Multiplicity of solutions of the classical Plateau problem: uniqueness, finiteness, and the number of minimal surfaces bounded by a given contour

Jesse Douglas's 1931 memoir 'Solution of the problem of Plateau' settled the existence of a least-area disk-type minimal surface for an arbitrary Jordan contour, and explicitly set aside four questions: uniqueness and the degree of multiplicity of the minimal surfaces bounded by a given contour; the sense in which a least-area characterization survives for contours that span no finite-area surface; a proof of the least-area property independent of Koebe's conformal mapping theorem; and extensions to several contours and to other topological types such as a Moebius leaf. The extension program was completed by Douglas and successors by 1939 (two contours, one-sided surfaces, arbitrary topological type; later Reifenberg's homological formulation), and the least-area items are either proved in the memoir itself (finite-area case) or vacuous and superseded in modern formulations. The surviving open core is the multiplicity question: for a given Jordan curve in $\mathbb{R}^3$, determine the number and structure of the disk-type minimal surfaces it bounds. Only narrow sufficient uniqueness criteria (Rado's convex-projection theorem; Nitsche's total-curvature bound $4\pi$), partial finiteness results (Tomi: absolute minima over real-analytic boundaries; Beeson's '$6\pi$ theorem': all minimal surfaces, stable or unstable, over real-analytic curves of total curvature at most $6\pi$; Tromba: generic finiteness), and isolated examples (Boehme's arbitrarily-many-solutions curves; Morgan's continuum of area-minimizing disks in $\mathbb{R}^4$) are known; finiteness for relative minima of area in $\mathbb{R}^3$ without a curvature, genericity, or boundary-regularity hypothesis, explicit multiplicity bounds beyond Enneper's wire, and the non-orientable (Moebius-type) analogue remain open.

Background

Plateau's problem - named after the physicist Joseph Plateau, whose soap-film experiments of 1849 motivated it and whose formulation goes back to Lagrange's 1760 calculus of variations - asks for a surface of least area bounded by a given closed curve. In its classical parametric form the boundary is a Jordan curve (a simple closed curve) $\Gamma$ in Euclidean space $\mathbb{R}^n$, and one seeks a minimal surface of disk type spanning $\Gamma$: a continuous map from the closed unit disk to $\mathbb{R}^n$ that is harmonic and angle-preserving (conformal) in the interior - the Euler-Lagrange condition for the area functional, equivalent to vanishing mean curvature - and that maps the unit circle monotonically, i.e. in cyclic order, onto $\Gamma$. A disk-type spanning surface is counted up to conformal automorphisms of the unit disk (equivalently, as a geometric minimal surface rather than as an individual parametrizing map). A minimal surface parametrization may degenerate at branch points, points where it behaves locally like $z\mapsto z^k$ for $k\ge 2$; excluding branch points is a separate regularity question.

Douglas's memoir 'Solution of the problem of Plateau' (Transactions of the AMS, 1931) gave the first solution for an arbitrary Jordan contour. Its method minimizes a functional $A(g)$ of the boundary parametrization $g$ of $\Gamma$ - a Dirichlet-type integral that is simpler to handle than area and satisfies $A(g)\le S(g)$, where $S(g)$ is the area of the harmonic surface with boundary values $g$, with equality exactly for minimal surfaces. Tibor Rado solved the problem independently at the same time, and the resulting least-area disk is called the Douglas-Rado solution. The memoir explicitly leaves open four issues: (1) uniqueness and the degree of multiplicity, the paper noting that examples show the solution need not be unique; (2) the least-area property for contours that span no surface of finite area, where that characterization becomes meaningless; (3) the fact that the Part V proof of the least-area property relies on the Koebe conformal mapping theorem (the uniformization statement that every simply-connected polyhedral surface can be mapped one-to-one and conformally onto a disk), which the author judged comparable in difficulty to Plateau's problem itself and planned to eliminate; and (4) the extension to several contours and to other topological structures of the surface, for instance a Moebius leaf (a one-sided, non-orientable surface; its standard model is obtained from a rectangle by identifying two opposite edges with a half twist) with prescribed boundary.

