Negatively curved Einstein metrics on Gromov-Thurston manifolds

Frieder J\"ackel; Ursula Hamenst\"adt

REVIEW 2 major objections 8 minor 1 cited by

Negatively curved Einstein metrics on Gromov-Thurston manifolds

T0 review · 2 major / 8 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read For every n≥4 there exist infinitely many closed manifolds with negatively curved Einstein metrics that are not locally symmetric.

desk verdict Solid extension of Fine-Premoselli to all n≥4 with a new rigidity theorem; two fillable gaps keep me from endorsing it unconditionally. read the letter →

arxiv 2411.12956 v2 pith:Q5KY2UND submitted 2024-11-20 math.DG

Ursula Hamenstädt , Frieder Jäckel This is my paper · ORCID

classification math.DG MSC 53C2553C2122E40

keywords EinsteinmetricsnegativecurvatureGromov–Thurstonmanifoldsbranchedcovershyperboliclocallysymmetricspaceslinearizedoperatorarithmeticlattices

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proves that in every dimension n≥4 there are infinitely many pairwise non-homeomorphic closed smooth manifolds that admit a negatively curved Einstein metric. The same manifolds also carry metrics with sectional curvature arbitrarily close to the constant −1, so they are Gromov–Thurston-style examples. The proof builds approximate Einstein metrics on cyclic branched covers of hyperbolic manifolds, shows their Einstein-tensor error tends to zero in L2, and then perturbs them to exact Einstein metrics using a uniform invertibility estimate for the linearized Einstein operator. A separate rigidity argument shows that at most one of the branched covers in the constructed family can be hyperbolic, which rules out locally symmetric metrics. This extends, from dimension four to all dimensions, the earlier result of Fine and Premoselli.

What carries the argument

The central object is the Gromov–Thurston manifold X_k, a cyclic d-fold branched cover of a hyperbolic manifold M_k along a null-homologous totally geodesic codimension-two submanifold Σ_k. The argument is carried by three tools: (1) the Fine–Premoselli approximate Einstein metric \bar g_k obtained by interpolating the model metric g = $du^{2}$/V(u) + V(u)$dθ^{2}$ + $u^{2}$ g_S (with V(u)=$u^{2}$−1+a(d)/$u^{{n−3}}$ chosen so the cone angle is 2π/d) with the hyperbolic metric; (2) the uniform invertibility of the linearized Einstein operator L = (1/2)Δ_L + (n−1)id acting on symmetric two-tensors, which follows from a uniform L2 spectral gap and De Giorgi–Nash–Moser estimates; and (3) a volume bound on the gluing region, obtained from subgroup separability and a geometric retraction, which makes the L2 norm of Ric(\bar g_k)+(n−1)\bar g_k tend to zero. The exclusion of locally symmetric metrics then uses Mostow rigidity applied to fixed point sets of deck-group isometries.

What would settle it

Compute the principal eigenvalue of the operator L = (1/2)Δ_L + (n−1)id on a family of explicit Gromov–Thurston manifolds with growing neck length; if it approaches zero, the uniform spectral gap in Lemma 4.1 fails and the inverse-function-theorem step collapses. Separately, exhibiting two distinct degrees d1,d2 ∈ 4Z for which the cyclic branched covers of the same hyperbolic manifold both admit hyperbolic metrics would disprove Theorem 5.5.

Watch

Extended reading notes

Core claim

Theorem 1 asserts that for any n≥4 and any ε>0 there are infinitely many pairwise non-homeomorphic closed n-manifolds X that admit a Riemannian metric with sectional curvature in [−1−ε,−1+ε], an Einstein metric with negative sectional curvature, and are not homeomorphic to any closed locally symmetric space. The examples are Gromov–Thurston manifolds: cyclic d-fold covers of closed arithmetic hyperbolic manifolds branched along null-homologous totally geodesic codimension-two submanifolds. The Einstein metric is found by showing that a sequence of approximate Einstein metrics, obtained by gluing a model Einstein metric to the hyperbolic metric, has Einstein-tensor error tending to zero in L2, and then applying a quantitative inverse function theorem to the Einstein operator in Bianchi gauge. This yields exact Einstein metrics close to the approximate ones, with negative sectional curvature. For n≥5 these are the first known negatively curved Einstein metrics on manifolds that are not locally symmetric.

