Regular Polygonal Arrangements

We can distribute six consecutive integers, such as 0 to 5, on the 
vertices and edges of a triangle in such a way that the sum along 
each edge is the same.  In fact, there are four essentially distinct
such arrangements, shown below.

       2              4              5                5
                                                       
     3   4          1   3          0   2            0   1
                                                    
   1   5   0      2   5   0      3   4   1        4   2   3

The common sums of these four triangular arrangements are 6, 7, 8,
and 9 respectively.  Of course, each of these arrangements has six
versions if we count rotations and reflections.

Similarly, arrangements of any 8 consecutive numbers can be constructed
on the vertices and edges of a square.  In this case there are six
essentially distinct arrangements, illustrated below.

     5  3  1      4  1  5      3  6  2       7  0  5
     4     6      6     2      1     5       3     1
     0  7  2      0  7  3      7  0  4       2  4  6


                  7  2  1      7  1  3
                  3     5      4     2
                  0  6  4      0  5  6

The common sums of the squares in each of these four columns are 9, 10,
11, and 12 respectively.  It's easy to show algebraically that, for
each square, the sums of opposite pairs of edges must be equal.  If
we let V and E denote the sums of the values on the vertices and edges
respectively, then the common sum is S = (E+2V)/n for an n-sided 
polygon.  We also know that there are a total of 2n vertices and 
edges, so we are distributing the integers 0 to 2n-1, which implies
that the sum of all the numbers is E+V = n(2n-1).  Solving these
equations for E and V, we have

         E = n[2(2n-1) - S]         V = n[S - (2n-1)]

This shows that both E and V must be divisible by n, and also that
the common sum S must be greater than (2n-1) and less than 2(2n-1).
In fact, since there are n edges, each containing n distinct non-
negative integers, the least possible value of E is n(n-1)/2, so
the quantity 2(2n-1) - S must not only be positive, it must be
greater than or equal to (n-1)/2, and thus S must be less than or
equal to 2(2n-1) - (n-1)/2 = (7n-3)/2.  Likewise V must be at least
n(n-1)/2, so S cannot be less than (2n-1) + (n-1)/2 = (5n-3)/2.
For n=4 this means S cannot be less than 8.5 and cannot exceed 12.5,
consistent with the fact that the smallest common sum is 9 and the
largest is 12.

We can place 10 consecutive numbers on the vertices and edges of a
pentagon such that the three numbers along any edge have the same 
sum.  This can be done in six essentially distinct ways.  Rather than
drawing the figures, we can just list the strings of numbers around 
the perimeter of the figure, beginning with a vertex.  The six
arrangements are listed below, along with their common sum.

           common
           sum      
           ------
             11       1738092546
             13       0765283194    6523814907
             14       3476185092    9234716805
             16       6283547091

As in the case of the square arrangements, the alternative patterns
for a single common sum differ by just two transpositions of
consecutive numbers, and these two transpositions are separated 
by a single position.  As always, the common sum must be no less
than (5n-3)/2 and no greater than (7n-3)/2, which gives the limits
of 11 and 16 for n=5.

For hexagonal arrangements we have n=6, so the only possible common
sums are in the range from 13.5 to 19.5, i.e., the set 14, 15, 16, 17,
18, and 19.  The essentially distinct solutions for each of these 
common sums are listed below, along with the corresponding values of
V and E, and (in the right-most columns) the sums of the squares of 
the vertex and edge numbers.

     S=14, V=18, E=48
      6  3  5  7  2 11  1  9  4 10  0  8      82 424
      5  3  6  7  1 11  2  8  4 10  0  9      82 424
      8  5  1 11  2  9  3  7  4 10  0  6      94 412
                                                     
     S=15, V=24, E=42
      0  6  9  2  4 10  1 11  3  5  7  8     156 350
                                                      
     S=16, V=30, E=36
      6  0 10  1  5  7  4  9  3 11  2  8      190 316
      7  3  6  8  2  4 10  1  5 11  0  9      214 292
      8  5  3 11  2  4 10  0  6  9  1  7      214 292
      2 10  4  1 11  5  0  9  7  3  6  8      226 280
      0  9  7  5  4  1 11  3  2  8  6 10      226 280
      7  9  0  6 10  4  2 11  3  5  8  1      226 280
                                                    
     S=17, V=36, E=30
      9  3  5  1 11  2  4  6  7 10  0  8      292 214
     11  2  4  8  5  3  9  1  7 10  0  6      292 214
     11  2  4 10  3  6  8  0  9  7  1  5      292 214
      9  0  8  6  3  4 10  2  5 11  1  7      280 226
      9  3  5  8  4  2 11  0  6 10  1  7      280 226
      6  4  7  2  8  0  9  3  5 11  1 10      256 250
                                                     
     S=18, V=42, E=24
      2  9  7  1 10  0  8  6  4  3 11  5      354 152
                                                     
     S=19, V=48, E=18
     11  1  7  2 10  0  9  4  6  8  5  3      412  94
     10  0  9  3  7  1 11  2  6  8  5  4      412  94
      3  5 11  1  7  4  8  2  9  0 10  6      424  82

These results show some interesting dualities.  For example, the first
solution with S=14 is the dual of the last solution with S=19, because
they have the same split of the numbers into vertex and edge numbers,
except that the types are reversed.  In other words, the numbers that
are assigned to vertices in the first solution are assigned to edges
in the last solution, and vice versa.  These two solutions are shown
pictorially below.

           6   3   5                   3   5  11
                                                       
        8             7             6             1
                                                
      0                 2        10                  7
                                                       
        10           11             0             4
                                                             
           4   9   1                   9   2   8


We can generalize the problem by allowing k numbers to be assigned 
to each edge, where k is greater than 1.  For example, we can 
construct triangles such as the two shown below

               1                          1
                                                           
            6     2                     6   10
                                                             
          8         9                 7        3
                                                              
        4    5   3    7             8           12
                                                               
                                  9   2   4   11   5                                 
                             

In general, to construct a triangle with k numbers on each edge (not
counting the vertices), let N = k+2 denote how many numbers are on
each edge counting the vertices, and then assign the numbers 1, N,
and 2N-1 to the three vertices.  Then, to fill in the edge numbers,
arrange the remaining numbers into three columns, ascending on each
row alternately from the left and the right.  For example, to 
construct a triangle with N=6 numbers on each side, we first assign
the numbers 1, 6, and 11 to the vertices, and then we arrange the
remaining numbers in three columns as shown below

                     2  3  4
                     8  7  5
                     9 10 12
                    15 14 13

Each column sums to 34, but we actually need columns that sum to
34-5, 34, and 34+5 in order to place these numbers on the edges [6,11],
[1,11], and [1,6] of the triangle, respectively.  We can do this by
leaving the middle column untouched, and making the transpositions
(8,5) and (15,13) in the outer columns.  This decreases the left-hand
column by 5, and increases the right hand column by 5.  As an alternate
solution, we could make the transpositions (2,4) and (9,12), which
would increase the left column by 5 and decrease the right column by
5.  Notice that this must always lead to a solution, because the two 
outer columns will always differ by 2 or 3, so we can form any 
required adjustment (greater than 1) by transposing pairs.

Return to MathPages Main Menu