Professional Documents
Culture Documents
Abstract
In this paper we take the concept of the magic square and extend
it to the Fano Plane. That is, as each point in the plane is assigned a
numerical value, we will investigate whether or not the sums along each
of the lines on the plane can ever be made to equal all of the others. More
specifically, we first explore the case where the values are allowed to be
any real numbers. We then relax the constraints by ignoring the sums
along one or more lines. Finally, we consider the case where the values
and sums are integers considered modulo n for any fixed integer n > 1.
1 Introduction
For thousands of years, magic squares have been a point of interest and topic
of study in many different cultures. Early magic squares were squares that
were divided into an n × n array of cells, with different letters placed into these
cells, so that the different order of these letters would result in the spelling of
different words. From these squares developed the kind that served as the main
inspiration of this paper. This new type of magic square has each cell is assigned
a different numerical (typically, integer) value, where the values in each row, in
each column, and along the two main diagonals, all sum to the same “magic
total”. For example, Figure 1 shows a 3 × 3 magic square with magic total 15.
4 9 2
3 5 7
8 1 6
1
Having been around for so long, it comes as no surprise that a substantial
amount of research has been done on the existence, properties, and formation of
magic squares (see e.g. [2]). Moreover, as time has passed, the idea of “magic”
shapes has been extended to many other polygons (see e.g. [4]). While not a
polygon, we aim to extend the notion of “magicness” to the Fano Plane.
Named after Italian mathematician, Gino Fano, the Fano Plane (also called
the finite projective plane of order two), is the smallest projective plane in pro-
jective geometry. (For some introductory material on finite projective geometry
the authors suggest [3] or [1].) The Fano Plane contains seven points and seven
lines, with each line containing exactly three points and each point lying on
exactly three lines. A graphical representation of the Fano Plane can be seen in
Figure 2. (Note that the circle also represents a line).
There are many well-known features of the Fano Plane which will become
useful in what follows. For one, observe that given any two points in the Fano
Plane, there exists a unique line containing both of them. Similarly, given
any two lines in the plane, there exists a unique point where they intersect.
Moreover, up to symmetry (collineation), we can say that any point in the Fano
Plane is equivalent to all of the others and similarly, any line in the Fano Plane
is equivalent to all of the others. In fact, because each pair of lines intersects at
a unique point and all points are equivalent up to symmetry, it follows that all
pairs of lines are equivalent up to symmetry as well. We will make use of these
facts in many of the arguments that follow.
We will assign numerical values to each of the points on the Fano Plane, just
as numerical values were assigned to the cells of the n × n magic squares that
we saw earlier. Then, we will call a shape quasi-magic if the sums of the entries
along each line are all equal to each other. Further, a shape will be called magic
if it is quasi-magic and all of its entries are distinct. To get a feel for the idea, let
us consider the following example. Traditionally, the integers 1, 2, . . . , n2 were
assigned to the n2 cells in an n × n magic square. Mimicking that practice, we
use the numbers 1 through 7 in our example, and assign them to random points
(see Figure 3 below).
In this case the sums along each of the lines are 14, 14, 14, 14, 12, 10, and
2
7
2 1
4
5 3 6
Figure 3: A Fano Plane with its points labeled by the values 1 through 7.
6. By definition, this Fano Plane is not magic (or even quasi-magic). As we will
see in Section 3, this attempt (and failure) at a magic Fano Plane is actually
representative of the general case. In fact, Theorem 1 shows that the Fano Plane
cannot even be quasi-magic unless all of the entries are equal. In view of that
fact, it becomes interesting to ask what constraints could be relaxed in order to
make the Fano Plane magic in some other sense.
One way to relax the constraints is to only require that some chosen subset
of the lines have equal sums. In Section 4 we explore this idea and determine
the minimum number of lines whose sums must be ignored in order to make
the Fano Plane magic. Another way to relax the constraints is to take the
numerical values to be integers and to consider the sums modulo n. That is,
for any integer n > 1, we say the Fano Plane is quasi-magic mod n if the sums
along each line are all congruent modulo n. Similarly, the Fano Plane is magic
mod n if it is quasi-magic mod n and all values are distinct when considered
modulo n. In Section 5, we describe all Fano Planes that are quasi-magic mod
n, for any integer n > 1.
2 Supporting Lemmas
As will soon be apparent, much of the work in discerning whether or not a given
Fano Plane is “magic,” in any of the senses described above, is contingent upon
the manipulation of large chains of equalities (or congruences). The following
two lemmas will be quite useful in making quicker work of this task.
Lemma 1 Let a, b, c, d ∈ R.
If a+b = c+d and a+c = b+d, then a=d and b=c.
Proof: If we subtract the second equation from the first, we are left with
b − c = c − b, which gives us 2b = 2c. Therefore, b = c. Substituting this into
our first equation, we can see that a = d. QED
3
Proof: Looking at the first two equations, we can cite Lemma 1 as proof that
a = d and b = c. Now, let us look at the third equation: a + d = b+c. We
can substitute the two equalities to obtain a + a = b + b ⇒ 2a = 2b ⇒ a = b.