Load-bearing premise

The entire perturbation step relies on a uniform L2 spectral gap for the linearized Einstein operator that is imported from Fine–Premoselli; if that gap degenerates as the approximate metrics develop longer and longer necks, the construction of exact Einstein metrics fails.

Editorial extensions

If this is right

In every dimension n≥4 and for every ε>0, one obtains closed n-manifolds that simultaneously have (1+ε)-pinched negatively curved metrics and genuine Einstein metrics, a combination not previously available outside dimension four.
For n≥5, the Einstein metrics produced are the first known examples of negatively curved Einstein metrics on manifolds that are not locally symmetric.
In dimension four, the family constructed here is different from the Fine–Premoselli examples, and the rigidity theorem of Besson–Courtois–Gallot shows these manifolds are not homotopy equivalent to hyperbolic manifolds.
As a by-product (Remark 4.4), the hyperbolic manifolds M_k themselves carry negatively curved Einstein metrics with conical singularities of cone angle 2π/d along Σ_k.
Theorem 5.5 implies that among the cyclic branched covers with degree d∈4Z of a fixed hyperbolic manifold, at most one can be hyperbolic, so the construction yields infinitely many distinct non-homeomorphic manifolds in any dimension.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A likely extension is that the same gluing-plus-perturbation scheme applies to branched covers along null-homologous codimension-two submanifolds in other non-positively curved Einstein spaces, provided analogous model metrics with the correct cone angle exist and the linearized operator has a uniform spectral gap.
If the imported spectral gap could be proved directly for the long-neck family rather than cited, the construction would become self-contained and might yield quantitative control on the Einstein metric in terms of the geometry of Σ_k.
Theorem 5.5 suggests a stronger statement, which the authors mention as forthcoming: no nontrivial branched cover of a closed hyperbolic n-manifold admits a hyperbolic metric; if true, the restriction to degree d∈4Z in the theorem is an artifact of the current proof.
One could test the construction numerically in dimension four on explicit arithmetic examples to estimate the size of the L2 error and the spectral gap; such data would indicate how large d may be for fixed M.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Ursula and Frieder have extended Fine-Premoselli from n=4 to all n≥4, constructing negatively curved Einstein metrics on Gromov-Thurston manifolds. The main analytic setup is a direct adaptation: build approximate Einstein metrics on branched covers, get an L2 bound on the error from a clever volume estimate, then invert the linearized Einstein operator. The genuinely new part is Section 5: Theorem 5.5, which says that for a branched cover of a hyperbolic manifold with a totally geodesic hypersurface fixed by an orientation-reversing involution, at most one degree d divisible by 4 can yield a hyperbolic manifold. That is a real result, and the proof using Mostow rigidity and fixed-point sets of isometries is intricate and, as far as I can tell, coherent.

The paper is well organized and honest about what is imported. The two soft spots are real but not fatal. First, Lemma 4.1, the uniform L2 spectral gap for the linearized Einstein operator, is cited to [FP20, Prop 4.3] rather than proved. The stress-test note worries that FP20 was written for dimension four and that the long neck in the approximate metrics could degenerate the gap. I can't rule that out from the text; the authors don't state the hypotheses of FP20's proposition or verify them for their sequence. That should be supplied. Second, the abstract promises 'mutually not homotopic' and Theorem 1 states 'pairwise non-homeomorphic', but the proof of Theorem 1 is only sketched as an immediate consequence of Theorem 4.3, Proposition 5.1, and Theorem 5.5. It is presumably true because the volumes of the covers tend to infinity, but the text never says so explicitly. A referee should ask for a sentence.

The curvature bounds needed for the Schauder estimates are also a bit loose—Proposition 2.3 gives an upper curvature bound but no lower bound, and the cited Schauder estimates require two-sided control. I suspect the lower bound follows from the explicit model, but it should be stated.

