---
title: 'Bol''s Web: Exceptional Planar Webs'
url: https://www.emergentmind.com/topics/bol-s-web
type: topic
---

# Bol's Web: Exceptional Planar Webs

Bol’s web is the classical exceptional planar \(5\)-web on \(\mathbb{C}^{2}\) with first integrals
\[
x,\qquad y,\qquad x/y,\qquad (y-1)/(x-1),\qquad x(y-1)/(y(x-1)).
\]
It was introduced by Bol as the first example of a maximal-rank planar web that is not linearizable, and it occupies a central position at the intersection of web geometry, dilogarithmic functional identities, moduli of marked points on \(\mathbb{P}^{1}\), cluster structures of type \(A_{2}\), and Gelfand–MacPherson constructions. In modern formulations, Bol’s web is identified with the \(5\)-web on \(\mathcal{M}_{0,5}\) given by the five forgetful maps, with the web by conics on the del Pezzo surface \(\mathrm{dP}_{5}\), and with a toric quotient of a natural web on \(G_{2}(\mathbb{C}^{5})\) [2401.06711] [2507.12180].

## 1. Classical definition and local web-theoretic setting

A planar \(k\)-web on a complex surface \(M\) is given by holomorphic submersions \(V_{1},\dots,V_{k}\) such that \(dV_{i}\wedge dV_{j}\neq 0\) generically for \(i\neq j\); the corresponding level sets define the foliations of the web. An abelian relation for \(\mathcal{W}(V_{1},\dots,V_{k})\) is a \(k\)-tuple of holomorphic functions \((F_{i})\) satisfying
\[
\sum_{i=1}^{k} F_{i}(V_{i})=0.
\]
The space of abelian relations has finite dimension, called the rank of the web, and this dimension is always bounded above by \((k-1)(k-2)/2\), the classical Bol bound [2401.06711].

Within this framework, Bol’s web \(\mathfrak{B}\) is the planar \(5\)-web
\[
\mathfrak{B}=\mathcal{W}\!\left(x,\ y,\ x/y,\ (y-1)/(x-1),\ x(y-1)/(y(x-1))\right).
\]
Its natural regularity domain is the complement of the line arrangement
\[
xy(x-1)(y-1)(x-y)=0.
\]
Bol introduced \(\mathfrak{B}\) as the first example of an exceptional planar web: a maximal-rank web that is not locally equivalent to a web of pencils of lines. In the classical hexagonality classification, any hexagonal planar \(5\)-web is either linearizable or equivalent to \(\mathfrak{B}\), which isolates Bol’s web as the unique non-linearizable model in that category [2401.06711].

## 2. Abelian relations, dilogarithms, and maximal rank

The abelian-relation structure of Bol’s web is governed by the Rogers dilogarithm
\[
R(x)=Li_{2}(x)+\tfrac12 Log(x)Log(1-x)-\pi^{2}/6.
\]
Its distinguished nonlinear abelian relation is Abel’s five-term identity:
\[
R(x)-R(y)-R(x/y)-R((1-y)/(1-x))+R(x(1-y)/(y(1-x)))=0.
\]
This identity supplies the unique, up to scale, dilogarithmic abelian relation of \(\mathfrak{B}\) [2401.06711].

The remaining abelian relations are logarithmic. More precisely, \(\mathfrak{B}\) has a \(5\)-dimensional logarithmic part \(AR_{\log}\), spanned by combinatorial \(3\)-term relations, and a \(1\)-dimensional dilogarithmic part generated by Abel’s identity. The decomposition is
\[
AR(\mathfrak{B})=AR_{\log}(\mathfrak{B})\oplus \langle \mathcal{A}b\rangle,
\]
with total rank \(6\). Since the Bol bound for a planar \(5\)-web is \((5-1)(5-2)/2=6\), Bol’s web has maximal rank. Because it is not linearizable, it is exceptional [2401.06711].

