

291: in this regard.

292: Nevertheless, it is striking that the traditional glitch paradigm ---

293: collective unpinning of quantized superfluid vortices interacting with

294: an inhomogeneous, slowly decelerating crust ---

295: conforms closely with (i)--(vi)

296: \citep{and75,alp96}.

297: So too does an alternative paradigm, based on crust fracture

298: \citep{alp96,mid06},

299: whose terrestrial counterpart (plate tectonics)

300: is renowned as an archetype of self-organized criticality

301: \citep{sor91}.

302: We elucidate the analogy briefly before continuing.

303: 

304: Consider a rectilinear array of quantized vortices, 

305: each carrying circulation $\kappa$,

306: spaced evenly according to Feynman's rule 

307: ($4\pi\nu/\kappa$ vortices per unit area) in the neutron superfluid

308: permeating the inner crust of a neutron star.

309: A small percentage of the vortices are pinned to defects and/or nuclei 

310: at random locations in the crustal lattice, clustered to varying degrees

311: \citep{alp96,won01}

312: As $\nu$ decreases gradually due to electromagnetic spin down,

313: most vortices move apart and the outermost ones are expelled.

314: However, the pinned vortices stay (nearly) fixed, 

315: in metastable reservoirs separated by relaxed zones [see (ii)],

316: creeping slowly between adjacent pinning sites in response to thermal fluctuations

317: \citep{lin93}.

318: The reservoirs are identical to the capacitive elements

319: (vortex traps surrounded by vortex depletion regions)

320: postulated by \citet{alp96} and \citet{won01}.

321: They may be seeded by starquakes,

322: which create large numbers of fresh lattice dislocations with deep

323: pinning potentials,

324: or they may emerge spontaneously in the self-organized critical state,

325: as successive generations of vortex avalanches traverse the crust.

326: As the pinned vortices increasingly lag the regular, unpinned array,

327: a gradient in vortex density is established, and the local Magnus force

328: on a pinned vortex rises.

329: When the pinning threshold is overcome, a pinned vortex unpins and

330: moves abruptly away from the pinning site (stick-slip motion), 

331: disturbing the local superfluid

332: velocity field (and hence the Magnus force) appreciably.

333: Often, this is enough to push neighboring, barely subcritical, pinned vortices 

334: over their thresholds, triggering an avalanche.

335: Most vortices in the avalanche rejoin the regular, unpinned array,

336: and the crust spins up proportionately to compensate.

337: The time-scale for a vortex to adjust locally to the Magnus and pinning forces

338: is much shorter than $\nu/\dot{\nu}$, in keeping with (iii).

339: 

340: Classic laboratory experiments on magnetic flux vortices

341: in a type II superconductor (e.g.\ NbTi) 

342: immersed in a slowly changing magnetic field,

343: an exactly analogous system, clearly exhibit properties (iv)--(vi)

344: \citep{fie95}.

345: Vortices are expelled mostly in a continuous flow (cf.\ steady spin down)

346: and occasionally in avalanches (cf.\ glitches).

347: The distribution of avalanche sizes

348: is measured to be a power law over several decades,

349: whose exponent depends on the strength of the applied magnetic field

350: (which controls the vortex spacing and hence the strength of

351: the vortex-vortex interaction).

352: The temporal fluctuation spectrum scales as an inverse power of

353: frequency at high frequencies.

354: After initial transients die away, the superconductor fluctuates around a 

355: self-organised critical state, called the Bean state,

356: where the Lorentz force acting on each vortex is everywhere equal

357: to the maximum pinning force.

358: 

359: If the pinning sites are sparsely distributed, 

360: so that global (mean-field) forces dominate local forces between 

361: pinned vortex clusters, scale invariance breaks down

362: \citep{jen98}.

363: Avalanches still occur, 

364: but they are distributed narrowly around a characteristic size and lifetime,

365: involving nearly all the vortices instead of small, independent subsets.

366: In this regime, avalanches recur quasiperiodically, not stochastically.

367: Similar behavior is observed when the external driver acts too rapidly,

368: but this situation never arises in pulsars.

369: 

370: Scale invariant avalanche dynamics and self-organised critical states 

371: are observed widely elsewhere, in sandpiles

372: \citep{bak87},

373: earthquakes

374: \citep{sor91},

375: solar flares

376: \citep{lu91,whe00},

377: and bursts from soft-gamma-ray

378: repeaters \citep{gwk00}.

379: The analogy with pulsar glitches has been pointed out by 

380: \citet{mor93a} and \citet{car98}

381: and modeled using a cellular automaton by \citet{mor96}.

382: 

383: \section{Data

384:  \label{sec:gli3}}

385: Table \ref{tab:gli1} lists all 285 glitches discovered 

386: up to the time of writing and known to the authors.

387: It is compiled from published sources

388: \citep{she96,lyn00,wan00,hob02,kra03,jan06,mid06,peralta06};

389: the Australia Telescope National Facility Pulsar Catalogue

390: \citep{man05},

391: which can be accessed on-line at 

392: {\tt http://www.atnf.csiro.au/research/pulsar/psrcat};

393: and unpublished data communicated privately by 

394: M.\ Kramer, D.\ Lewis, and A.\ G.\ Lyne.

395: For each pulsar, the table lists its J2000 coordinates,

396: and the number of glitches detected

397: ($N_{\rm g}$). The earliest and latest epochs observed

398: ($t_{\rm min}$ and $t_{\rm max}$ respectively) are

399: recorded separately in Table \ref{tab:gli1b} for the

400: nine pulsars with $N_{\rm g} > 5$.

401: An asterisk signifies that segmented data spans are not specified

402: in the cited references;

403: in this situation, $t_{\rm min}$ and $t_{\rm max}$ are estimated

404: by eye from spin-down histories graphed in the cited references,

405: where available,

406: or else from the first and last glitches by default.

407: For each glitch, Table \ref{tab:gli1} lists its epoch,

408: the fractional increase in spin frequency $\Delta\nu/\nu$,

409: and one or more bibliographic references.

410: Uncertainties are quoted as a trailing integer in parentheses,

411: corresponding to an absolute number of days for $t$

412: [e.g.\ MJD 51141(248) means MJD $51141\pm 248$]

413: and an uncertainty in the last signifcant digit for $\Delta\nu/\nu$

414: [e.g.\ 0.04(2) means $0.04\pm0.02$].

415: For some newly discovered glitches, the information is incomplete.

416: Epochs and sizes have been measured for 271 and 250 glitches respectively.

417: Other parameters, like the healing fraction and

418: post-glitch relaxation time-scale,

419: are omitted as they are not analyzed in this paper;

420: please consult \citet{peralta06} and references therein for a full catalog.

421: 

422: 

423: 

424: \section{Size distribution

425:  \label{sec:gli4}}

426: 

427: \subsection{Scale invariance

428:  \label{sec:gli4a}}

429: If pulsar glitches are the result of an avalanche process,

430: their size distribution should be scale invariant in any

431: individual pulsar, with probability density function

432: \begin{equation}

433:  p(\Delta\nu/\nu) \propto (\Delta\nu/\nu)^{-a}~.

434:  \label{eq:gli1}

435: \end{equation}

436: The exponent $a$ is set by the dimensionality

437: \footnote{The effective dimension need not equal three.

438: For example, it may equal two in a rectilinear vortex array

439: or faulting crust,

440: where the local forces act in the transverse plane,

441: and three in a turbulent vortex tangle \citep{per05,per06}.}

442: and symmetries of the local forces,

443: which are likely to be universal,

444: and the strength and level of conservation of these forces,

445: which are functions of temperature and therefore not universal

446: (see \S\ref{sec:gli2}).

447: One therefore expects $a$ to differ from pulsar to pulsar.

448: As a corollary, the aggregate size distribution

449: drawn from all pulsars

450: is not expected to be a simple power law of the form (\ref{eq:gli1}).

451: 

452: To test these ideas, we construct the observed cumulative size distributions

453: of the nine known pulsars with $N_{\rm g} > 5$.

454: The selection criterion $N_{\rm g}>5$ is arbitrary;

455: it seeks to limit the impact of random errors while testing as many objects

456: from Table \ref{tab:gli1} as possible.

457: We then compare the data against the theoretical cumulative distribution

458: \begin{equation}

459:  P(\Delta\nu/\nu)

460:  =

461:  \frac{ (\Delta\nu/\nu)^{1-a} - (\Delta\nu/\nu)_{\rm min}^{1-a} }

462:   { (\Delta\nu/\nu)_{\rm max\phantom{i}}^{1-a} - (\Delta\nu/\nu)_{\rm min}^{1-a} }

463:  \label{eq:gli2}

464: \end{equation} 

465: derived from (\ref{eq:gli1}).

466: The theoretical distribution is normalized after restricting it

467: to the domain 

468: $(\Delta\nu/\nu)_{\rm min} \leq \Delta\nu/\nu 

469:  \leq (\Delta\nu/\nu)_{\rm max}$,

470: where 

471: $(\Delta\nu/\nu)_{\rm min}$ and $(\Delta\nu/\nu)_{\rm max}$ 

472: are the smallest and largest glitches observed in that pulsar

473: respectively, quoted in Table \ref{tab:gli1}.

474: There are more sophisticated ways to choose

475: $(\Delta\nu/\nu)_{\rm min}$ and $(\Delta\nu/\nu)_{\rm max}$,

476: which we consider further below, but this is a conservative starting point.

477: 

478: For each object, we choose $a$ to minimize the Kolmogorov-Smirnov (K-S)

479: statistic $D$, i.e.\ the maximum unsigned distance between the curves.

480: The numerical results are recorded in Table \ref{tab:gli2},

481: while the measured and theoretical cumulative distributions are plotted

482: together in Figure \ref{fig:gli1}.

483: (Cumulative distributions are free of binning bias.)

484: The goodness of the fit at the optimal value of $a$ is characterized

485: by $P_{\rm KS}$, defined such that $1-P_{\rm KS}$ equals

486: the probability that the K-S null hypothesis

487: (that the two data sets are drawn from the same underlying distribution) 

488: is false.

489: \footnote{The K-S test is most sensitive to discrepancies near the median.

490: An alternative test, based on Kuiper's statistic, mitigates this bias

491: \citep{pre86}. 

492: It will be worth implementing when more data become available.

493: The K-S test is also inefficient if the underlying probability density function

494: contains a narrow notch, where the probability vanishes.

495: Again, there is insufficient data at hand to look for such a notch;

496: it is difficult to find in a cumulative distribution, 

497: and the probability density function is biased by binning when $N_{\rm g}$ is small.}

498: The 1-$\sigma$ lower and upper bounds $a_-$ and $a_+$ mark the 

499: range of $a$ where the null hypothesis is rejected with less than

500: $68\%$ confidence. Note that the interval $[a_-,a_+]$ is 

501: asymmetric about the optimal $a$ and widest for the best fits.

502: 

503: 

504: The results in Table \ref{tab:gli2} confirm what is apparent by eye

505: from Figure \ref{fig:gli1}: the null hypothesis that

506: the size distribution is described by a power law for all nine pulsars

507: with $N_{\rm g} > 5$ is not ruled out at the 1-$\sigma$ level of confidence.

508: In turn, this is consistent with the avalanche hypothesis.

509: However, in two objects, namely PSR J0537$-$6910 and PSR J0835$-$4510,

510: the agreement is marginal.

511: Interestingly, these two objects are also the only ones discovered so far

512: that are believed to glitch quasiperiodically

513: \citep{lyn96,mid06}.

514: 

515: \subsection{PSR J0537$-$6910 and PSR J0835$-$4510

516:  \label{sec:gli4b}}

517: Quasiperiodicity is a

518: natural feature of avalanche dynamics when mean-field forces 

519: overwhelm local interactions, as described in \S\ref{sec:gli2}.

520: We explore its manifestation in glitch waiting times in \S\ref{sec:gli5}.

521: With respect to glitch sizes, we note that avalanches in the

522: quasiperiodic regime tend to be distributed narrowly around a 

523: characteristic size

524: \citep{jen98}.

525: This can be modeled crudely by adding a term proportional to 

526: $\delta[\Delta\nu/\nu - (\Delta\nu/\nu)_{\rm c}]$

527: to (\ref{eq:gli1}), viz.\

528: \begin{equation}

529:  p(\Delta\nu/\nu)

530:  =

531:  C_{\rm s} (\Delta\nu/\nu)^{-a} + 

532:  C_{\rm p} \delta[\Delta\nu/\nu - (\Delta\nu/\nu)_{\rm c}]~,

533:  \label{eq:gli3a}

534: \end{equation}

535: where $(\Delta\nu/\nu)_{\rm c}$ denotes the characteristic size,

536: and the scale invariant and quasiperiodic components are weighted by

537: the constants $C_{\rm s}$ and $C_{\rm p}$ respectively.

538: Normalization fixes $C_{\rm s}$ in terms of $C_{\rm p}$

539: (or vice versa), with $C_{\rm s} + C_{\rm p} \neq 1$ in general.

540: The associated cumulative size distribution is given by

541: \begin{eqnarray}

542:  P(\Delta\nu/\nu)

543:  & = &

544:  C_{\rm p} H[ \Delta\nu/\nu - (\Delta\nu/\nu)_{\rm c} ]

545:  \nonumber \\

546:  & &

547:  + 

548:  \frac{ (1-C_{\rm p}) [ (\Delta\nu/\nu)^{1-a} - (\Delta\nu/\nu)_{\rm min}^{1-a} ] }

549:   { (\Delta\nu/\nu)_{\rm max\phantom{i}}^{1-a} - (\Delta\nu/\nu)_{\rm min}^{1-a} }~,

550:  \label{eq:gli3b}

551: \end{eqnarray}

552: where $H(\cdot)$ denotes the Heaviside step function.

553: 

554: Parameters determined by fitting (\ref{eq:gli3b}) to the data are recorded

555: in Table \ref{tab:gli3},

556: while the corresponding measured and theoretical cumulative distributions

557: are plotted together in Figure \ref{fig:gli2}.

558: The fits are much improved,

559: with $C_{\rm p}\approx 0.2$ in both objects

560: --- although, to be fair, the delta-distributed component is not 

561: strictly required, at least not 

562: at the 1-$\sigma$ level.

563: Importantly, the delta-distributed component contains

564: only $\sim20\%$ of the glitches, not all of them.

565: This is consistent with the historical interpretation of the pulsar data

566: \citep{lyn95,mar04}.

567: It is also seen in self-organized critical systems like sandpiles, 

568: where large, system-spanning, quasiperiodic avalanches

569: of a characteristic size are interspersed with small, randomly timed

570: avalanches, which are power-law distributed

571: \citep{ros93,jen98}.

572: 

573: 

574: \subsection{Upper and lower cut-offs

575:  \label{sec:gli4c}}

576: Strictly speaking, it is incorrect to normalize $P(\Delta\nu/\nu)$

577: by choosing 

578: $(\Delta\nu/\nu)_{\rm min}$ and $(\Delta\nu/\nu)_{\rm max}$ 

579: to be the smallest and largest glitches observed in a pulsar

580: respectively.

581: A better choice of $(\Delta\nu/\nu)_{\rm min}$ is the actual

582: resolution of the timing experiment, 

583: which varies with object and epoch. 

584: \citet{jan06} simulated detection of a microglitch in a noisy time series

585: and obtained $(\Delta\nu/\nu)_{\rm min}=1\times 10^{-11}$.

586: Usually, this information is not provided explicitly 

587: and must be estimated from the size uncertainties quoted for

588: detected glitches.

589: All the same, the smallest glitch observed is likely to be a

590: reasonable estimate of $\Delta\nu_{\rm min}$,

591: because the occurrence probability increases steeply as $\Delta\nu$ decreases, 

592: according to Table \ref{tab:gli2}.

593: On the other hand, $(\Delta\nu/\nu)_{\rm max}$ is limited by the

594: total observing time.

595: Its true value exceeds the largest glitch observed, but not by much, 

596: because the occurrence probability decreases steeply as $\Delta\nu$ increases.

597: 

598: To quantify these effects, we allow 

599: $(\Delta\nu/\nu)_{\rm min}$ to vary between 0.5 and 1.0 times the

600: smallest glitch size observed,

601: $(\Delta\nu/\nu)_{\rm max}$ to vary between 1.0 and 2.0 times the

602: largest glitch size observed,

603: and fit equation (\ref{eq:gli3b}) again to the data.