Three of these four issues were settled within a decade. Douglas himself published the two-contour case in 1931, the one-sided (Moebius-type) case in 'One-sided minimal surfaces with a given boundary' (1932), the case of any finite number of assigned boundaries in 1936, and 'The Most General Form of the Problem of Plateau' in 1939; Shiffman gave an independent treatment in 1939, Morse and Tompkins constructed unstable solutions of higher topological structure in 1941, and the Encyclopedia of Mathematics records that by 1936-1940 sufficient solvability conditions for the 'Douglas problem' - a minimal surface of prescribed Euler characteristic and orientability, bounded by any finite number of Jordan curves - had been established. Reifenberg (1960) later solved the problem for surfaces of varying topological type in arbitrary dimension by homological methods, and the 'general problem of Douglas' is now textbook material (Courant's monograph; Dierkes-Hildebrandt-Tromba's 'Global Analysis of Minimal Surfaces'). The least-area items dissolved as research questions: for contours spanning some surface of finite area, least area among all spanning surfaces is proved in the memoir itself (Part V, via Koebe's theorem); for contours spanning no finite-area surface, every disk-type spanning surface has infinite area, so a least-area characterization is vacuous, and the memoir defines its minimal surface through the Weierstrass representation (the classical explicit representation of minimal surfaces by a holomorphic function and a holomorphic differential) as a limit of polygonal-contour solutions. Modern geometric measure theory - Federer and Fleming's integral currents and Reifenberg's homological formulation - produces mass-minimizing spanning objects by construction, with no conformal-mapping input, which defused Douglas's methodological concern about Koebe's theorem (Douglas also published a dedicated 1931 study, 'The Mapping Theorem of Koebe and the Problem of Plateau').

What survives is the multiplicity question. Uniqueness of the disk-type solution fails in general, and multiplicity can be arbitrarily large: Boehme proved that for every $n$ there is a real-analytic Jordan curve of total curvature less than $4\pi+\epsilon$ (total curvature is $\int_\Gamma|\kappa|\,ds$, the integral of the absolute curvature along the curve) bounding more than $n$ minimal surfaces, and that boundary curves in a fixed knot class can have many solutions. The positive theory is narrow. Uniqueness is known under sufficient conditions such as Rado's theorem (the contour admits a one-to-one projection onto the boundary of a plane convex set) and Nitsche's theorem (a real-analytic curve of total curvature at most $4\pi$, a sharp bound), with Ruchert's refinement for Enneper's wire, the image of a circle of radius $r$ under Enneper's classical minimal surface. Finiteness is known only in pieces: Tomi proved that a real-analytic boundary bounds only finitely many absolute minima of area (an absolute minimum is a surface of globally least area; a relative minimum is one whose area does not decrease under sufficiently small perturbations, the mathematical model of a physically stable soap film); Beeson's '$6\pi$ theorem' proves that a real-analytic Jordan curve of total curvature at most $6\pi$ bounds only finitely many minimal surfaces of disk type, stable or unstable, removing a no-branch-point hypothesis from Nitsche's earlier version of the same conclusion (appendix A29 of his Lectures on Minimal Surfaces); Tromba proved finiteness for generic boundary curves; Boehme and Tomi described the structure of the solution set; Jakob treated polygonal contours. The hypotheses are essential: Morgan exhibited a smooth curve in $\mathbb{R}^4$ bounding a continuum of area-minimizing disks, and a boundary of four circles in $\mathbb{R}^3$ bounding a continuum of unstable minimal surfaces of arbitrarily high genus. Whether a real-analytic or $C^\infty$ Jordan curve in $\mathbb{R}^3$ can bound infinitely many disk-type minimal surfaces furnishing relative minima of area remains unresolved - the only claimed finiteness theorem at that level of generality (no curvature, genericity, boundary-regularity, or topological hypothesis) is an unpublished preprint of Beeson, whose companion paper on Enneper's wire was withdrawn in 2016 with the note that 'the problems posed by Nitsche remain open' - and, apart from curves covered by uniqueness theorems, the only Jordan curve with any known explicit bound on the number of disk-type solutions is Enneper's wire, with bound three (Ruchert 1981; Beeson-Tromba 1984). For non-orientable spanning surfaces - the Moebius-strip case whose existence Douglas settled in 1932 - the multiplicity theory is even less developed: the simplest open question there is whether a one-parameter family of absolute minima of Moebius type can exist, because Tomi's loop argument breaks down when the first eigenfunction of the Jacobi (second-variation) operator has a nodal line rather than a fixed sign; the branch-point regularity of least-area non-orientable surfaces was still being settled in 2014 (Gulliver).