Overall, this is a significant paper. The construction is not routine, and the rigidity theorem is new and interesting. The analytic argument is standard modulo the imported spectral gap. I would send it to a good referee, and I expect the gaps to be fillable.

Referee Report

2 major / 8 minor

Summary. The paper constructs, for every n≥4 and ε>0, infinitely many pairwise non-homeomorphic closed n-manifolds X that admit a Riemannian metric with sectional curvature in [−1−ε,−1+ε], an Einstein metric with negative sectional curvature, and no locally symmetric metric. The examples are cyclic branched covers of a sequence of arithmetic hyperbolic manifolds, following Gromov–Thurston. The proof combines a Fine–Premoselli-type approximate Einstein metric, an inverse function theorem for the Einstein operator with uniform a priori estimates, and a rigidity theorem (Theorem 5.5) showing that at most one branched cover in a relevant family can be hyperbolic. The paper also contains a result on conical Einstein metrics (Remark 4.4).

Significance. If the analytic gaps are closed, this is a major advance: it extends the four-dimensional examples of Fine–Premoselli to all dimensions n≥4 and gives the first negatively curved Einstein metrics on non-locally-symmetric manifolds in dimensions n≥5. The algebraic construction via subgroup separability and virtual retractions is novel and carefully executed, and Theorem 5.5 is a result of independent interest. The paper is honest about its reliance on prior work, explicit about the L2 estimate, and the inverse function theorem framework is standard. However, the central analytic input, the uniform spectral gap, is imported without verification for this specific family.

major comments (2)

[§4.1, Lemma 4.1] The uniform L2 spectral gap for L on the approximate metrics ¯g_k is the central analytic input, but it is not proved. The text refers to [FP20, Proposition 4.3] as 'a bit more general,' yet no verification is given that the hypotheses of that result hold for the sequence here, where the gluing parameter is U_glue=(U_max)^{1/2}, the neck length R_k=R^ν_k/2 tends to infinity, and the branch locus may be disconnected. If the spectral gap degenerates as the neck grows, Proposition 4.2 and the inverse function theorem in Section 4.2 collapse, and Theorem 4.3 (hence Theorem 1) is unsupported. The authors should provide a proof of Lemma 4.1 in this setting, or at minimum a detailed verification that the constants in [FP20, Proposition 4.3] depend only on n and d and not on the neck length.
[§4.1, Proposition 4.2] The a priori estimate uses Lemma 2.2, which requires a two-sided sectional curvature bound |sec|≤Λ and a uniform lower injectivity radius bound inj≥i0 for all k. Proposition 2.3 records only sec≤−c<0 and no injectivity bound for the approximate metrics. Since the sequence develops a long neck, the uniform lower injectivity radius and uniform upper curvature bound are not automatic, and without them the C0-estimate (4.5) is not justified. The uniform invertibility of L therefore rests on a second unverified geometric input. The authors should prove these uniform geometric bounds for the approximate metrics or modify the argument to avoid them.

minor comments (8)

[Abstract] 'mututally' should be 'mutually'.
[Introduction] 'Perelmann' should be 'Perelman'.
[§2.2] The section title 'C0-.' appears truncated; it should read 'C0-estimates'.
[§2.2, (2.8)] There is an inconsistency in the Hölder exponent: (2.8) states ||g^φ_{ij}||_{C^{1,α}} ≤ C, but the following sentence refers to the C^{2,α} norm.
[§4.1, (4.1)-(4.2)] The definitions of the hybrid norms leave implicit the harmonic charts used to define the Hölder norms; uniformity in k of the chart size is needed and should be stated explicitly.
[§5, Theorem 5.5] The statement 'at most one d ∈ 4Z' should read 'at most one d ∈ 4N', since d is a positive covering degree.
[§5.2] The notation M^{2π/d_i}_cut is introduced without a precise definition; the cone structure and the path isometric boundary should be described more carefully.
[Theorem 1 proof] The infiniteness of the family X_k is not explicitly argued; it follows from the volumes of M_k (and hence X_k) tending to infinity, but this should be stated.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: the Einstein-metric construction is an independent perturbation argument; the uniform spectral gap is imported from Fine–Premoselli and the self-citations are auxiliary, not load-bearing.