604: For every object,

605: $(\Delta\nu/\nu)_{\rm min}$ and $(\Delta\nu/\nu)_{\rm max}$

606: shift only slightly,

607: and $a$ stays within the range $[a_-,a_+]$ in Tables \ref{tab:gli2} and \ref{tab:gli3}.

608: This confirms that the smallest and largest glitches

609: provide reasonable estimates of

610: $(\Delta\nu/\nu)_{\rm min}$ and $(\Delta\nu/\nu)_{\rm max}$.

611: At the 1-$\sigma$ level, the constrained and unconstrained fits

612: are both consistent with the data. 

613: 

614: \subsection{Aggregate distribution

615:  \label{sec:gli4d}}

616: Figure \ref{fig:gli3} displays the cumulative size distribution for

617: the glitch population in aggregate,

618: together with the best power-law fit of the form (\ref{eq:gli2}).

619: The fit is poor.

620: When all 250 glitches with measured sizes are included, 

621: the best fit corresponds to

622: $a=0.96$, $(\Delta\nu/\nu)_{\rm min}=9.5\times 10^{-12}$,

623: $(\Delta\nu/\nu)_{\rm max}=2.0\times 10^{-5}$, 

624: and $P_{\rm KS}=7.1\times 10^{-4}$.

625: When the glitches from PSR J0537$-$6910 and PSR J0835$-$4510 are excluded,

626: the best fit corresponds to

627: $a=0.98$ and $P_{\rm KS}=3.2\times 10^{-4}$,

628: with $(\Delta\nu/\nu)_{\rm min}$ and $(\Delta\nu/\nu)_{\rm max}$

629: as before. 

630: Either way, we can state confidently that

631: the theoretical and observed data are drawn from different

632: underlying distributions.

633: This is not surprising; the results in Figure \ref{fig:gli1}

634: and Table \ref{tab:gli2} demonstrate clearly that the size distribution

635: in individual pulsars is consistent with being scale invariant, but that $a$

636: differs from object to object.

637: Hence the aggregate distribution is expected to be a weighted sum of 

638: power laws, not a pure power law.

639: Accordingly, the size distribution in individual pulsars

640: is a more direct probe of glitch physics than the aggregate distribution

641: \citep{lyn00}.

642: The aggregate distribution can be inverted, in principle,

643: to determine how $a$ is distributed across the pulsar population.

644: We defer this exercise until better historical estimates 

645: of $(\Delta\nu/\nu)_{\rm min}$ and $(\Delta\nu/\nu)_{\rm max}$, 

646: and more data, become available.

647: 

648: \citet{jan06} claimed that the glitches in PSR J1740$-$3015 are

649: drawn from a flat size distribution in $\log (\Delta\nu/\nu)$,

650: i.e. $a=1$,

651: with $P_{\rm KS}=0.902$.

652: This agrees with the results in Table \ref{tab:gli2}.

653: 

654: \citet{lyn00} noted some evidence for an excess of large glitches,

655: which is corroborated to some extent by Figure \ref{fig:gli3}a.

656: However, the excess largely disappears when the quasiperiodic

657: glitchers are excluded, as in Figure \ref{fig:gli3}b.

658: Large glitches do not originate preferentially from any

659: particular class of object.

660: While it is true that the most active objects

661: (e.g.\ PSR J0537$-$6910 and PSR J0835$-$4510)

662: experience relatively large and narrowly distributed glitches,

663: with $\Delta\nu/\nu > 10^{-7}$,

664: other active objects (e.g. PSR J1740$-$3015) experience a mix of small

665: and large events,

666: and there are several objects (e.g.\ PSR J1806$-$4212) which have only

667: glitched once, with $\Delta\nu/\nu > 10^{-5}$ for that single glitch.

668: Furthermore, although PSR J0534$+$2200 is sometimes portrayed as unusual

669: for not experiencing large glitches, its size distribution is actually

670: relatively flat ($a\approx 1.2$).

671: There is every reason to expect that it will experience large glitches 

672: in the future, but these events will be slow in coming,

673: because PSR J0534$+$2200 builds up 

674: differential rotation between the crust and superfluid

675: at a relatively slow rate,

676: as we show in \S\ref{sec:gli5}. 

677: \footnote{

678: \citet{won01} argued that the relative angular acceleration of the crust

679: and superfluid in the Crab, inferred from the activity parameter,

680: is much smaller than expected given the large 

681: $\Delta\dot{\nu}/\dot{\nu}\sim 10^{-4}$

682: observed during glitches.

683: This paradox is resolved if most of the differential rotation

684: is being stored temporarily, in advance of a large glitch in the future.

685: }

686: 

687: \section{Waiting-time distribution

688:  \label{sec:gli5}}

689: 

690: \subsection{Poisson process

691:  \label{sec:gli5a}}

692: If pulsar glitches are the result of an avalanche process,

693: they should be statistically independent events.

694: To understand why, recall that a system in a self-organized 

695: critical state configures itself into many metastable stress reservoirs

696: insulated by relaxed zones (\S\ref{sec:gli2}).

697: Every avalanche relaxes one reservoir,

698: typically occupying a small fraction of the system, 

699: and the next avalanche occurs at random, typically far from its predecessor.

700: There is essentially no interference between successive avalanches;

701: this is verified empirically in tests with cellular automata

702: \citep{jen98}.

703: Avalanches in the tail of the size distribution, which relax the whole system,

704: are an (extremely rare) exception.

705: 

706: Given statistical independence, and assuming that the system is driven 

707: at a constant (mean) rate, the avalanche model predicts that the

708: time between successive glitches, $\Delta t$, termed the waiting time,

709: obeys Poisson statistics

710: \citep{jen98,whe00}.

711: Hence, in any individual pulsar, the waiting-time distribution is exponential,

712: with probability density function

713: \begin{equation}

714:  p(\lambda, \Delta t)

715:  =

716:  \lambda \exp(-\lambda \Delta t)~.

717: \label{eq:gli4}

718: \end{equation}

719: The mean glitching rate $\lambda$ is different for every pulsar. 

720: It depends on the rate at which differential rotation builds up

721: between the superfluid and the crist ($\propto \dot{\nu}$)

722: as well as the capacity to store the differential rotation

723: (e.g.\ strength of pinning, rate of vortex creep, shear modulus

724: of the crust).

725: The storage capacity is presumably controlled by thermodynamic variables

726: like temperature, as well as the inhomogeneous nuclear structure of the crust.

727: We do not expect $\lambda$ to change appreciably during four decades

728: of pulsar timing. In principle, however, as more data are collected in future,

729: this claim can be tested by using a Bayesian blocks algorithm

730: to divide the time series into a sequence of Poisson processes with

731: piecewise-constant rates

732: \citep{sca98,whe00,con03}.

733: \footnote{

734: It is tempting to assume that $\lambda$ is constant over decades,

735: because the thermodynamic variables that control

736: storage capacity (e.g.\ temperature) are nearly constant on

737: such a time-scale. 

738: Yet the Sun provides a cautionary counterexample:

739: the dynamics of subphotospheric turbulence, and hence the rate of

740: solar flaring, vary with the 11-yr solar cycle \citep{whe00}.

741: }

742: 

743: The avalanche model makes a further powerful prediction.

744: Suppose the system tends to a stationary, self-organised critical state,

745: in which global driving balances local release in a time-averaged sense

746: [property (vi), \S2],

747: i.e.\ there is no secular accumulation or leakage of stress.

748: Stationarity implies that the mean waiting time 

749: $\langle \Delta t \rangle = \lambda^{-1}$, 

750: multiplied by the rate at which crust-superfluid differential rotation 

751: builds up ($\epsilon\dot{\nu}$),

752: equals the mean glitch size $\langle \Delta\nu \rangle$,

753: i.e.\ 

754: \begin{equation}

755:  \lambda = - \epsilon\dot{\nu} / \langle \Delta\nu \rangle~.

756:  \label{eq:gli5}

757: \end{equation}

758: Here, $2\pi\epsilon\dot{\nu}$ is the relative angular acceleration

759: of the crust and superfluid. 

760: Importantly, glitch data allow $\epsilon$ to be measured directly

761: in principle \citep{won01}. However, there is a serious

762: question as to whether stationarity is achieved

763: in practice, during the $40$ yr that a typical

764: pulsar has been observed. We discuss this issue

765: further below.

766: 

767: To test the above ideas, we compare the 

768: observed cumulative waiting-time distributions

769: of the same nine pulsars as in \S\ref{sec:gli4}, with $N_{\rm g} > 5$,

770: against the theoretical cumulative distribution.

771: In order to make the comparison fairly, 

772: we must first adjust for observational selection effects.

773: Any given obesrvation can detect waiting times in a range

774: $\Delta t_{\rm min} \leq \Delta t \leq \Delta t_{\rm max}$.

775: The upper limit $\Delta t_{\rm max}$ is set by the total data span

776: available for that pulsar,

777: i.e.\ $\Delta t_{\rm max} = t_{\rm max} - t_{\rm min}$.

778: The lower limit $\Delta t_{\rm min}$ is different at different epochs.

779: It is set by the gap between data spans in which a glitch

780: is localized.

781: For small glitches, the glitch epoch is determined by requiring

782: continuity of pulse phase across the glitch.

783: For larger glitches, where the phase winding number is ambiguous,

784: the epoch is taken to be halfway between the bounding observations

785: \citep{wan00}.

786: Either way, $\Delta t_{\rm min}$ is different for each glitch,

787: and is twice the absolute value of the epochal uncertainty 

788: quoted in Table \ref{tab:gli1}

789: \citep{lyn00,wan00,jan06,mid06}.

790: Let $f(\Delta t_{\rm min})d(\Delta t_{\rm min})$ 

791: be the observing-time-weighted probability that, when a glitch occurs,

792: $\Delta t_{\rm min}$ lies in the range

793: $[ \Delta t_{\rm min},

794:  \Delta t_{\rm min} + d(\Delta t_{\rm min}) ]$,

795: and let the smallest and largest values of $\Delta t_{\rm min}$

796: be $\Delta t_{\rm min}^{(<)}$ and $\Delta t_{\rm min}^{(>)}$ respectively,

797: recorded in Table \ref{tab:gli4} for the nine pulsars

798: with $N_{\rm g} > 5$.

799: Then the cumulative waiting-time distribution is given by

800: \begin{eqnarray}

801:  P(\lambda, \Delta t) 

802:  & = &

803:  \int_{\Delta t_{\rm min}^{(<)}}^{\Delta t_{\rm min}^{(>)}}

804:  d(\Delta t_{\rm min}') \, f(\Delta t_{\rm min}') \nonumber \\

805: & & \times \int_{\Delta t_{\rm min}'}^{\Delta t_{\rm max}} 

806:  d(\Delta t') \, p(\lambda, \Delta t')~,

807:  \label{eq:gli6}

808:  \\

809:  & = &

810:  \frac{1}{N_{\rm g}}

811:  \sum_{\Delta t_{\rm min}=\Delta t_{\rm min}^{(<)}}^{\Delta t_{\rm min}^{(>)}}

812:  \frac{ \exp(-\lambda\Delta t_{\rm min}) - \exp(-\lambda\Delta t) }

813:   { \exp(-\lambda\Delta t_{\rm min}) - \exp(-\lambda\Delta t_{\rm max}) }~,

814: \nonumber \\

815:  \label{eq:gli7}

816: & & 

817: \end{eqnarray}

818: where each glitch is weighted equally in the sum in (\ref{eq:gli7})

819: as a first approximation. 

820: 

821: In Figure \ref{fig:gli4}, we plot as cumulative histograms the measured

822: waiting-time distributions of the nine pulsars in Figure \ref{fig:gli1}.

823: The theoretical curves (\ref{eq:gli7}) are overlaid,

824: with $\Delta t_{\rm min}$ and $\Delta t_{\rm max}$ 

825: chosen according to the second through fourth columns in Table \ref{tab:gli4}.

826: For each object, we choose $\lambda$ to minimize the K-S statistic $D$.

827: The fitting parameters are displayed in the fifth through eighth columns

828: in Table \ref{tab:gli4}.

829: As in \S\ref{sec:gli4},

830: the goodness of the fit is characterized by the K-S probability $P_{\rm KS}$,

831: with $P_{\rm KS} < 0.32$ in the interval $[\lambda_-,\lambda_+]$.

832: 

833: 

834: For all nine pulsars in Figure \ref{fig:gli4} and Table \ref{tab:gli4},

835: the null hypothesis that

836: the waiting-time distribution is described by Poisson statistics

837: is not ruled out at the 1-$\sigma$ level.

838: The data are therefore consistent with an avalanche process.

839: An exponential waiting-time distribution was first postulated by

840: \citet{won01}

841: for PSR J0534$+$2200, 

842: based on timing data up to and including the glitch on MJD 51452.

843: These authors obtained 

844: $\lambda = 0.53\,{\rm yr^{-1}}$ and $P_{\rm KS}=0.7$,

845: marginally outside the 1-$\sigma$ range in Table \ref{tab:gli4}.

846: The data analyzed here confirm that waiting times are

847: consistent with Poisson statistics

848: in several glitching pulsars,

849: affording a key insight into the physics of the glitch mechanism. 

850: The implications of this result are discussed further in \S\ref{sec:gli6}.

851: 

852: \subsection{Quasiperiodicity

853:  \label{sec:gli5b}}

854: For seven pulsars in Figure \ref{fig:gli4}, 

855: the Poisson distribution affords an excellent fit, 

856: both formally and by eye.

857: However, for PSR J0537$-$6910 and PSR J0835$-$4510,

858: the fits are marginal at the 1-$\sigma$ level.

859: [Indeed, in an earlier analysis of PSR J0835$-$4510,

860: \citet{won01} excluded a Poisson distribution with $96\%$ confidence,

861: on the basis of fewer data.]

862: These are the same two objects whose size distributions are exceptional,

863: and which are observed to glitch quasiperiodically.

864: 

865: Taking the same approach as in \S\ref{sec:gli4}, 

866: we model the quasiperiodicity crudely by adding a periodic

867: component to (\ref{eq:gli4}), viz.

868: \begin{equation}

869:  p(\lambda,\Delta t) 

870:  =

871:  C_{\rm s}' \lambda \exp(-\lambda\Delta t)

872:  + C_{\rm p}' \delta ( \Delta t - \Delta t_{\rm c} )~.

873:  \label{eq:gli8}

874: \end{equation}

875: In (\ref{eq:gli8}), $C_{\rm s}'$ and $C_{\rm p}'$ are the 

876: normalized relative weights

877: of the Poisson and periodic components respectively,

878: and $\Delta t_{\rm c}$ is the characteristic period.

879: The associated cumulative distribution, weighted by $\Delta t_{\rm min}$,

880: is obtained by substituting (\ref{eq:gli8}) into (\ref{eq:gli6}),

881: yielding

882: \begin{eqnarray}

883:  P(\lambda,\Delta t)

884:  & = &

885:  \frac{1}{N_{\rm g}}

886:  \sum_{\Delta t_{\rm min}=\Delta t_{\rm min}^{(<)}}^{\Delta t_{\rm min}^{(>)}}

887:  \left\{ 

888:   C_{\rm p}' H(\Delta t - \Delta t_{\rm c})

889:   \phantom{\frac{1}{2}}

890:  \right.

891:  \nonumber \\

892:  & & 

893:  \left.

894:  + 

895:  \frac{ (1-C_{\rm p}') [ \exp(-\lambda\Delta t_{\rm min}) - \exp(-\lambda\Delta t) ] }

896:    { \exp(-\lambda\Delta t_{\rm min}) - \exp(-\lambda\Delta t_{\rm max}) }

897:  \right\}~. \nonumber \\

898: & & 

899:  \label{eq:gli9}

900: \end{eqnarray}

901: 

902: The two-component model (\ref{eq:gli8}) yields improved fits 

903: to the data, with $C_{\rm p}' \approx 0.25$ for both objects.

904: The fits are graphed with the data in Figure \ref{fig:gli5},

905: and the best-fit parameters are recorded in Table \ref{tab:gli5}.

906: We obtain $\Delta t_{\rm c} = (0.3\pm0.1)\,{\rm yr}$ 

907: and $\Delta t_{\rm c} = (2.8\pm0.1)\,{\rm yr}$ 

908: for PSR J0537$-$6910 and PSR J0835$-$4510 respectively,

909: in accord with previous authors

910: \citep{lyn96,mid06}.

