1: %\documentstyle[12pt,aaspp4]{article}

2: \documentstyle[emulateapj,onecolfloat]{article}

3: \input epsf

4: 

5: \def\degrees{\hbox{${}^\circ$\hskip-3pt .}}

6: \newcommand{\simlt}{\lesssim}

7: \newcommand{\simgt}{\gtrsim}

8: \newcommand{\zem}{z_{\rm s}}

9: \newcommand{\aem}{a_{\rm s}}

10: \newcommand{\clk}{C_\ell}

11: \newcommand{\smooth}{\sigma}

12: \renewcommand{\topfraction}{1.0}

13: \renewcommand{\bottomfraction}{1.0}

14: \renewcommand{\textfraction}{0.00}

15: \renewcommand{\dbltopfraction}{1.0}

16: 

17: \begin{document}

18: 

19: \twocolumn[

20: \title{Constraints on the long-range properties of gravity from weak

21: gravitational lensing}

22: \author{Martin White \& C.S. Kochanek}

23: \affil{ Harvard-Smithsonian Center for Astrophysics, Cambridge, MA 02138}

24: \authoremail{mwhite, ckochanek@cfa.harvard.edu}

25: 

26: \begin{abstract}

27: \noindent

28: \rightskip=0pt

29: Weak gravitational lensing provides a means of testing the long-range

30: properties of gravity.  Current measurements are consistent with standard

31: Newtonian gravity and inconsistent with substantial modifications on

32: Mpc scales.  The data allows long range gravity to deviate from a $1/r$

33: potential only on scales where standard cosmology would use normal

34: gravity but be dominated by dark matter.  

35: Thus, abnormal gravity theories must introduce two fine-tuning scales -- an

36: inner scale to explain flat rotation curves and an outer scale to force a

37: return to Newtonian gravity on large scales -- and these scales must

38: coincidently match the scales produced by the dark matter theory after

39: evolving the universe for 10 billion years starting from initial conditions

40: which are exquisitely determined from the cosmic microwave background.  

41: \end{abstract}

42: \keywords{cosmology:theory -- gravitational lensing}

43: ]

44: 

45: %\rightskip=0pt

46: \section{Introduction} \label{sec:intro}

47: 

48: Weak lensing of background galaxies by foreground large-scale structure

49: offers an opportunity to directly probe the mass distribution on large

50: scales over a wide range of redshifts.  As first pointed out by

51: Blandford et al.~(\cite{Blaetal91}) and Miralda-Escude~(\cite{Mir91}),

52: these effects are of order a few percent in adiabatic cold dark matter

53: models making their observation challenging but feasible.

54: Early predictions for the power spectrum of the shear and convergence were

55: made by Kaiser~(\cite{Kai92}) on the basis of linear perturbation theory.

56: Jain \& Seljak~(\cite{JaiSel97}) estimated the effect of

57: non-linearities in the density through analytic fitting formulae

58: (Peacock \& Dodds~\cite{PeaDod96}) and showed they substantially increase

59: the power in the convergence below the degree scale.  Because weak

60: lensing can measure the matter power spectrum without many of the

61: problems of approaches based on the distributions of galaxies or 

62: clusters (e.g. bias), it may ultimately provide as clean a cosmological 

63: probe as the microwave background.  Recently, several observational

64: groups have reported convincing evidence of the effect

65: (van Waerbeke et al.~\cite{Ludo};

66:  Bacon et al.~\cite{BRE};

67:  Kaiser et al.~\cite{KWL};

68:  Wittman et al.~\cite{WTKDB};

69:  Maoli et al.~\cite{Maoli};

70:  Rhodes et al.~\cite{RRG};

71:  van Waerbeke et al.~\cite{Ludo2})

72: 

73: All these theoretical and observational studies are primarily motivated by

74: standard theories of gravity and cosmology.  Despite the tremendous 

75: overall success of these theories, there has been a recent resurgence

76: of interest in non-standard theories of gravity, largely motivated by

77: the possibility that the standard paradigm has difficulty matching the

78: dynamical structure of galaxies (e.g. Flores \& Primack~\cite{Flores94};

79: Moore~\cite{Moore94}; Navarro \& Steinmetz~\cite{Navarro00}).

