A MATHEMATICIAN'S PROGRESS

Author(s): U. ALFRED
Source: The Mathematics Teacher, Vol. 59, No. 8 (DECEMBER 1966), pp. 722-727
Published by: National Council of Teachers of Mathematics
Stable URL: https://www.jstor.org/stable/27957464
Accessed: 03-05-2020 01:52 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

National Council of Teachers of Mathematics is collaborating with JSTOR to digitize,

preserve and extend access to The Mathematics Teacher

This content downloaded from 128.32.10.230 on Sun, 03 May 2020 01:52:11 UTC
All use subject to https://about.jstor.org/terms
 "One thing leads to another/' and
exploration in mathematics gives insight

A MATHEMATICIAN'S PROGRESS

By BROTHER U. ALFRED most unspecialized case is that of lines of

St, Mary's College which none are parallel and no three pass
St. Mary's College, California through one point.
Starting with this simplest case one can
STUDENTS often note that mathe draw figures and prepare a table as fol
matics in its full dress suit of postulates, lows :
definitions, theorems, and formal proofs is No. of lines No. of regions
more of a fossil than a living organism. The 0 1
reader who scans such material has no idea 1 2
2 4
of the hours and days of cogitation spent
3 7
on even minor points, the misdirections 4 11
that led to blind alleys, the sudden inspira 5 16
tions, and all the various thrills and dis
appointments that make mathematics a It is possible to arrive at a tentative for
truly human experience. Furthermore, mula in a number of ways from this data.
with no particular idea of just how the For example, one might note that the
result was achieved, the student loses a successive differences in the number of
great deal of useful information regarding regions are 1, 2, 3, 4, 5, etc. From this it
methods of procedure?and it is precisely could be seen that the total is 1 + 1 + 2
such information that would prove most + 3 + 4 + 5 + ? ? ? and hence for lines
valuable to him and others who are trying in a plane, the total number of regions
to penetrate the domains of mathematics would be
by personal thought. n(n + 1)
With this in mind, we find justification S - 1 +
for presenting the peregrinations of one
student relative to some related questions While there is every reason to believe
of fairly elementary proportions. As is that this formula holds in general, due to
usual in mathematics, the work began the way in which the numbers progress,
with a question : Into how many regions is the result calls for a proof. When the
a plane cut by a given number of straight fourth line, for example, is drawn, it cuts
lines? three lines in three points, so that there are
Probably the first thing a mathemati four different regions on this fourth line.
cian learns to do in trying to find an an These correspond to four regions in the
swer to a question that seems to involve plane through which the fourth line passes.
some complications is to restrict the prob As it divided these regions each into two
lem to its most elementary form. Lines in a parts, putting in the fourth line added four
plane could have certain specializations regions to those already in the plane. In
such as being parallel, or they could form a general, by the same line of reasoning, the
pencil of lines in which three or more lines nth line adds spaces. Hence, the given
pass through one point. The simplest and formula is justified.

722 The Mathematics Teacher | December 1966

This content downloaded from 128.32.10.230 on Sun, 03 May 2020 01:52:11 UTC
All use subject to https://about.jstor.org/terms
 Lines in the Plane/ sections in which three lines meet and
in General three in which two lines meet. Hence, the
The result, however, does not apply to equivalent number of intersections would
be:
the case of parallel lines or to situations
involving pencils of lines. For example, / = 4(3 - 1) +3(2 - 1) = 11
L = 6
five parallel lines divide the plane into six
regions, whereas the formula gives 16 as S = L + J + 1 = 6 + 11 +1 =18.
the number of regions when is 5. Evi
dently, the single variable, "number of A count of the actual number of regions as
lines," is not sufficient to determine the shown by the accompanying figure agrees
number of regions into which the plane is with this result. It was found in all cases
divided. What other variable could be considered that the relation held, using
used? How about taking into account the the interpretation that the equivalent
number of intersections as well? number of intersections corresponding to A;
For the general case of lines where intersecting lines is fe ? 1.
none are parallel and no three go through
any one point, the number of lines, L, is
equal to n; the number of intersections is
the number of ways of taking things two
at a time, so that if I is the number of
intersections,
/== n(n - l) ?
2
Since
n(n - 1) . n(n + 1)
?i
- equals-?
Thus there was an empirical relation prov
it follows that able in some particular instances, but as
yet not completely confirmed for the gen
s = l+nCn+_l)=/+L + l eral case of any set of lines in a plane.