911: Significantly, the data imply $C_{\rm p} \approx C_{\rm p}'$.

912: In other words,

913: the delta-distributed component accounts for the same fraction

914: of the size and waiting-time distributions, even though the sizes

915: and waiting times are statistically independent.

916: This raises confidence in the model and suggests that a

917: quasiperiodic component is indeed present and distinct.

918: It also suggests that the quasiperiodic component coexists with

919: the Poisson component, instead of completely displacing it.

920: Vela, for example, is likely to possess an extensively connected network

921: of capacitive elements, but it also contains smaller subnetworks

922: that are disconnected from the main network;

923: cf.\ \citet{alp96}.

924: This is natural for an avalanche process,

925: as noted in \S\ref{sec:gli4b}

926: \citep{ros93,jen98}.

927: 

928: 

929: \subsection{Mean rate

930:  \label{sec:gli5c}}

931: Stationarity of the avalanche model over long time intervals

932: implies a relation between

933: the Poisson rate, driving rate, and mean glitch size,

934: given by (\ref{eq:gli5}).

935: Unfortunately, for $a < 2$,

936: $\langle \Delta \nu \rangle$ is dominated by large glitches

937: near the upper cut-off of $P(\Delta \nu/\nu)$:

938: \footnote{For $1 < a < 2$, $\langle \Delta \nu \rangle$

939: is dominated by the upper cut-off, but the normalization

940: of $p(\Delta \nu/\nu)$ is dominated by the lower cut-off.}

941: \begin{equation}

942: \label{eq:new_equation}

943: \langle \Delta \nu \rangle  = \left|\frac{a-1}{a-2}\right| \left\{ \begin{array}{ll}

944:        \displaystyle  \Delta \nu_{\rm upper} & a > 2 \\

945:         \displaystyle  \left(\frac{\Delta \nu_{\rm upper}}{\Delta \nu_{\rm lower}}\right)^{a-1} \Delta \nu_{\rm upper} & 1 < a < 2 \\

946: \displaystyle \Delta \nu_{\rm upper} & a < 1.

947:      \end{array}\right.

948: \end{equation}

949: In (\ref{eq:new_equation}), $\Delta \nu_{\rm lower}$

950: and $\Delta \nu_{\rm upper}$ are the physical lower and upper 

951: cut-offs of the probability distribution function (\ref{eq:gli1}).

952: As large glitches are rare, stationarity is not achieved during

953: the $40$-yr interval over which a typical pulsar is observed; the

954: largest observed size, $(\Delta \nu/\nu)_{\rm max}$, cannot be equated

955: reliably with the maximum size allowed physically.

956: Likewise, $\langle \Delta \nu\rangle$ is approximated poorly by the

957: mean of the observed glitches. In practice, therefore, it is impossible

958: to estimate $\langle \Delta \nu\rangle$ without much longer monitoring.

959: 

960: 

961: Physically, $\epsilon\dot{\nu}$ is the rate at which

962: differential rotation builds up between the crust and superfluid.

963: Hence, in the vortex unpinning model, $\epsilon$ gives the time-averaged

964: fraction of pinned vortices or capacitive elements.

965: We can use equation (\ref{eq:new_equation}) to place limits

966: on $\epsilon$, at least in principle. \footnote{The error in

967: $\Delta \nu$ is $d \langle \Delta \nu \rangle /da$ multiplied

968: by the error in $a$.}

969: For example, the inequalities $\Delta \nu_{\rm lower} < \Delta \nu_{\rm min}$

970: and $\Delta \nu_{\rm max} < \Delta \nu_{\rm upper} < \nu$

971: must always be satisfied. For PSR J0358$+$5413, assuming

972: $a=2.4$, we find $\langle \Delta \nu\rangle/\nu \leq 1.1 \times 10^{-10}$

973: and hence $\epsilon \leq 7 \times 10^{-5}$. This is lower

974: than previous estimates of the pinned fraction for objects

975: of that age, but in line with previous estimates of the pinned

976: fraction in young objects like the Crab \citep{lyn00,won01}.

977: For the remaining eight objects, the above inequalities lead

978: to upper limits on $\epsilon$ which are greater

979: than unity and hence not useful. As a crude experiment,

980: we check the result of setting $\Delta \nu_{\rm upper} = 2 \times 10^{-4}$,

981: the largest glitch observed in any pulsar over the last $40$ years,

982: for every pulsar.

983: We obtain five slightly more useful upper limits

984: ($\epsilon \leq 4 \times 10^{-2}$, $0.2$, $0.8$, $0.8$, and $0.7$ for

985: PSR J0534$+$2200, PSR J0631$+$1036, PSR J0835$-$4510, PSR J1341$-$6220, and

986: PSR J1740$-$3015 respectively). However, we emphasize that these values

987: are still problematic, because there is no guarantee that a total effective

988: observation interval of $40$ yr $\times$ $101$ pulsars 

989: is long enough in aggregate for

990: stationarity to be observed. Moreover, these values are based

991: on assuming that all pulsars have the same physical $\Delta \nu_{\rm upper}$,

992: which is not necessarily the case.

993: 

994: 

995: 

996: Figure \ref{fig:gli6} displays the cumulative histogram of 

997: Poisson rates derived from the best-fit waiting-time distributions 

998: in Figures \ref{fig:gli4} and \ref{fig:gli5}.

999: Let $q(\lambda)$ denote the rate probability density function,

1000: such that $q(\lambda)d\lambda$ is the probability that the mean rate

1001: lies in the interval $[\lambda,\lambda+d\lambda]$ in a given pulsar.

1002: There is no obvious theoretical reason to prefer a particular 

1003: analytic form of $q(\lambda)$,

1004: which is controlled by the physics of the global driver,

1005: not the scale invariant avalanche dynamics.

1006: In addition, the data in Figure \ref{fig:gli6} are insufficient to 

1007: specify the analytic form of $q(\lambda)$ uniquely.

1008: However, motivated by the rate distribution observed in solar flares,

1009: which is measured reliably to be exponential

1010: \citep{whe00},

1011: we find that a distribution of the form

1012: \begin{equation}

1013:  q(\lambda) 

1014:  =

1015:  \langle \lambda \rangle^{-1}

1016:  \exp(-\lambda / \langle \lambda \rangle )

1017:  \label{eq:gli10}

1018: \end{equation}

1019: fits the data satisfactorily, with

1020: $\langle \lambda \rangle = 1.3_{-0.6}^{+0.7}\,{\rm yr^{-1}}$

1021: including quasiperiodic glitchers ({\em left panel})

1022: or $\langle \lambda \rangle = 1.2_{-0.4}^{+0.5}\,{\rm yr^{-1}}$

1023: excluding quasiperiodic glitchers ({\em right panel}).

1024: Formally, the K-S probabilities are $P_{\rm KS} = 0.9946$

1025: and $0.82$ respectively.

1026: 

1027: The distribution is incompletely sampled below an effective

1028: minimum rate $\lambda_{\rm min}$,

1029: which is set by $\Delta t_{\rm max}$.

1030: To illustrate, if we proclaim five glitches (arbitrarily) to be 

1031: the minimum number needed for a reliable determination of $\lambda$,

1032: we obtain 

1033: $\lambda_{\rm min} = 5 \Delta t_{\rm max}^{-1} 

1034:  \sim 0.2\,{\rm yr^{-1}}$.

1035: Careful modeling of this observational bias is deferred

1036: to a future paper. We describe a first attempt in \S\ref{sec:gli5d}.

1037: 

1038: \subsection{Aggregate distribution

1039:  \label{sec:gli5d}}

1040: The nine pulsars in Figure \ref{fig:gli4} account for only 108 out of a 

1041: total of 285 observed glitches.

1042: Most glitching pulsars have only glitched once or twice,

1043: but they still contribute statistical information regarding 

1044: waiting times, the former via lower limits on $\Delta t$.

1045: While these data cannot usefully constrain $P(\lambda,\Delta t)$

1046: in individual pulsars, they feed into the aggregate waiting-time distribution

1047: and hence constrain $q(\lambda)$ more tightly than in \S\ref{sec:gli5c}.

1048: 

1049: In Figure \ref{fig:gli7}, 

1050: we present the aggregate waiting-time distribution $P_{\rm agg}(\Delta t)$

1051: including ({\em left panel}) and excluding ({\em right panel})

1052: the quasiperiodic glitchers PSR J0537$-$6910 and PSR J0835$-$4510.

1053: The histogram is constructed to include all 182 waiting times in those

1054: objects that have glitched more than once.

1055: The K-S test confirms that the aggregate distribution is fitted poorly

1056: by a single, constant-rate, Poisson distribution of the form

1057: (\ref{eq:gli4}). Furthermore, when we weight (\ref{eq:gli4}) by

1058: the exponential rate distribution (\ref{eq:gli10}),

1059: \footnote{We compute $P_{\rm agg}(\Delta t)$ theoretically as a weighted sum of

1060: independent Poisson processes. In the same way, the waiting-time distribution

1061: for decays observed from a mixture of radioactive isotopes is a sum of

1062: constant-rate Poisson distributions, one per isotope,

1063: weighted by isotopic abundance.} viz.

1064: \begin{eqnarray}

1065:  P_{\rm agg}(\Delta t)

1066:  & = &

1067:  \int_0^{\Delta t} d(\Delta t')

1068:  \int_{0}^{\infty}

1069:  d\lambda' \,

1070:  q(\lambda') p(\lambda',\Delta t'),

1071:  \label{eq:gli11}

1072: \end{eqnarray}

1073: the fit remains poor. For example, the dotted curves in Figure \ref{fig:gli7} are

1074: computed by evaluating (\ref{eq:gli11})

1075: with the mean values

1076: $\langle \lambda \rangle = 1.3_{-0.6}^{+0.7}\,{\rm yr^{-1}}$ ({\em left panel})

1077: and $\langle \lambda \rangle = 1.2_{-0.4}^{+0.5}\,{\rm yr^{-1}}$ ({\em right panel})

1078: extracted from Figure \ref{fig:gli6}. They yield $P_{\rm KS} = 4.3 \times 10^{-2}$

1079: and $2.4 \times 10^{-2}$ respectively. If, instead,

1080: we adjust $\langle \lambda \rangle$ to maximize

1081: $P_{\rm KS}$ while fitting (\ref{eq:gli10}) and (\ref{eq:gli11}) to the observed $P_{\rm agg}(\Delta t)$,

1082: as shown by the dashed curves in Figure \ref{fig:gli7}, we obtain

1083: $\langle \lambda \rangle = 1.1 \,{\rm yr^{-1}}$, $P_{\rm KS} = 0.18$ ({\em left panel})

1084: and $\langle \lambda \rangle = 0.92\,{\rm yr^{-1}}$, $P_{\rm KS} = 0.31$ ({\em right panel}) respectively.

1085: 

1086: We can exploit the extra information in Figure \ref{fig:gli7}

1087: from objects with $2 \leq N_{\rm g}  \leq 5$ to determine $q(\lambda)$

1088: more accurately.

1089: To do this, we assume a rate

1090: probability density function of the form

1091: \begin{equation}

1092: \label{eq:14_new} q_{\lambda} = \langle \lambda \rangle^{-1}

1093: H(\lambda - \lambda_{\rm min}) \exp[-(\lambda - \lambda_{\rm min})/\langle \lambda \rangle],

1094: \end{equation}

1095: normalize $P(\lambda,\Delta t)$ over the range $[\Delta t_{\rm min}, \Delta t_{\rm max}]$,

1096: and evaluate (\ref{eq:gli11}) to obtain 

1097: \begin{eqnarray}

1098:  P_{\rm agg}(\Delta t)

1099:  & = &

1100: \langle \lambda \rangle^{-1} \int_{\lambda_{\rm min}}^{\infty}

1101:  d\lambda' \, \exp[-(\lambda'-\lambda_{\rm min})/\langle \lambda \rangle] 

1102: \nonumber \\

1103: &  & 

1104: \times \frac{1 - \exp(-\lambda' \Delta t)}{1 - \exp(-\lambda' \Delta t_{\rm max})}

1105: \label{eq:gli12}

1106: \end{eqnarray}

1107: In (\ref{eq:gli12}), we neglect for simplicity the observational bias

1108: introduced by uncertainties in glitch epoch discussed in \S\ref{sec:gli5a};

1109: lacking fuller information, 

1110: we take $\Delta t_{\rm min}=0$ and $\Delta t_{\rm max} = 28.03$ {\rm yr} for all pulsars.

1111: Excellent fits are obtained using (\ref{eq:gli12}).

1112: We find $\langle \lambda \rangle = 0.54$ {\rm yr}$^{-1}$, 

1113: $\lambda_{\rm min}=0.25$ {\rm yr}$^{-1}$, and $P_{\rm KS} = 0.76$ for

1114: all nine pulsars with $N_{\rm g} > 5$,

1115: and $\langle \lambda \rangle = 0.43$ {\rm yr}$^{-1}$,

1116: $\lambda_{\rm min}=0.25$ {\rm yr}$^{-1}$, and $P_{\rm KS} = 0.98$

1117: when the quasiperiodic glitchers are excluded. 

1118: The fits are plotted as solid curves in Figure \ref{fig:gli7}\

1119: ({\em left} and {\em right panels} respectively).

1120: We verify the results

1121: by referring back to the measured $q(\lambda)$ distribution.

1122: Substituting the fitted values of $\langle \lambda \rangle$ and

1123: $\lambda_{\rm min}$ into (\ref{eq:14_new}), we obtain

1124: the solid curves in Figure \ref{fig:gli6_redo}, with

1125: $P_{\rm KS} = 0.52$ ({\em left panel})

1126: and $0.25$ ({\em right panel}) respectively. 

1127: Alternatively, the previous fits $\langle \lambda \rangle=1.1$ {\rm yr}$^{-1}$,

1128: $\lambda_{\rm min}=0$ and $\langle \lambda \rangle=0.54$ {\rm yr}$^{-1}$,

1129: $\lambda_{\rm min}=0$ yield

1130: $P_{\rm KS} = 0.72$ ({\em left panel}, dashed curve)

1131: and $0.52$ ({\em right panel}, dashed curve). 

1132: 

1133: In all cases, $P_{\rm agg} (\Delta t)$

1134: points to the existence of more low-rate objects than 

1135: the $N_{\rm g}>5$ sample in Figure \ref{fig:gli6}

1136: predicts. Specifically, up to $\sim 30$ \% of the population

1137: of glitching pulsars can have $\lambda < 0.25$ {\rm yr}$^{-1}$ while

1138: still reproducing $P_{\rm agg} (\Delta t)$.

1139: We emphasize again that Figure \ref{fig:gli7} constrains $q(\lambda)$ 

1140: more tightly than Figure \ref{fig:gli6}, 

1141: because it contains information about $\Delta t$ from 1.3 times 

1142: as many glitches, including useful information from pulsars which have 

1143: glitched twice.

1144: 

1145: Undetectable microglitches probably occur between detected glitches

1146: without our knowledge, given that $p(\Delta\nu/\nu)$ is scale invariant.

1147: This effect subtracts from the lower end of the $\Delta t$ distribution 

1148: and adds to the upper end.

1149: We do not correct for it here, because it is hard

1150: to quantify without better statistics.

1151: On two occasions, a pair of glitches occurred on the same date,

1152: once in the same pulsar, and once in different pulsars

1153: \citep{kra05}.

1154: We take $\Delta t = 0$ for these pairs.

1155: Phase connected timing mitigates duty cycle biases,

1156: but it does not eliminate them. 

1157: 

1158: \subsection{Fluctuation spectrum

1159:  \label{sec:gli5e}}

1160: The temporal fluctuations in a stochastic signal $x(t)$ carry 

1161: independent statistical information about the underlying physical process.

1162: The power spectrum $S(f)$, where $f$ denotes the Fourier frequency,

1163: is related to the temporal autocorrelation function

1164: $G(\tau) = \langle x(t) x(t+\tau) \rangle - \langle x(t) \rangle^2$

1165: (where the average $\langle \dots \rangle$ is performed over $t$ 

1166: for a stationary process)

1167: through the cosine transform

1168: \begin{equation}

1169:  S(f) =

1170:  2\int_0^\infty d\tau \, G(\tau) \cos(2\pi f \tau)~.

1171:  \label{eq:gli13}

1172: \end{equation}

1173: 

1174: In general, for an avalanche process, the power spectrum depends

1175: jointly on the size, waiting-time, and lifetime distributions

1176: of the avalanches

1177: \citep{jen98}.

