1: \documentstyle[aps,prl,epsfig,twocolumn,floats]{revtex}

2: %\documentstyle[aps,prl,floats]{revtex}

3: \begin{document}

4: \draft

5: \tighten

6: \twocolumn[\hsize\textwidth\columnwidth\hsize\csname@twocolumnfalse%

7: \endcsname

8: 

9: \title{Galactic Constraints on the Sources of Ultra-High Energy Cosmic

10: Rays}

11: \author{Abraham Loeb$^\star$ \& Eli Waxman$^\dagger$} 

12: \address{$\star$ Astronomy Department, Harvard University, 60 Garden

13: Street, Cambridge, MA 02138, USA; aloeb@cfa.harvard.edu\\ $\dagger$ 

14: Department of Condenced Matter Physics, Weizmann Institute, Rehovot 76100,

15: Israel; waxman@wicc.weizmann.ac.il } 

16: \date{\today} 

17: \maketitle

18: \begin{abstract}

19: We show that if the sources of ultra-high energy cosmic rays (UHECRs) with

20: energies $E\ge 10^{19}$~eV are associated with galaxies, then the

21: production of UHECRs must occur in transient events.  Our galaxy is

22: currently at a dim state in between transients.  The time interval between

23: transients is $\gtrsim 10^{4.5} \min\{1,(\gamma_{\rm

24: min}/10^3)^{-0.6}\}$~yr for $\gamma_{\rm min}\lesssim 10^7$, where

25: $\gamma_{\rm min}$ is the minimum Lorentz factor to which protons are

26: accelerated in the transients. This constraint is satisfied by $\gamma$-ray

27: bursts.

28: 

29: \end{abstract}

30: 

31: \pacs{PACS numbers: 98.70.Sa, 96.40.-z, 98.62.-g}

32: ]

33: 

34: \narrowtext

35: 

36: The origin of the observed cosmic rays at different energies is still

37: unknown (see \cite{Axford94,Nagano00,Halzen02} for recent reviews). As

38: illustrated in Fig. 1, the cosmic ray spectrum changes its qualitative

39: behavior as a function of particle energy; it steepens around $\sim 5\times

40: 10^{15}$~eV (the ``knee'') and flattens around $5\times 10^{18}$~eV (the

41: ``ankle'').  Below $\sim 10^{15}$~eV, the cosmic rays are thought to

42: originate from Galactic supernovae. The composition is dominated by protons

43: at the lowest energies, and the fraction of heavy nuclei increases with

44: energy. The proton fraction at $\sim 10^{15}$~eV is reduced to $\sim15\%$

45: \cite{Burnett90,Bernlohr98}. At yet higher energies, there is evidence that

46: the fraction of light nuclei increases, and that the cosmic-ray flux above

47: $5\times 10^{18}$~eV is again dominated by protons \cite{composition}. This

48: composition change and the flattening of the spectrum around $10^{19}$~eV

49: (see Fig. 2) suggest that the flux above and below this energy is dominated

50: by different sources. Since the Galactic magnetic field can not confine

51: protons above $10^{19}$~eV, it is believed that the nearly isotropic cosmic

52: ray flux at $E>5\times 10^{18}$~eV, originates from extragalactic sources.

53: The small, but statistically significant, enhancement of the flux at

54: $E<3\times 10^{18}$~eV near the Galactic plane \cite{Bird99,Hayashida99},

55: suggests a Galactic origin at these lower energies.

56: 

57: The Milky-Way disk shows prominently relative to the extragalactic

58: background in electromagnetic radiation ranging from radio to X-ray

59: wavelengths. This is a consequence of the fact that our location within the

60: Galaxy is not representative of a random point in the universe, where the

61: observed radiation would be nearly isotropic.  The Galactic prominence must

62: also apply to UHECRs with energies $\gtrsim 10^{19}$ eV. At these energies

63: the Galactic magnetic field is unable to isotropize the particle orbits,

64: and the Galactic disk should be on average brighter than the extragalactic

65: background by orders of magnitude. Contrary to this expectation, the actual

66: anisotropy of the observed UHECR is $\lesssim 4\%$ in the direction of the

67: Galactic disk \cite{Bird99,Hayashida99,Nagano00}.  In this paper, we

68: therefore argue that any generic galactic production of UHECRs must occur

69: in transient events with a short duty cycle.  Using existing data for the

70: spectrum and the confinement time of cosmic rays at lower energies, we set

71: limits on the event rate and the minimum Lorentz factor of the accelerated

72: particles in these transients.

