A Kobon triangle is a triangular region in an arrangement of straight lines. The Kobon triangle problem asks for the largest number of nonoverlapping triangles
that can be constructed using
lines (Fujimura 1978; Gardner 1983,
p. 170). While it appears to be very difficult to find an analytic expression
for the general
th
term, upper bounds are known. The first few known terms for
, 4, ... are 1, 2, 5, 7, 11, 15, 21, 25, 32, 38, 47,
... (OEIS A006066).
A. Wajnberg (pers. comm., Nov. 18, 2005) found a configuration for containing 25 triangles (left figure).
A different 10-line, 25-triangle construction was found by Grünbaum (2003, p. 400),
and a third configuration is referenced by Honma. Two other distinct solutions were
found in 1996 by Grabarchuk and Kabanovitch (Kabanovitch 1999, Pegg 2006).
Honma illustrated an 11-line, 32-triangle configuration and another solution was found by Kabanovitch (1999; Pegg 2006). Savchuk (2025) ruled out 33 triangles
for simple arrangements. Parpalak and Utkin (2026b) also ruled out 33 when three
or more lines may meet at a point, establishing .
Kabanovitch (1999) also found a 12-line, 38-triangle configuration, and a 13-line 47-triangle configuration (Pegg 2006).
Maiorana (2026) and, independently, Parpalak and Utkin (2026b) found 14-line arrangements with 54 triangular faces. The figure shows Maiorana's
first arrangement using coordinates supplied by E. Pegg, Jr. (pers. comm., Nov. 6,
2026). The highlighted points are the two triple points.
Parpalak and Utkin (2026b) ruled out 55 or more such faces even for arrangements
of pseudolines, so .
T. Suzuki (pers. comm., Oct. 2, 2005) found the above configuration for .
A number of other configurations that are either maximal or the best known are summarized and linked to on the page for the sequence OEIS A006066
in the Online Encyclopedia of Integer Sequences. Savchuk (2025) enumerated all possible
Kobon arrangements for , 5, 9, 15, and 17 and also
found new optimal Kobon arrangements for 23 and 27 lines.
The simplest bound was found by Saburo Tamura, who showed that for ,
|
(1)
|
(Felsner and Kriegel 1999, Clément and Bader 2007), where is the floor function.
Clément and Bader (2007) proposed tightening Tamura's bound by one for
. Their proof has a gap. The common point
of the three medians of a triangle
belongs to six pairs of triangles sharing a side in
the arrangement of the medians and sides, whereas their argument allows at most two.
The expression
for
was also printed incorrectly as
in their paper.
For a simple arrangement every pair of lines intersects and no three meet at one point. For even , the maximum number
of triangular regions in
such an arrangement satisfies the bound of Bartholdi et al. (2007) and the
stronger bound of Blanc (2011),
|
(2)
|
and
|
(3)
|
Blanc's bound is exceeded by some arrangements with triple points. In particular, Parpalak and Utkin (2026b) constructed arrangements with
for
having two triple points
and
triangles, one more than Blanc's bound.
The following table compares Tamura's general upper bound, the proposed Clément-Bader bound, and the two upper bounds for simple arrangements.
| OEIS | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | |
| Tamura bound | A032765 | 1 | 2 | 5 | 8 | 11 | 16 | 21 | 26 | 33 | 40 | 47 | 56 | 65 | 74 | 85 | 96 | 107 | 120 |
| Clément-Bader proposed | 1 | 2 | 5 | 7 | 11 | 15 | 21 | 26 | 33 | 39 | 47 | 55 | 65 | 74 | 85 | 95 | 107 | 119 | |
| Bartholdi et al. bound | 2 | 7 | 15 | 25 | 38 | 54 | 72 | 94 | 117 | ||||||||||
| Blanc bound | 2 | 7 | 14 | 25 | 38 | 53 | 72 | 93 | 116 |
Exact values are known for some special forms of , with
a nonnegative integer (Bartholdi
et al. 2007, Parpalak and Utkin 2026a).
Parpalak and Utkin (2026b) also established unrestricted upper bounds matching the known constructions for 6, 8, 10, and 12 lines.
Bader's 16-line construction with 72 triangles reaches the simple bound (Bader, Bartholdi et al. 2007, Blanc 2011). For unrestricted
arrangements, these results give , using Tamura's
bound for the upper limit. Parpalak and Utkin (2026b, Fig. B.8) constructed
an arrangement of 20 lines that gives the lower bound
.
Known exact values of
for
are summarized in the following table.
| comment | reference | ||
| 3 | 1 | ||
| 4 | 2 | upper limit construction | |
| 5 | 5 | ||
| 6 | 7 | upper limit construction | |
| 7 | 11 | ||
| 8 | 15 | upper limit construction | |
| 9 | 21 | ||
| 10 | 25 | upper limit construction | Kabanovitch (1999), Grabarchuk, Grünbaum (2003), Wajnberg (pers. comm., 2005) |
| 11 | 32 | exhaustive upper bound | Savchuk (2025), Parpalak and Utkin (2026b) |
| 12 | 38 | upper limit construction | Kabanovitch (1999) |
| 13 | 47 | ||
| 14 | 54 | exhaustive upper bound | Maiorana (2026), Parpalak and Utkin (2026b) |
| 15 | 65 | Suzuki (pers. comm., 2005) | |
| 17 | 85 | Clément and Bader (2007) | |
| 19 | 107 | upper limit construction | Wood |
| 21 | 133 | upper limit construction | Savchuk (2025) |
| 23 | 161 | upper limit construction | Savchuk (2025) |
| 25 | 191 | ||
| 27 | 225 | upper limit construction | Savchuk (2025) |
| 29 | 261 | ||
| 31 | 299 | upper limit construction | Wood |
| 33 | 341 | ||
| 35 | 385 | upper limit construction | Utkin and Parpalak (2026) |
| 37 | 431 | Parpalak and Utkin (2026a) | |
| 41 | 533 | upper limit construction | Parpalak and Utkin (2026a) |
| 43 | 587 | upper limit construction | Utkin and Parpalak (2026) |
| 45 | 645 | upper limit construction | Parpalak and Utkin (2026a) |
| 49 | 767 |