1178: For glitches, however, the lifetimes are too short to measure

1179: with current technology (see \S\ref{sec:gli6}).

1180: If, furthermore, we restrict attention to the unit-impulse signal

1181: $x(t) = \sum_i \delta(t-t_i)$,

1182: where $t_i$ denotes the epoch of the $i$-th glitch,

1183: then the sizes drop out of the problem too.

1184: The power spectrum then carries exactly the same information as the

1185: waiting-time distributions 

1186: $P(\lambda,\Delta t)$ and $P_{\rm agg}(\Delta t)$,

1187: with

1188: \begin{equation}

1189:  S(f) \propto 

1190:  \frac{\lambda}{\lambda^2 + (2\pi f)^2}

1191:  \label{eq:gli14}

1192: \end{equation}

1193: for any individual pulsar, and 

1194: \begin{equation}

1195:  S(f) \propto

1196:  \int_{\lambda_{\rm min}}^\infty

1197:  \frac{d\lambda' \, \lambda' q(\lambda')}

1198:   {\lambda'^2 + (2\pi f)^2}

1199:  \label{eq:gli15}

1200: \end{equation}

1201: for the pulsar population in aggregate.

1202: 

1203: At high frequencies $f\gg\langle\lambda\rangle$, 

1204: equations (\ref{eq:gli14}) and (\ref{eq:gli15}) 

1205: [with $q(\lambda)$ given by (\ref{eq:gli10})]

1206: scale as $f^{-2}$, with $O(f^{-4})$ and $O(f^{-4} \sin 2 f)$ corrections. 

1207: These scalings are modified if the delta function in $x(t)$

1208: is replaced by a nonsingular window function that embodies the

1209: shape of the signal from an individual avalanche.

1210: It will be instructive to revisit this question when it becomes

1211: possible to resolve the lifetimes of individual avalanches,

1212: e.g.\ in single- or giant-pulse timing experiments,

1213: 

1214: \section{Discussion

1215:  \label{sec:gli6}}

1216: In this paper, we analyze the size and waiting-time distributions

1217: of pulsar glitches, taking advantage of the

1218: enlarged data set produced by the Parkes Multibeam Survey.

1219: We conclude that the data are consistent with the hypothesis

1220: that pulsar glitches arise from an avalanche process.

1221: In each of seven pulsars with $N_{\rm g} > 5$,

1222: the size distribution is consistent with being 

1223: scale invariant across the observed range

1224: of $\Delta\nu$ (up to four decades),

1225: and the waiting-time distribution is consistent

1226: with being Poissonian.

1227: These features are

1228: natural if the system is driven globally at a constant rate

1229: (as the pulsar spins down),

1230: and each glitch corresponds to a locally collective, threshold activated

1231: relaxation of one of the many spatially independent,

1232: metastable stress reservoirs in the system

1233: (e.g. via a vortex unpinning or crust cracking avalanche).

1234: In two pulsars, PSR J0537$-$6910 and PSR J0835$-$4510,

1235: the dynamics may include a second, quasiperiodic component,

1236: comprising $\sim 20\%$ of the events.

1237: The size and waiting-time distributions of the quasiperiodic

1238: component are narrowly peaked,

1239: as expected for rare, system-spanning avalanches, which relax

1240: a large fraction of the total stress accumulated in the system.

1241: This two-component behavior is observed widely in self-organised critical systems,

1242: including experiments on magnetic flux vortices in type II superconductors,

1243: which are closely analogous to neutron star superfluids

1244: \citep{fie95}.

1245: 

1246: The power-law exponent of the size probability density function

1247: differs from pulsar to pulsar, spanning the range

1248: $-0.13 \leq a \leq 2.4$.

1249: Calculating $a$ theoretically from first principles is a deep problem

1250: which has not yet been solved for other self-organised critical systems,

1251: let alone glitching pulsars,

1252: although some progress has been made on two-dimensional sandpiles using 

1253: renormalization group techniques

1254: \citep{pie94,jen98}.

1255: In the mean-field approximation, which is exact in four or more dimensions, 

1256: theoretical calculations on sandpiles (and other systems in their

1257: universality class) yield $a = 1.5$,

1258: whereas three-dimensional cellular automata output $a=1.3$

1259: \citep{jen98}.

1260: 

1261: The size distribution transmits two important lessons

1262: concerning the microphysics of glitches.

1263: First, the fact that $a$ differs from pulsar to pulsar implies

1264: that the strength and level of conservation of the local 

1265: (e.g.\ pinning and intervortex) forces also differs

1266: \citep{ola92,fie95}.

1267: By contrast, in equilibrium critical systems like ferromagnets,

1268: $a$ depends only on the dimensionality of the system and its

1269: order parameter and is therefore universal.

1270: Second, except for the two pulsars which show quasiperiodicity,

1271: $a$ appears to vary smoothly with spin-down age,

1272: with $a\approx 1.2$ for the youngest pulsars (e.g.\ the Crab).

1273: Figure \ref{fig:gli9} depicts the trend between $a$ and $\tau_{\rm c}$.

1274: It is suggestive; after all, local pinning forces do depend on

1275: temperature and hence $\tau_{\rm c}$.

1276: Interestingly, however, there is no clear trend between $a$ and $\nu$,

1277: even though the mean vortex spacing (and hence intervortex force)

1278: is proportional to $\nu^{1/2}$.

1279: It will pay to study these trends more thoroughly

1280: as more glitch data is collected.

1281: 

1282: An avalanche process predicts a specific relation between

1283: the distributions of glitch sizes $\Delta\nu$ and lifetimes $T$

1284: (as opposed to waiting times $\Delta t$).

1285: Specifically, in a self-organized critical state,

1286: the lifetime probability density function is also a power law,

1287: $p(T) \propto T^{-b}$,

1288: with

1289: \begin{equation}

1290:  b = 1 + (a-1)\gamma_2 / \gamma_3~.

1291:  \label{eq:gli16}

1292: \end{equation}

1293: The constants $\gamma_2$ and $\gamma_3$ are defined such that the cardinality 

1294: of an avalanche scales with its linear extent ($L$) as $L^{\gamma_2}$

1295: and its lifetime (i.e.\ duration) scales as $L^{\gamma_3}$

1296: \citep{jen98}.

1297: Both $\gamma_2$ and $\gamma_3$ depend on the effective dimensionality of the 

1298: local forces

1299: and can be calculated numerically using a cellular automaton.

1300: In two dimensions, avalanches are compact, not fractal,

1301: and one has $\gamma_2=2$;

1302: in three dimensions, one has $2 < \gamma_2 < 3$.

1303: At present, radio timing experiments cannot measure $T$;

1304: most glitches are detected as unresolved, discontinuous, spin-up events 

1305: with $T < 120\,{\rm s}$ \citep{mcc90}.

1306: \footnote{In the Crab, some spin-up events seem to be resolved,

1307: e.g.\ at epochs MJD 50260 ($T \approx 0.5\,{\rm d}$)

1308: and MJD 50489 (secondary spin up, $T \approx 2\,{\rm d}$)

1309: \citep{won01}.

1310: If these are rare but otherwise standard glitches originating from the long-$T$

1311: tail of the lifetime distribution, it is puzzling that other, shorter,

1312: but still resolved (and presumably more common) spin-up events 

1313: are not observed, e.g.\ with $T\sim 0.1\,{\rm d}$ or $0.01\,{\rm d}$.

1314: Alternatively, the events at MJD 50620 and MJD 50489

1315: may have been triggered by a different physical mechanism.}

1316: In the future, however, single- and/or giant-pulse timing experiments

1317: with more sensitive instruments (e.g.\ the Square Kilometer Array) 

1318: will test this prediction.

1319: If confirmed, it will independently corroborate the avalanche hypothesis.

1320: 

1321: The mean glitching rates of the nine pulsars studied here

1322: are fairly narrowly distributed,

1323: spanning the range

1324: $0.35\,{\rm yr^{-1}} \leq \lambda \leq 2.6\,{\rm yr^{-1}}$.

1325: The probability density function for $\lambda$ is adequately

1326: fitted by an exponential, as for solar flare avalanches \citep{whe00},

1327: with $\langle\lambda\rangle = 1.3_{-0.6}^{+0.7}$ {\rm yr}$^{-1}$,

1328: or by an exponential with a lower cutoff, at $\lambda_{\rm min} \approx 0.25$ {\rm yr}$^{-1}$.

1329: A theoretical derivation of $\langle\lambda\rangle$ from first principles

1330: is currently lacking,

1331: although estimates of how long it takes to crack the crust locally

1332: predict reasonable rates,

1333: if the critical strain angle approaches that of imperfect terrestrial metals

1334: \citep{alp96,mid06}.

1335: 

1336: Figure \ref{fig:gli10} plots $\lambda$ versus $\tau_c$ for the nine

1337: pulsars examined individually in this paper. There is no

1338: significant trend. The data are consistent with the notion that

1339: old pulsars glitch less frequently than young pulsars

1340: \citep{she96}, but they are equally consistent with the notion

1341: that the glitching rate is independent of age.

1342: 

1343: Many authors have searched for a correlation between waiting time

1344: and the size of the next glitch.

1345: Such a correlation appears to be absent from the data,

1346: e.g.\ Figure 17 in \citet{wan00}

1347: and Figure 10 in \citet{won01}.

1348: At first blush, this is surprising: the vortex unpinning and

1349: crust fracture paradigms, which are driven by the accumulation

1350: of differential rotation and mechanical stress respectively,

1351: seem to be natural candidates for a `reservoir effect'.

1352: Avalanche dynamics resolves this apparent paradox.

1353: The reservoir effect does operate locally, 

1354: but the star contains many reservoirs,

1355: insulated from each other by relaxed zones,

1356: whose storage capacities evolve stochastically in response to

1357: the slow driver and avalanche history.

1358: During a glitch, a single reservoir

1359: (often small but sometimes large) relaxes at random via an avalanche,

1360: releasing its stored $\Delta\nu$

1361: (and destabilizing neighboring reservoirs in preparation for

1362: the next glitch).

1363: Some of the $\Delta\nu$ is accumulated since the previous glitch, 

1364: but the remainder is `borrowed' from earlier epochs,

1365: when some other reservoir relaxed instead.

1366: All self-organized critical systems share these dynamics;

1367: the waiting time is uncorrelated with the size of the next avalanche

1368: \citep{jen98}.

1369: The only exceptions are large, system-spanning avalanches,

1370: which always have roughly the same sizes and waiting times,

1371: and which account for $\sim 20\%$ of the 

1372: glitches in PSR J0537$-$6910 and PSR J0835$-$4510.

1373: 

1374: A corollary of the previous paragraph is that the total $\Delta\nu$

1375: released in glitches up to some epoch is less than the total crust-superfluid

1376: differential rotation accumulated since that epoch, viz.

1377: \begin{equation}

1378:  \sum_{i=1}^{N_{\rm g}} \Delta\nu_i

1379:  \leq 

1380:  \epsilon |\dot{\nu}|  \sum_{i=1}^{N_{\rm g}} \Delta t_{i},

1381:  \label{eq:gli17}

1382: \end{equation}

1383: where $\epsilon\dot{\nu}$ is the relative angular acceleration of the

1384: crust and superfluid due to electromagnetic spin down.

1385: The `staircase' described by (\ref{eq:gli17}) has been noted previously 

1386: \citep{she96,lyn00}, 

1387: both in quasiperiodic glitchers like PSR J0537$-$6910

1388: [e.g.\ Figure 8 in \citet{mid06}],

1389: where the reservoir effect is obvious,

1390: and in Poisson glitchers like PSR J0534$+$2200,

1391: [e.g.\ Figure 12 in \citet{won01}],

1392: where the trend is more subtle because it reverts to the mean

1393: over long times, not after every glitch.

1394: Upon dividing (\ref{eq:gli17}) by $N_{\rm g}$,

1395: and averaging over long times,

1396: the inequality becomes an equality (provided there is no secular

1397: accumulation of differential rotation in the system)

1398: and we recover (\ref{eq:gli5}). 

1399: 

1400: It is fundamentally impossible to measure $\epsilon$

1401: for individual pulsars with current data, because

1402: $\langle \Delta \nu \rangle$ is dominated by large

1403: (and therefore rare) glitches for $a < 2$. It is therefore wrong

1404: to assume stationarity over a typical, $40$-yr observation

1405: interval. Consequently, we are prompted to reassess

1406: the familiar correlation between activity and spin-down age

1407: \citep{she96}.

1408: Our definition of $\epsilon\dot{\nu}$ is identical to 

1409: $\dot{\nu}_{\rm glitch}$ in \citet{lyn00}

1410: (but for individual objects, not in aggregate)

1411: and $A_{\rm g}$ in \citet{won01}.

1412: It is closely related to the original activity parameter

1413: defined by \citet{mck90},

1414: which equals $N_{\rm g} \nu^{-1} \epsilon |\dot{\nu}|$.

1415: For PSR J0358$+$5413, we measure

1416: $\epsilon\leq 7 \times 10^{-5}$, lower than

1417: the {\it aggregate} value $0.017\pm0.002$

1418: measured by \citet{lyn00} for objects with 

1419: $\tau_{\rm c} > 10\,{\rm kyr}$ (binned by semi-decades in $\dot{\nu}$).

1420: \footnote{The aggregate value $\dot{\nu}_{\rm glitch}$

1421: \citep{lyn00}, binned over semi-decades in $\dot{\nu}$, effectively averages

1422: together different pulsars. While this approach reduces the formal

1423: error bar on $\dot{\nu}_{\rm glitch}$, its physical interpretation

1424: is less straightforward, given the likelihood that $\epsilon$

1425: is different in different pulsars.} 

1426: Interpreted in terms of the vortex unpinning model,

1427: this result suggests that $0.007$--$2$ \% of the angular momentum outflow

1428: during spin down may be stored in metastable reservoirs on average

1429: over time. On the other hand, five other objects have

1430: $0.04 \leq \epsilon_{\rm max} \leq 0.8$, under the questionable

1431: assumption that the maximum physical size is 

1432: $\Delta \nu_{\rm upper} = 2 \times 10^{-4}$

1433: in all pulsars. Our data are therefore inadequate to update usefully

1434: the value $A_{\rm g}/|\dot{\nu}| = 1\times 10^{-5}$ measured by \citet{won01}

1435: for PSR J0534$+$2200.

1436: 

1437: In the context of vortex unpinning, it has been argued that

1438: the aggregated $\epsilon$ measured by \citet{lyn00}

1439: partly corroborates 

1440: the hypothesis that younger pulsars are still in the process of

1441: forming their capacitive elements,

1442: e.g.\ by creating pinning centers through crust fracture,

1443: while older pulsars have mostly completed the task

1444: \citep{alp96,won01}.

1445: However, the full picture is more complicated.

1446: Vela's quasiperiodic avalanches point to a richly

1447: connected network of reservoirs

1448: \citep{alp96},

1449: yet its aggregated value $\dot{\nu}_{\rm glitch}$

1450: is relatively low.

1451: On the other hand, the other quasiperiodic glitcher,

1452: PSR J0537$-$6910, is relatively young 

1453: ($\tau_{\rm c} = 4.9\,{\rm kyr}$);

1454: how did it form a richly connected reservoir network so quickly?

1455: And, if its network is so richly connected, why is its 

1456: aggregated $\dot{\nu}_{\rm glitch}$ value so low? Likewise,

1457: PSR J0358$+$5413 is the oldest object in

1458: the sample ($\tau_{\rm c} = 560 \,{\rm kyr}$),

1459: yet its $\epsilon$ value arguably points to a dearth

1460: of capacitive elements, characteristic of a young object.

1461: There are no obvious grounds (e.g.\ quasiperiodicity) 

1462: on which to treat PSR J0358$+$5413 as exceptional.

1463: 

1464: Do all pulsars glitch eventually?

1465: It has been speculated in the past that there is something special

1466: physically about the minority of pulsars that do glitch.

1467: While it is impossible to reject this hypothesis unequivocally

1468: with the data at hand,

1469: the results presented here suggest

1470: that all pulsars are capable of glitching.

1471: However, most do so infrequently (low $\lambda$)

1472: and hence have not been detected

1473: during the last four decades of timing experiments.