73: \begin{figure}[th]

74: %\centerline{\epsfig{file=Figure1.eps, height=2.0in, clip=}}

75: \centerline{\epsfig{file=fig1.eps, height=2.5in}}

76: \caption{Data points from [2],

77: %\cite{Nagano00}, 

78: showing the differential flux of all cosmic-ray particles multiplied by

79: $E^3$.  The dotted line is a power-law fit to the observed proton flux

80: [3]. The solid line shows the expected volume-averaged, extragalactic

81: cosmic-ray flux based on the observed UHECR flux (see text and Fig. 2). The

82: corresponding minimum, time-averaged, Galactic flux is shown as the dashed

83: line.}

84: \end{figure}

85: 

86: \paragraph*{The extragalactic cosmic-ray flux.}

87: In Fig. 2 we compare the UHECR spectrum measured by the three experiments

88: with the largest current exposure, Yakutsk, Fly's Eye and AGASA, with that

89: expected from a homogeneous cosmological distribution of sources, each

90: generating a power-law, differential number flux of high energy protons,

91: $dn/dE\propto E^{-2}$. This power-law slope is expected for astrophysical

92: sources which accelerate particles in strong collisionless shocks

93: \cite{Blandford87,RelShock}, and is found in supernovae \cite{SN} and

94: $\gamma$-ray bursts \cite{WAG97a}. Experimental differences in the absolute

95: flux calibration at $3\times10^{18}$~eV yield a systematic over-estimation

96: of the event energies by $\simeq20\%$ ($\simeq10\%$) in the Yakutsk (AGASA)

97: experiment as compared to the Fly's Eye experiment; in Fig. 2 we use the

98: Yakutsk energy normalization.  The calculation of the model flux follows

99: ref. \cite{W95b}, and assumes a flat universe with $\Omega_{\rm m}=0.3$,

100: $\Omega_\Lambda=0.7$ and $H_0=65~{\rm km~s^{-1}~Mpc^{-1}}$. The generation

101: rate of cosmic rays per unit comoving volume is assumed to trace the

102: redshift evolution of the luminosity density of bright quasars\cite{QSO},

103: which is also similar to the evolution of the cosmic star formation rate

104: \cite{SFR}, namely $\dot n_{CR}(z)\propto(1+z)^{\alpha}$ with

105: $\alpha\approx3$ at low redshifts $z<1.9$, $\dot n_{CR}(z)={\it const}$~

106: for intermediate redshifts $1.9<z<2.7$, and an exponential decay at high

107: redshifts $z>2.7$. The cosmic-ray spectrum at energies $> 10^{19}$~eV is

108: only weakly dependent on the cosmological parameters or the assumed

109: redshift evolution, because the particles at these energies originate from

110: distances shorter than a few hundred Mpc.  The spectrum and flux at

111: $E>10^{19}$~eV is mainly determined by the present ($z=0$) generation rate

112: and spectrum, which for the model shown in Fig. 2 is given by

113: \begin{equation}

114: E^2 \left({d\dot n \over dE}\right)_{z=0}= 0.8\times10^{44}~

115: {\rm {erg \over Mpc^3~yr}}.

116: \label{eq:cr_rate}

117: \end{equation}

118: 

119: \begin{figure}

120: \centerline{\epsfig{file=fig2.eps, height=2.5in}} 

121: \caption{ The UHECR flux expected in a cosmological model for which

122: high-energy protons are produced at the rate given by Eq. (1) (solid line)

123: compared to the Fly's Eye [9], Yakutsk [10] and AGASA

124: [11] data (adopted from [12]).  The flux error bars

125: correspond to $1\sigma$. The highest energy points are derived assuming

126: that the detected events represent a uniform flux over the energy range

127: $10^{20}$~eV~--~$3\times10^{20}$~eV.  The dashed line is the sum of the

128: cosmological model flux and the Fly's Eye fit to the Galactic heavy nuclei

129: component, $J\propto E^{-3.5}$ [9].  }

130: \label{fig:fig2}

131: \end{figure}

132: 