Could this more general, Euler-like, Relations in Three-Dimensional Space

formula be shown to apply to other cases At this juncture, the corresponding
as well? We studied various situations. For problem in three dimensions was attacked :
parallel lines, there are +- 1 regions. Into how many regions is geometric space
L = n, 7 = 0, so that S = 1 + + 0 of three dimensions divided by a given
gives the correct result. For a pencil of number of planes?
rays, there are 2n spaces. This would Once again, the divide-and-conquer
mean that the number of intersections approach was used. In this instance, the
would have to be interpreted as ? 1. following restrictions were made in order
Does this hold in general for all possible to arrive at the simplest case:
configurations of lines in the plane? Ini
1. Each set of three planes meets in a
tially, there was no way of proving this.
point, but not more than three planes pass
However, various configurations of lines
through any one point.
could be tried out to see whether a counter
2. Each pair of planes meets in a line,
example could be found.
but not more than two planes pass through
One of the many cases we considered
any one line.
consisted of a triangle and three lines
through the vertices meeting in a point Two thoughts were carried over from
within the triangle. There are four inter the two-dimensional problem :

A mathematician's progress 723

This content downloaded from 128.32.10.230 on Sun, 03 May 2020 01:52:11 UTC
All use subject to https://about.jstor.org/terms
 1. Just as the intersection of a line with and a combination of the two middle
the previous lines provided the means of terms proves the equivalence with S, as
solving the plane problem, so the intersec determined above.
tion of a plane with the previously given
Dimensional Generalization
planes should furnish the solution of the
space problem. The form o?P + L+ V + 1 suggestec
2. Would there be any relation in space another idea. Going back to a line divided
corresponding to the relation in the plane, by points, we see that
8 = L + I + 1? S=1+.
The first idea led to a rapid solution of For a plane, divided by lines (in the un
the problem in the simples case as out specialized case)
lined, for the nth plane will cut the ? 1 n(n ? 1)
S = 1 + n --?
planes already in space in ? 1 lines.
These form
For three-dimensional space in th
n(n ? 1) specialized case
+
S = 1 + -1
n(n - 1) n(n - l)(n - 2)
regions in the nth plane corresponding to
3!
the regions of space that have been di It would take little imagination to con
vided by it. Hence this quantity is the jecture that for a four-dimensional space
number of regions added by the nth plane in the nonspecialized case, the number of
when it intersects the ? 1 planes previ regions would be
ously established. n(n - 1) n(n - l)(n - 2)
The first plane adds one region; the S = 1 + -1
2 3!
second, two; the third, four. All these n(n - l)(n - 2)(n - 3)
+
figures check with ^~4!
n(n ? 1)
l + ?2 or, in words: The number of regions equals
the number of three-spaces plus the num
The total number of regionsberformed by
of planes of intersection plus the num
nonspecialized planes in space
ber of would thus
lines of intersection plus the number
be of points of intersection plus 1. Another

-|[^-]
way of writing this result is

, (n - l)n(n + 1) ,
= 1+-3j-+ ?.
For the more general space relation, the
where
+
0+0?
following was tried :
S = +L + V +1
where is the number of planes; L, the
o
(pronounced un ch
number of lines of intersection of the
of combinations o
planes; and 7, the number of points time,
of or
intersection (vertices). This relation works
out as follows:
n(n ? 1)
(n\ n \-k)\'
k) ~ kl(n
+ L + 7 + 1=n+
The Problem in Space Resumed
(n - l)(n - 2) Having found the formula for the sim
+ n-+ 1,
plest case of planes intersecting in three