80: Most of these proposed modifications aim to make gravity a longer-ranged

81: force on scales comparable to the sizes of galaxies in order to explain

82: the flat rotation curves of galaxies on scales larger than the apparent

83: distribution of matter (e.g. Sellwood \& Kosowsky~\cite{Sellwood00}, 

84: Sanders~\cite{San98,San99,San00}; McGaugh~\cite{McGaugh99,McGaugh00},

85: but see Scott et al.~\cite{SWCP} and Aguirre et al.~\cite{ABFN} for an

86: opposite perspective).

87: As has been noted before (e.g.~Krisher~\cite{Kri}; Walker~\cite{Wal};

88: Bekenstein \& Sanders~\cite{BekSan}; Zhytnikov \& Nester~\cite{ZhyNes};

89: Edery~\cite{Ede}; Kinney \& Brisudova~\cite{Kinney01};

90: Uzan \& Bernardeau~\cite{UzaBer}; Mortlock \& Turner~\cite{MorTur})

91: any longer ranged gravitational force, if it also affects photons, should

92: have implications for gravitational lensing.

93: In particular it should profoundly affect the strength of weak lensing shears

94: on large scales.  Many of the above authors, however, consider gravitational

95: lensing by isolated objects.

96: To understand the lensing effects of modifying gravity on large scales it is

97: necessary to use the weak lensing formalism, summing over the contributions

98: {}from all density perturbations.

99: 

100: \section{The model} \label{sec:model}

101: 

102: We base our models on the discussion by Zhytnikov \& Nester~(\cite{ZhyNes})

103: of modified gravity theories within the context of linearized relativity

104: (see also Edery~\cite{Ede}).

105: This framework provides a relativistic gravity model which automatically

106: obeys the equivalence principle and within which definite calculations can be

107: made, while at the same time being as unrestrictive as possible.

108: Further discussion of the experimental foundations for the assumptions can

109: be found in Zhytnikov \& Nester~(\cite{ZhyNes}) and in

110: Weinberg~(\cite{Wei}), Misner, Thorne \& Wheeler~(\cite{MTW}) and

111: especially Will~(\cite{Wil}, \S\S2-3).

112: 

113: For any such model, the important change in the formalism for the propagation

114: of light through such a weak field metric is to change the Poisson equation

115: relating the density to the potential whose derivative is used to determine

116: the bend angle of photons.

117: The angular power spectrum of the convergence, $\kappa$, can be written as

118: an integral over the line-of-sight of the power spectrum of the density

119: fluctuations (Kaiser~\cite{Kai92}).  For sources at a distance $D_s$,

120: \begin{eqnarray}

121:   \ell(\ell+1)C_\ell/(2\pi) = {9\pi\over 4\ell}

122:     \left[\Omega_m H_0^2 D_s^2\right]^2 \int {d D\over D_s} t^3 (1-t)^2 

123: \nonumber \\

124:     \qquad\times

125:     \left[{\Delta_{\rm mass}^2(k=\ell/D,a)\over a^2}\right] 

126: f^2(k=\ell/D),

127: \label{eqn:semianalytic}

128: \end{eqnarray}

129: where $t\equiv D/D_s$, $\Delta_{\rm mass}^2(k)=k^3P(k)/(2\pi^2)$

130: is the contribution to the mass variance per logarithmic interval

131: physical wavenumber and $\ell(\ell+1)C_\ell/(2\pi)$ is the contribution to

132: $\kappa_{\rm rms}^2$  per logarithmic interval in angular wavenumber

133: (or equivalently multipole) $\ell$. 

134: The only change from the standard result is that the Poisson equation 

135: relating the potential to the density perturbations is modified from

136: $f(k)=1$ to a functional form determined by the Poisson equation of the

137: modified theory of gravity.  On small physical scales (large wavenumber

138: $k$), $f(k)=1$ is required to be consistent with the known properties

139: of gravity.

140: 

141: If the sources have a range of redshifts then one simply integrates the

142: above expression over the redshift distribution of the sources.  We shall

143: assume throughout that

144: \begin{equation}

145:   {dn\over dD} \propto D\exp[-(D/D_*)^4]

146: \end{equation}

147: and fix $D_*$ by the requirement that $\langle z_{\rm src}\rangle=1$.

148: In evaluating Eq.~(\ref{eqn:semianalytic}), we will use the method of

149: Peacock \& Dodds~(\cite{PeaDod96}) to compute the non-linear power spectrum

150: as a function of scale-factor.  Throughout we shall use the concordance

151: cosmology of Ostriker \& Steinhardt~(\cite{OstSte}) since it provides a

152: reasonable fit to recent CMB, weak lensing and large-scale structure data.