1474: We find that up to $\sim 30 \%$ of the pulsar population

1475: can glitch at rates lower than $\lambda_{\rm min}= 0.25$ {\rm yr}$^{-1}$

1476: and still conform with the measured aggregate waiting-time distribution.

1477: 

1478: Once verified, the claimed Poissonian nature of the glitch mechanism can be

1479: invoked to exclude broad classes of glitch theories,

1480: e.g.\ those that rely on `defects' or `turbulence' at special locations

1481: (like the pole), 

1482: or that involve a pair of dependent events

1483: (A.\ Martin, private communication).

1484: It is important to interpret aftershocks carefully in this light

1485: \citep{won01}.

1486: In self-organized critical systems,

1487: the excess number of avalanches following a large avalanche

1488: (over and above the Poissonian baseline following a small avalanche)

1489: scales inversely with the time elapsed, a property known as 

1490: Omori's law for earthquakes

1491: \citep{jen98}.

1492: 

1493: In this paper, we do not analyze post-glitch relaxation times 

1494: and glitch-activated changes in $\dot{\nu}$ in the context of

1495: avalanche processes,

1496: e.g.\ the correlation between $\Delta\dot{\nu}$ and the transient

1497: component of $\Delta\nu$

1498: \citep{won01}.

1499: We also assume implicitly that the quantized superfluid vortices 

1500: in the vortex unpinning model are organized in a rectilinear array,

1501: even though recent work suggests that meridional circulation

1502: destabilizes the array and converts it into a turbulent tangle

1503: \citep{per05,per06}. Further study of

1504: these matters is deferred to future work.

1505: 

1506: We acknowledge the computer time and system support

1507: supplied by the Australian Partnership for

1508: Advanced Computation (APAC) and the

1509: Victorian Partnership for

1510: Advanced Computation (VPAC).

1511: We thank Andre Trosky for illuminating conversations on

1512: self-organized criticality and cellular automata.

1513: This research was supported by

1514: a postgraduate scholarship from the University of Melbourne.

1515: It makes use of the Australia Telescope National Facility 

1516: Pulsar Catalogue

1517: \citep{man05}, 

1518: which can be accessed on-line at

1519: {\tt http://www.atnf.csiro.au/research/pulsar/psrcat}.

1520: 

1521: %\bibliographystyle{apj}

1522: %\bibliography{glitchstat}

1523: 

1524: \begin{thebibliography}{71}

1525: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi

1526: 

1527: \bibitem[{{Alpar} {et~al.}(1996){Alpar}, {Chau}, {Cheng}, \& {Pines}}]{alp96}

1528: {Alpar}, M.~A., {Chau}, H.~F., {Cheng}, K.~S., \& {Pines}, D. 1996, \apj, 459,

1529:   706

1530: 

1531: \bibitem[{{Anderson} \& {Itoh}(1975)}]{and75}

1532: {Anderson}, P.~W., \& {Itoh}, N. 1975, \nat, 256, 25

1533: 

1534: \bibitem[{{Andersson} {et~al.}(2007){Andersson}, {Sidery}, \& {Comer}}]{and07}

1535: {Andersson}, N., {Sidery}, T., \& {Comer}, G.~L. 2007, astro-ph/0703257

1536: 

1537: \bibitem[{{Bak} {et~al.}(1987){Bak}, {Tang}, \& {Wiesenfeld}}]{bak87}

1538: {Bak}, P., {Tang}, C., \& {Wiesenfeld}, K. 1987, Physical Review Letters, 59,

1539:   381

1540: 

1541: \bibitem[{{Camilo} {et~al.}(2000){Camilo}, {Kaspi}, {Lyne}, {Manchester},

1542:   {Bell}, {D'Amico}, {McKay}, \& {Crawford}}]{Camilo00}

1543: {Camilo}, F., {Kaspi}, V.~M., {Lyne}, A.~G., {Manchester}, R.~N., {Bell},

1544:   J.~F., {D'Amico}, N., {McKay}, N.~P.~F., \& {Crawford}, F. 2000, \apj, 541,

1545:   367

1546: 

1547: \bibitem[{{Camilo} {et~al.}(2004){Camilo}, {Manchester}, {Lyne}, {Gaensler},

1548:   {Possenti}, {D'Amico}, {Stairs}, {Faulkner}, {Kramer}, {Lorimer},

1549:   {McLaughlin}, \& {Hobbs}}]{camilo04}

1550: {Camilo}, F., {Manchester}, R.~N., {Lyne}, A.~G., {Gaensler}, B.~M.,

1551:   {Possenti}, A., {D'Amico}, N., {Stairs}, I.~H., {Faulkner}, A.~J., {Kramer},

1552:   M., {Lorimer}, D.~R., {McLaughlin}, M.~A., \& {Hobbs}, G. 2004, \apjl, 611,

1553:   L25

1554: 

1555: \bibitem[{{Carroll}(1998)}]{car98}

1556: {Carroll}, B. 1998, APS Meeting Abstracts, 302

1557: 

1558: \bibitem[{{Cognard} \& {Backer}(2004)}]{cognard04}

1559: {Cognard}, I., \& {Backer}, D.~C. 2004, \apjl, 612, L125

1560: 

1561: \bibitem[{{Connors} \& {Carrami{\~n}ana}(2003)}]{con03}

1562: {Connors}, A., \& {Carrami{\~n}ana}, A. 2003, {Finding gamma-ray pulsars with

1563:   sparse Bayes Blocks} (Statistical Challenges in Astronomy), 403--408

1564: 

1565: \bibitem[{{Cordes} \& {Downs}(1985)}]{cor85}

1566: {Cordes}, J.~M., \& {Downs}, G.~S. 1985, \apjs, 59, 343

1567: 

1568: \bibitem[{{Cordes} {et~al.}(1988){Cordes}, {Downs}, \&

1569:   {Krause-Polstorff}}]{cordes88}

1570: {Cordes}, J.~M., {Downs}, G.~S., \& {Krause-Polstorff}, J. 1988, \apj, 330, 847

1571: 

1572: \bibitem[{{D'Alessandro} {et~al.}(1995){D'Alessandro}, {McCulloch}, {Hamilton},

1573:   \& {Deshpande}}]{dal95}

1574: {D'Alessandro}, F., {McCulloch}, P.~M., {Hamilton}, P.~A., \& {Deshpande},

1575:   A.~A. 1995, \mnras, 277, 1033

1576: 

1577: \bibitem[{{Dall'Osso} {et~al.}(2003){Dall'Osso}, {Israel}, {Stella},

1578:   {Possenti}, \& {Perozzi}}]{dal03}

1579: {Dall'Osso}, S., {Israel}, G.~L., {Stella}, L., {Possenti}, A., \& {Perozzi},

1580:   E. 2003, \apj, 599, 485

1581: 

1582: \bibitem[{{Dodson} {et~al.}(2004){Dodson}, {Buchner}, {Reid}, {Lewis}, \&

1583:   {Flanagan}}]{dbrlf04}

1584: {Dodson}, R., {Buchner}, S., {Reid}, B., {Lewis}, D., \& {Flanagan}, C. 2004,

1585:   I.A.U. Circ., 8370, 4

1586: 

1587: \bibitem[{{Dodson} {et~al.}(2002){Dodson}, {McCulloch}, \& {Lewis}}]{dml02}

1588: {Dodson}, R.~G., {McCulloch}, P.~M., \& {Lewis}, D.~R. 2002, \apjl, 564, L85

1589: 

1590: \bibitem[{{Downs}(1981)}]{downs81a}

1591: {Downs}, G.~S. 1981, \apj, 249, 687

1592: 

1593: \bibitem[{{Field} {et~al.}(1995){Field}, {Witt}, {Nori}, \& {Ling}}]{fie95}

1594: {Field}, S., {Witt}, J., {Nori}, F., \& {Ling}, X. 1995, \prl, 74, 1206

1595: 

1596: \bibitem[{{Flanagan} \& {Buchner}(2006)}]{fb06}

1597: {Flanagan}, C.~S., \& {Buchner}, S.~J. 2006, Central Bureau Electronic

1598:   Telegrams, 595, 1

1599: 

1600: \bibitem[{{G{\"o}{\u g}{\"u}{\c s}} {et~al.}(2000){G{\"o}{\u g}{\"u}{\c s}},

1601:   {Woods}, {Kouveliotou}, {van Paradijs}, {Briggs}, {Duncan}, \&

1602:   {Thompson}}]{gwk00}

1603: {G{\"o}{\u g}{\"u}{\c s}}, E., {Woods}, P.~M., {Kouveliotou}, C., {van

1604:   Paradijs}, J., {Briggs}, M.~S., {Duncan}, R.~C., \& {Thompson}, C. 2000,

1605:   \apjl, 532, L121

1606: 

1607: \bibitem[{{Hessels} {et~al.}(2004){Hessels}, {Roberts}, {Ransom}, {Kaspi},

1608:   {Romani}, {Ng}, {Freire}, \& {Gaensler}}]{hessels04}

1609: {Hessels}, J.~W.~T., {Roberts}, M.~S.~E., {Ransom}, S.~M., {Kaspi}, V.~M.,

1610:   {Romani}, R.~W., {Ng}, C.-Y., {Freire}, P.~C.~C., \& {Gaensler}, B.~M. 2004,

1611:   \apj, 612, 389

1612: 

1613: \bibitem[{{Hobbs}(2002)}]{hob02}

1614: {Hobbs}, G. 2002, PhD thesis, University of Manchester

1615: 

1616: \bibitem[{{Hobbs} {et~al.}(2004{\natexlab{a}}){Hobbs}, {Faulkner}, {Stairs},

1617:   {Camilo}, {Manchester}, {Lyne}, {Kramer}, {D'Amico}, {Kaspi}, {Possenti},

1618:   {McLaughlin}, {Lorimer}, {Burgay}, {Joshi}, \& {Crawford}}]{psurv_2004}

1619: {Hobbs}, G., {Faulkner}, A., {Stairs}, I.~H., {Camilo}, F., {Manchester},

1620:   R.~N., {Lyne}, A.~G., {Kramer}, M., {D'Amico}, N., {Kaspi}, V.~M.,

1621:   {Possenti}, A., {McLaughlin}, M.~A., {Lorimer}, D.~R., {Burgay}, M., {Joshi},

1622:   B.~C., \& {Crawford}, F. 2004{\natexlab{a}}, \mnras, 352, 1439

1623: 

1624: \bibitem[{{Hobbs} {et~al.}(2002){Hobbs}, {Lyne}, {Joshi}, {Kramer}, {Stairs},

1625:   {Camilo}, {Manchester}, {D'Amico}, {Possenti}, \& {Kaspi}}]{hobbs02}

1626: {Hobbs}, G., {Lyne}, A.~G., {Joshi}, B.~C., {Kramer}, M., {Stairs}, I.~H.,

1627:   {Camilo}, F., {Manchester}, R.~N., {D'Amico}, N., {Possenti}, A., \& {Kaspi},

1628:   V.~M. 2002, \mnras, 333, L7

1629: 

1630: \bibitem[{{Hobbs} {et~al.}(2004{\natexlab{b}}){Hobbs}, {Lyne}, {Kramer},

1631:   {Martin}, \& {Jordan}}]{hobbs04}

1632: {Hobbs}, G., {Lyne}, A.~G., {Kramer}, M., {Martin}, C.~E., \& {Jordan}, C.

1633:   2004{\natexlab{b}}, \mnras, 353, 1311

1634: 

1635: \bibitem[{{Janssen} \& {Stappers}(2006)}]{jan06}

1636: {Janssen}, G.~H., \& {Stappers}, B.~W. 2006, \aap, 457, 611

1637: 

1638: \bibitem[{{Jensen}(1998)}]{jen98}

1639: {Jensen}, H.~J. 1998, Self-Organized Criticality: Emergent Complex Behavior in

1640:   Physical and Biological Systems (Cambridge University Press, Cambridge UK)

1641: 

1642: \bibitem[{{Jones}(1998)}]{jon98}

1643: {Jones}, P.~B. 1998, \mnras, 296, 217

1644: 

1645: \bibitem[{{Kaspi} \& {Gavriil}(2003)}]{kas03}

1646: {Kaspi}, V.~M., \& {Gavriil}, F.~P. 2003, \apjl, 596, L71

1647: 

1648: \bibitem[{{Kaspi} {et~al.}(2003){Kaspi}, {Gavriil}, {Woods}, {Jensen},

1649:   {Roberts}, \& {Chakrabarty}}]{kaspi03b}

1650: {Kaspi}, V.~M., {Gavriil}, F.~P., {Woods}, P.~M., {Jensen}, J.~B., {Roberts},

1651:   M.~S.~E., \& {Chakrabarty}, D. 2003, \apjl, 588, L93

1652: 

1653: \bibitem[{{Kaspi} {et~al.}(2000){Kaspi}, {Lackey}, \& {Chakrabarty}}]{kaspi00}

1654: {Kaspi}, V.~M., {Lackey}, J.~R., \& {Chakrabarty}, D. 2000, \apjl, 537, L31

1655: 

1656: \bibitem[{{Kramer} \& {Lyne}(2005)}]{kra05}

1657: {Kramer}, M., \& {Lyne}, A. 2005, private communication

1658: 

1659: \bibitem[{{Krawczyk} {et~al.}(2003){Krawczyk}, {Lyne}, {Gil}, \&

1660:   {Joshi}}]{kra03}

1661: {Krawczyk}, A., {Lyne}, A.~G., {Gil}, J.~A., \& {Joshi}, B.~C. 2003, \mnras,

1662:   340, 1087

1663: 

1664: \bibitem[{{Lewis}(2005)}]{lew05}

1665: {Lewis}, D. 2005, private communication

1666: 

1667: \bibitem[{{Link} \& {Epstein}(1996)}]{lin96}

1668: {Link}, B., \& {Epstein}, R.~I. 1996, \apj, 457, 844

1669: 

1670: \bibitem[{{Link} {et~al.}(1993){Link}, {Epstein}, \& {Baym}}]{lin93}

1671: {Link}, B., {Epstein}, R.~I., \& {Baym}, G. 1993, \apj, 403, 285

1672: 

1673: \bibitem[{{Lohsen}(1975)}]{lohsen75}

1674: {Lohsen}, E. 1975, \nat, 258, 688

1675: 

1676: \bibitem[{{Lohsen}(1981)}]{lohsen81}

1677: {Lohsen}, E.~H.~G. 1981, \aaps, 44, 1

1678: 

1679: \bibitem[{{Lu} \& {Hamilton}(1991)}]{lu91}

1680: {Lu}, E.~T., \& {Hamilton}, R.~J. 1991, \apjl, 380, L89

1681: 

1682: \bibitem[{{Lyne} {et~al.}(1996){Lyne}, {Pritchard}, {Graham-Smith}, \&

1683:   {Camilo}}]{lyn96}

1684: {Lyne}, A.~G., {Pritchard}, R.~S., {Graham-Smith}, F., \& {Camilo}, F. 1996,

1685:   \nat, 381, 497

1686: 

1687: \bibitem[{{Lyne} {et~al.}(1995){Lyne}, {Pritchard}, \& {Shemar}}]{lyn95}

1688: {Lyne}, A.~G., {Pritchard}, R.~S., \& {Shemar}, S.~L. 1995, J. Astrophys.