full rationale

The paper's Theorem 1 is obtained by a genuine perturbation construction: Proposition 3.1 produces a sequence of branched covers with diam(Σ_k)/Rν_k→0; Proposition 2.3, imported from Fine–Premoselli, gives an approximate Einstein metric with Ric error O(U^{-(n-1)}); Corollary 3.2 uses the volume bound to show the L² error tends to 0; Proposition 4.2 proves uniform invertibility of the linearized Einstein operator; and the inverse function theorem then yields exact Einstein metrics. None of these steps defines the target quantity in terms of itself. The only non-proved input with real mathematical weight is the uniform L² spectral gap in Lemma 4.1, which is quoted from Fine–Premoselli [FP20, Proposition 4.3]; that is an external result by different authors, and even in the worst case its failure would be a gap in the proof concerning uniformity in k or injectivity-radius assumptions, not circularity. The self-citations [HJ22, Proposition 2.5] and [HJ24, Lemma 2.2] are technical lemmas from the authors' earlier work, but they do not encode the conclusion of Theorem 1: [HJ22] supplies Schauder estimates and [HJ24] supplies a group-action lemma used in the auxiliary non-local-symmetricity argument. They are auxiliary, parameter-free lemmas, not fitted inputs or restatements of the target theorem. The integer d and the gluing radii are chosen explicitly, for example U_glue=(U_max)^{1/2}, not fitted to force the conclusion. Hence no step reduces by construction to its input; the score of 1 reflects only the presence of minor self-citations, not circular dependence.

Assumptions & free parameters 0 free parameters · 8 assumptions · 0 invented entities

No empirical constants are fitted; the construction is deterministic. The proof relies on standard results in hyperbolic geometry, arithmetic lattices, and elliptic PDE theory, plus imported analytic lemmas from FP20 and two lemmas from the authors' prior papers.

assumptions (8)

domain assumption Subgroup separability in standard arithmetic hyperbolic lattices (Bergeron, BHW11 Theorem 2.7/2.8)
Used in Proposition 3.1 to enlarge the normal injectivity radius of the codimension-two submanifold while keeping it fixed.
domain assumption Virtual retraction of geometrically finite subgroups (BHW11 Theorem 1.4)
Used in Proposition 3.1 to extend homomorphisms and to pass to covers with prescribed properties.
domain assumption Uniform L2-spectral gap for the linearized Einstein operator on Fine-Premoselli approximate metrics (Lemma 4.1, from FP20 Prop 4.3)
Needed for the uniform invertibility of L in Proposition 4.2; assumed rather than proved here.
standard math Mostow rigidity
Used throughout Section 5 to replace homotopy equivalences of hyperbolic manifolds by unique isometries.
standard math Sampson's theorem on harmonic maps from Kähler manifolds to real hyperbolic spaces
Used in Proposition 5.1 to exclude complex hyperbolic homeomorphisms.
standard math Novikov's topological invariance of rational Pontryagin classes and Hirzebruch proportionality
Used in Proposition 5.1 to exclude quaternionic and Cayley hyperbolic homeomorphisms.
standard math Wolf's theorem on abelian subgroups of higher rank lattices and Preissmann's theorem on abelian subgroups of negatively curved manifolds
Used in Proposition 5.1 to reduce locally symmetric candidates to rank one.
domain assumption Lemma 2.2 from the authors' prior paper [HJ24] on finite order diffeomorphisms and fixed point free quotients
Used in two places in the proof of Proposition 5.8 to control fixed point sets of admissible diffeomorphisms.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Negatively curved Einstein metrics on Gromov-Thurston manifolds." pith.science (2026). https://pith.science/paper/Q5KY2UND

@misc{pith2026241112956,
  author       = {Pith},
  title        = {Pith review of: Negatively curved Einstein metrics on Gromov-Thurston manifolds},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Q5KY2UND}},
  note         = {Machine review of arXiv:2411.12956}
}

abstract

For every $n\geq 4$ we construct infinitely many mutually not homotopic closed manifolds of dimension $n$ which admit a negatively curved Einstein metric but no locally symmetric metric.