153: For this choice of parameters the lensing kernel peaks at $z\simeq 0.43$

154: at a (comoving) angular diameter distance of $1150h^{-1}\,$Mpc.

155: 

156: \begin{figure}

157: \begin{center}

158: \leavevmode

159: \epsfxsize=3.5in \epsfbox{f1.eps}

160: \end{center}

161: \caption{\footnotesize%

162: The angular power spectrum, $\ell(\ell+1)C_\ell/(2\pi)$, vs.~multipole

163: moment $\ell$ for models with $\alpha=1.0$ and $m=0$, 0.3, 1.0 and

164: $3.0h\,{\rm Mpc}^{-1}$.  The sources are assumed to have

165: $\langle z_{\rm src}\rangle=1$.  Spectra for other values of $\alpha$ can

166: be roughly obtained by averaging the $m=0$ spectrum and the appropriate

167: $\alpha=1$ spectrum (plotted here) with the relevant weights.}

168: \label{fig:cl}

169: \end{figure}

170: 

171: In our calculation we only consider the propagation of rays through a known

172: density distribution, and we model that known density distribution using a

173: standard cosmological model viewed as a means to interpolate the evolution

174: of structure with redshift.  We do not attempt to self-consistently form

175: the observed structures using the modified gravitational potential\footnote{In

176: the model described below, a linear fluctuation analysis suggests that

177: long-wavelength modes would grow more slowly than the standard model would

178: predict.  Thus neglect of this effect is conservative if we start from an

179: initially scale-invariant spectrum.}.

180: If we assume that all theories must match the local density distribution,

181: the only consequence of this assumption is that the evolution of structure

182: implicit in Eq.~(\ref{eqn:semianalytic}) uses the standard growth rates

183: rather than those of the modified gravity.

184: 

185: Examining the effects of modified gravity simply becomes a question of

186: considering different structures for the function $f(k)$.   

187: In 4D, the metric, being symmetric, contains 10 functions.  The 4 constraints

188: of energy-momentum conservation reduce the number of free functions to 6.

189: These 6 free functions can be decomposed under rotations as 2 scalar

190: (density perturbations), 2 vector (vortical motions) and 2 tensor (gravity

191: wave) modes.

192: Within the linearized theory there are a number of propagating modes, which

193: have the form of Yukawa (exponential) potentials

194: \begin{equation}

195:   U(\vec{r};m) = G\int {\rho(\vec{r'}) d^3r'\over

196:                \left| \vec{r}-\vec{r}' \right| }

197:                e^{-m|\vec{r}-\vec{r}'|} \qquad .

198: \end{equation}

199: Under a variety of reasonable assumptions Zhytnikov \& Nester~(\cite{ZhyNes})

200: conclude that the most general metric describes forces mediated by massive and

201: massless scalar and tensor particles.

202: We follow Zhytnikov \& Nester~(\protect\cite{ZhyNes}) in neglecting the vector

203: modes, however we will allow arbitrary couplings for the scalar and tensor

204: modes.

205: %[in possible violation of the theorem of Jagannathan \&

206: %Singh~(\cite{JagSin})].

207: In general relativity in the weak field limit

208: \begin{eqnarray}

209: g_{00} &=& \left(-1 + 2U\right)\\

210: g_{ij} &=& \left(\phantom{-}1+2U\right)\delta_{ij}

211: \end{eqnarray}

212: where $U$ is the usual Newtonian potential.  The metric of

213: Zhytnikov \& Nester~(\cite{ZhyNes}) has the same form, but with Yukawa

214: potentials in addition to the Newtonian one.