133: The suppression of the flux above $10^{19.7}$~eV is caused by the energy

134: loss of high energy protons in their interaction with the microwave

135: background, i.e.  the ``GZK cutoff'' \cite{GZK}. The available data does

136: not allow to determine the existence (or absence) of the ``cutoff'' with

137: high confidence. The AGASA results show an excess of events (at a

138: $\sim2.5\sigma$ confidence level) compared to the model prediction above

139: $10^{20}{\rm eV}$. This excess is not confirmed, however, by other

140: experiments; preliminary results from the new HiRes experiment are

141: consistent with the Fly's Eye data which does not show this excess

142: \cite{HiRes}.  Since the flux at $10^{20}{\rm eV}$ is dominated by sources

143: at distances $<100\ {\rm Mpc}$ over which the distribution of known

144: astrophysical systems (such as galaxies or galaxy clusters) is

145: inhomogeneous, significant deviations are expected at this energy relative

146: to the model predictions presented in Fig. 2 under the assumption of a

147: uniform source distribution \cite{W95b,CR_clustering}.

148: 

149: 

150: \paragraph*{Galactic enhancement of observed flux.}

151: If the production rate of high energy protons given in

152: Eq. (\ref{eq:cr_rate}) extends to lower energies, then a universal flux of

153: extragalactic protons would be produced as shown by the solid line in

154: Fig. 1.  If extragalactic cosmic rays originate in galaxies similar to our

155: own, then the average Galactic flux would be enhanced relative to this

156: extragalactic background by two factors. The first factor is cosmological

157: and is independent of the confinement of cosmic rays by the Galactic

158: magnetic field.  Ignoring confinement for the moment, we expect that the

159: extragalactic flux would be diluted relative to the Galactic flux by the

160: volume filling factor of galaxies $n_{\rm G} V_{\rm G}$ (where $n_{\rm G}$

161: is the number density of galaxies and $V_{\rm G}$ is the volume of a

162: galaxy), and be amplified by the ratio between the emission time (which is

163: of order the Hubble time, $t_{\rm H}$) and the escape time from the

164: Galactic disk (which is of order the vertical crossing time of the

165: scale--height of the Galactic disk, $h/c$). The combination of these ratios

166: yields a Galactic enhancement factor of the cosmic ray flux $J(E)$,

167: \begin{equation}

168: f_1={J_{\rm G}\over J_{\rm C}}\approx {1\over c t_{\rm H} n_{\rm G} A_{\rm

169: G}}\sim 10^3,

170: \label{eq:enhance1}

171: \end{equation}

172: where $A_{\rm G}\equiv V_{\rm G}/h$ is the typical cross-sectional area of

173: a galactic disk, and the subscripts $G$ and $C$ are used to denote the

174: Galactic and cosmological values. The numerical value of $f_1\sim 10^3$ is

175: obtained for the characteristic $A_{\rm G}\sim \pi (5~{\rm kpc})^2$ and

176: $n_{\rm G}\sim 3\times 10^{-3}~{\rm Mpc^{-3}}$ of $L_\star$ galaxies

177: similar to the Milky-Way \cite{Madgwick01}.

178: 

179: A second enhancement factor originates from the confinement of cosmic rays

180: by the Galactic magnetic field. The time that a cosmic ray spends in the

181: disk is larger than $h/c$ by a factor $f_2$. This factor is directly

182: determined by observations of daughter nuclei produced in interactions of

183: cosmic ray nuclei with H and He nuclei in the Galactic disk. The abundance

184: ratio of daughter to parent nuclei determines the average grammage $\Sigma$

185: (defined as the path length integral of the gas density) that cosmic ray

186: nuclei traverse prior to their escape.  The factor $f_2$ is given by the

187: ratio $\Sigma/\Sigma_{\rm disk}$, where $\Sigma_{\rm disk}=2\times

188: 10^{-3}~{\rm g~cm^{-3}}$ is the surface mass density of the Galactic disk.