Figures

Figures reproduced from arXiv: 2411.12956 by the authors.

**Figure 1.** The cyclic 4-fold branched cover. The involutions ι and j act via reflection along the colored submanifolds and ζ via rotation around Σ. which associates to a homeomorphism the unique isometry homotopic to it is a group homomorphism. The following result relates the finite group of admissible diffeomorphisms of X to the corresponding finite group of isometries for the hyperbolic metric. It seems be known to the expe… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

An explicit description of the K\"{a}hler-Einstein metrics of Guenancia-Hamenst\"{a}dt
math.DG 2025-05 accept novelty 6.0 of 10

A model Einstein metric on complex hyperbolic branched covers, built the way Fine and Premoselli built theirs in the real hyperbolic case, is shown to equal the model Kähler-Einstein metric of Guenancia and Hamenstädt...

Reference graph

Works this paper leans on

35 extracted references · 26 canonical work pages · cited by 1 Pith paper

[1]

M. T. Anderson. Convergence and rigidity of manifolds under Ricci curvature bounds . Inventiones mathematicae , 102(2):429--446, 1990

work page 1990
[2]

M. T. Anderson. Dehn Filling and Einstein Metrics in Higher Dimensions . Journal of Differential Geometry , 73(2):219--261, 2006

work page 2006
[3]

M. T. Anderson. A survey of Einstein metrics on 4-manifolds , Adv. Lect. Math. (ALM), 14, International Press, Somerville, MA 2010, 1--39

2010
[4]

Bangert and V

V. Bangert and V. Schroeder
[5]

Barthels and W

A. Barthels and W. L\"uck
[6]

Bergeron

N. Bergeron. Premier nombre de Betti et spectre du Laplacien des certaines variétés hyperboliques . L'Enseignement Mathématique , 46:109--137, 2000

work page 2000
[7]

Bergeron, F

N. Bergeron, F. Haglund, and D. T. Wise. Hyperplane sections in arithmetic hyperbolic manifolds . Journal of the London Mathematical Society , 83(2): 431--448, 2011

work page 2011
[8]

A. L. Besse. Einstein manifolds . Classics in Mathematics. Reprint of the 1987 edition. Springer-Verlag , 2008

1987

Show all 35 references

[9]

Besson, G

G. Besson, G. Courtois, and S. Gallot. Entropies and rigidit\'es des espaces local\'ement symm\'etriques de courbure strictement n\'egative , Geometric and Functional Analysis , 5:731--799, 1995

1995
[10]

O. Biquard. M\'etriques d'Einstein asymptotiquement sym\'etriques . Number 265 in Ast\'erisque. Soci\'et\'e math\'ematique de France, 2000

2000
[11]

K.S. Brown. Cohomology of groups . Graduate Texts in Mathematics. Springer, 1982

1982
[12]

Cappell, A

S. Cappell, A. Lubotzky, and S. Weinberger. A trichotomy for transformation groups of locally symmetric manifolds and topological rigidity . Advances Math. , 327:25--46, 2018

2018
[13]

M. P. do Carmo. Riemannian Geometry . Birkh\"auser, 2. edition, 1992

1992
[14]

Eells and J

J. Eells and J. H. Sampson. Harmonic Mappings of Riemannian Manifolds . American Journal of Mathematics , 86(1):106--160, 1964

1964
[15]

V. Emery. Arithmetic groups . Lecture notes on an introductory course on arithmetic lattices , arXiv:2308.03752

arXiv
[16]

Fine and B

J. Fine and B. Premoselli. Examples of compact Einstein four-manifolds with negative curvature . Journal of the American Mathematical Society , 33(4):991--1038, 2020

2020
[17]

Gilbarg and N

D. Gilbarg and N. S. Trudinger. Elliptic Partial Differential Equations of Second Order . Classics in Mathematics. Springer-Verlag, 2001. Reprint of the 1998 edition

2001
[18]