215: 

216: \begin{figure}

217: \begin{center}

218: \leavevmode

219: \epsfxsize=3.5in \epsfbox{f2a.eps}

220: \end{center}

221: \begin{center}

222: \leavevmode

223: \epsfxsize=3.5in \epsfbox{f2b.eps}

224: \end{center}

225: \caption{\footnotesize%

226: The rms shear, smoothed with a $5'$ FWHM gaussian (top) or a $10'$ FWHM

227: gaussian (bottom), predicted for the ``concordance'' cosmology with

228: $\Omega_{\rm mat}=0.3$, $\Omega_\Lambda=0.7$, $h=0.67$ and $\sigma_8=0.9$

229: as a function of $\alpha$ and $m$ ($h$/Mpc).  Contours are spaced every

230: $0.001$ with bold contours indicating $0.005$ (top), $0.01$ and $0.015$

231: (bottom). The stippled regions are consistent (at $1\sigma$) with the 

232: van Waerbeke et al.~(\protect\cite{Ludo2}) measurements. }

233: \label{fig:sig5-10}

234: \end{figure}

235: 

236: For test particles with $v\ll c$ or fluids with $p\ll \rho c^2$ only the

237: time-time part of the metric is relevant, the contribution of the $g_{ij}$

238: terms being suppressed by ${\cal O}(v^2/c^2)$.

239: However, for light, the bend angle due to the potential is actually the

240: arithmetic mean of the coefficients in $g_{00}$ and $g_{ij}$.

241: Though the extra scalar and tensor modes can enter into the space-space and

242: time-time part of the metric differently, we shall consider the 1 parameter

243: family of models where these coefficients are equal.

244: As Kinney \& Brisudova~(\cite{Kinney01}) discuss,

245: %in suggesting that many features of MOND could be duplicated by a force law

246: %coupled only to baryon number,

247: the requirement that cluster mass estimates from galaxy dynamics,

248: pressure equilibrium of the X-ray gas and gravitational lensing agree means

249: that any modified gravity law must affect photon propagation in roughly the

250: same was as it affects particle orbits.  A modified gravity which

251: differentially affects particles and photons will almost always lead to a

252: discrepancy between these three cluster mass estimates.

253:  

254: \begin{figure}

255: \begin{center}

256: \leavevmode

257: \epsfxsize=3.5in \epsfbox{f3.eps}

258: \end{center}

259: \caption{\footnotesize%

260: The ratio of the rms shear on $5'$ and $10'$ scales for the same cosmology

261: as the previous figure.  Contours are spaced in steps of 0.05, increasing to

262: top left.  Thick contours are spaced every 0.5, starting at 1.5.

263: The stippled region is consistent (at $1\sigma$) with the

264: van Waerbeke et al.~(\protect\cite{Ludo2}) measurements.}

265: \label{fig:sigratio}

266: \end{figure}

267: 

268: Thus in our model, in the weak field limit, the propagation of light is

269: the same as in standard general relativity, except that the potential is

270: \begin{equation}

271:   U(\vec{r}) = (1-\alpha)U(\vec{r},0) + \alpha U(\vec{r},m) + \cdots

272: \end{equation}

273: where $\cdots$ represents possible other terms of the same form as

274: the second.  We shall further simplify our calculation by considering only 1

275: correction term in what follows.

276: In such a theory with one additional ``field'', the function appearing in

277: the estimate of the weak lensing power spectrum is

278: \begin{equation}

279:   f(k) = (1-\alpha) + \alpha\ {k^2\over k^2+m^2},

280: \end{equation}

281: where $\alpha=0$ for standard gravity and $\alpha \simeq -0.9$ and

282: $m^{-1} \sim 50\,$kpc in order to produce flat rotation curves without

283: dark matter (e.g.~Sanders~\cite{Sanders86}).

284: The corresponding potential for an object of mass $M$ simplifies to the

285: Newtonian result, $-GM/r$, on small scales where $mr\ll 1$, and has a

286: different effective coupling constant, $-GM(1-\alpha)/r$ on large scales,

287: $mr\gg 1$.

288: 

289: Fig.~\ref{fig:cl} shows the anisotropy spectrum predicted for a range of

290: models.  If we limit the range of gravity ($\alpha >0$) then the shear

291: fluctuations on large angular scales are suppressed, and if we extend the

292: range they are enhanced.

293: This should be a generic feature of any modification to the long-range force

294: law.  To obtain limits on the parameters in our model we calculated the rms

295: shear expected in Gaussian windows with FWHM of $5'$ and $10'$ as a function

296: of $\alpha$ and $m$ (Fig.~\ref{fig:sig5-10}).  These predictions are consistent

297: with the rms shear measured on these scales by

298: van Waerbeke et al.~(\cite{Ludo2}) only for models with parameters close

299: to those of standard gravity.