Therefore, by the transitive property, we see that a = b = c = d. QED
B F
D
C E G
In order for this generic Fano Plane to be magic it must also be quasi-magic,
so the sums along each of the lines must all be equal. That is, we must have
4
Theorem 1 now allows us to answer our original question with the following
immediate corollary:
4 Ignoring Lines
We have shown that when all seven lines are checked against each other, the
Fano Plane can never be magic. However, we cannot necessarily say the same
thing if we only require that six of the sums be equal, or five, or four, etc. A
line in the Fano Plane is ignored if its sum is not required to be equal to the
others. Recall that, by symmetry, each line of the Fano Plane is equivalent to
all of the others. Therefore, if ignoring a single line, it does not matter which
line is actually ignored. Using the same generic model from Figure 4, we can
show that the Fano Plane, with one line ignored, is never magic.
Theorem 2 If the Fano Plane, with one line ignored, is quasi-magic, then the
three nodes on the ignored line will all have the same value and the four that do
not appear on that line will all have the same value. Moreover, the two values
are independent of each other.
Proof: Without loss of generality, we will ignore ABC. Then, the following
chain of equalities must hold:
Corollary 2 The Fano Plane with real entries and one line ignored is never
magic.
Next, we consider the case of ignoring two lines. It is important to recall
that given any two lines in the Fano Plane, they intersect at a unique point.
Since all points are equivalent by symmetry, it follows that similarly, all ways
of ignoring two lines are equivalent. We now show that the Fano Plane, with
two lines ignored, is also never magic.
Theorem 3 If the Fano Plane, with real entries and two lines ignored, is quasi-
magic then at most four different values can be assigned to its nodes. Moreover,
if (for example) the two ignored lines are ABC and AF G, then B = C, F = G,
D = E, and A = B+D-F.
5
Proof: As discussed above, we may choose to ignore ABC and AF G without
loss of generality and thus we have A + D + E = C + E + G = C + D + F =
B+D+G = B+E+F . The following implications can be drawn: E+G = D+F
and D + G = E + F . By Lemma 1, we know that D = E and F = G. Applying
these to lines BEF and CDF we find that B +D+F = C +D+F and therefore,
B = C. Moreover, lines ADE and BEF imply that A + 2D = B + D + F and
hence, A = B − D + F . Now, observe that for any values of B, D, and F , if
A = B − D + F , then all sums are equal to B + D + F . QED
Again, we get an immediate corollary.
Corollary 3 The Fano Plane with real entries and two lines ignored is never
magic.
Theorem 4 The Fano Plane, with three lines ignored, can be made magic if and
only if the three lines do not share a common point of intersection. Moreover, if
the three lines ignored do intersect at a common node, for example lines ABC,
BDG, and BEF , then being quasi-magic implies that A = C, D = G, E = F,
and B can be assigned any value.
6
Theorem 5 The Fano Plane with real entries and four, five, or six lines ig-
nored, can be made magic.
Proof: Choose any four (or five, or six) lines to be ignored. Since every point is
on exactly three lines, it follows that there exist three lines among those selected
that do not all intersect at a common point. Therefore, by Theorem 4, we know
that the Fano Plane can be made magic when ignoring only these three lines.
Ignoring additional lines only makes it easier to find a magic labeling. QED
5 Magic Modulo n
The final possibility that we would like to explore is whether the Fano Plane
can be made magic if the entries and sums are integers considered modulo n for
some integer n > 1. As it turns out, whether the Fano Plane can be quasi-magic
in a non-trivial way depends on whether n is even or odd.
In the case when n is odd, the Fano Plane can only be quasi-magic if all of
the entries are congruent to each other modulo n. The case when n is even has a
little bit more freedom. There, we find that the Fano Plane can be quasi-magic
with up to two distinct non-congruent entries. Moreover, when those values are
distinct, they must differ by exactly n2 when compared modulo n.
Proof: Certainly if all values are congruent modulo n, then the Fano Plane
will be quasi-magic mod n. Next, consider the case where n is even and let k
be any residue modulo n. Then without loss of generality we may take A ≡
B ≡ C ≡ k (mod n) and D ≡ E ≡ F ≡ G ≡ k + n2 (mod n). Clearly, the sum
for line ABC is congruent to 3k (mod n). Since every other line must intersect
with ABC at a exactly one point, it follows that each of the other six sums are
congruent to k + (k + n2 ) + (k + n2 ) ≡ 3k (mod n) as well.
Conversely, suppose that the Fano Plane is quasi-magic mod n so that
If all values are congruent modulo n then we are done, so assume without
loss of generality that B 6≡ C (mod n). Therefore, B ≡ C + i for some i ∈
{1, 2, . . . , n − 1}. Applying this to the sums from lines CDF and BDG we find
that C + D + F ≡ C + i + D + G, which implies that F = G + i(mod n).
Applying both criteria to the lines CEG and BEF we find that C + E + G ≡
C + i + E + G + i(mod n), which implies that 2i ≡ 0(mod n). Since 0 < i < n
it follows that n must be even and i = n2 .
7
So B ≡ C + n2 (mod n) and F ≡ G + n2 (mod n). Applying these to the
lines CEG and BDF we find that C + E + G ≡ C + D + G + n2 (mod n)
and hence E ≡ D + n2 (mod n). Now we can apply all of these facts to the
lines ABC and ADE to find that A + 2C + n2 ≡ A + 2D + n2 , which implies
that 2C ≡ 2D(mod n). There are now two possible cases to consider: either
C ≡ D(mod n) or C ≡ D + n2 (mod n)
Case 1: Suppose C ≡ D(mod n). Since B ≡ C + n2 (mod n), and E ≡
D + n2 (mod n) it now follows that B ≡ E(mod n). Applying all of our known
congruences to lines ABC and BEG we find that A + 2C + n2 ≡ G + 2C +
n
2 (mod n), which implies that A ≡ G(mod n). By transitivity, we now have
that F ≡ A + n2 (mod n) as well. Now considering lines ABC and AF G we find
that 3A + n2 ≡ A + 2C + n2 (mod n) and hence 2A ≡ 2C(mod n). Once again,
we have two cases to consider: either A ≡ C(mod n) or A ≡ C + n2 (mod n).
Subcase 1.1: Suppose A ≡ C(mod n). By the transitive property, we now
have that A ≡ C ≡ D ≡ G(mod n). We can also now say that C + n2 ≡ D + n2 ≡
G + n2 (mod n) and hence B ≡ E ≡ F (mod n). Observe that BEF is, in fact,
one of the lines in the Fano Plane and that the values on and off the line differ
by exactly n2 .
Subcase 1.2: Suppose A ≡ C + n2 (mod n). We know that B ≡ C +
n
2 (mod n) and thus, A ≡ B(mod n). By the transitive property, we now see
that A ≡ B ≡ E ≡ G(mod n). Therefore, B + n2 ≡ E + n2 ≡ G + n2 (mod n) and
thus, C ≡ D ≡ F (mod n). Observe that CDF is, in fact, one of the lines in the
Fano Plane and that the values on and off the line again differ by exactly n2 .
Case 2: Suppose C ≡ D + n2 (mod n). We already know that C ≡ B +
n n
2 (mod n), so B ≡ D(mod n). Similarly, we know that E ≡ D + 2 (mod n),
so we can say that C ≡ E(mod n). Substituting all of our known congruences
into lines ADE and BDG we find that A + 2D + n2 ≡ 2D + G(mod n). This
implies that A ≡ G + n2 (mod n) and therefore, A ≡ F (mod n). Substituting
all of these congruences into the lines AF G and ABC we find that 3A + n2 ≡
A + 2D + n2 (mod n) which implies that 2A ≡ 2D(mod n). Once again, we are
left with two cases to consider: either A ≡ D(mod n) or A ≡ D + n2 (mod n).
Subcase 2.1: Suppose A ≡ D(mod n). By the transitive property, we have
that A ≡ B ≡ D ≡ F (mod n). Thus, B + n2 ≡ D + n2 ≡ F + n2 (mod n) and
therefore, C ≡ E ≡ G(mod n). Observe that CEG is a line in the Fano Plane
and that the values on and off the line again differ by exactly n2 .
Subcase 2.2: Suppose A ≡ D + n2 (mod n). We already know that C ≡
D + n2 (mod n). This implies that A ≡ C(mod n) and thus, by transitivity,
A ≡ C ≡ E ≡ F (mod n). Therefore, C + n2 ≡ E + n2 ≡ F + n2 (mod n) which
implies that B ≡ D ≡ G(mod n). Observe that BDG is a line in the Fano
Plane and that the values on and off the line again differ by exactly n2 . QED
The theorem leads to one final corollary:
Corollary 4 The Fano Plane is never magic modulo n for any n > 1.
8
References
[1] Albert, A., Finite planes for the high school, Mathematics Teacher, Vol.
55, pp 165-169, 1962.
[2] Farrar, Mark S., “Magic Squares”, BookSurge LLC, 2006.
[3] Malkevitch, Joe Finite Geometries?, AMS Feature Column, Sept. 2006.
[4] Trenkler, Marián, Magic Stars, ΠME Journal, Vol. 11, No. 10, pp 549-
554, 2004.
David Nash
David is an assistant professor in the Department of Mathematics and Com-
puter Science at Le Moyne College. He received his B.S. in Mathematics and
Physics from Santa Clara University (‘04) and his M.S. and Ph.D. in Mathe-
matics from the University of Oregon (‘10).
Department of Mathematics and Computer Science, Le Moyne College 1419
Salt Springs Rd Syracuse, NY 13214 nashd@lemoyne.edu