1689:   Astron., 16, 179

1690: 

1691: \bibitem[{{Lyne} {et~al.}(2000){Lyne}, {Shemar}, \& {Smith}}]{lyn00}

1692: {Lyne}, A.~G., {Shemar}, S.~L., \& {Smith}, F.~G. 2000, \mnras, 315, 534

1693: 

1694: \bibitem[{{Manchester} {et~al.}(2005){Manchester}, {Hobbs}, {Teoh}, \&

1695:   {Hobbs}}]{man05}

1696: {Manchester}, R.~N., {Hobbs}, G.~B., {Teoh}, A., \& {Hobbs}, M. 2005, \aj, 129,

1697:   1993

1698: 

1699: \bibitem[{{Manchester} \& {Taylor}(1974)}]{manchester74}

1700: {Manchester}, R.~N., \& {Taylor}, J.~H. 1974, \apjl, 191, L63+

1701: 

1702: \bibitem[{{Marshall} {et~al.}(2004){Marshall}, {Gotthelf}, {Middleditch},

1703:   {Wang}, \& {Zhang}}]{mar04}

1704: {Marshall}, F.~E., {Gotthelf}, E.~V., {Middleditch}, J., {Wang}, Q.~D., \&

1705:   {Zhang}, W. 2004, \apj, 603, 682

1706: 

1707: \bibitem[{{McCulloch} {et~al.}(1990){McCulloch}, {Hamilton}, {McConnell}, \&

1708:   {King}}]{mcc90}

1709: {McCulloch}, P.~M., {Hamilton}, P.~A., {McConnell}, D., \& {King}, E.~A. 1990,

1710:   \nat, 346, 822

1711: 

1712: \bibitem[{{McKenna} \& {Lyne}(1990)}]{mck90}

1713: {McKenna}, J., \& {Lyne}, A.~G. 1990, \nat, 343, 349

1714: 

1715: \bibitem[{{Melatos} \& {Peralta}(2007)}]{mel07}

1716: {Melatos}, A., \& {Peralta}, C. 2007, \apjl, in press

1717: 

1718: \bibitem[{{Middleditch} {et~al.}(2006){Middleditch}, {Marshall}, {Wang},

1719:   {Gotthelf}, \& {Zhang}}]{mid06}

1720: {Middleditch}, J., {Marshall}, F.~E., {Wang}, Q.~D., {Gotthelf}, E.~V., \&

1721:   {Zhang}, W. 2006, \apj, 652, 1531

1722: 

1723: \bibitem[{{Morii} {et~al.}(2005){Morii}, {Kawai}, \& {Shibazaki}}]{morii05}

1724: {Morii}, M., {Kawai}, N., \& {Shibazaki}, N. 2005, \apj, 622, 544

1725: 

1726: \bibitem[{{Morley} \& {Garc{\'{\i}}a-Pelayo}(1993{\natexlab{a}})}]{mor93a}

1727: {Morley}, P.~D., \& {Garc{\'{\i}}a-Pelayo}, R. 1993{\natexlab{a}}, Europhysics

1728:   Letters, 23, 185

1729: 

1730: \bibitem[{{Morley} \& {Garc{\'{\i}}a-Pelayo}(1993{\natexlab{b}})}]{mor93b}

1731: ---. 1993{\natexlab{b}}, Europhysics Letters, 24, 235

1732: 

1733: \bibitem[{{Morley} \& {Schmidt}(1996)}]{mor96}

1734: {Morley}, P.~D., \& {Schmidt}, I. 1996, Europhys. Lett., 33, 105

1735: 

1736: \bibitem[{{Olami} {et~al.}(1992){Olami}, {Feder}, \& {Christensen}}]{ola92}

1737: {Olami}, Z., {Feder}, H.~J.~S., \& {Christensen}, K. 1992, \prl, 68, 1244

1738: 

1739: \bibitem[{{Peralta} {et~al.}(2005){Peralta}, {Melatos}, {Giacobello}, \&

1740:   {Ooi}}]{per05}

1741: {Peralta}, C., {Melatos}, A., {Giacobello}, M., \& {Ooi}, A. 2005, \apj, 635,

1742:   1224

1743: 

1744: \bibitem[{{Peralta} {et~al.}(2006){Peralta}, {Melatos}, {Giacobello}, \&

1745:   {Ooi}}]{per06}

1746: ---. 2006, \apj, 651, 1079

1747: 

1748: \bibitem[{{Peralta}(2006)}]{peralta06}

1749: {Peralta}, C.~A. 2006, PhD thesis, University of Melbourne

1750: 

1751: \bibitem[{{Pietronero} {et~al.}(1994){Pietronero}, {Vespignani}, \&

1752:   {Zapperi}}]{pie94}

1753: {Pietronero}, L., {Vespignani}, A., \& {Zapperi}, S. 1994, Physical Review

1754:   Letters, 72, 1690

1755: 

1756: \bibitem[{{Press} {et~al.}(1986){Press}, {Flannery}, \& {Teukolsky}}]{pre86}

1757: {Press}, W.~H., {Flannery}, B.~P., \& {Teukolsky}, S.~A. 1986, {Numerical

1758:   recipes. The art of scientific computing} (Cambridge: University Press, 1986)

1759: 

1760: \bibitem[{{Rosendahl} {et~al.}(1993){Rosendahl}, {Veki{\'c}}, \&

1761:   {Kelley}}]{ros93}

1762: {Rosendahl}, J., {Veki{\'c}}, M., \& {Kelley}, J. 1993, \pre, 47, 1401

1763: 

1764: \bibitem[{{Scargle}(1998)}]{sca98}

1765: {Scargle}, J.~D. 1998, \apj, 504, 405

1766: 

1767: \bibitem[{{Shabanova}(1998)}]{shabanova98}

1768: {Shabanova}, T.~V. 1998, \aap, 337, 723

1769: 

1770: \bibitem[{{Shabanova}(2005)}]{shabanova05}

1771: ---. 2005, \mnras, 356, 1435

1772: 

1773: \bibitem[{{Shemar} \& {Lyne}(1996)}]{she96}

1774: {Shemar}, S.~L., \& {Lyne}, A.~G. 1996, \mnras, 282, 677

1775: 

1776: \bibitem[{{Sornette} {et~al.}(1991){Sornette}, {Sornette}, \&

1777:   {Vanneste}}]{sor91}

1778: {Sornette}, D., {Sornette}, A., \& {Vanneste}, C. 1991, in Lecture Notes in

1779:   Physics, Berlin Springer Verlag, Vol. 392, Large Scale Structures in

1780:   Nonlinear Physics, ed. J.-D. {Fournier} \& P.-L. {Sulem}, 275--277

1781: 

1782: \bibitem[{{Torii} {et~al.}(2000){Torii}, {Gotthelf}, {Vasisht}, {Dotani}, \&

1783:   {Kinugasa}}]{torii00}

1784: {Torii}, K., {Gotthelf}, E.~V., {Vasisht}, G., {Dotani}, T., \& {Kinugasa}, K.

1785:   2000, \apjl, 534, L71

1786: 

1787: \bibitem[{{Urama} \& {Okeke}(1999)}]{ura99}

1788: {Urama}, J.~O., \& {Okeke}, P.~N. 1999, \mnras, 310, 313

1789: 

1790: \bibitem[{{Wang} {et~al.}(2000){Wang}, {Manchester}, {Pace}, {Bailes}, {Kaspi},

1791:   {Stappers}, \& {Lyne}}]{wan00}

1792: {Wang}, N., {Manchester}, R.~N., {Pace}, R.~T., {Bailes}, M., {Kaspi}, V.~M.,

1793:   {Stappers}, B.~W., \& {Lyne}, A.~G. 2000, \mnras, 317, 843

1794: 

1795: \bibitem[{{Wang} {et~al.}(2001){Wang}, {Wu}, {Manchester}, {Zhang}, {Lyne}, \&

1796:   {Yusup}}]{wwmzly01}

1797: {Wang}, N., {Wu}, X.-J., {Manchester}, R.~N., {Zhang}, J., {Lyne}, A.~G., \&

1798:   {Yusup}, A. 2001, Chinese J. Astron. Astrophys., 1, 195

1799: 

1800: \bibitem[{{Wheatland}(2000)}]{whe00}

1801: {Wheatland}, M.~S. 2000, \apjl, 536, L109

1802: 

1803: \bibitem[{{Wong} {et~al.}(2001){Wong}, {Backer}, \& {Lyne}}]{won01}

1804: {Wong}, T., {Backer}, D.~C., \& {Lyne}, A.~G. 2001, \apj, 548, 447

1805: 

1806: \bibitem[{{Zhang} {et~al.}(2001){Zhang}, {Marshall}, {Gotthelf}, {Middleditch},

1807:   \& {Wang}}]{zhang01}

1808: {Zhang}, W., {Marshall}, F.~E., {Gotthelf}, E.~V., {Middleditch}, J., \&

1809:   {Wang}, Q.~D. 2001, \apjl, 554, L177

1810: 

1811: \end{thebibliography}

1812: 

1813: 

1814: \newpage

1815: 

1816: \renewcommand{\thefootnote}{\alph{footnote}}

1817: 

1818: \begin{deluxetable}{cccclcc}

1819: %\rotate

1820: %\tabletypesize{\scriptsize}

1821: \tablewidth{0pt}

1822: \tablecaption{Parameters of pulsar glitches}

1823: \tablehead{PSR J & $ N_{\rm g}$ &	Epoch	& $\Delta \nu/\nu$&  Ref.\\

1824: &  & (MJD) & $(10^{-9})$ & }

1825: %\hline\\[-1mm]

1826: %\endfoot

1827: 

1828: %Now the data...