724 The Mathematics Teacher | December 1966

This content downloaded from 128.32.10.230 on Sun, 03 May 2020 01:52:11 UTC
All use subject to https://about.jstor.org/terms
space and determined likewise that S points, such as the vertices? The thought
= P + L+ V + 1 in that instance, the came of a sweep plane that would start
next step was to ascertain whether some far out with all vertices on one side of it
interpretation could be given so that the and proceed parallel to itself.
latter formula would continue to apply for Consider, for example, the simplest case
all cases. Here a major difficulty arose, that we studied above. If there were
for whereas in the plane it is relatively planes, then the sweep plane would cut
simple to draw lines in certain configura them in lines. These, according to the
tions and count the number of regions in formula obtained for the plane, would di
order to check one's work, there is no cor vide the sweep plane into :
responding process in space.
Such problems as the following, for
example, are baffling: What happens
regions that would be in one-to-one corre
when a pencil of four planes (through one
spondence with regions into which the
line) is met by five other planes in general
planes were dividing space. This plane
position? The simple vertices can be
would maintain the same general con
counted as one. But what equivalent value
figuration until it came to one of the
should be given the vertices formed by the
vertices in space. At such a point one of its
intersection of a plane and the axis of the
regions would dwindle to a point, and
pencil? Also, to how many lines of intersec then on the other side of the vertex the
tion is a pencil of four planes equivalent?
plane would emerge into a new region of
Reasonable hypotheses can be made and
space. Thus, for each vertex, there would
answers worked out. But how is it possible
to check these results if a count cannot be be one additional region. Since the number
of vertices is
made?
Actually, by dint of diligent efforts ap n(n - \){n - 2)(n - 3)
plied to special cases, we surmised that a 3!
pencil of k planes would have to be equiva the total number of regions would be
lent to k ? 1 lines of intersection; that the
, , n(n - 1) n(n - l)(n - 2)(n - 3)
vertex f ormed by the axis of a pencil of k
planes and another plane would be equiva
?+-8 --
lent to k ? 1 vertices; that the vertex It becomes apparent
formed by k planes, no three of which go tions why the form
through the same line, would be equiva hold. To cover the g
lent to need be done is to kee
(k - l)(k -2) ber of regions as dete
2
section of the sweep p
vertices. But whether these relations
beyond all vertices a
would hold universally in any such com
how many regions d
binations was not in the least cleareach
andvertex.
th?re seemed to be no way of checking in
any but the simplest cases. Completion of the Solution
for the Plane
Solution by Inspiration This development is a striking confirma
tion of the idea, brought out by P?lya,
In the midst of this apparently hopeless
involvement, there entered an idea thatthat very often the way to solve a simpler
problem
solved the question in one stroke. Since it is to try to solve one that is more
was impossible to count the regions,difficult
the ! The problem of lines in the plane
thought was this: Why not try to find now becomes
a crystal clear. Since there are
lines,
method that utilizes change at certain key a sweep line that is not parallel to