189: For cosmic rays with energies of up to 1~TeV per nucleon \cite{Sigma}

190: \begin{equation}

191: f_2\equiv{\Sigma\over \Sigma_{\rm disk}}\approx 10^4 \left({E/Z\over {\rm

192: GeV}}\right)^{-\alpha},

193: \label{eq:enhance2}

194: \end{equation}

195: where $Z$ is the atomic charge and $\alpha\approx0.6$.  The extrapolation

196: of this scaling relation to a proton energy $E\sim 10^{16}$~eV gives

197: $\Sigma\sim \Sigma_{\rm disk}$ or $f_2\approx1$, which implies that the

198: extrapolation can not be extended to yet higher proton energies since

199: $\Sigma$ must be larger or equal to $\Sigma_{\rm disk}$. Note that the

200: extragalactic background is also enhanced by a factor $f_2$ if measured

201: within the Milky-Way galaxy. Hence the ratio between the local values of

202: the mean Galactic and extragalactic fluxes remains $f_1$ even in the

203: presence of confinement.

204: 

205: Figure 1 shows that the observed energy flux $E^2 J(E)$ at $10^{15}$~eV is

206: larger by a factor of $\sim 10^3$ than its value at $10^{19}$~eV. This

207: factor is coincidentally comparable to the enhancement factor $f_1$. If, as

208: commonly assumed, the cosmic rays at $10^{15}$~eV are Galactic in origin

209: while those at $10^{19}$~eV are extragalactic, then {\it the energy

210: production rate per galaxy at these two cosmic-ray energies must be

211: comparable}. Since the particle spectrum expected for acceleration in

212: astrophysical shocks, $E^2 dN/dE= const$, implies equal amount of energy

213: per logarithmic energy interval, the above coincidence may seem to suggest

214: that both the $10^{15}$~eV and $10^{19}$~eV cosmic rays are produced by the

215: same source population \cite{Milgrom96}.  However, the associated Galactic

216: component of cosmic rays with energies between $10^{15}$~eV and

217: $10^{19}$~eV {\it appears to be missing}.  The spectrum steepens instead of

218: maintaining a $J(E)\propto E^{-2}$ slope in Fig. 1, as expected if the same

219: sources were producing the observed cosmic-ray flux at $10^{15}$~eV and

220: $10^{19}$~eV.  The lack of strong anisotropy in the direction of the

221: Galactic disk at $10^{19}$~eV supports this inference.  {\it Where are the

222: Galactic counterparts to the extragalactic $10^{19}$~eV cosmic rays?} To

223: reconcile the data in Fig. 1 with the enhancement factors in

224: Eqs.~(\ref{eq:enhance1}) and~(\ref{eq:enhance2}), we argue that the UHECRs

225: must be produced in transient events and that our galaxy is currently at a

226: dim state in between transients.

227: 

228: 

229: \paragraph*{Constraints on transient properties.} The

230: minimum flux of lower energy cosmic rays produced by the UHECR source

231: population corresponds to an extrapolation with a spectral slope of

232: $J(E)\propto E^{-2}$, which is the shallowest slope possible in shock

233: acceleration (analysis of the observed spectrum at $E>10^{19}$~eV implies

234: that the source spectral slope can not be shallower than $J(E)\propto

235: E^{-1.8}$ \cite{W95b}).  The dashed line in Fig. 1 shows this extrapolation

236: together with the expected enhancement factors $f_1$ and $f_2$ from

237: Eqs.~(\ref{eq:enhance1}) and~(\ref{eq:enhance2}), assuming conservatively

238: that Eq. (\ref{eq:enhance2}) with $\alpha=0.6$ holds for protons up to

239: $E\sim 10^{16}$~eV. At energies exceeding $\sim1$~TeV, the expected mean

240: Galactic flux of protons exceeds the observed value. The fact that the bulk

241: of the Galactic cosmic rays in this energy range are missing is supported

242: by the isotropy of the observed cosmic rays at $10^{19}$~eV.  The Galactic

243: field is unable to isotropize the distribution to the observed level since

244: the Larmor radius of a particle of atomic charge $Z$ in the Galactic field

245: $B$ is $\sim 10~{\rm kpc}(E/10^{19}~{\rm eV})/(B/\mu{\rm G})$,

246: significantly larger than the Galactic scale height of this magnetic field.