Problem Statement

The problem is the multiplicity question that Douglas's 1931 memoir explicitly set aside ('Examples show that the solution of the Plateau problem may not be unique. The question of the degree of multiplicity of the solution is not dealt with here.'). Given a Jordan curve $\Gamma\subset\mathbb{R}^3$, determine the multiplicity structure of the set of minimal surfaces of disk type bounded by $\Gamma$:

(i) Uniqueness classification: characterize geometrically the contours for which the disk-type solution of Plateau's problem is unique. The known sufficient conditions - one-to-one projection onto the boundary of a plane convex set (Rado), real-analyticity with total curvature (the integral of the absolute curvature along $\Gamma$) at most $4\pi$ (Nitsche, sharp) - leave the general classification open.

(ii) Finiteness: decide whether a real-analytic (respectively $C^\infty$) Jordan curve in $\mathbb{R}^3$ can bound infinitely many minimal surfaces of disk type that furnish relative minima of area. Infinitely many absolute minima are excluded for real-analytic boundaries by Tomi's theorem, and finiteness holds for real-analytic curves of total curvature at most $6\pi$ by Beeson's '$6\pi$ theorem' (which bounds all disk-type minimal surfaces, stable or unstable) and for generic curves by Tromba's theorem; Morgan's example of a smooth curve in $\mathbb{R}^4$ bounding a continuum of area-minimizing disks shows that codimension one is essential, so the question is genuinely about $\mathbb{R}^3$.

(iii) Explicit bounds: for contours not covered by uniqueness theorems, bound the number of disk-type solutions in terms of geometric data of $\Gamma$ (for example, total curvature). Currently only Enneper's wire - the image of a circle of radius $r$ under Enneper's classical minimal surface - carries any explicit bound (three, for radii slightly above the uniqueness threshold $r=1$ established by Ruchert).

(iv) Non-orientable analogue: extend items (i)-(iii) to surfaces of the type of the Moebius strip, whose least-area representatives exist by Douglas (1932); in particular, decide whether a one-parameter family of absolute minima of area of Moebius type can exist.

The problem is the classical parametric one - minimal surfaces of disk or Moebius type spanning a Jordan curve - not the reformulated mass-minimization problem for currents or the homological Reifenberg problem, although those frameworks may supply tools. The four parts (i)-(iv) are independent sub-questions of the multiplicity question.

References

1. Jesse Douglas, 'Solution of the problem of Plateau', Transactions of the American Mathematical Society 33 (1931), 263-321, DOI 10.1090/s0002-9947-1931-1501590-9, https://doi.org/10.1090/s0002-9947-1931-1501590-9
2. Jesse Douglas, 'The Problem of Plateau for Two Contours', Journal of Mathematics and Physics 10 (1931), 315-359, DOI 10.1002/sapm1931101315, https://doi.org/10.1002/sapm1931101315
3. Jesse Douglas, 'The Mapping Theorem of Koebe and the Problem of Plateau', Journal of Mathematics and Physics 10 (1931), 106-130, DOI 10.1002/sapm1931101106, https://doi.org/10.1002/sapm1931101106
4. Jesse Douglas, 'One-sided minimal surfaces with a given boundary', Transactions of the American Mathematical Society 34 (1932), 731-756, DOI 10.1090/s0002-9947-1932-1501661-8, https://doi.org/10.1090/s0002-9947-1932-1501661-8
5. Jesse Douglas, 'Minimal Surfaces of General Topological Structure with Any Finite Number of Assigned Boundaries', Journal of Mathematics and Physics 15 (1936), 105-123, DOI 10.1002/sapm1936151105, https://doi.org/10.1002/sapm1936151105
6. Jesse Douglas, 'The Most General Form of the Problem of Plateau', American Journal of Mathematics 61 (1939), 590-608, DOI 10.2307/2371315, https://doi.org/10.2307/2371315
7. Tibor Radó, 'On the Problem of Plateau', Ergebnisse der Mathematik und ihrer Grenzgebiete, 2. Folge, Band 2, Springer, Berlin (1933), 109 pp., DOI 10.1007/978-3-642-99118-9, https://doi.org/10.1007/978-3-642-99118-9
8. Max Shiffman, 'The Plateau Problem for Minimal Surfaces of Arbitrary Topological Structure', American Journal of Mathematics 61 (1939), 853-882, DOI 10.2307/2371631, https://doi.org/10.2307/2371631
9. Marston Morse and C. Tompkins, 'Unstable minimal surfaces of higher topological structure', Duke Mathematical Journal 8 (1941), no. 2, 350-375, DOI 10.1215/s0012-7094-41-00828-1, https://doi.org/10.1215/s0012-7094-41-00828-1
10. E. R. Reifenberg, 'Solution of the Plateau problem for m-dimensional surfaces of varying topological type', Acta Mathematica 104 (1960), 1-92, DOI 10.1007/bf02547186, https://doi.org/10.1007/bf02547186
11. Herbert Federer and Wendell H. Fleming, 'Normal and Integral Currents', Annals of Mathematics (2) 72 (1960), 458-520, DOI 10.2307/1970227, https://doi.org/10.2307/1970227
12. Richard Courant, 'Dirichlet's Principle, Conformal Mapping, and Minimal Surfaces', Interscience (1950), Springer edition, DOI 10.1007/978-1-4612-9917-2, https://doi.org/10.1007/978-1-4612-9917-2
13. R. Osserman, 'A Proof of the Regularity Everywhere of the Classical Solution to Plateau's Problem', Annals of Mathematics (2) 91 (1970), 550-569, DOI 10.2307/1970637, https://doi.org/10.2307/1970637
14. R. Gulliver and F. Lesley, 'On boundary branch points of minimizing surfaces', Archive for Rational Mechanics and Analysis 52 (1973), 20-25, DOI 10.1007/bf00249089, https://doi.org/10.1007/bf00249089
15. J. C. C. Nitsche, 'A new uniqueness theorem for minimal surfaces', Archive for Rational Mechanics and Analysis 52 (1973), 319-329, DOI 10.1007/bf00247466, https://doi.org/10.1007/bf00247466
16. Reinhold Böhme and Friedrich Tomi, 'Zur Struktur der Lösungsmenge des Plateauproblems', Mathematische Zeitschrift 133 (1973), 1-29, DOI 10.1007/bf01226239, https://doi.org/10.1007/bf01226239
17. F. Tomi, 'On the finite solvability of Plateau's problem', in Geometry and Topology (Proc. Latin American School of Math., Rio de Janeiro 1976), Lecture Notes in Mathematics 597, Springer (1977), 679-695, DOI 10.1007/bfb0085380, https://doi.org/10.1007/bfb0085380
18. A. J. Tromba, 'On the number of simply connected minimal surfaces spanning a curve', Memoirs of the American Mathematical Society 12, no. 194 (1977), DOI 10.1090/memo/0194, https://doi.org/10.1090/memo/0194
19. R. Böhme and A. J. Tromba, 'The number of solutions to the classical Plateau problem is generically finite', Bulletin of the American Mathematical Society 83 (1977), 1043-1044, DOI 10.1090/s0002-9904-1977-14373-5, https://doi.org/10.1090/s0002-9904-1977-14373-5
20. Frank Morgan, 'A smooth curve in R4 bounding a continuum of area minimizing surfaces', Duke Mathematical Journal 43 (1976), 867-870, DOI 10.1215/s0012-7094-76-04366-0, https://doi.org/10.1215/s0012-7094-76-04366-0