1829: \startdata

1830: 0142+61 \phantom{12}&		1	&	51141	&	650	&	6 \\ \tableline

1831: 0157+6212	&	1	&	48504	&	2.46	&	6 \\ \tableline

1832: 0358+5413	&   	6	&	46077(2)	&	5.5(1)	&	1 \\ 

1833: 		&   	&	46497(4)	&	4368(1)	&	1 \\

1834: 	&   	&	51673(15)	&	0.04(2)	&	31 \\

1835: 	&   		&	51965(14)	&	0.030(2)	&	31 \\

1836: 	&   	&	52941(9)	&	0.04(1)	&	31 \\

1837: 	&   	&	53216(11)	&	0.10(2)	&	31 \\ \tableline

1838: 0528+2200	&		3	&	42057	&	1.3	&	1 \\

1839: 	&			&	52289	&	1.46	&	31 \\

1840: 	&			&	53379	&	0.17	&	31 \\ \tableline

1841: 0534+2200	&    	26	&	40493.4(1)	&	4(2)	&	1 \\

1842: 	&			&	41163(1)	&	2.2(1)	&	24 \\

1843: 	&			&	41250(1)	&	2(1)	&	23 \\

1844: 	&			&	42448(1)	&	44.0(6)	&	1 \\

1845: 	&			&	43023(1)	&	1.1(1)	&	24 \\

1846: 	&			&	43768(1)	&	2.8(1)	&	24 \\

1847: 	&			&	46664.42(5)	&	4.1(1)	&	1 \\

1848: 	&			&	47768.4(2)	&	85.0(4)	&	1 \\

1849: 	&			&	48945.5(2)	&	4.5(7)	&	1 \\

1850: 	&			&	50020.6(3)	&	2.7(7)	&	4 \\

1851: 	&			&	50259.93(2)	&	22(1)	&	4 \\

1852: 	&			&	50459.1(5)	&	7.67(3)	&	4 \\

1853: 	&			&	50489(2)	&	6.67	&	4 \\

1854: 	&			&	50812.9(1)	&	8.67(2)	&	4 \\

1855: 	&			&	51452.3(1)	&	9.67(2)	&	4 \\

1856: 	&			&	51741(5)	&	24(5)	&	25 \\

1857: 	&			&	51805.03(3)	&	3.3(2)	&	29 \\

1858: 	&			&	52083.969(2)	&	23.6(6)	&	29 \\

1859: 	&			&	52146.757(9)	&	8(1)	&	29 \\

1860: 	&			&	52498.22(6)	&	2.6(2)	&	29 \\

1861: 	&			&	52587.84(3)	&	1.1(2)	&	29 \\

1862: 	&			&	53067.059(1)	&	210(1)	&	29 \\

1863: 	&			&	53254.039(1)	&	4.84(8)	&	29 \\

1864: 	&			&	53331(1)	&	n/a	&	29 \\

1865: 	&			&	53463.72(3)	&	n/a	&	29 \\

1866: 	&			&	53476.7(8)	&	n/a	&	29 \\ \tableline

1867: 0537$-$6910	&   	23	&	51285(8.6)	&	681(65)	&	33 \\

1868: 	&			&	51568(6.8)	&	449(8)	&	33 \\

1869: 	&			&	51711(6.7)	&	315(9)	&	33 \\

1870: 	&			&	51826(7.1)	&	140(7)	&	33 \\

1871: 	&			&	51880(5.5)	&	141(20)	&	33 \\

1872: 	&			&	51959(4.9)	&	456(46)	&	33 \\

1873: 	&			&	52171(8.3)	&	185(6)	&	33 \\

1874: 	&			&	52242(7.8)	&	427(6)	&	33 \\

1875: 	&			&	52386(5.7)	&	168(20)	&	33 \\

1876: 	&			&	52453(6.9)	&	217(30)	&	33 \\

1877: 	&			&	52546(6.2)	&	421(18)	&	33 \\

1878: 	&			&	52740(5.3)	&	144(6)	&	33\\

1879: 	&			&	52819(3.6)	&	256(16)	&	33 \\

1880: 	&			&	52887(4.5)	&	234(23)	&	33 \\

1881: 	&			&	53014(9.5)	&	338(10)	&	33 \\

1882: 	&			&	53125(2.8)	&	18(14)	&	33 \\

1883: 	&			&	53145(2.7)	&	392(8)	&	33 \\

1884: 	&			&	53288(2.4)	&	395(10	&	33 \\

1885: 	&			&	53446(1.7)	&	259(16)	&	33 \\

1886: 	&			&	53551(4.4)	&	322(26)	&	33 \\

1887: 	&			&	53699(3.9)	&	402(8)	&	33 \\

1888: 	&			&	53860(1.5)	&	236(20)	&	33 \\

1889: 	&			&	53951(1.5)	&	18(20)	&	33 \\ \tableline

1890: 0540$-$6919			&	1	&	51325	&	1.9	&	8 \\ \tableline

1891: 0601$-$0527	&	1	&	51662	&	0.19	&	28,29 \\ \tableline

1892: 0614+2229	&			1	&	51339	&	n/a	&	28,29 \\ \tableline

1893: 0631+1036	&     	9	&	50185.711(6)	&	5.1(1)	&	28,29 \\

1894: 	&			&	50479.74(7)	&	3.7(2)	&	29 \\

1895: 	&			&	50608.246(2)	&	57.9(3)	&	29 \\

1896: 	&			&	50730(2)	&	1662.8(4)	&	29 \\

1897: 	&			&	51911.133(8)	&	1.33(8)	&	29 \\

1898: 	&			&	52852.586(2)	&	17.4(2)	&	29 \\

1899: 	&			&	53228.387(2)	&	1.9(1)	&	29 \\

1900: 	&			&	53359.27(1)	&	1.9(3)	&	29 \\ 

1901: 	&			&	53621(2)	&	n/a	&	29 \\ \tableline

1902: 0659+1414			&	2	&	50185	&	0.39	&	28,29 \\ 

1903: 	&			&	51039	&	1.4	&	29 \\ \tableline

1904: 0729$-$1448	&		1	&	52149.6	&	31	&	28,29 \\ \tableline

1905: 0742$-$2822	&		5	&	n/a	&	n/a	&	28 \\

1906: 	&				&	51770	&	1	&	31 \\

1907: 	&				&	52027	&	2.1	&	31 \\

1908: 	&				&	53090.2	&	2.9	&	31 \\

1909: 	&				&	53469.7	&	1.1	&	31 \\ \tableline

1910: 0745$-$5353	&		1	&	n/a	&	n/a	&	 \\ \tableline

1911: 0758$-$1528	&		1	&	49948	&	0.13	&	28,29 \\ \tableline

1912: 0826+2637	&		1	&	n/a	&	n/a	&	28 \\ \tableline

1913: 0835$-$4510	&     	17	&	40280(4)	&	2340(10)	&	1 \\

1914: 	&			&	41192(8)	&	2050(30)	&	1 \\

1915: 	&			&	41312(4)	&	12	&	26 \\

1916: 	&			&	42683(3)	&	1990(10)	&	1 \\

1917: 	&			&	43693(12)	&	3060(60)	&	1 \\

1918: 	&			&	44888.0707(4)	&	1145(3)	&	1 \\

1919: 	&			&	45192(5)	&	2050(10)	&	1 \\

1920: 	&			&	46257.2284(2)	&	1601(1)	&	1 \\

1921: 	&			&	47519.803(8)	&	1807.1(8)	&	1 \\

1922: 	&			&	48457.382(10)	&	2715(2)	&	1 \\

1923: 	&			&	48550(1)	&	5.6	&	3 \\

1924: 	&			&	49559.057(2)	&	835(2)	&	1 \\

1925: 	&			&	49591.158(2)	&	199(2)	&	1 \\

1926: 	&			&	50369.345(2)	&	2110(17)	&	3 \\

1927: 	&			&	51559.345(5)	&	3120	&	27 \\

1928: 	&			&	53195.09(5)	&	2100	&	30 \\

1929: 	&			&	53959.93(2)	&	2620	&	34 \\ \tableline

1930: 1016$-$5857	&	1	&	52550	&	n/a	&	29,32 \\ \tableline

1931: 1048$-$5832		&	4	&	48944	&	19	&	3 \\

1932: 	&			&	49034	&	3000	&	3 \\

1933: 	&			&	50788	&	769	&	3 \\

1934: 	&			&	52733	&	n/a	&	29 \\ \tableline

1935: 1105$-$6107	&		2	&	50417	&	279.7	&	3 \\

1936: 	&				&	50610	&	2.1	&	3 \\ \tableline

1937: 1112$-$6103	&		1	&	51513	&	n/a	&	29 \\ \tableline

1938: 1119$-$6127	&		2	&	51398	&	4.4	&	9 \\

1939: 	&			&	53300	&	100	&	29 \\ \tableline

1940: 1123$-$6259		&	1	&	49705.87	&	749.31	&	3 \\ \tableline

1941: 1141$-$3322		&	1	&	50551	&	0.7	&	28,29 \\ \tableline

1942: 1302$-$6350		&	1	&	50690.7	&	3.2	&	10 \\ \tableline

1943: 1328$-$4357		&	1	&	43590	&	1.16	&	1 \\ \tableline

1944: 1341$-$6220		&	12	&	47989(24)	&	1507(1)	&	3 \\

1945: 	&		&	48453(12)	&	24.2(9)	&	3 \\

1946: 	&		&	48645(10)	&	990(3)	&	3 \\

1947: 	&		&	49134(22)	&	10(2)	&	3 \\

1948: 	&		&	49363(130)	&	142(21)	&	3 \\

1949: 	&		&	49523(17)	&	33(3)	&	3 \\

1950: 	&		&	49766(2)	&	11(1)	&	3 \\

1951: 	&		&	49904(16)	&	16(7)	&	3 \\

1952: 	&		&	50008(16)	&	1636(13)	&	3 \\

1953: 	&		&	50321.7(6)	&	27(1)	&	3 \\

1954: 	&		&	50528.9(8)	&	20(4)	&	3 \\

1955: 	&		&	50683(13)	&	703(4)	&	3 \\ \tableline

1956: 1357$-$6429		&	1	&	52021	&	2425	&	11 \\ \tableline

1957: 1401$-$6357		&	1	&	48305	&	2.49	&	28,29 \\ \tableline

1958: 1413$-$6141		&	1	&	n/a	&	n/a	&	29 \\ \tableline

1959: 1437$-$6146		&	1	&	51614	&	n/a	&	29,32 \\ \tableline

1960: 1509+5531		&	1	&	41732	&	0.22	&	11 \\ \tableline

1961: 1532+2745		&	1	&	n/a	&	n/a	&	28 \\ \tableline

1962: 1539$-$5626		&	1	&	48165	&	2790.8	&	1 \\ \tableline

1963: 1603$-$2531		&	1	&	n/a	&	n/a	&	 \\ \tableline

1964: 1614$-$5048		&	1	&	49803	&	6460	&	3 \\ \tableline

1965: 1617$-$5055		&	1	&	49960	&	600	&	13 \\ \tableline

1966: 1644$-$4559		&	3	&	43390	&	191	&	1 \\

1967: 	&		&	46453	&	803.6	&	1 \\

1968: 	&		&	47589	&	1.61	&	1 \\ \tableline

1969: 1705$-$1906		&	1	&	48888	&	0.44	&	28,29 \\ \tableline

1970: 1705$-$3423		&	2	&	50060	&	n/a	&	28,29 \\

1971: 	&		&	51940	&	0.6	&	28,29 \\ \tableline

1972: 1708$-$4008		&	2	&	51459	&	620	&	14 \\

1973: 	&		&	52014	&	140	&	15 \\ \tableline

1974: 1709$-$4429	&	1	&	48780	&	2050	&	1 \\ \tableline

1975: 1717$-$3425	&	1	&	49868	&	n/a	&	29 \\ \tableline

1976: 1720$-$1633	&	1	&	n/a	&	n/a	&	 \\ \tableline

1977: 1721$-$3532	&	1	&	49969.7	&	8	&	29 \\ \tableline

1978: 1726$-$3530	&	1	&	n/a	&	n/a	&	29 \\ \tableline

1979: 1730$-$3350	&		2	&	47990	&	3080	&	1 \\

1980: 	&		&	52139	&	3190	&	29 \\ \tableline

1981: 1731$-$4744	&	2	&	49397.3	&	139.2	&	3 \\

1982: 	&			&	50703	&	3.1	&	3 \\ \tableline

1983: 1737$-$3137	&	2	&	51553	&	4	&	29 \\

1984: 	&			&	53052.8	&	236	&	29 \\ \tableline

1985: 1739$-$2903	&	1	&	46956	&	3.09	&	1 \\

1986: 1739$-$3131	&	1	&	n/a	&	n/a	&	28 \\ \tableline

1987: 1740+1311	&	1	&	n/a	&	n/a	&	28 \\ \tableline

1988: 1740$-$3015	&	30	&	47003(50)	&	420(20)	&	1 \\

1989: 	&	&	47281(2)	&	33(5)	&	1 \\

1990: 	&	&	47332(16)	&	7(5)	&	1 \\

1991: 	&	&	47458(2)	&	30(8)	&	1 \\

1992: 	&	&	47670.2(2)	&	600.9(6)	&	1 \\

1993: 	&	&	48149(2)	&	4(2)	&	29 \\

1994: 	&	&	48186(6)	&	642(16)	&	5 \\

1995: 	&	&	48218(2)	&	48(10)	&	5 \\

1996: 	&	&	48431(2)	&	15.7(5)	&	5 \\

1997: 	&	&	49046(4)	&	9.1(2)	&	1 \\

1998: 	&	&	49239(2)	&	169.7(2)	&	1 \\

1999: 	&	&	49451.7(4)	&	9.5(5)	&	2 \\

2000: 	&	&	49543.9(8)	&	3(6)	&	2 \\

2001: 	&	&	50574.5497(4)	&	439.3(2)	&	2 \\

2002: 	&	&	50941.6182(2)	&	1443(3)	&	2 \\

2003: 	&	&	51334(2)	&	1.1(6)	&	29 \\

2004: 	&	&	51685(24)	&	0.7(4)	&	31 \\

2005: 	&	&	51822(7)	&	0.8(3)	&	31 \\

2006: 	&	&	52007(6)	&	0.7(1)	&	31 \\

2007: 	&	&	52235(2)	&	42.1(9)	&	31 \\

2008: 	&	&	52240.2(2)	&	5(1)	&	29 \\

2009: 	&	&	52266.8(2)	&	14.3(7)	&	29 \\

2010: 	&	&	52271(2)	&	444(5)	&	31 \\

2011: 	&	&	52344(2)	&	220.6(9)	&	31 \\

2012: 	&	&	52603(5)	&	1.5(1)	&	31 \\

2013: 	&	&	52759(5)	&	1.6(3)	&	31 \\

2014: 	&	&	52859(2)	&	17.6(3)	&	31 \\

2015: 	&	&	52943.5(2)	&	22.1(4)	&	31 \\

2016: 	&	&	53023.512(2)	&	1850.0(8)	&	29 \\

2017: 	&	&	53741(2)	&	n/a	&	29 \\ \tableline

2018: 1743$-$3150	&	1	&	49553	&	1.6	&	28,29 \\ \tableline

2019: 1755$-$2534	&	1	&	52170	&	n/a	&	29 \\ \tableline

2020: 1759$-$2205	&	1	&	51800	&	28	&	29 \\ \tableline

2021: 1801$-$0357	&	1	&	48016	&	2.9	&	2 \\ \tableline

2022: 1801$-$2304	&	9	&	46907(40)	&	200(30)	&	1 \\

2023: 	&	&	47855(50)	&	231.2(9)	&	1 \\

2024: 	&	&	48454(3)	&	347.68(8)	&	1 \\

2025: 	&	&	49709(32)	&	64(2)	&	5 \\

2026: 	&	&	50055(4)	&	22.6(9)	&	2 \\

2027: 	&	&	50363.414(4)	&	80.6(6)	&	2 \\

2028: 	&	&	50938(2)	&	4(1)	&	29 \\

2029: 	&	&	52126(100)	&	651(3)	&	29 \\

2030: 	&	&	53356(100)	&	499(4)	&	29 \\ \tableline

2031: 1801$-$2451	& 	5	&	49476	&	1988	&	3 \\

2032: 	&	&	50651	&	1247	&	3 \\

2033: 	&	&	52567	&	n/a	&	29 \\

2034: 	&	&	52791	&	n/a	&	29 \\

2035: 	&	&	53030.51	&	16.1	&	29 \\ \tableline

2036: 1803$-$2137	&	4	&	48245	&	4075	&	1 \\

2037: 	&	&	50269.4	&	5.3	&	2 \\

2038: 	&	&	50765	&	3185	&	3 \\

2039: 	&	&	50765	&	27	&	3 \\

2040: 	&	&	53429	&	3943	&	29 \\ \tableline

2041: 1806$-$2125	&	1	&	51063	&	15615	&	16 \\ \tableline

2042: 1809$-$1917	&	1	&	53250	&	1629.1	&	29 \\ \tableline

2043: 1812$-$1718	&	2	&	49926	&	1.6	&	28,29 \\

2044: 	&	&	53105.68	&	14.7	&	28,29 \\ \tableline

2045: 1814$-$1744	&	5	&	51384	&	9	&	29 \\

2046: 	&			&	51700	&	5	&	31 \\

2047: 	&			&	52094.96	&	27	&	29 \\

2048: 	&			&	52117	&	33	&	31 \\

2049: 	&			&	53302	&	7	&	31 \\ \tableline

2050: 1824$-$1118	&	1	&	52402	&	1.3	&	28,29 \\ \tableline

2051: 1824$-$2452	&	1	&	51980	&	0.0095	&	17 \\ \tableline

2052: 1825$-$0935	&	8	&	49615(8)	&	0.2(1)	&	21 \\

2053: 	&			&	49857(8)	&	12.6	&	22 \\

2054: 	&			&	49940(2)	&	5.21(7)	&	21 \\

2055: 	&			&	50557(6)	&	12.6(2)	&	21 \\

2056: 	&			&	51060(8)	&	20	&	22 \\

2057: 	&			&	51879(8)	&	31.4(2)	&	22 \\

2058: 	&			&	52058(2)	&	29(1)	&	29 \\

2059: 	&			&	52802.6(2)	&	1.8(7)	&	29 \\ \tableline

2060: 1826$-$1334	&	3	&	46507	&	2700	&	1 \\

2061: 	&			&	49014	&	3060	&	1 \\

2062: 	&			&	53738	&	n/a	&	29 \\ \tableline

2063: 1827$-$0958	&	1	&	n/a	&	n/a	&	28 \\ \tableline

2064: 1833$-$0559	&	1	&	52200	&	n/a	&	29 \\ \tableline

2065: 1833$-$0827	&	1	&	48041	&	1864.8	&	1 \\ \tableline

2066: 1835$-$1106	&	1	&	52265	&	27	&	28,29 \\ \tableline

2067: 1838$-$0453	&	1	&	52000	&	26	&	29 \\ \tableline

2068: 1841$-$0425	&	1	&	53356	&	578.5	&	29 \\ \tableline

2069: 1844$-$0538	&	2	&	47438	&	0.8	&	29 \\

2070: 	&			&	47955	&	0.5	&	29 \\ \tableline

2071: 1845$-$0316	&	1	&	52212.9	&	30	&	29 \\ \tableline

2072: 1856+0113			&	1	&	n/a	&	n/a	&	28 \\ \tableline

2073: 1901+0716	&			1	&	46859	&	30	&	1 \\ \tableline

2074: 1902+0615	&			4	&	48645.11	&	0.45	&	29 \\

2075: 	&			&	49441	&	0.23	&	29 \\

2076: 	&			&	50311	&	0.31	&	29 \\

2077: 	&			&	51165.9	&	0.47	&	29 \\ \tableline

2078: 1903+0135	&	1	&	48634	&	n/a	&	28,29 \\ \tableline

2079: 1905$-$0056	&	2	&	49385	&	n/a	&	28,29 \\

2080: 	&			&	49695.3	&	0.21	&	28,29 \\ \tableline

2081: 1908+0909	&	2	&	52240	&	11.8	&	29 \\

2082: 	&			&	53340	&	1.7	&	29 \\ \tableline

2083: 1909+0007	&	1	&	49491.9	&	0.72	&	29 \\ \tableline

2084: 1910+0358	&	1	&	52331	&	1.4	&	29 \\ \tableline

2085: 1910$-$0309	&	3	&	48241	&	0.6	&	2 \\

2086: 	&			&	49219.85	&	1.84	&	2 \\

2087: 	&			&	53232.75	&	2.66	&	29 \\ \tableline

2088: 1913+0446	&	1	&	53500	&	n/a	&	29 \\ \tableline

2089: 1918+1444	&	1	&	52285	&	2.2	&	28,29 \\

2090: 1919+0021	&	1	&	50174	&	1.29	&	2 \\ \tableline

2091: 1922+2018	&	1	&	n/a	&	n/a	&	1 \\ \tableline

2092: 1932+2220	&	4	&	46900	&	4450	&	29 \\

2093: 	&			&	50264	&	4457	&	2 \\

2094: 	&			&	52210	&	12	&	29 \\

2095: 	&			&	52394	&	12	&	29 \\ \tableline

2096: 1946+2611	&	1	&	53326	&	70	&	29 \\ \tableline

2097: 1952+3252	&	5	&	n/a	&	n/a	&	28 \\

2098: 	&			&	51967	&	2.25	&	31 \\

2099: 	&			&	52385	&	0.72	&	31 \\

2100: 	&			&	52912	&	1.29	&	31 \\

2101: 	&			&	53305	&	0.51	&	31 \\ \tableline

2102: 2021+3651	&	1	&	52630.07	&	2587	&	18 \\ \tableline

2103: 2040+1657	&	1	&	53142	&	n/a	&	29 \\ \tableline

2104: 2116+1414	&	3	&	47972	&	0.2	&	28,29 \\

2105: 	&			&	49950	&	0.07	&	29 \\

2106: 	&			&	51357	&	0.11	&	29 \\ \tableline

2107: 2225+6535	&	4	&	43072	&	1707	&	1 \\

2108: 	&			&	51900	&	0.14	&	31 \\

2109: 	&			&	52950	&	0.08	&	31 \\

2110: 	&			&	53434	&	0.19	&	31 \\ \tableline

2111: 2229+6114	&	1	&	53064	&	1133	&	29 \\ \tableline

2112: 2257+5909	&	1	&	49463.2	&	0.92	&	2 \\ \tableline

2113: 2301+5852	&	1	&	52443.9	&	4100	&	19 \\ \tableline

2114: 2330$-$2005	&	1	&	n/a	&	n/a	&	1 \\ \tableline

2115: 2337+6151	&	1	&	53639	&	20000	&	29 \\ \tableline

2116: \enddata

2117: \tabletypesize{\normal}

2118: \tablerefs{[1] \citet{lyn00}, [2] \citet{kra03} [3] \citet{wan00}, [4] \citet{won01}, [5] \citet{she96}, [6] \citet{morii05},

2119: [7] \citet{mar04},

2120: [8] \citet{zhang01}, [9] \citet{Camilo00}, [10] \citet{wan00},

2121: [11] \citet{camilo04}, [12] \citet{manchester74}, [13] \citet{torii00}, [14] \citet{kaspi00}, [15] \citet{kas03},

2122: [16] \citet{hobbs02}, [17] \citet{cognard04}, [18] \citet{hessels04},

2123: [19] \citet{kaspi03b}, [20] \citet{cordes88}, [21] \citet{shabanova98}, [22] \citet{shabanova05}, [23] \citet{lohsen75},

2124: [24] \citet{lohsen81}, [25] \citet{wwmzly01}, [26] \citet{downs81a}, [27] \citet{dml02}, [28] \citet{hobbs04}, 

2125: [29] \citet{kra05}, [30] \citet{dbrlf04} [31] \citet{jan06} [32] \citet{psurv_2004} [33] \citet{mid06} [34] \citet{fb06}}

2126: \tablecomments{n/a: not available}

2127: \label{tab:gli1}

2128: \end{deluxetable}

2129: 

2130: 

2131: \begin{deluxetable}{cccclcc}

2132: %\rotate

2133: %\tabletypesize{\scriptsize}

2134: \tablewidth{0pt}

2135: \tablecaption{Observing time intervals for pulsars with $N_{\rm g}>5$.}

2136: \tablehead{PSR J & $t_{\rm min}$ & $t_{\rm max}$\\

2137: & (MJD) &  (MJD) }

2138: 

2139: %Now the data...