300: We can minimize the model dependence of the result by examining the ratio

301: of the power at $5'$ and $10'$, as this largely removes any dependence of

302: the result on the matter density and the normalization of the power spectrum.

303: In Fig.~\ref{fig:sigratio}, we see that the data are consistent with standard 

304: gravity and a broad range of alternate theories.  These theories

305: are acceptable because our alternate gravity model has a $1/r$ potential 

306: on large scales so that when the $5'$ scale corresponds to a physical

307: scale larger than $m^{-1}$, the change in the coupling constant $\alpha$

308: is degenerate with a change in the enclosed mass. For sources with a 

309: mean redshift of unity, the $5'$ scale corresponds to a length scale

310: at the peak of the lensing kernel of approximately $1h^{-1}\,$Mpc.

311: 

312: Theories which do not return to a $1/r$ potential on large scales are

313: relatively easy to rule out (see Walker~\cite{Wal}).  Assuming that the

314: bend angle of light remains proportional to the gradient of the projected

315: gravitational potential, such theories predict that random lines of sight

316: would be highly sheared and (de)magnified\footnote{For example, for a $\log r$

317: potential and a Poisson distribution of lenses the convergence, $\kappa$, of a

318: source at $D_s$ (assumed to be much larger than the scale, $r_0$, beyond which

319: gravity is $\log r$) is

320: $\kappa\simeq \pi\alpha_0\left(n r_0^2 D_s\right)\gg 1$ for any reasonable

321: source density $n$.}

322: in contradiction with observations.

323: This problem can be traced to the lack of degeneracy between renormalizing the

324: mass and adjusting the coupling constant.

325: For example, ignoring the Kinney \& Brisudova~(\cite{Kinney01}) {\it ansatz\/}

326: for permissible forms of alternate gravity, we could use the force law

327: \begin{equation}

328:     -\phi'(r)/GM = - { 1 \over r^2 } - { \exp(-m r) \over r r_0 } 

329: \end{equation}

330: which is Keplerian for $r\ll r_0$ and $r\gg m^{-1}$ but is a $1/r$ force law,

331: producing a flat rotation curve, in between.

332: The potential corresponding to this force law is

333: \begin{equation}

334:   \phi/GM= -1/r + {\rm Ei}[-mr]

335: \end{equation}

336: where ${\rm Ei}[x]$ is the exponential integral.

337: The corresponding kernel for the weak lensing integral is

338: \begin{equation}

339:   f(k) = 1 - { k m - (k^2+m^2) \tan^{-1}(k/m) \over r_0 k (k^2+m^2) }.

340: \end{equation}

341: Figure~\ref{fig:cl2} shows the angular power spectrum in this model for a

342: range of scales $r_0$ and a large outer cutoff $m^{-1}=50h^{-1}\,$Mpc.  

343: Compared to normal gravity, the modified theories have enormously enhanced

344: large scale power and very different shapes.

345: 

346: \begin{figure}

347: \begin{center}

348: \leavevmode

349: \epsfxsize=3.5in \epsfbox{f4.eps}

350: \end{center}

351: \caption{\footnotesize%

352: The angular power spectrum, $\ell(\ell+1)C_\ell/(2\pi)$, vs.~multipole moment

353: $\ell$ for our second model with $m^{-1}=50h^{-1}\,$Mpc and $r_0=\infty$, $20$,

354: $10$ and $5h^{-1}\,$Mpc.  As $r_0\rightarrow\infty$ the model becomes standard

355: gravity.  Note the change in shape and the enormous enhancement in the power on

356: large scales.}

357: \label{fig:cl2}

358: \end{figure}

359: 

360: \section{Discussion} \label{sec:discussion}

361: 

362: Current modified gravity theories tuned to explain the rotation curves of

363: galaxies work in a standard cosmology because we measure rotation curves only

364: where there are baryons.  We can see that the rotation curve is flat out to

365: the limit where there are no more baryons to measure, but we cannot see that

366: it is Keplerian as we approach the edge of the more extended dark matter

367: distribution.

368: If we could continue to trace rotation curves on larger scales we would see

369: a growing difference between standard cosmological models and theories using

370: modified gravitational physics.  