Gromov and W

M. Gromov and W. Thurston. Pinching constants for hyperbolic manifolds . Inventiones Mathematicae , 89:1--12, 1987

1987
[19]

adt and F. J\

U. Hamenst\"adt and F. J\"ackel. Stability of Einstein Metrics and effective hyperbolization in large Hempel distance . Preprint , arXiv:2206.10438 https://arxiv.org/abs/2206.10438, 2022

2022 arXiv
[20]

adt and F. J\

U. Hamenst\"adt and F. J\"ackel. Rigidity of geometric structures . Geometriae Dedicata , 218(1): Paper no. 16, 2024

2024
[21]

Hirzebruch

F. Hirzebruch. Automorphe Formen und der Satz von Riemann-Roch . Symposium International de Topologia Algebraica, Mexico , 129--144, 1956

1956
[22]

Jost and H

J. Jost and H. Karcher. Geometrische Methoden zur Gewinnung von A-Priori-Schranken f \"u r harmonische Abbildungen . Manuscripta Mathematica , 40:27--77, 1982

1982
[23]

Kerckhoff, and P

S. Kerckhoff, and P. Storm, Local rigidity of hyperbolic manifolds with geodesic boundary . Journal of Topology , 5:757--784, 2012

2012
[24]

N. Koiso. Nondeformability of Einstein metrics . Osaka Journal of Mathematics , 15(2):419--433, 1978

1978
[25]

J. F. Lafont and R. Roy. A note on the characteristic classes of non-positively curved manifolds . Expositiones Mathematicae , 25(1):21--35, 2007

2007
[26]

T.Y. Lam. Algebraic Theory of Quadratic Forms . Benjamin Inc. , 1973

1973
[27]

R. C. Lyndon and P. E. Schupp. Combinatorial Group Theory . Classics in Mathematics. Reprint of the 1977 edition. Springer Verlag , 2001

1977
[28]

S. P. Novikov. Topological invariance of rational Pontrjagin classes . Soviet Mathematics. Doklady , 6:921--923, 1965

1965
[29]

Ontaneda

P. Ontaneda. Riemannian hyperbolization . Publications mathématiques de l'IHÉS , 131:1--72, 2020

2020
[30]

Petersen

P. Petersen. Riemannian Geometry . Springer, 3. edition, 2016

2016
[31]

Stover, D

M. Stover, D. Toledo,
[32]

Preissmannn

A. Preissmannn. Quelques propriétés globales des espaces de Riemann . Commentarii Mathematici Helvetici , 15:175--216, 1942

1942
[33]

J.H. Sampson. Applications of harmonic maps to K\"ahler geometry . Contemp. Math. , 49:125--133, 1986

1986
[34]

P. Topping. Lectures on the Ricci Flow . London Mathematical Society Lecture Note Series. Cambridge University Press, 2006

2006
[35]

J.A. Wolf. Discrete groups, symmetric spaces, and global holonomy . Amer. J. Math. , 84:527--542, 1962

1962

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

M. T. Anderson. Convergence and rigidity of manifolds under Ricci curvature bounds . Inventiones mathematicae , 102(2):429--446, 1990

work page 1990

[2] [2]

M. T. Anderson. Dehn Filling and Einstein Metrics in Higher Dimensions . Journal of Differential Geometry , 73(2):219--261, 2006

work page 2006

[3] [3]

M. T. Anderson. A survey of Einstein metrics on 4-manifolds , Adv. Lect. Math. (ALM), 14, International Press, Somerville, MA 2010, 1--39

2010

[4] [4]

Bangert and V

V. Bangert and V. Schroeder

[5] [5]

Barthels and W

A. Barthels and W. L\"uck

[6] [6]

Bergeron

N. Bergeron. Premier nombre de Betti et spectre du Laplacien des certaines variétés hyperboliques . L'Enseignement Mathématique , 46:109--137, 2000

work page 2000

[7] [7]

Bergeron, F

N. Bergeron, F. Haglund, and D. T. Wise. Hyperplane sections in arithmetic hyperbolic manifolds . Journal of the London Mathematical Society , 83(2): 431--448, 2011

work page 2011

[8] [8]

A. L. Besse. Einstein manifolds . Classics in Mathematics. Reprint of the 1987 edition. Springer-Verlag , 2008

1987