247: 

248: Our assumption that Eq. (\ref{eq:enhance2}) holds for protons up to $E\sim

249: 10^{16}$~eV is conservative since it implies that $\Sigma=\Sigma_{\rm

250: disk}$ at $E\sim 10^{16}$~eV. If $\Sigma$ is higher, e.g. due to a smaller

251: value of $\alpha$ beyond 1~TeV per nucleon, then the expected Galactic

252: counterpart to the extragalactic flux will be higher than shown in Fig. 1.

253: 

254: The confinement time of cosmic rays with an energy of $\sim1$~GeV per

255: nucleon is measured to be $\sim 10^{7.5}$~yr, based on the survival

256: fraction of radioactive cosmic-ray nuclei \cite{Webber98}. As the

257: cosmic-ray energy increases, the decline in $\Sigma$ described in

258: Eq. (\ref{eq:enhance2}) may be caused either by a decrease in the

259: confinement time or by a decrease in the time-averaged gas density through

260: which the cosmic rays propagate (e.g., due to an increase in the thickness

261: $h_{\rm CR}$ of the disk in which the cosmic rays are confined to a value

262: larger than the thickness $h$ of the Galactic gaseous disk).  Assuming,

263: conservatively, that the decrease in $\Sigma$ is only due to a decrease in

264: the confinement time, we get

265: \begin{equation}

266: \tau_{\rm conf}\approx 10^{7.5}~{\rm yr}

267: \left({E/Z\over {\rm GeV}}\right)^{-\alpha}.

268: \end{equation}

269: In order that the Galactic counterpart to the extragalactic $\sim

270: 10^{19}$~eV cosmic ray protons will not exceed the observed proton flux

271: above $\sim1$~TeV, the time between Galactic transients must satisfy

272: $\tau_{\rm trans}> f_{\rm vis}\times \min\{\tau_{\rm conf}(E=\gamma_{\rm

273: min}{\rm GeV}), \tau_{\rm conf}(E=1~{\rm TeV})\}$. Here, $\gamma_{\rm min}$

274: is the minimum Lorentz factor of the accelerated protons and $f_{\rm vis}$

275: is the fraction of all Galactic transients which are visible to us. Since

276: diffusion perpendicular to the Galactic disk drains the cosmic-ray flux

277: from a distant source, a transient would be visible to us only if it has

278: occurred within a distance comparable to the scale height $h_{\rm CR}$ of

279: the cosmic-ray disk.  The estimated scale-height, based on the

280: time-averaged density through which the cosmic rays propagate, is $\sim 3$

281: kpc \cite{Webber98} implying $f_{\rm vis}\sim 0.1$.  We therefore obtain

282: \begin{equation}

283: \tau_{\rm trans}\gtrsim 10^{4.5} \left({f_{\rm vis}\over 0.1}\right)

284: \min\left[\left({\gamma_{\rm min}\over10^3}\right)^{-0.6},1\right]~{\rm

285: yr},

286: \label{eq:t_burst}

287: \end{equation} 

288: as long as $\gamma_{\rm min}\lesssim 10^7$. At higher values of

289: $\gamma_{\rm min}$ for which $\Sigma\sim \Sigma_{\rm disk}$ and the cosmic

290: rays escape freely from the Galactic disk, the only constraints are that

291: $\tau_{\rm trans}$ must be longer than the exposure time of modern

292: experiments (several decades) and that the transients have a short duty

293: cycle.

294: 

295: Radioactive dating provides only a conservative lower limit on the

296: confinement time due to the potential existence of an extended cosmic-ray

297: halo.  We note that the above conclusions remain unchanged even if the

298: $10^{19}$~eV cosmic rays are heavy nuclei since the factor $f_1$ is

299: independent of composition.

300: 

301: \paragraph*{Discussion.}

302: Among the most likely astrophysical sources for the unsteady production of

303: UHECRs are Gamma-Ray Bursts (GRBs) and active galactic nuclei (AGNs).

304: While AGNs (such as SgrA* in the Milky Way galaxy) are known to be

305: transient just as GRBs, the transient properties of GRBs are

306: better-determined and allow a more detailed comparison with the required

307: properties of UHECR sources. The GRB rate per comoving volume today is

308: $\sim 0.5\times 10^{-9}~{\rm Mpc^{-3}~yr^{-1}}$ \cite{Schmidt01}.  This

309: rate is derived based on the observed value at a redshift $z\sim1$,

310: assuming that the rate follows the redshift evolution of the cosmic star

311: formation rate and that the emission of $\gamma$--rays by the bursts is

312: isotropic.  If the emission is confined to a double sided jet with a small

313: angular radius $\theta$, then the burst rate is increased by a factor

314: $4\pi/2\pi\theta^2$.  GRB observations indicate a characteristic opening

315: angle $\theta\gtrsim0.1$ \cite{Freedman01,Frail01}.  The implied temporal

316: separation between bursts in $L_\star$ galaxies is thus $\tau_{\rm GRB}\sim

317: 10^{4.5}(\theta/0.1)^2$~yr, consistent with the constraint in

318: Eq.~(\ref{eq:t_burst}).

319: 

320: Under the assumption that supernovae dominate the cosmic-ray production

321: below $10^{15}$~eV, the value of $\gamma_{\rm min}$ for GRBs with

322: $\tau_{\rm GRB}\sim10^{4.5}$~yr cannot be substantially lower than $\sim

323: 10^3$. This constraint is naturally satisfied in GRBs since the plasma

324: within which particles are accelerated, is inferred to expand with a bulk

325: Lorentz factor $\Gamma\sim10^{2.5}$ (e.g. \cite{Dermer01}).

326: 

327: The required energy production rate at $10^{19}$~eV is $\sim 10^{44}~{\rm

328: erg~Mpc^{-3}~yr^{-1}}$ \cite{W95b,W01Rev}, implying that the cosmic-ray

329: energy per burst is $10^{51}(\tau_{\rm burst}/10^{4.5}~{\rm yr})~{\rm

330: ergs}$.  This energy release is surprisingly close to the energy carried by

331: $\gamma$-rays in GRBs for $\tau_{\rm trans}=10^{4.5}~{\rm yr}$

332: (corresponding to $\theta\sim 0.1=6^\circ$) \cite{Frail01}, implying that

333: the local energy generation rate of $\gamma$-rays in GRBs is similar to the

334: UHECR energy generation rate \cite{W95}\footnote{The AGASA excess above

335: $10^{20}$~eV has been used to argue that GRBs can not produce the observed

336: cosmic rays above $10^{20}$~eV \cite{Stecker00}. As explained above, the

337: excess is not seen in the Fly's Eye, the Yakutsk and the HiRes data, and if

338: real is likely associated with source inhomogeneities.}. This similarity,

339: the agreement with the transient rate constraint, and the fact that protons

340: may be accelerated to energies $>10^{20}$~eV by collisionless internal

341: shocks of GRBs \cite{W95} support the suggested association between GRBs

342: and UHECR sources \cite{W95,Vietri95}.

343: %

344: %\footnote{It has been claimed that the conditions at the highly

345: %relativistic external shock driven by the GRB explosion into the

346: %surrounding gas are not likely to allow proton acceleration up to

347: %$10^{20}$~eV.  Regardless of the validity of this claim, it is irrelevant

348: %to the GRB model for UHECR production, where protons are accelerated in

349: %mildly-relativistic internal shocks within the expanding plasma

350: %\cite{W95,W01Rev}.}.

351: %Although extragalactic GRBs may account for UHECRs at $>10^{19}$~eV,

352: %neither Galactic nor extragalactic GRBs can account for the cosmic rays at

353: %$\sim 10^{15}$--$10^{18}$~eV, in difference from the suggestions in

354: %\cite{Milgrom96}. Extragalactic bursts are short of producing the

355: %observed flux and Galactic GRBs would overproduce the observed flux 

356: %or make it anisotropic.

357: 

358: The shocks produced due to converging flows during the formation of large

359: scale structure in the intergalactic medium are also expected to accelerate

360: cosmic rays \cite{Loeb00}. For the characteristic

361: magnetic field strength and scale of these shocks, we estimate that the

362: maximum proton energy is $\sim 10^{17}~{\rm eV}$ and that the intergalactic

363: cosmic rays may approach the observed flux around this energy but fall

364: significantly below the observed flux at lower energies.