A mathematician's progress 725

This content downloaded from 128.32.10.230 on Sun, 03 May 2020 01:52:11 UTC
All use subject to https://about.jstor.org/terms
any of the given lines will cut these lines in + 8(3 + 9 + 6) + 3(9 + 6) + 9 ? 6
points and be divided into + 1 sec = 347. Hence
tions by these points. These + 1 sec S = 30 + 25 + 347 + 1 = 403.
tions on the line indicate that the line is
Example 3
passing through + 1 regions. Let the Let there be points in a plane, no three
sweep line move parallel to itself. As it of them collinear. If they are connected in
approaches a simple vertex, one region all possible ways and no three of the lines
dwindles and disappears. But on going go through any point other than the
through the vertex a new region is found
given points, and if none of the lines are
on the other side. Thus to a simple vertex
parallel, determine the number of regions
corresponds one additional region. For a into which the plane is divided.
multiple vertex through which pass k There would be
lines, there are k ? I regions which disap
n(n ? 1)
pear at the vertex as the sweep line passes 2
through it, while k ? 1 new regions are or
entered on the other side of the vertex.
Thus to a vertex with k lines through it
there correspond k ? 1 regions. Conse
o
lines in all. L lines would normal
quently, the formula S = L + I + 1 is
sect in
justified if I is interpreted to be the equiva
lent number of vertices. HL- 1)
2
Example 1
Let there be four sets of parallel lines in simple intersections. In the present case it
would be necessary to take into account
the plane, containing respectively 5, 7, 4,
and 10 lines. Let there be no multiple the equivalent number of simple intersec
intersections. Into how many regions do tions represented by the lines going
these lines divide the plane? through the given points. Since there are
The number of lines, L, is 26. The num ? 1 lines through a vertex and since
ber of intersections can be found by de these would normally intersect in
termining the intersections for the sets of (n-l)(n-2)
2
parallel lines taken in pairs. Thus I is
5(7 + 4 + 10) + 7(4 + 10) + 4 ? 10, or points, the number of simple vertices
243. Hence the number of regions into missing as a result of the lines going
which the plane is divided is 26 + 243 through the given points is
+ 1 = 270. n(n ? l)(n ? 2) /n\
Example 2 -2- or 3U
Let there be five pencils of lines in the
Hence the number ot simple intersect
plane, containing respectively 4, 8, 3, 9, would be
and 6 lines. Let none of the lines be
parallel and let no line of one pencil pass
through the center of another pencil. Into
how many regions do these pencils divide
the plane? Thus the number of spaces would be
L is 30. The equivalent number of ver
tices corresponding to the centers of the
pencils is given by (4 ? 1) + (8 ? 1)
+ (3 - 1) + (9 - 1) + (6 - 1) = 25. The
number of intersections of the pencils +3(I)+n(n -2)'
among themselves is 4(8 + 3 + 9 + 6) or

726 The Mathematics Teacher | December 1966

This content downloaded from 128.32.10.230 on Sun, 03 May 2020 01:52:11 UTC
All use subject to https://about.jstor.org/terms
 Conclusion
These examples are sufficient to illus
trate the simplicity and power of this
Eulerian-type formula. More important,
perhaps, is the fact that this investigation
opens the door to many other avenues of
research. The following are offered for
consideration: 5. The division of three-dimensional
space by planes in various configurations.
1. The number of regions into which an This has been solved in general, but many
area enclosed by a convex curve is divided details still require consideration.
by straight lines. 6. Given points in space in general
2. The number of regions into which an position, into how many regions is space
area enclosed by a polygon of sides is divided by connecting these points by all
divided by its diagonals. the planes they determine?
3. The problem of division of a plane by 7. Given points in the plane in gen
circles. eral position, let circles be determined by
4. The number of regions into which a sets of three points in all possible ways.
plane is divided by any closed curve, such Into how many regions is the plane
as the one shown in the illustration. divided?

A Christmas test?[continued from p. 716]

In XMAS, ?X = iy + 10, M = y, ?A = y Given: AB || DC,

+ 10,andZS - 2y - 65. BC J_ AB,
Find the measure of each angle in this AD _L DC;
quadrilateral. Prove: AD \\ BC.

shaped so as to form a very elegant test that is work. The test met with approval by students
fun to take and still retains a level of measure of all levels of accomplishment, and everyone
that may be necessary in the plans of a quarter's wore a smile?including me.
Eileen Cox
Billings West High School
Billings, Montana

A mathematician's progress 727

This content downloaded from 128.32.10.230 on Sun, 03 May 2020 01:52:11 UTC
All use subject to https://about.jstor.org/terms