371: 

372: Weak lensing allows us to do this experiment, although on such large scales

373: we must sum over the contributions of all of the mass rather than consider the

374: rotation curves of discreet objects.  As we would expect qualitatively,

375: increasing the strength of the gravitational field at long ranges predicts

376: stronger weak lensing signals on large scales than standard cosmological

377: models.  Current measurements of the rms shear on scales of $5'$-$10'$ 

378: rule out the theories we consider in the parameter ranges where they 

379: could explain rotation curves without dark matter unless the deviation from

380: normal gravity is limited to a restricted range of spatial scales from

381: $10h^{-1}\,{\rm kpc}\la r\la 1h^{-1}\,$Mpc.

382: On larger scales the models must return to the $r^{-2}$ force law of normal

383: gravity in order to be consistent with measurements.

384: 

385: In standard cosmological models, once we postulate the existence of dark 

386: matter, the inner and outer scales appear naturally.  On small scales the

387: cooling of the baryons concentrates the baryons relative to the dark matter

388: and renders them luminous and detectable.  Thus, normal matter combined with

389: normal gravity naturally explain dynamics on scales $\la 10h^{-1}\,$kpc.

390: On intermediate scales, dark matter provides an additional source of density,

391: which can be interpreted as an abnormal gravitational theory using only the

392: visible baryons as sources.  On large scales the universe returns to

393: homogeneity, and the special properties of the $1/r^2$ force law make the weak

394: lensing power slowly diminish on large scales.  Abnormal, longer ranged

395: theories lose the cancellation properties of the $1/r^2$ force law on large

396: scales, despite the increasing homogeneity of the density on these scales,

397: leading to enormous enhancements in the strength of the weak lensing shear.

398: Such strong shears are in gross disagreement with even the first generation

399: of weak lensing measurements on these scales

400: (van Waerbeke et al.~\cite{Ludo};

401:  Bacon et al.~\cite{BRE};

402:  Kaiser et al.~\cite{KWL};

403:  Wittman et al.~\cite{WTKDB};

404:  Maoli et al.~\cite{Maoli};

405:  Rhodes et al.~\cite{RRG};

406:  van Waerbeke et al.~\cite{Ludo2}).

407: Thus, abnormal gravity theories must introduce two fine-tuning scales -- an

408: inner scale to explain flat rotation curves and an outer scale to force a

409: return to Newtonian gravity on large scales -- and these scales must

410: coincidently match the scales produced by the dark matter theory after

411: evolving the universe for 10 billion years starting from initial conditions

412: which are exquisitely determined from the cosmic microwave background.  

413: 

414: Finally, although we lack a formalism for estimating weak lensing in

415: non-potential theories such as MOND, Mortlock \& Turner~(\cite{MorTur})

416: have emphasized that weak lensing results should be generic, as it requires

417: only that photons and particles have similar responses to gravitational fields.

418: This similarity of behavior is observed on the relevant scales (Mpc) through

419: the near equivalence of weak lensing, dynamical, and X-ray determinations of

420: cluster masses (Kinney \& Brisudova~\cite{Kinney01}).  

421: 

422: \bigskip

423: \acknowledgments  

424: We would like to thank L.V. van Waerbeke for providing more details of their

425: VIRMOS survey results.

426: M.W. was supported by NSF-9802362 and a Sloan Fellowship.

427: C.S.K. was supported by the Smithsonian Institution and NASA grants NAG5-8831

428: and NAG5-9265.

429: 

430: \begin{thebibliography}{99}

431: \bibitem[2001]{ABFN}

432: Aguirre A., Burgess C.P., Friedland A., Nolte D., 2001, preprint

433:   [hep-ph/0105083]

434: \bibitem[2000]{BRE}

435: Bacon D., Refrigier A., Ellis R., 2000, \mnras, 318, 625

436: \bibitem[1991]{Blaetal91}

437: Blandford R.D., Saust A.B., Brainerd T.G., Villumsen J.V.,

438:   1991, MNRAS, 251, 600

439: \bibitem[1994]{BekSan}

440: Bekenstein J.D., Sanders R.H., 1994, \apj, 429, 480

441: \bibitem[1999]{Ede}

442: Edery A., 1999, Phys. Rev. Lett., 83, 3990

443: \bibitem[1994]{Flores94}

444: Flores, R., \& Primack, J.R., 1994, ApJ, 427, L1

445: %\bibitem[1986]{JagSin}

446: %Jagannathan K., Singh L.P.S., 1986, Phys. Rev. D33, 2475

447: \bibitem[1997]{JaiSel97}

448: Jain B., Seljak U., 1997, ApJ, 484, 560 [astro-ph/9611077]