2140: \startdata

2141: 0358+5413 &  41807	& 53546	\\

2142: 0534+2200&   \phantom{1} 40466 $\ast$	& \phantom{12}53476 $\ast$ \\

2143: 0537$-$6910& 51197	& 53952 \\

2144: 0631+1036 &  \phantom{1} 50186 $\ast$ & \phantom{12}53621 $\ast$ \\

2145: 0835$-$4510& \phantom{1} 40140 $\ast$ & \phantom{12}53960 $\ast$ \\

2146: 1341$-$6220 & 47915& 51022 \\

2147: 1740$-$3015 &\phantom{1} 46770 $\ast$ & 53190 \\

2148: 1801$-$2304& 46697 & \phantom{12}53356 $\ast$ \\

2149: 1825$-$0935& \phantom{1} 48300 $\ast$& \phantom{12}52803 $\ast$ \\ \tableline

2150: \enddata

2151: \tabletypesize{\normal}

2152: \tablecomments{$\ast$: Segmented data spans are not published.

2153: $t_{\rm min}$ and $t_{\rm max}$ are estimated

2154: by eye from graphed spin-down histories,

2155: where available, or else from the first and last glitches by default.}

2156: \label{tab:gli1b}

2157: \end{deluxetable}

2158: 

2159: 

2160: \begin{table}

2161: \caption{\label{tab:gli2} Power-law size distribution parameters for pulsars

2162: with $N_{\rm g} > 5$.}

2163: \begin{center}

2164: \begin{tabular}{lrrrr} 

2165: %\begin{tabular}{ccccc}

2166: \tableline

2167: \tableline

2168: PSR J & $a_-$ & $a$ & $a_+$ & $P_{\rm KS}$ \\ \tableline

2169: 0358$+$5413 & 1.5 & 2.4 & 5.2 & 0.9913 \\

2170: 0534$+$2200 & 1.1 & 1.2 & 1.4 & 0.86 \\

2171: 0537$-$6910 & 0.39 & 0.42 & 0.43 & 0.36 \\

2172: 0631$+$1036 & 1.2 & 1.8 & 2.7 & 0.99896 \\

2173: 0835$-$4510 & $-0.20$ & $-0.13$ & 0.18 & 0.908 \\

2174: 1341$-$6220 & 1.2 & 1.4 & 2.1 & 0.77 \\

2175: 1740$-$3015 & 0.98 & 1.1 & 1.3 & 0.9920 \\

2176: 1801$-$2304 & 0.092 & 0.57 & 1.1 & 0.99968 \\

2177: 1825$-$0935 & $-0.30$ & 0.36 & 1.0 & 0.99904 \\ 

2178: \tableline

2179: \tableline

2180: \end{tabular}

2181: \end{center}

2182: \end{table}

2183: 

2184: \begin{table}

2185: \caption{\label{tab:gli3}

2186: Two-component size distribution parameters for quasiperiodic glitchers.}

2187: \begin{center}

2188: \begin{tabular}{lrrrrrr}

2189: \tableline

2190: \tableline

2191: PSR J & $a_-$ & $a$ & $a_+$ & 

2192:   $C_{\rm p}$ & $(\Delta\nu/\nu)_{\rm c}$ & $P_{\rm KS}$

2193: \\ \hline

2194: 0537$-$6910 & 0.22 & 0.44 & 0.68 & 

2195:   $0.25\pm0.05$ & $(3.0\pm0.5)\times10^{-7}$ & 0.81

2196: \\

2197: 0835$-$4510 & $-0.49$ & 0.11 & 0.44 &

2198:   $0.15\pm0.05$ & $(2.5\pm0.5)\times10^{-6}$ & 0.970 \\

2199: \tableline

2200: \tableline

2201: \end{tabular}

2202: \end{center}

2203: 

2204: \end{table}

2205: 

2206: \begin{deluxetable}{crrrrrrr}

2207: \rotate

2208: %\tabletypesize{\scriptsize}

2209: \tablewidth{0pt}

2210: \tablecaption{ \label{tab:gli4}

2211: Poissonian waiting-time distribution parameters for pulsars

2212: with $N_{\rm g} > 5$.}

2213: \tablehead{

2214: PSR J & $\Delta t_{\rm min}^{(<)}$ & $\Delta t_{\rm min}^{(>)}$

2215:   & $\Delta t_{\rm max}$

2216:   & $\lambda_-$ & $\lambda$ & $\lambda_+$ & $P_{\rm KS}$

2217: \\

2218:   & (d) & (d) & (d)

2219:   & $({\rm yr^{-1}})$ & $({\rm yr^{-1}})$ & $({\rm yr^{-1}})$ }

2220: \startdata

2221: 0358$+$5413 & 4 & 30 & 11739 & 0.21 & 0.57 & 1.3 & 0.999960

2222: \\

2223: 0534$+$2200 & 2 & 18 & 13010 & 0.57 & 0.91 & 1.3 & 0.982

2224: \\

2225: 0537$-$6910 & 3 & 19 & 2755 & n/a & 2.6 & n/a & 0.31

2226: \\

2227: 0631$+$1036 & 2 & 16 & 3435 & 0.55 & 0.95 & 1.9 & 0.9970

2228: \\

2229: 0835$-$4510 & 2 & 24 & 13820 & 0.33 & 0.35 & 0.42 & 0.45

2230: \\

2231: 1341$-$6220 & 4 & 260 & 3107 & 1.2 & 1.8 & 5.6 & 0.980

2232: \\

2233: 1740$-$3015 & 4 & 100 & 6330 & 1.2 & 1.5 & 2.5 & 0.928

2234: \\

2235: 1801$-$2304 & 4 & 200 & 6659 & 0.35 & 0.55 & 0.88 & 0.962

2236: \\

2237: 1825$-$0935 & 4 & 16 & 4503 & 0.48 & 0.91 & 1.8 & 0.9989

2238: \\ \tableline

2239: \enddata

2240: \tablecomments{n/a: not applicable, as the best fit yields $P_{\rm KS} < 0.32$.}

2241: \end{deluxetable}

2242: 

2243: \begin{deluxetable}{lrrrrrrrccc}

2244: \rotate

2245: %\tabletypesize{\scriptsize}

2246: \tablewidth{0pt}

2247: %\begin{table}

2248: \tablecaption{\label{tab:gli5}

2249: Two-component waiting-time distribution parameters for quasiperiodic glitchers.}

2250: \tablehead{PSR J & $\lambda_-$ & $\lambda$ & $\lambda_+$

2251:   & $C_{\rm p}'$ & $\Delta t_{\rm c}$ & $P_{\rm KS}$ 

2252: \\

2253:  & $({\rm yr^{-1}})$ & $({\rm yr^{-1}})$ & $({\rm yr^{-1}})$ & 

2254:  & (yr) &} 

2255: \startdata

2256: 0537$-$6910 & 2.0 & 2.6 & 3.3 & 

2257:  $0.25\pm0.05$ & $0.3\pm0.1$ & 0.80 

2258: \\

2259: 0835$-$4510 & 0.27 & 0.43 & 0.62 & 

2260:  $0.25\pm0.05$ & $2.8\pm0.1$ & 0.968

2261: \\ \tableline

2262: \enddata

2263: \end{deluxetable}

2264: 

2265: 

2266: \newpage

2267: \begin{figure}

2268: \includegraphics[scale=0.9]{f1.ps}

2269: \caption{Cumulative distribution of fractional glitch sizes $\Delta\nu/\nu$ in

2270: the nine pulsars that have glitched more than five times. The observational data ({\em histogram})

2271: are plotted together with the best power-law fit ({\em solid curve}) given by (\ref{eq:gli2}), with

2272: $(\Delta\nu/\nu)_{\rm min}$ and $(\Delta\nu/\nu)_{\rm max}$ taken from Table \ref{tab:gli1}

2273: and $a$ taken from Table \ref{tab:gli2}.}

2274: \label{fig:gli1}

2275: \end{figure}

2276: 

2277: \begin{figure}

2278: \plottwo{f2a.ps}{f2b.ps}

2279: \caption{Cumulative distribution of fractional glitch sizes $\Delta\nu/\nu$ for 

2280: the two pulsars which have a quasiperiodic component.

2281: The observational data ({\em histogram}) are plotted together with the best two-component fit ({\em solid curve}) given by (\ref{eq:gli3b}), with

2282: $(\Delta\nu/\nu)_{\rm min}$ and $(\Delta\nu/\nu)_{\rm max}$ taken from Table \ref{tab:gli1},

2283: and $a$, $C_p$ and $(\Delta\nu/\nu)_c$ taken from Table \ref{tab:gli3}.

2284: }

2285: \label{fig:gli2}

2286: \end{figure}

2287: 

2288: \begin{figure}

2289: \plottwo{f3a.ps}{f3b.ps}

2290: \caption{Aggregate cumulative distribution of fractional glitch sizes $\Delta\nu/\nu$ 

2291: for all glitching pulsars ({\em left panel}) and for all glitching pulsars

2292: except PSR J0537$-$6910 and PSR J0835$-$4510, which have a quasiperiodic

2293: component ({\em right panel}). The observational data ({\em histogram}) 

2294: are plotted together with the best power-law fit ({\em solid curve})

2295: given by (\ref{eq:gli2}), for the best fit parameters in \S\ref{sec:gli4d}.}

2296: \label{fig:gli3}

2297: \end{figure}

2298: 

2299: \begin{figure}

2300: \includegraphics[scale=0.9]{f4.ps}

2301: \caption{

2302: Cumulative distribution of 

2303: glitch waiting times $\Delta t$ (measured in {\rm yr}) for

2304: the nine pulsars that have glitched more than five times.

2305: The observational data ({\em histogram}) are plotted together with the best Poisson fit ({\em solid curve}) given by 

2306: (\ref{eq:gli7}), with

2307: $\Delta t_{\rm min}$ taken from Table \ref{tab:gli1} (twice

2308: the epochal uncertainty), $\Delta t_{\rm max}$ from Table \ref{tab:gli1b}, and

2309: $\lambda$ from Table \ref{tab:gli5}.}

2310: \label{fig:gli4}

2311: \end{figure}

2312: 

2313: \begin{figure}

2314: \plottwo{f5a.ps}{f5b.ps}

2315: \caption{

2316: Cumulative distribution of

2317: glitch waiting times $\Delta t$ (measured in {\rm yr}) for the two pulsars which have a quasiperiodic component.

2318: The observational data ({\em histogram}) are plotted together with the best two-component fit ({\em solid curve}), given 

2319: by (\ref{eq:gli9}), with $\Delta t_{\rm min}$ taken from Table \ref{tab:gli1}

2320: (twice the epochal uncertainty),  $\Delta t_{\rm max}$ from Table \ref{tab:gli1b},

2321: and $\lambda$, $C_p'$, and $\Delta t_c$ taken from Table \ref{tab:gli1}.  }

2322: \label{fig:gli5}

2323: \end{figure}

2324: 

2325: \begin{figure}

2326: \plottwo{f6a.ps}{f6b.ps}

2327: \caption{Cumulative distribution of mean glitching rate $\lambda$ (measured in {\rm yr}$^{-1}$)

2328: for the nine pulsars that have glitched more than five times

2329: ({\em left panel}) and excluding the two quasiperiodic glitchers ({\em right panel}),

2330: showing the observational data ({\em histogram}) and the best fit obtained from (\ref{eq:gli10}) ({\em solid curve}),

2331: with $\langle \lambda \rangle = 1.3_{-0.6}^{+0.7}\,{\rm yr^{-1}}$ ({\em left panel}) and 

2332: $\langle \lambda \rangle = 1.2_{-0.4}^{+0.5}\,{\rm yr^{-1}}$ ({\em right panel}).}

2333: \label{fig:gli6}

2334: \end{figure}

2335: 

2336: \begin{figure}

2337: \plottwo{f7a.ps}{f7b.ps}

2338: \caption{Aggregate cumulative waiting-time distribution $P_{\rm agg}(\Delta t)$  

2339: for all pulsars with $N_{\rm g} \geq 2$, 

2340: including ({\em left panel}) and excluding ({\em right panel})

2341: the two quasiperiodic glitchers. The histograms represent

2342: the observational data. The

2343: dotted curves are obtained

2344: by evaluating (\ref{eq:gli11}) with (\ref{eq:gli10}), using

2345: $\langle \lambda \rangle =  1.3_{-0.6}^{+0.7}\,{\rm yr^{-1}}$ ({\em left panel})

2346: and $\langle \lambda \rangle =  1.2_{-0.4}^{+0.5}\,{\rm yr^{-1}}$ ({\em right panel})

2347: extracted from Figure \ref{fig:gli6}. The dashed curves are obtained by

2348: evaluating (\ref{eq:gli11}) with (\ref{eq:gli10}) and adjusting $\langle \lambda \rangle$ to maximize

2349: $P_{\rm KS}$ when fitting $P_{\rm agg}(\Delta t)$.

2350: The solid curves are obtained from (\ref{eq:gli12}),

2351: with $\langle \lambda \rangle = 0.54$ {\rm yr}$^{-1}$,

2352: $\lambda_{\rm min}=0.25$ {\rm yr}$^{-1}$ ({\em left panel})

2353: and $\langle \lambda \rangle = 0.43$ {\rm yr}$^{-1}$,

2354: $\lambda_{\rm min}=0.25$ ({\em right panel}).}

2355: \label{fig:gli7}

2356: \end{figure}

2357: 

2358: \begin{figure}

2359: \plottwo{f8a.ps}{f8b.ps}

2360: \caption{Cumulative distribution of mean glitching rate $\lambda$ (measured in {\rm yr}$^{-1}$)

2361: for the nine pulsars that have glitched more than five times

2362: ({\em left panel}) and excluding the two quasiperiodic glitchers ({\em right panel}),

2363: showing the observational data ({\em histogram}) and 

2364: the theoretical rate distribution (\ref{eq:14_new})

2365: corresponding to the dashed and solid curves in Figure \ref{fig:gli7}.}

2366: \label{fig:gli6_redo}

2367: \end{figure}

2368: 

2369: \begin{figure}

2370: \includegraphics[angle=270,scale=0.9]{f8.ps}

2371: \caption{Spin-down age $\tau_c$ (in {\rm kyr}) versus

2372: power-law exponent $a$ for the glitch size distribution.

2373: The error bars indicate the 1-$\sigma$ range of allowable

2374: fits according to the K-S test. Systematic differences between

2375: $\tau_c$ and true age are not quantified here. Solid (open)

2376: triangles symbolize quasiperiodic glitchers with $N_{\rm g} > 5$,

2377: to which we fit a two-component (one-component) $P(\Delta \nu/\nu)$, as in

2378: Table \ref{tab:gli3} (\ref{tab:gli2}). Squares symbolize aperiodic glitchers with $N_{\rm g} > 5$,

2379: to which we fit a power-law $P(\Delta \nu/\nu)$, as in 

2380: Table \ref{tab:gli2}.}

2381: \label{fig:gli9}

2382: \end{figure}

2383: 

2384: \begin{figure}

2385: \begin{center}

2386: \includegraphics[angle=270,scale=0.9]{f9.ps}

2387: \caption{Mean glitching rate $\lambda$ (in {\rm yr}$^{-1}$)

2388: versus spin-down age $\tau_c$ (in {\rm kyr}).

2389: The error bars indicate the 1-$\sigma$ range of allowable

2390: fits according to the K-S test. Systematic differences between

2391: $\tau_c$ and true age are not quantified

2392: here. Solid (open) triangles symbolize quasiperiodic 

2393: glitchers with $N_{\rm g} > 5$,

2394: to which we fit a two-component (one-component) $P(\lambda,\Delta t)$, as in

2395: Table \ref{tab:gli5} (\ref{tab:gli4}). Squares symbolize aperiodic glitchers with $N_{\rm g} > 5$,

2396: to which we fit a Poissonian $P(\lambda,\Delta t)$ as in Table \ref{tab:gli4}.}

2397: \label{fig:gli10}

2398: \end{center}

2399: \end{figure}

2400: 

2401: \end{document}

2402: 

2403: