Minimal triangulations of circle bundles, circular permutations and binary Chern cocycle

Nikolai Mn\"ev

REVIEW 3 major objections 4 minor 16 references

Minimal triangulations of circle bundles, circular permutations and binary Chern cocycle

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A circle bundle over a triangulated base is triangulable exactly when its Chern class admits a 0/1-valued simplicial cocycle.

desk verdict A new cohomological criterion for triangulability of circle bundles—clean and plausible, but two load-bearing lemmas are asserted rather than proved. read the letter →

arxiv 1908.04029 v2 pith:4IL2TU6A submitted 2019-08-12 math.GT math.ATmath.CO

Nikolai Mnëv This is my paper · ORCID

classification math.GTmath.ATmath.CO MSC 55R1057Q1557R2055U10

keywords circlebundlestriangulationssemi-simplicialsetsChernclassbinarycocyclecircularpermutationscyclicorderstransitivityaxiom

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

This paper asks a concrete PL topology question: given a fixed triangulation of the base space, which circle bundles over it can themselves be triangulated compatibly with the base? The answer claimed is that triangulability is exactly a cohomological 0/1 condition: a bundle over a finite semi-simplicial base can be semi-simplicially triangulated precisely when its integer Chern class is representable by a simplicial 2-cocycle whose value on every 2-simplex is 0 or 1. The route to this answer passes through minimal triangulations, which are encoded by local systems of circular permutations of the vertices of each base simplex, and through the observation that the classical transitivity axiom for cyclic orders is exactly the universal binary Chern cocycle. A sympathetic reader would care because this reduces an existence question about triangulations to a small check on the base cohomology, and on closed oriented surfaces it yields an explicit construction of triangulated circle bundles with any Chern number up to half the number of triangles.

What carries the argument

The load-bearing objects are circular permutations and their associated minimal elementary bundles, together with the universal binary Chern cocycle. A circular permutation of $[k]$ is an oriented necklace with one bead of each color, and it determines a minimal semi-simplicial circle bundle over the $k$-simplex; minimal bundles over a base are exactly local systems of such permutations, i.e. simplicial maps $B \to \mathrm{SSSC}$, where $\mathrm{SSSC}$ is the simplicial set of all circular permutations. The universal binary Chern cocycle is the parity formula $c_{01}(0,1,2)=0$, $c_{01}(2,1,0)=1$ on $\mathrm{SSSC}$, giving a $0/1$-valued integer 2-cocycle. The argument then hangs on Proposition 7, which identifies the simplicial cocycle condition on the boundary of a 3-simplex with the classical transitivity axiom for cyclic orders: once triples are cyclically ordered compatibly, the orders extend uniquely over higher skeleta by the unique-extension property of $\mathrm{SSSC}$. A spindle-contraction trick reduces an arbitrary triangulated bundle to a minimal one by deleting all but one bead from each fiber circle, preserving strong concordance and hence the isomorphism class.

What would settle it

Enumerate the 16 binary 2-cochains on the boundary of the ordered 3-simplex and check Proposition 7 directly: for every cochain with $c(f)=0$, verify that the induced cyclic orders on the four faces extend to a cyclic order on $\{0,1,2,3\}$; for every cochain with $c(f)\neq 0$, verify that extension fails. A single counterexample in either direction would break the only-if direction of Theorem 1.

Watch

Extended reading notes

Core claim

The central claim is Theorem 1: for a finite semi-simplicial set $B$, an oriented circle bundle $p$ over $|B|$ admits a semi-simplicial triangulation over $B$ if and only if $c_1(p)\in H^2(|B|;\mathbb{Z})$ can be represented by a binary simplicial cocycle, i.e. a 2-cocycle taking only the values $0$ and $1$ on 2-simplices. For ordinary classical simplicial triangulations the same condition is necessary but not sufficient. The 'if' direction is built from a spindle-contraction trick reducing any triangulated bundle to a minimal one while preserving the bundle class, combined with a universal local formula $c_{01}$ assigning to a circular permutation of three elements the parity $0$ or $1$; the 'only if' direction uses a binary cocycle to define cyclic orders on every triple of vertices of each base simplex, and then uses the classical transitivity axiom for cyclic orders (verified by an explicit case check over the hexagram of the 3-skeleton, Proposition 7) to extend them to a global system of circular permutations, i.e. a minimal triangulation. The paper also proves Theorem 2: over an oriented closed surface triangulated with $N$ 2-simplices, every circle bundle with Chern number $c$ satisfying $|c| \le N/2$ can be semi-simplicially triangulated, and at the extreme $|c| = N/2$ the triangulation cannot be classical simplicial.

Load-bearing premise

The only-if direction depends on an unproved case check: that a pattern of zeroes and ones on the four faces of a tetrahedron produces a consistent cyclic ordering of the four vertices exactly when the pattern is a cocycle, and the paper's proof of this is an experimental enumeration, not a formal proof.

Editorial extensions

If this is right

A circle bundle over a finite semi-simplicial base is triangulable exactly when its Chern class has a $0/1$-valued simplicial representative, so triangulability becomes a local arithmetic condition on the base rather than a search for a triangulation.
Over an oriented closed surface with $N$ 2-simplices, every circle bundle with Chern number $c$ satisfying $|c| \le N/2$ admits a semi-simplicial triangulation over that surface triangulation.
The extreme values $|c| = N/2$ require genuinely semi-simplicial triangulations; classical simplicial triangulations cannot realize them.
Minimally triangulated circle bundles coincide with local systems of circular permutations of base-vertex orders, so the construction is purely local: one assigns a circular permutation to each 2-simplex and the face maps fit automatically.
The simplicial set of circular permutations is a $K(\mathbb{Z},2)$, giving a combinatorial universal object for minimal circle bundles with a universal binary Chern cocycle.

Reading between the lines

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

This is an editorial inference: if Theorem 1 is accepted, deciding triangulability of a circle bundle over a fixed finite base reduces to a finite enumeration—one can check all binary cochains for the cocycle condition and compare cohomology classes.
The author leaves open a rigorous proof of Proposition 7; an automated enumeration of the 16 cases would be a direct check and would close the only-if direction without relying on the experimental verification.
The same binary-cocycle mechanism suggests a combinatorial analogue of prequantization: an integral symplectic form represented by a $0/1$ cocycle should correspond to a minimally triangulated prequantum circle bundle with an explicit piecewise-linear connection, giving local formulas for its curvature.
The author notes that crossed simplicial groups and generalized orders are the natural context; if pursued, the classification might extend to higher sphere bundles by replacing circular permutations with cyclic structures on higher-dimensional spheres.

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

What you should know first: this paper gives a clean iff—a circle bundle over a finite semi-simplicial set can be triangulated exactly when its Chern class is representable by a 0/1-valued simplicial cocycle. That's a new and useful reduction to purely cohomological data, and the paper also gives an elegant bound on Chern numbers over oriented surfaces. The machinery is genuinely interesting: minimal triangulations become local systems of circular permutations, and the simplicial set SSSC of circular permutations is the classifying object. The spindle contraction argument that cuts any triangulation down to a minimal one is a nice trick and is proved.

The soft spots are real but specific. The proof of the 'if' direction relies on Propositions 6 and 7. Proposition 6 asserts a unique extension property for SSSC in all dimensions except 2 and 3, but it is not proved—just a homotopy heuristic. Proposition 7, the 16-case check that a binary cocycle condition is exactly Huntington transitivity for triples, is explicitly labeled 'experimental' and 'pseudoscientific.' The table of 16 cochains is all the cases, so a referee can verify it by hand in an hour, but the paper doesn't do that verification, and it certainly doesn't state a formal proof. Proposition 6 is more serious because it's an infinite family of dimensions and there's no proof sketch. Until both are established, Theorem 1 is a research announcement with a convincing outline, not a finished theorem.

That said, I don't see any fundamental circularity. The cocycle condition is not tautological, the 1/2 shift is a harmless normalization, and the self-citations are to prior work that actually does the local formulas. The paper is honest about its own gaps.

This is definitely worth taking seriously. I'd send it to a good referee who can check Table 1 and push on Proposition 6. If those close, the result is solid.

Referee Report

3 major / 4 minor

Summary. The paper studies which circle bundles over a given semi-simplicial base admit a semi-simplicial triangulation over that base. Its main result, Theorem 1, states that a circle bundle p can be semi-simplicially triangulated over a finite semi-simplicial set BBB if and only if the integer Chern class c1(p) ∈ H^2(|BBB|; Z) can be represented by a binary simplicial cocycle with values 0 and 1 on 2-simplices. The proof strategy is to encode minimal triangulations as local systems of circular permutations, represented by simplicial maps to a classifying object SSSC, then to connect the simplicial cocycle condition to Huntington's transitivity axiom for cyclic orders. The paper also gives a surface version, Theorem 2, bounding the possible Chern numbers of triangulable bundles over a fixed triangulated oriented closed surface.

Significance. If the main theorem is established, it would be a clean and striking characterization: triangulability of a circle bundle over a fixed semi-simplicial base becomes a purely cohomological 0/1 condition. The manuscript introduces a natural combinatorial object, SSSC, the simplicial set of circular permutations, and proposes a useful spindle-contraction reduction from arbitrary triangulations to minimal ones. The universal binary Chern cocycle c01 is a concrete local formula that could be of independent interest. However, the paper explicitly labels its central bridge as an 'experimental fact,' and the two key propositions that make the converse direction of Theorem 1 work are not proved. The manuscript is therefore best viewed as a promising research announcement whose central claim is not yet established as written.

major comments (3)

[Section 6.2, Proposition 7] Proposition 7 is the load-bearing step that lets a binary 2-cochain on ∂⟨3⟩ satisfying the cocycle condition (equation (8)) be extended to a transitive Huntington cyclic order on [3]. Its proof is explicitly called 'experimental' and 'a pseudoscientific check of cases during meditation over the hexagram,' and it rests entirely on Table 1, which is not machine-checked and is not accompanied by a derivation from Huntington's axioms. A single incorrect or omitted row in the 16-case table would invalidate the extension argument. This proposition is used in Proposition 8 and in the 'if' direction of Theorem 1, so the converse direction of Theorem 1 is not rigorously established.
[Section 6.1, Proposition 6] Proposition 6 asserts the unique extension property for maps ∂⟨k⟩ → SSSC for k = 0, 1 and k ≥ 4, with a gap in dimension 3. No proof is given; the surrounding text says 'we can see that SSSC ≈ K(Z, 2)' and that π2 follows 'by inspection of the hexagram on Figure 5.' This proposition is required in the proof of Proposition 8 to extend a local system of circular permutations from the 3-skeleton to the whole base, so the gap is load-bearing for Theorem 1. A rigorous proof or a precise reference for the Kan properties of SSSC is needed.
[Section 7, proof of Proposition 8] The proof of Proposition 8, which is the core of the 'if' direction of Theorem 1, reduces the construction of a minimal bundle from a binary cocycle to 'general Huntington theory and Proposition 7' and 'the Kan property of cyclic orders from Proposition 6.' Since both Proposition 7 and Proposition 6 are unproved in the manuscript, Proposition 8 is not independently established. The proof as written does not supply a complete argument that a binary 2-cocycle uniquely defines a local system of circular permutations over the entire base.

minor comments (4)

[Section 6.2, Table 1] The layout of Table 1 is difficult to read: the row and column labels are not clearly separated, and the correspondence between the 16 binary cochains and the listed circular permutations is not immediately evident. A clearer enumeration with explicit labels for the faces of ∂⟨3⟩ would help the reader verify the statement.
[Section 3.6] The text says 'We don't prove this fact in this paper' regarding the universality of the minimal bundle over SSSC. If this fact is used later, it should either be proved or clearly marked as an assumption; if it is not used, the sentence could be removed for clarity.
[Section 5] The definition of the universal binary Chern cocycle c01 states that it is the rational local formula from [MS17] shifted by the universal 2-coboundary 1/2, but the verification that this cochain is indeed a cocycle on SSSC is not carried out in the paper. A direct check or a more detailed reference would strengthen the exposition.
[Section 8, Lemma 9] The proof of Lemma 9 uses a 1-cochain 11 taking value 1 on every 1-simplex. The notation is understandable but a brief explanation of why the resulting 2-cochain evaluates to 1 on every 2-simplex would help readability.

Circularity Check

1 steps flagged · score 4.0 of 10

Load-bearing pieces of Theorem 1 are imported from the authors' [MS17] and an explicitly experimental 16-case check; the central statement is not a tautology but its converse direction is not independently proved here.

self citation load bearing [Section 3.6 (definition of SSSC and minimal bundles; used in Proposition 8)]
"Actually, a minimal elementary s.c. bundle eeec(ϑ) is the stalk of the simplicial map SSS ⟳−→SSSC over the base simplex ⟨k⟩ ϑ−→SSSC. Therefore ⟳ is the universal minimal s.c. bundle over SSSC. We don’t prove this fact in this paper."

The identification of minimally triangulated circle bundles with local systems of circular permutations is the premise of Proposition 8 and of both directions of Theorem 1. The paper explicitly declines to prove it here and inherits it from the author's [MS17]; Section 5 likewise takes the binary Chern cocycle as 'the rational local formula from [MS17]'. Since [MS17] is the same author's prior work and is not machine-checked or reproduced in this paper, the theorem's classification rests on a load-bearing self-citation rather than on a derivation contained in the text. This is not a fitted-parameter reduction, but it is exactly the 'uniqueness/representation imported from authors' pattern that the rubric treats as partial circularity.

full rationale

No prediction in the paper is obtained by fitting a parameter and then re-predicting the fitted data, and the main equivalence is not definitional: a binary 2-cocycle is not simply a renamed triangulable bundle, and the passage through Huntington cyclic orders and the extension over 3-simplices is genuine mathematical work. However, two load-bearing supports are not independently established in this paper. First, the representation of minimal s.c. bundles by simplicial maps to SSSC is explicitly stated to be unproved here ('We don’t prove this fact in this paper') and comes from [MS17]; Proposition 8 and the converse direction of Theorem 1 use it as input. Second, Proposition 7—the only bridge from the cocycle condition (8) to transitivity, hence to extension over each 3-simplex—has a proof the author labels 'experimental' and 'a pseudoscientific check of cases' resting on Table 1; a wrong or missing row would invalidate Proposition 8's inverse direction. These are correctness/rigor gaps and self-citation dependencies, not reductions by construction, and the paper contains independent content (spindle concordance, Theorem 2's explicit surface construction, the hexagram analysis), so the score is 4 rather than 6 or higher. The experimental status of Proposition 7 is flagged here as a limitation explicitly stated in the manuscript (Section 6.2); it does not by itself make the derivation circular.

Assumptions & free parameters 1 free parameters · 7 assumptions · 1 invented entities

The theorem rests on new combinatorial objects (SSSC, circular-permutation local systems) and several unproved or experimentally checked statements: the necklace-stalk description from [MS17], the spindle contraction reduction (Proposition 4), the unique extension property (Proposition 6), and the transitivity-cocycle equivalence (Proposition 7). The only hand-chosen normalization is the 1/2 shift of the local formula. No physical entities or fitted data are introduced.

free parameters (1)

universal coboundary shift 1/2 (binary normalization) = 1/2
In Section 5, the local rational Chern formula from [MS17] is shifted by the universal 2-coboundary 1/2 to produce a binary 0/1 cocycle. The shift is chosen by hand; it is not fitted to data, but it is a normalization the proof depends on for the binary statement.

assumptions (7)

standard math Weil-Kostant correspondence: H^1(B;S^1) approximately H^2(B;Z) identifies circle bundles with Chern classes
Invoked in Sections 1.1 and 3.1 to attach a first Chern class to a circle bundle; classical background.
standard math Oriented PL S1 bundles can be made into U(1) principal bundles by a flat Euclidean fiber metric, uniquely up to gauge
Section 3.1; needed to pass from combinatorial triangulation data to a Chern class.
domain assumption Necklace description: stalks of an s.c. circle bundle over ordered k-simplex correspond to necklaces colored by [k]; minimal stalks are circular permutations
Sections 3.3 and 3.4, taken from [MS17]; not fully reproved in this paper but load-bearing for the local-system encoding.
domain assumption Any s.c. bundle is strongly concordant to a minimal s.c. bundle (Proposition 4)
Section 4; proof is sketched via spindle contraction and bead removal in the necklace language, not given in full combinatorial detail.
ad hoc to paper SSSC has unique extension over ∂⟨k⟩ for k=0,1 and k≥4 (Proposition 6)
Section 6.1; asserted by a simplicial homotopy argument, with no detailed proof supplied.
ad hoc to paper Binary cochain on ∂⟨3⟩ is a cocycle iff the associated cyclic orders are transitive (Proposition 7)
Section 6.2; proof is explicitly an 'experimental' table check of 16 cases, not a derivation.
ad hoc to paper π2(SSSC)=Z follows from homology of the 2-sphere by inspecting the hexagram (Figure 5)
Section 6.1; finite computation presented as an observation, without a formal verification.

invented entities (1)

SSSC, the simplicial set of circular permutations
purpose: Classifying object for minimal triangulated circle bundles and local systems of circular permutations; central to the iff criterion.
It is a new combinatorial object introduced in Section 3.6. Its claimed homotopy type K(Z,2) is supported only by arguments inside this paper, not by an external benchmark.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Minimal triangulations of circle bundles, circular permutations and binary Chern cocycle." pith.science (2026). https://pith.science/paper/4IL2TU6A

@misc{pith2026190804029,
  author       = {Pith},
  title        = {Pith review of: Minimal triangulations of circle bundles, circular permutations and binary Chern cocycle},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4IL2TU6A}},
  note         = {Machine review of arXiv:1908.04029}
}

read the original abstract

We investigate a PL topology question: which circle bundles can be triangulated over a given triangulation of the base? The question got a simple answer emphasizing the role of minimal triangulations encoded by local systems of circular permutations of vertices of the base simplices. The answer is based on an experimental fact: classical Huntington transitivity axiom for cyclic orders can be expressed as the universal binary Chern cocycle.

Figures

Figures reproduced from arXiv: 1908.04029 by the authors.

**Figure 1.** Elementary simplicial circle bundle over the interval. simplex is uniquely named by a vertex of the base where its collapsing edge collapses. This creates a coloring of the n intervals by k + 1 ordered vertices of base simplex. Thus we got a necklace N (e) out of the combinatorics of e ([MS17, §16]). The process is illustrated on [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗

**Figure 4.** [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗

**Figure 5.** Skeleton SC(3) depicted as a hexagram. All the circles are oriented clockwise. p is a semi-simplicial oriented circle consisting from vertices and oriented arcs contracting to a vertex of the base by the bundle projection p. Proposition 4. Any s.c. bundle is strongly concordant to a minimal s.c. bundle, uniquely determined by a free choice of a single arc in every circle of the bundle. We will prove the proposition … view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 15 canonical work pages

[3]

Sheaves, Cosheaves and Applications

Justin Curry . Sheaves, Cosheaves and Applications . arXiv e-prints , arXiv:1303.3255, Mar 2013

arXiv 2013
[1]

Loop spaces, characteristic classes and geometric quantization

Jean-Luc Brylinski . Loop spaces, characteristic classes and geometric quantization. Reprint of the 1993 edition. Basel: Birkh\"auser, reprint of the 1993 edition edition, 2008

work page 1993
[2]

Circle bundles

Shiing-shen Chern. Circle bundles. In Jacob Palis and Manfredo do Carmo, editors, Geometry and Topology , volume 597 of Lecture Notes in Mathematics , pages 114--131. Springer Berlin Heidelberg, 1977

work page 1977
[4]

Crossed simplicial groups and structured surfaces

Tobias Dyckerhoff and Mikhail Kapranov . Crossed simplicial groups and structured surfaces . arXiv e-prints , arXiv:1403.5799, Mar 2014

work page Pith review arXiv 2014
[5]

Crossed simplicial groups and their associated homology

Zbigniew Fiedorowicz and Jean-Louis Loday . Crossed simplicial groups and their associated homology. Trans. Am. Math. Soc. , 326(1):57--87, 1991

work page 1991
[6]

B. L. Feigin and B. L. Tsygan . Additive $K$-theory. $K$-theory, arithmetic and geometry, Semin., Moscow Univ. 1984-86, Lect. Notes Math. 1289, 67-209 (1987). , 1987

work page 1987
[7]

E. V. Huntington . A set of independent postulates for cyclic order. Nat. Acad. Proc. 2, 630-631 (1916). , 1916

work page 1916
[8]

Huntington

Edward V. Huntington . Inter-relations among the four principal types of order. Trans. Am. Math. Soc. , 38:1--9, 1935

work page 1935

Show all 16 references

[9]

Dmitry N. Kozlov . Combinatorial algebraic topology. Eur. Math. Soc. Newsl. , 68:13--16, 2008

2008
[10]

Krasauskas

R. Krasauskas . Skew-simplicial groups. Lith. Math. J. , 27(1):47--54, 1987

1987
[11]

N. Mn\"ev . Which circle bundles can be triangulated over ^3 ? arXiv e-prints , arXiv:1807.06842, Jul 2018

2018 arXiv
[12]

K. V. Madahar and K. S. Sarkaria. A minimal triangulation of the H opf map and its application. Geom. Dedicata , 82(1-3):105--114, 2000

2000
[13]

Mn \"e v and G

N. Mn \"e v and G. Sharygin . On local combinatorial formulas for Chern classes of a triangulated circle bundle. J. Math. Sci., New York , 224(2):304--327, 2017

2017
[14]

Cyclically ordered sets

Vitezslav Novak . Cyclically ordered sets. Czech. Math. J. , 32:460--473, 1982

1982
[15]

C. P. Rourke and B. J. Sanderson. -sets. I . H omotopy theory. Quart. J. Math. Oxford Ser. (2) , 22:321--338, 1971

1971
[16]

Spaces of PL Manifolds and Categories of Simple Maps (AM-186)

Friedhelm Waldhausen, Bj rn Jahren, and John Rognes. Spaces of PL Manifolds and Categories of Simple Maps (AM-186) . Princeton University Press, 2013

2013

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [3]

Sheaves, Cosheaves and Applications

Justin Curry . Sheaves, Cosheaves and Applications . arXiv e-prints , arXiv:1303.3255, Mar 2013

arXiv 2013

[2] [1]

Loop spaces, characteristic classes and geometric quantization

Jean-Luc Brylinski . Loop spaces, characteristic classes and geometric quantization. Reprint of the 1993 edition. Basel: Birkh\"auser, reprint of the 1993 edition edition, 2008

work page 1993

[3] [2]

Circle bundles

Shiing-shen Chern. Circle bundles. In Jacob Palis and Manfredo do Carmo, editors, Geometry and Topology , volume 597 of Lecture Notes in Mathematics , pages 114--131. Springer Berlin Heidelberg, 1977

work page 1977

[4] [4]

Crossed simplicial groups and structured surfaces

Tobias Dyckerhoff and Mikhail Kapranov . Crossed simplicial groups and structured surfaces . arXiv e-prints , arXiv:1403.5799, Mar 2014

work page Pith review arXiv 2014

[5] [5]

Crossed simplicial groups and their associated homology

Zbigniew Fiedorowicz and Jean-Louis Loday . Crossed simplicial groups and their associated homology. Trans. Am. Math. Soc. , 326(1):57--87, 1991

work page 1991

[6] [6]

B. L. Feigin and B. L. Tsygan . Additive \(K\)-theory. \(K\)-theory, arithmetic and geometry, Semin., Moscow Univ. 1984-86, Lect. Notes Math. 1289, 67-209 (1987). , 1987

work page 1987

[7] [7]

E. V. Huntington . A set of independent postulates for cyclic order. Nat. Acad. Proc. 2, 630-631 (1916). , 1916

work page 1916

[8] [8]

Huntington

Edward V. Huntington . Inter-relations among the four principal types of order. Trans. Am. Math. Soc. , 38:1--9, 1935

work page 1935