A particularly effective reformulation uses cluster \(A_{2}\) coordinates. Writing
\[
X_{1}=u_{1},\quad
X_{2}=u_{2},\quad
X_{3}=(1+u_{2})/u_{1},\quad
X_{4}=(1+u_{1}+u_{2})/(u_{1}u_{2}),\quad
X_{5}=(1+u_{1})/u_{2},
\]
one obtains the cluster web
\[
\mathcal{W}_{\mathcal{X},A_{2}}
=
\mathcal{W}\!\left(
u_{1},\
u_{2},\
(1+u_{2})/u_{1},\
(1+u_{1}+u_{2})/(u_{1}u_{2}),\
(1+u_{1})/u_{2}
\right).
\]
The logarithmic \(3\)-term relations become
\[
Log(X_{\ell-1})-Log(1+X_{\ell})+Log(X_{\ell+1})=0,\qquad \ell=1,\dots,5,
\]
and the antisymmetric symbolic identity is
\[
\sum_{\ell=1}^{5}\nu_{\ell,1}\wedge \nu_{\ell,2}=0,
\qquad
\nu_{\ell,1}=dLog(X_{\ell}),\ \nu_{\ell,2}=dLog(1+X_{\ell}).
\]
In these variables Abel’s identity takes the cluster \(A_{2}\) form
\[
R(u_{1})+R(u_{2})+R((1+u_{2})/u_{1})+R((1+u_{1}+u_{2})/(u_{1}u_{2}))+R((1+u_{1})/u_{2})=\pi^{2}/2,
\]
which makes explicit the relation between Bol’s web and finite-type cluster combinatorics [2401.06711].

## 3. Geometric realizations: \(\mathcal{M}_{0,5}\), \(\mathrm{dP}_{5}\), and \(G_{2}(\mathbb{C}^{5})\)

Bol’s web admits several equivalent geometric incarnations. On the del Pezzo surface \(\mathrm{dP}_{5}\), realized as the blow-up of \(\mathbb{P}^{2}\) at
\[
p_{1}=[1\!:\!0\!:\!0],\quad
p_{2}=[0\!:\!1\!:\!0],\quad
p_{3}=[0\!:\!0\!:\!1],\quad
p_{4}=[1\!:\!1\!:\!1],
\]
the five conic fibrations define a planar \(5\)-web \(\mathcal{W}_{\mathrm{dP}_{5}}\), and one has
\[
\mathcal{W}_{\mathrm{dP}_{5}}
\simeq
\mathcal{W}\!\left(
x,\ y,\ x/y,\ (1-y)/(1-x),\ x(1-y)/(y(1-x))
\right),
\]
hence \(\mathfrak{B}\simeq \mathcal{W}_{\mathrm{dP}_{5}}\) [2401.06711].

The same web is modular. Via the map
\[
(x,y)\longmapsto [\infty,0,1,x,y],
\]
it identifies with the \(5\)-web on \(\mathcal{M}_{0,5}\) determined by the five forgetful morphisms to \(\mathcal{M}_{0,4}\simeq \mathbb{P}^{1}\). This realization is the moduli-theoretic form of Bol’s web and explains its role in the geometry of \(5\) marked points on \(\mathbb{P}^{1}\) [2401.06711].

A third model is provided by Gelfand–MacPherson theory. Let \(X=G_{2}(\mathbb{C}^{5})\), and let \(H\subset GL_{5}(\mathbb{C})\) be the Cartan torus. The weight polytope is the hypersimplex \(\Delta_{2,5}\subset [0,1]^{5}\) with \(\sum t_{i}=2\), and the web-relevant facets are
\[
F_{i}=\Delta_{2,5}\cap \{t_{i}=0\}.
\]
Each face map
\[
\Pi_{F_{i}}:G_{2}(\mathbb{C}^{5})\dashrightarrow G_{2}(\mathbb{C}^{5}/\langle e_{i}\rangle)\simeq G_{2}(\mathbb{C}^{4})
\]
is \(H\)-equivariant and descends, on suitable stable open sets, to the forgetful morphism
\[
\pi_{F_{i}}:\mathcal{M}_{0,5}\to \mathcal{M}_{0,4}.
\]
Consequently, the Gelfand–MacPherson \(5\)-web on \(\mathcal{M}_{0,5}\) is equivalent to Bol’s web [2507.12180].

| Realization | Ambient space | Defining maps |
|---|---|---|
| Affine model | \(\mathbb{C}^{2}\) | \(x,\ y,\ x/y,\ (y-1)/(x-1),\ x(y-1)/(y(x-1))\) |
| Modular model | \(\mathcal{M}_{0,5}\) | five forgetful morphisms |
| del Pezzo model | \(\mathrm{dP}_{5}\) | five conic fibrations |
| Gelfand–MacPherson model | \(G_{2}(\mathbb{C}^{5})//H\) | descended facet maps \(\pi_{F_{i}}\) |

These realizations are not merely equivalent descriptions. They organize distinct aspects of the same object: local web geometry in affine coordinates, modularity on \(\mathcal{M}_{0,5}\), birational surface geometry on \(\mathrm{dP}_{5}\), and torus-quotient representation theory in the Gelfand–MacPherson model.

## 4. Symmetry, canonical reconstruction, and representation theory

Bol’s web carries a natural Weyl-group action. On \(\mathrm{dP}_{5}\), the Weyl group \(W(A_{4})=S_{5}\) acts on the five conic fibrations and therefore on the space of abelian relations. In this action,
\[
AR_{\log}\simeq V^{5}_{[221]},
\qquad
\langle \mathcal{A}b\rangle \simeq \mathrm{sign},
\]
so the logarithmic component is an irreducible \(5\)-dimensional representation, whereas the Abel line is the signature representation [2401.06711].

A second structural feature is canonical reconstruction from slopes. For a web \(\mathcal{W}(V_{1},\dots,V_{k})\), one forms the slope functions \(\zeta_{V_{i}}\) and the canonical map
\[
\Phi_{\mathcal{W}}:\Omega\to \mathcal{M}_{0,k},
\qquad
\Phi_{\mathcal{W}}=[\zeta_{V_{1}},\dots,\zeta_{V_{k}}].
\]
For Bol’s web, one has
\[
\mathcal{W}=\Phi_{\mathcal{W}}^{*}(\mathcal{W}_{\mathcal{M}_{0,5}}),
\]
which gives one of the “canonical algebraizations” of \(\mathfrak{B}\). The other proceeds through the space of combinatorial abelian relations. These two reconstructions formalize a central fact: Bol’s web is rigidly encoded both by its abelian-relation algebra and by its slope geometry [2401.06711].

This rigidity underlies the classical uniqueness theorem for hexagonal planar \(5\)-webs. In that theorem, the exceptional case is not a deformation family but a single local equivalence class, represented by Bol’s web. A plausible implication is that the coexistence of maximal rank, non-linearizability, and modular realization is unusually restrictive in low-dimensional web geometry.

## 5. Generalizations: from \(\mathrm{dP}_{4}\) to the spinor tenfold

A major modern development is the extension of Bol’s web to higher-cardinality webs on del Pezzo surfaces. For a smooth quartic del Pezzo surface \(\mathrm{dP}_{4}\), there are \(\kappa_{5}=10\) conic fibrations, hence a planar \(10\)-web \(\mathcal{W}_{\mathrm{dP}_{4}}\). This web generalizes almost all remarkable features of Bol’s web: it is not linearizable, has maximal rank, is exceptional, and all its abelian relations are hyperlogarithmic of weights \(1,2,3\). Its decomposition is
\[
AR(\mathcal{W}_{\mathrm{dP}_{4}})
=
AR^{1}\oplus \big(AR^{2}_{\mathrm{sym}}\oplus AR^{2}_{\mathrm{asym}}\big)\oplus \langle HLog^{3}\rangle,
\]
with
\[
\dim AR^{1}=20,\qquad
\dim AR^{2}_{\mathrm{sym}}=5,\qquad
\dim AR^{2}_{\mathrm{asym}}=10,
\]
and \(AR^{3}\) \(1\)-dimensional, spanned by \(HLog^{3}\). The weight-\(3\) relation is an explicit antisymmetric hyperlogarithmic identity
\[
\sum_{i=1}^{10} AH_{i}^{3}(U_{i})=0,
\]
where the ten first integrals \(U_{1},\dots,U_{10}\) are rational functions on an affine chart of \(\mathrm{dP}_{4}\), and the residues of \(HLog^{3}\) span exactly \(AR^{2}_{\mathrm{asym}}\) [2401.06711].

The Weyl-group pattern also persists. For \(\mathcal{W}_{\mathrm{dP}_{4}}\), \(W(D_{5})\) acts on the abelian relations, and the decomposition into irreducibles is
\[
AR^{1}\simeq V^{20}_{[2,21]},\qquad
AR^{2}_{\mathrm{asym}}\simeq V^{10}_{[11,111]},\qquad
AR^{2}_{\mathrm{sym}}\simeq V^{5}_{[.221]},\qquad
AR^{3}\simeq \mathrm{sign}.
\]
This is the quartic-del-Pezzo analogue of the \(S_{5}\)-module decomposition for Bol’s web [2401.06711].

An even more structural generalization is the Gelfand–MacPherson web \(\mathcal{W}^{GM}_{\mathcal{Y}_{5}}\) on
\[
Y_{5}=S_{5}^{sf}/H,
\]
the Cartan torus quotient of a stable open subset of the spinor tenfold \(S_{5}=\mathrm{Spin}_{10}/P_{4}\). Here \(Y_{5}\) is a \(5\)-dimensional quasi-projective variety, \(\mathrm{Aut}(Y_{5})\simeq W_{D_{5}}\), and the web is a codimension-\(2\) \(10\)-web defined by ten rational first integrals \(U_{i}:Y_{5}\dashrightarrow \mathbb{P}^{2}\). The key theorem is that \(\mathcal{W}^{GM}_{\mathcal{Y}_{5}}\) is a uniquely defined rank-\(5\) generalization of Bol’s web: its virtual \(2\)-rank is \(11\), its actual \(2\)-rank is also \(11\), and
\[
AR^{2}(\mathcal{W}^{GM}_{\mathcal{Y}_{5}})
=
AR^{2}_{C}(\mathcal{W}^{GM}_{\mathcal{Y}_{5}})\oplus \langle HLOG_{\mathcal{Y}_{5}}\rangle,
\]
with
\[
\dim AR^{2}_{C}=10,\qquad rk^{2}=11=\rho^{2}.
\]
The master \(2\)-abelian relation \(HLOG_{\mathcal{Y}_{5}}\) is unique up to scale, transforms by the signature under \(W_{D_{5}}\), and its residues along the weight divisors generate the combinatorial \(2\)-abelian relations. Pulling \(HLOG_{\mathcal{Y}_{5}}\) back by the Skorobogatov–Serganova embedding \(f_{SS}:\mathrm{dP}_{4}\to Y_{5}\) recovers the weight-\(3\) identity \(HLog^{3}_{\mathrm{dP}_{4}}\) [2507.12180].

This construction places Bol’s web at the first nontrivial stage of a hierarchy: the \(A_{4}\) Grassmannian model yields Bol’s \(5\)-web, while the \(D_{5}\) spinor model yields a codimension-\(2\) \(10\)-web whose master \(2\)-abelian relation generalizes Abel’s five-term identity. The higher-rank theorem for \(r=4,5,6,7\) further suggests a uniform Gelfand–MacPherson pattern across types \(A_{4},D_{5},E_{6},E_{7}\) [2507.12180].

## 6. Bol’s name in circular \(3\)-web geometry: the Blaschke–Bol problem

A distinct classical strand attached to Bol concerns hexagonal circular \(3\)-webs. In that setting, a planar \(3\)-web \(W_{3}\) is given by \(1\)-forms \(\sigma_{1}=0,\sigma_{2}=0,\sigma_{3}=0\) with pairwise transverse kernels at regular points. Under the Blaschke normalization
\[
\sigma_{1}+\sigma_{2}+\sigma_{3}=0,
\]
the Chern, or Blaschke, connection \(\gamma\) is defined by
\[
d\sigma_{i}+\gamma\wedge \sigma_{i}=0,\qquad i=1,2,3,
\]
and the web is hexagonal if and only if its Blaschke curvature vanishes:
\[
d\gamma=0.
\]
This is the analytic form of the classical Blaschke–Bol closure condition [2306.11707].

In Lie sphere geometry, circular \(3\)-webs are encoded by polar curves in \(\mathbb{RP}^{3}\). Recent work has resolved the nonplanar reducible degree-\(3\) cases. If the polar curve is the union of three non-coplanar lines, equivalently three pencils of circles, there are exactly nine Möbius orbits. If the polar curve is a smooth conic plus a line not lying in the conic’s plane, there are exactly fifteen Möbius-equivalence types. By contrast, there is no hexagonal circular \(3\)-web whose polar curve is a rational normal cubic. The same work also classifies webs with a \(1\)-parameter Möbius symmetry and shows that no genuine loxodromic-symmetric hexagonal circular \(3\)-webs exist [2306.11707].

This circular \(3\)-web literature is not the same object as the exceptional planar \(5\)-web usually called Bol’s web, but it extends Bol’s influence in classical web geometry. A plausible interpretation is that Bol’s name now marks two complementary traditions: exceptional maximal-rank planar webs on one side, and the curvature-based classification of hexagonal circular webs on the other. The conjecture proposed in the circular setting—that the polar curve of a hexagonal circular \(3\)-web is algebraic and each irreducible component is a planar curve of degree at most \(3\)—indicates that the Bol program remains active in a modern algebro-geometric form [2306.11707].

Source: https://www.emergentmind.com/topics/bol-s-web