365: 

366: In closing, we emphasize that the Earth was exposed to a time-averaged

367: UHECR flux that is greater by three orders of magnitude than the currently

368: measured value.  During transients the enhancement is much

369: larger\footnote{The transient event rate and intensity might depend on

370: particle energy. For beamed GRBs, the bursts at $>10^{19}~{\rm eV}$ will be

371: less frequent but more intense than those at much lower particle energies,

372: since the collimation of UHECRs at the source will not be entirely smeared

373: by the Galactic magnetic field.  When averaged over long timescales the

374: mean flux is independent of beaming; for a fixed energy output the observed

375: flux increases due to beaming but the event rate decreases by the inverse

376: of the same factor.}.  Any fossil record of this intense cosmic-ray

377: bombardment (e.g. through the detection of stable or radioactive daughter

378: nuclei in ancient rocks) would provide an important test of our

379: conclusions.

380: 

381: 

382: \paragraph*{Acknowledgments.}

383: This work was supported in part by grants from the Israel-US BSF

384: (BSF-9800343), NASA (NAG 5-7039, 5-7768), and NSF (AST-9900877,

385: AST-0071019). AL acknowledges support from the Einstein Center during his

386: visit to the Weizmann Institute.

387: 

388: \begin{references}

389: 

390: \bibitem{Axford94}

391: W. I. Axford, Astrophys. J. Supp. {\bf 90}, 937 (1994).

392: 

393: \bibitem{Nagano00} 

394: M. Nagano, \& A. A. Watson, Rev. Mod. Phys. {\bf 72}, 689 (2000).

395: 

396: \bibitem{Halzen02} F. Halzen, \& Hooper, D., submitted to Reports on

397: Progress in Physics (2002); astro-ph/0204527

398: 

399: 

400: \bibitem{Burnett90}

401: T. H. Burnett, et al., Astrophys. J. {\bf 349}, L25 (1990).

402: 

403: \bibitem{Bernlohr98}

404: K. Bernlohr, et al., Astropar. Phys. {\bf 8}, 253 (1998).

405: 

406: \bibitem{composition} 

407: T. K. Gaisser {\it et al.}, Phys. Rev. {\bf D47},

408: 1919 (1993); B. R. Dawson, R. Meyhandan, K. M. Simpson,

409: Astropar. Phys. {\bf 9}, 331 (1998).

410: 

411: \bibitem{Bird99} 

412: D. J. Bird {\it et al.}, Astrophys. J. {\bf 511}, 739 (1999).

413: 

414: \bibitem{Hayashida99}

415: N. Hayashida {\it et al.}, Astropar. Phys. {\bf 10}, 303 (1999).

416: 

417: \bibitem{Bird94} 

418: D. J. Bird {\it et al.}, Astrophys. J. {\bf 424}, 491 (1994).

419: 

420: \bibitem{Yakutsk} 

421: N. N. Efimov {\it et al.}, in {\it Proceedings of the

422: International Symposium on Astrophysical Aspects of the Most Energetic

423: Cosmic-Rays}, edited by M. Nagano and F. Takahara (World Scientific,

424: Singapore, 1991), p. 20.

425: 

426: \bibitem{agasa}

427: N. Hayashida {\it et al.}, Astrophys. J. {\bf 522}, 225 (1999); 

428: see also astro-ph/0008102.

429: 

430: \bibitem{W01Rev} 

431: E. Waxman, to appear in ICTP Lecture Notes Series:

432: Astroparticle physics and Cosmology, Eds. G. Senjanovic and A. Smirnov;

433: astro-ph/0103186 (2001).

434: 

435: \bibitem{Blandford87}

436: R. Blandford, \& D., Eichler, Phys. Rep. {\bf 154}, 1 (1987).

437: 

438: \bibitem{RelShock}

439: J. Bednarz \& M. Ostrowski, Phys. Rev. Lett. {\bf 80}, 3911 (1998);

440: A. Achterberg et al., MNRAS {\bf 328}, 393 (2001).

441: 

442: \bibitem{SN}

443: e.g., S. P. reynolnds \& D. C. Ellison, Astrophys. J. {\bf 399}, L75 (1992).