449: \bibitem[1992]{Kai92}

450: Kaiser N., 1992, \apj, 388, 272

451: \bibitem[2000]{KWL}

452: Kaiser N., Wilson G., Lupino G., 2000, preprint [astro-ph/0003338]

453: \bibitem[2001]{Kinney01}

454: Kinney W.H., Brisudova M., 2001, in ``Proceedings of the 15th Florida

455: Workshop in Nonlinear Astronomy and Physics'' [astro-ph/0006453]

456: \bibitem[1988]{Kri}

457: Krisher T.P., 1988, \apj, 331, L135

458: \bibitem[2001]{Maoli}

459: Maoli R., et al., 2001, A\&A, in press [astro-ph/0011251]

460: \bibitem[1999]{McGaugh99}

461: McGaugh S., 1999, ApJ, 523, L99 [astro-ph/9907409]

462: \bibitem[2000]{McGaugh00}

463: McGaugh S., 2000, ApJ, 541, L33 [astro-ph/0008188]

464: \bibitem[1991]{Mir91}

465: Miralda-Escude J., 1991, ApJ, 380, 1

466: \bibitem[1973]{MTW}

467: Misner C.W., Thorne K.S., Wheeler J.A., 1973, ``Gravitation'',

468: Freeman, New York.

469: \bibitem[1994]{Moore94}

470: Moore, B., 1994, Nature, 370, 629

471: \bibitem[2001]{MorTur}

472: Mortlock D.J., E.L. Turner, 2001, preprint [astro-ph/0103208]

473: \bibitem[2000]{Navarro00}

474: Navarro, J.F., \& Steinmetz, M., 2000, ApJ, 528, 607

475: \bibitem[1995]{OstSte}

476: Ostriker J., Steinhart P.J., 1995, Nature, 377, 600

477: \bibitem[1996]{PeaDod96}

478: Peacock J. A., Dodds S. J., 1996, MNRAS, 280, 19

479: \bibitem[2001]{RRG}

480: Rhodes J. Refregier A., Groth E., 2001, preprint [astro-ph/0101213]

481: \bibitem[1986]{Sanders86}

482: Sanders, R.H., 1986, MNRAS, 223, 539

483: \bibitem[1998]{San98}

484: Sanders R.H., 1998, MNRAS, 296, 1009

485: \bibitem[1999]{San99}

486: Sanders R.H., 1999, ApJ, 512, L23

487: \bibitem[2000]{San00}

488: Sanders R.H., 2000, astro-ph/0011439

489: \bibitem[2001]{SWCP}

490: Scott D., White M., Cohn J.D., Pierpaoli E., 2001, preprint

491:   [astro-ph/0104435]

492: \bibitem[2000]{Sellwood00}

493: Sellwood, J.A., \& Kosowsky, A., 2000, in Gas \& Galaxy Evolution,

494:   Hibbar, Rupen \& van Gorkom, eds., astro-ph/0009074

495: \bibitem[2001]{UzaBer}

496: Uzan J.-P., Bernardeau F., 2001, preprint [hep-ph/0012011]

497: \bibitem[1998]{WaeBerMel98}

498: van Waerbeke L.V., Bernardeau F., Mellier Y., 1999, A\&A, 342, 15  

499: [astro-ph/9807007] 

500: \bibitem[2000]{Ludo}

501: van Waerbeke L.V., et al., 2000, A\&A, 358, 30

502: \bibitem[2001]{Ludo2}

503: van Waerbeke L.V., et al., 2001, A\&A submitted, astro-ph/0101511

504: \bibitem[1994]{Wal}

505: Walker M.A., 1994, \apj, 430, 463

506: \bibitem[1972]{Wei}

507: Weinberg S., 1972, ``Gravitation and cosmology'', Wiley, New York.

508: \bibitem[1993]{Wil}

509: Will C.M., 1993, ``Theory and experiment in gravitational physics'',

510: Revised edition, Cambridge University Press, New York.

511: \bibitem[2000]{WTKDB}

512: Wittman D.M., et al., 2000, Nature, 405, 143

513: \bibitem[1994]{ZhyNes}

514: Zhytnikov V.V., Nester J.M., 1994, Phys. Rev. Lett., 73, 2950

515: \end{thebibliography}

516: \end{document}

517: 

518: