This page is about another numerical experiment. In 1997 Horace Crater
sent me a study that he had done concerning the D&M pyramid in which
he considered a certain infinite family of pentagons. One of these pentagons
was the hypothetical model for the base of the D&M pyramid proposed
by Erol Torun. For a description of this infinite family of pentagons,
click here. Each pentagon is determined completely
by choosing one specific angle D. For example, D = 69.4 corresponds
to Torun's model. If one changes the choice of the angle D, the shape
of the pentagon changes, although it will still share some specific properties
with the pentagon in Torun's Model.

In the Fall of 1997, I asked Dr. Crater to choose another angle D *completely
at random* so that I could see what kinds of numerical relationships
could be discovered in the corresponding pentagon. Since there were
so many possibilities for the kinds of relationships that might exist,
I felt quite confident that I would find lots of striking relationships.
The angle he chose was D=84.3. Here is a rough picture of that
pentagon, which is the subject of this page.

This pentagon, which we will refer to by the letter **P**, is a precisely
specified pentagon. All of the angles, lengths, and areas associated with
the pentagon could, in principle, be computed to any desired accuracy.
Over a period of many months, mostly on Sundays, I devoted quite a bit
of time to searching for numerical relationships. It was actually a lot
of fun, perhaps something like a mathematical analogue of doing the Sunday
New York Times Crossword Puzzle. Instead of a dictionary, all that I needed
was my hand calculator. I started with lengths and areas, and never
got around to considering relationships involving the angles of the above
pentagon. I looked exclusively for relationships where the percentage
error was less than .1% and found about 100 relationships which struck
me as rather elegant and natural, many of which were impressively
accurate. On this page, we will give a selection of those relationships.
To indicate the accuracy of the relationships, I will use the following
notation. If x and y are two positive numbers, then I will write x
~
y to mean that x and y are approximately
equal. In parentheses, the error in the approximation will be indicated
by both the absolute value |y-x|
of the difference between y and x and the percentage error
(in comparison to the number y). For example, if x=10.2
and y=10, then I will write 10.2
~
10 and put in parentheses
( .2 , 2%) since |10-
10.2| = .2 and .2
is 2% of the number 10.

*Relationships Involving the
Golden Section.*

The Golden Section is the number (1+\/5)/2,
which is often called *phi*. There is one relationship
involving
*phi* which is valid for all of the pentagons in the family
considered by Crater, no matter what the angle D is chosen to be.
This relationship states that d/b
= *phi *, where b
is the length of the base of the pentagon as pictured above, and d
is
the length of the diagonal of the pentagon joining the vertex at the top
to either of the endpoints of the base. In the illustration which is found
here,
d
is the length of either one of the green line segments, i.e., the distance
between the vertex A and either of the vertices B or B', and** **b
is the distance between B and B'. This relationship is intentional
and is implicit in part 1of the description of Crater's family of pentagons
given
here. But for the specific pentagon
pictured above (with D=84.3^{o}), there is another interesting
relationship involving *phi* which
is simply an unintentional consequence of the randomlychosen value of D.
It involves the length f of either one of
the two red line segments in the illustration. The point M is the
midpoint of the side of the pentagon joining B to C (and M'
is the midpoint of the side B'C'). Thus f
is the distance between M and C' (or between M' and C).

f /b
~
*phi
*(.00018 , .012%)

Thus, if b is taken as the
unit of measurement, then the length d is
exactly equal to *phi* and the length
f
is very close to *phi*.

*Relationships Involving Areas
and the Number e.*

There are many rather striking numerical relationships involving the areas
of various parts of this pentagon. The relationships that I will give here
were chosen because they are quite accurate, they seem rather elegant to
me, and they show some thematic unity in that they all involve areas coming
from the above pentagon and areas of squares with sides of length
2/e, 3/e, 4/e, 5/e, and 6/e.

**1**. Let's denote the above pentagon
by the letter **P. **There is another pentagon **Q **
which has precisely the same perimeter as **P**, namely
the pentagon formed by flipping the two top edges (which have length c)
inward. (See the picture below.) Then not only are the perimeters
equal, but the areas have a ratio which is approximately equal to
the area of a square with sides of length 2/e
. The two pentagons
**P** and **Q** share four vertices. The
remaining fifth vertex of
**Q** is precisely the "*special interior
point*" which plays a role in the definition of the pentagon **P.**

Area(**Q**)
/Area(**P**)
~
(2/e)^{2 }^{ }
( .000002 , .004%)

**2**. The base of the pentagon has length b
. The two lower sides of the pentagon each have length a
. This relationship involves the areas of three squares: one with
sides of length b, one with sides of length
a,
and one with sides of length 3/e.

a^{2}/
b^{2
}~
(3/e)^{2}
( .0008 , .07% )

**3**. This relationship involves the areas of
two bilaterally symmetric subregions of the pentagon and the area of a
square with sides of length 4/e. Note that the four vertices
of the quadrilateral
**T** are precisely specified points: the apex
of the pentagon, the two endpoints of the base, and the *special interior
point* . (See the picture below.) I was quite excited when
I discovered this extremely accurate relationship one Sunday. But
the following Sunday, I realized (with a little disappointment) that all
of the pentagons in Crater's family satisfy this very same relationship.
Even though the quantities Area(**S**)
and Area(**T**) will change, the ratio
Area(**S**)/Area(**T**)
turns out to always equal the largest of the four roots of the polynomial
x^{4}+2x^{3}-6x^{2}-7x+1.
It just so happens that x = (4/e)^{2}
is an extremely good approximation to this root.

Area(**S**)
/Area(**T**)
~
(4/e)^{2}
( .000013 , .0006% )

**4**. Now we give a Pythagorean type relationship
involving the areas of the squares on the five sides of the pentagon **P
**and
a square whose sides have length 5/e .

a^{2}
+ a^{2} + c^{2}
+ c^{2} - b^{2 }~
(5/e)^{2}
( .0005 , .01% )

**5**. This relationship is a double
one involving the area and the perimeter of the pentagon **P**
and the area and perimeter (which is usually called the circumference)
of a certain circle **C**. This circle **C** is tangent
to the base and the upper two sides of the pentagon, but *not* tangent
to the lower two sides. (See the picture below. The picture is slightly
deceptive because it almost makes it appear that the circle **C **is
tangent to all five sides of the pentagon **P**.) The areas of
squares with sides of length 2 and of length
6/e
are also involved.

Area(**C**)
/
Area(**P**) + 2^{2
}~ (6/e)^{2}
( .0000047 , .0006% )

Peri(**C**) /
Peri(**P**) ~ \/3
/
2
(.000098 , .01% )

COMMENTARY: This experiment is a good illustration
of "The Power of Randomness." Coincidences abound!! They are
there to be found if one spends enough time looking for them. And
it seems that some people spend a lot of time looking for them in various
contexts, and, sadly, fool themselves and others into thinking that
they have discovered something significant.