21. Frank Morgan, 'A smooth curve in R3 bounding a continuum of minimal manifolds', Archive for Rational Mechanics and Analysis 75 (1981), 193-197, DOI 10.1007/bf00250480, https://doi.org/10.1007/bf00250480
22. Hartmut Ruchert, 'A uniqueness result for Enneper's minimal surface', Indiana University Mathematics Journal 30 (1981), no. 3, 427-431, DOI 10.1512/iumj.1981.30.30033, https://doi.org/10.1512/iumj.1981.30.30033
23. Reinhold Böhme and Anthony J. Tromba, 'The Index Theorem for Classical Minimal Surfaces', Annals of Mathematics (2) 113 (1981), 447-499, DOI 10.2307/2006993, https://doi.org/10.2307/2006993
24. M. Beeson and A. J. Tromba, 'The cusp catastrophe of Thom in the bifurcation of minimal surfaces', Manuscripta Mathematica 46 (1984), 273-308, DOI 10.1007/bf01185204, https://doi.org/10.1007/bf01185204
25. Brian White, 'Classical area minimizing surfaces with real-analytic boundaries', Acta Mathematica 179 (1997), 295-305, DOI 10.1007/bf02392746, https://doi.org/10.1007/bf02392746
26. Robert Gulliver, 'Branch points of area-minimizing projective planes and a question of Courant', Mathematische Annalen 362 (2014), 389-400, DOI 10.1007/s00208-014-1121-8, https://doi.org/10.1007/s00208-014-1121-8
27. Michael Beeson, 'A real-analytic Jordan curve cannot bound infinitely many relative minima of area', arXiv:math/0612001v7 (2006, latest revision 2015; unpublished preprint), https://arxiv.org/abs/math/0612001
28. Michael Beeson, 'The number of minimal surfaces bounded by Enneper's wire', arXiv:1506.03064 (2015; withdrawn by the author in 2016), https://arxiv.org/abs/1506.03064
29. Ruben Jakob, 'About the finiteness of the set of solutions of Plateau's problem for polygonal boundary curves', Analysis 29 (2009), no. 4, 365-385, DOI 10.1524/anly.2009.1030, https://doi.org/10.1524/anly.2009.1030
30. 'Plateau problem', Encyclopedia of Mathematics (article adapted from I. Kh. Sabitov), permanent link http://encyclopediaofmath.org/index.php?title=Plateau_problem&oldid=52158, http://encyclopediaofmath.org/index.php?title=Plateau_problem&oldid=52158
31. J. C. C. Nitsche, 'Lectures on Minimal Surfaces, Volume 1: Introduction, Fundamentals, Geometry and Basic Boundary Value Problems', Cambridge University Press, Cambridge (1989), xxv + 563 pp., ISBN 0-521-24427-7; review: Robert Osserman, review in The American Mathematical Monthly 98 (1991), no. 1, 72-74, DOI 10.1080/00029890.1991.11995707, https://doi.org/10.1080/00029890.1991.11995707
32. Ulrich Dierkes, Stefan Hildebrandt, and Anthony J. Tromba, 'Global Analysis of Minimal Surfaces', Grundlehren der mathematischen Wissenschaften 341, Springer (2nd ed. 2010), incl. the chapter 'The General Problem of Plateau: Another Approach', DOI 10.1007/978-3-642-11706-0_4, https://doi.org/10.1007/978-3-642-11706-0_4
33. Anthony J. Tromba, 'A Theory of Branched Minimal Surfaces', Springer Monographs in Mathematics (2012), DOI 10.1007/978-3-642-25620-2, https://doi.org/10.1007/978-3-642-25620-2
34. Camille Labourie, 'Solutions of the (free boundary) Reifenberg Plateau problem', Advances in Calculus of Variations 15 (2022), no. 4, 913-927, published online 27 November 2020, DOI 10.1515/acv-2020-0067, https://doi.org/10.1515/acv-2020-0067
35. Reinhold Böhme, 'A Plateau problem with many solutions for boundary curves in a given knot class', in Global Differential Geometry and Global Analysis (Proc. Colloq., Berlin 1979), Lecture Notes in Mathematics 838, Springer, Berlin (1981), 36-41, DOI 10.1007/bfb0088840, https://doi.org/10.1007/bfb0088840
36. Michael Beeson, 'The 6π theorem about minimal surfaces', Pacific Journal of Mathematics 117 (1985), no. 1, 17-25, DOI 10.2140/pjm.1985.117.17, https://doi.org/10.2140/pjm.1985.117.17
37. Robert Gulliver, 'A minimal surface with an atypical boundary branch point', in Differential Geometry: A Symposium in Honour of Manfredo do Carmo (Proc. Int. Conf., Rio de Janeiro 1988), Pitman Monographs and Surveys in Pure and Applied Mathematics 52, Longman, Harlow (1991), 211-228, zbMATH Open document 4208973, https://zbmath.org/4208973