444: 

445: \bibitem{WAG97a}

446: E. Waxman, Astrophys. J. {\bf 485}, L5 (1997).

447: 

448: \bibitem{W95b} 

449: E. Waxman, Astrophys. J. {\bf 452}, L1 (1995).

450: 

451: \bibitem{QSO} 

452: M. Schmidt, D. P. Schneider \& J. E. Gunn, Astron. J. {\bf

453: 110}, 68 (1995); P. C. Hewett, C. B. Foltz \& F. Chaffee,

454: Astrophys. J. {\bf 406}, 43 (1993); B. J. Boyle \& R. J. Terlevich, MNRAS

455: {\bf 293}, L49 (1998).

456:  

457: \bibitem{SFR} 

458: S. J. Lilly, O. Le Fevre, F. Hammer \& D. Crampton,

459: Astrophys. J. {\bf 460}, L1 (1996); P. Madau {\it et al.} MNRAS {\bf 283},

460: 1388 (1996).

461: 

462: \bibitem{GZK} 

463: K. Greisen, Phys. Rev. Lett. {\bf 16}, 748 (1966);

464: G. T. Zatsepin \& V. A. Kuzmin, JETP lett., {\bf 4}, 78 (1966).

465: 

466: \bibitem{HiRes}

467: J. Matthews for the HiRes collaboration, priv. comm. (2001)

468: 

469: \bibitem{CR_clustering}

470: J. N. Bahcall \& E. Waxman, Astrophys. J. {\bf 542}, 542 (2000).

471: 

472: \bibitem{Madgwick01} 

473: P. Norberg, et al., Mon. Not. R. Astr. Soc., submitted (2001);

474: astro-ph/0111011 .

475: 

476: \bibitem{Sigma}

477: J. J. Engelmann et al., Astron. Astrophys. {\bf 233}, 96 (1990);

478: D. M\"uller et al., Astrophys. J. {\bf 374}, 356 (1991); S. A. Stephens \&

479: R. E. Streitmatter, Astrophys. J. {\bf 505}, 266 (1998).

480: 

481: \bibitem{Milgrom96}

482: M. Milgrom, \& V. Usov, Astroparticle Physics, {\bf 4}, 365 (1996);

483: C. D. Dermer, astro-ph/0005440 (2000).

484: 

485: \bibitem{Webber98} 

486: W. R. Webber, \& A. Soutoul, Astroph. J., {\bf 506}, 335 (1998).

487: 

488: \bibitem{Schmidt01} 

489: M. Schmidt, Astrophys. J. {\bf 552}, 36 (2001).

490: 

491: \bibitem{Freedman01}

492: D. L. Freedman, \& E. Waxman, Astrophys. J. {\bf 547}, 922 (2001).

493: 

494: \bibitem{Frail01}

495: D. A. Frail, et al., Astrophys. J., {\bf 562}, L55 (2001).

496: 

497: \bibitem{Dermer01}

498: C. D. Dermer, \& M. Humi, Astrophys. J., {\bf 556}, 479 (2001).

499: 

500: \bibitem{Stecker00}

501: F. W. Stecker, Astroparticle Phys. {\bf 14}, 207 (2000).

502: 

503: \bibitem{Gallant98}

504: Y. A. Gallant, \& A. Achterberg, MNRAS {\bf 305}, L6 (1999).

505: 

506: \bibitem{W95} 

507: E. Waxman, Phys. Rev. Lett. {\bf 75}, 386 (1995).

508: 

509: \bibitem{Vietri95}

510:   M. Vietri, Astrophys. J., {\bf 453}, 883 (1995);

511:   M. Milgrom \& V. Usov, Astrophys. J., {\bf 449}, L37 (1995).

512: 

513: 

514: \bibitem{Loeb00} 

515: A. Loeb, \& E. Waxman, Nature, {\bf 405}, 156 (2000); E. Waxman, \&

516: A. Loeb, Astrophys. J. Lett., {\bf 545}, L11 (2000); F. Miniati, D. Ryu,

517: H. Kang, \& T. W. Jones, {\bf 559}, 59 (2001).

518: 

519: 

520: 

521: \end{references}

522: 

523: 

524: \end{document}

525: