CHAPTER 12

Foundations of Neutral
Geometry

The play is independent of the pages on which it is printed,

and “pure geometries” are independent of lecture rooms, or
of any other detail of the physical world.
– G. H. Hardy in A Mathematician’s Apology [10] (1877 –
1947)

12.1 TRIANGLES AND PARALLELS

As mentioned at the end of Chapter 11, Hilbert’s incidence, betweenness,
congruence, and Dedekind axioms form a basis for Euclidean, Hyper-
bolic, and Elliptic geometry. As such we say these axioms form a basis
for Universal geometry. As shown in Chapter 11, the theorems of Uni-
versal geometry include Euclid’s Propositions 1-15, 23, and ASA triangle
congruence. Continuing from Proposition 15, there is another set of re-
sults that are not universal, but interestingly enough, are still common
to two of our geometries —Euclidean and Hyperbolic. Traditionally, the
set of results common to these two geometries has been called Neutral
or Absolute geometry.

12.1.1 Exterior Angle Theorem

The first of Euclid’s Propositions that does not hold in Universal geom-
etry is the Exterior Angle Theorem (Proposition 16).

75
76  Exploring Geometry - Web Chapters

Definition 12.1. Given triangle ∆ABC, the angles ∠CBA,

∠BAC, and ∠ACB are called interior angles of the triangle. Their
supplementary angles are called exterior angles. The two angles of
a triangle that are not supplementary to an exterior angle are called
remote interior angles relative to the exterior angle.

The statement of the Exterior Angle Theorem is as follows:

Theorem 12.1. (Exterior Angle Theorem) An exterior angle of a

triangle is greater than either remote interior angle.

Let’s consider a traditional proof of this theorem:

Proof: Let ∠CAD be an exterior angle to ∠BAC in triangle ∆ABC.
We can assume AD ∼ = BC. We want to show that ∠CAD is greater
than the remote interior angles ∠ACB and ∠ABC.

Suppose that ∠CAD was congru-

ent to ∠ACB. Triangles ∆CAD
and ∆ACB have AD ∼ = BC
and they share side AC. By SAS
these triangles are congruent and
∠ACD ∼ = ∠CAB.

Since ∠CAD and ∠CAB are supplementary, and ∠CAD ∼ = ∠ACB,

then the supplementary angle to ∠ACB must be congruent to ∠CAB.
We are assuming that ∠CAB ∼ = ∠ACD. Thus, by angle transitivity we
have that the supplementary angle to ∠ACB is congruent to ∠ACD,
but since these two angles share a side, then the supplementary angle
←→
must be ∠ACD. This implies that D lies on BC, which is impossible.
Thus, either ∠CAD < ∠ACB or ∠CAD > ∠ACB.

Suppose ∠CAD < ∠ACB. If we

copy the exterior angle ∠CAD so
−→
that one side is on CA and the
−→
other is on the same side of CA
as B, then the other side will lie
in the interior of ∠ACB, and thus
by the Crossbar Theorem will in-
tersect AB at a point B 0 .
 Foundations of Neutral Geometry  77

Then, ∠CAD will be an exterior angle of triangle AB 0 C that is

congruent to a remote interior angle, which we just proved is impossible.
Thus, ∠CAD > ∠ACB and a similar argument can be made to show
∠CAD > ∠ABC.2
On the face of it, this proof looks solid. It only uses results from ear-
lier sections in this chapter. But, we know this theorem is false in Elliptic
geometry, as we showed in chapter 8 that it is possible to construct a
triangle with more than one right angle in Elliptic geometry.
So, there must be some subtle assumption that we are making in
this proof that is different in Euclidean and Hyperbolic geometry, as
compared to Elliptic geometry. The first part of the proof involves a proof
by contradiction. We start by assuming that one of the interior angles is
congruent to an exterior angle. This leads to the conclusion that point D
←→
must be on BC, which is claimed to be a contradiction. However, this is
only a contradiction if we are assuming that lines cannot “double back”
on themselves. That is, if lines have infinite extent. However, this is not
the case in Elliptic geometry.

Here we have an illustration of a

triangle in Elliptic geometry that
has three right angles. In chap-
ter 8 we saw that this is possi-
ble. In this case, it happens that
the point D constructed so that
AD ∼ = BC will be in exactly the
same position as point B, so there
is no contradiction!

For Neutral geometry, we will assume the Exterior Angle Theorem

holds. Thus, we are implicitly assuming that lines are not bounded,
that they are of infinite extent. In Neutral geometry, we then have that
Euclid’s Propositions 1-15, 23, ASA, and the Exterior Angle Theorem
(Proposition 16) all hold. Also, we have all of the results on betweenness,
separation, congruence, angle and segment ordering, and numerical mea-
sure of segments and angles from Chapter 11. As we have not yet proven
the circle-circle intersection property (section 11.10), we will be careful
to not use this property (or Euclid’s Proposition 1) in this section.
78  Exploring Geometry - Web Chapters

12.1.2 Triangles —Angles and Sides

Let’s continue on our review of Euclid’s Propositions, assuming that
we are working in Neutral geometry. Euclid’s version of Proposition 17
states

In any triangle two angles taken together in any manner are

less than two right angles.

We could use angle measure to translate what Euclid means by “less

than two right angles” as a comparison to a given angle. However, we will
prove the following Theorem, which is equivalent to Euclid’s statement.

Theorem 12.2. (Proposition 17 substitute) In triangle ∆ABC, for

any pair of angles, say ∠ABC and ∠BAC, we have that ∠ABC is
less than the supplementary angle to ∠BAC.

The proof of this theorem follows directly from the Exterior Angle
Theorem and is left as an exercise.
Propositions 18 and 19 deal with relative comparisons of sides and
angles in a triangle.

Theorem 12.3. (Proposition 18) In a triangle, the larger side is

opposite the larger angle.

Theorem 12.4. (Proposition 19) In a triangle, the larger angle is

opposite the larger side.

The proofs of these two theorems are left as exercises.

Proposition 20 has traditionally been called the Triangle Inequality.

Theorem 12.5. (Proposition 20) In triangle ∆ABC, the sum of

two sides of a triangle is always greater than the third side.
 Foundations of Neutral Geometry  79

Proof: We have defined segment ordering and addition of segments in

section 11.5.2. Without loss of generality, let us assume that the two
sides in question are AB and BC. We need to show that the sum of
these segments is greater than AC.
−→
On the ray opposite to BA, there D
is a point D such that BD ∼ = BC
(Axiom III-1 in section 11.5 ).
Then, B is between A and D and B
AD is the sum of AB and BD.
By exercise 11.5.6 we can also say
that AD is the sum of AB and
BC. A C

Since B is between A and D, then by Theorem 11.18 and the def-

inition of angle inequality (definition 11.20), we have that ∠ACD >
∠BCD. Since ∆CBD is an isosceles triangle we have that ∠BCD ∼ =
∠BDC. Thus, by angle ordering (Theorem 11.38), we have that
∠ACD > ∠BDC. By Theorem 12.3 we than have that AD > AC.
Since AD is the sum of AB and BC, then the sum of AB and BC is
greater than AC. 2
The triangle inequality can be used to prove some very useful results,
such as the following:

←→
Theorem 12.6. Given a line BC and a point A not on the line, the
←→
perpendicular AD from A to a point D on the line has the shortest
←→
distance among all segments from A that intersect BC.

←→
Proof: Let E be a point not equal to D on BC.

We can find a point A0 on the ray A

−−→
opposite to DA such that A0 D ∼=
AD. Then, triangles ∆ADE and
∆A0 DE will be congruent by
B D E C
SAS. Also, AE + A0 E > AA0 by
the Triangle Inequality Theorem.
Thus, 2AE > 2AD and we are A’
done. 2
80  Exploring Geometry - Web Chapters

Continuing in our review of Euclid’s propositions, we come to Propo-

sition 21. This proposition deals with the relative comparison of side
lengths between a triangle and a triangle created in its interior.

Theorem 12.7. (Proposition 21) Given triangle ∆ABC, if we con-

struct triangle ∆DBC with D an interior point of ∆ABC, then
BD + CD < BA + CA. Also, ∠BDC > ∠BAC.

Proof:

Since D is interior to the triangle

−−→ A
then BD will intersect side AC
at some point E with A ∗ E ∗ C. E
In triangle ∆ABE we have that
D
AB + AE > BE by the Triangle
Inequality theorem.
B C

Then, AB+AE+EC > BE+EC. Since A∗E∗C we have AE+EC =

AC and thus AB + AC > BE + EC.
Similarly in triangle ∆CDE we can get CE + ED > CD. Add BD
to both sides and combine to get CE + BE > CD + BD.
Then, by segment ordering we have that since AB + AC > BE + EC
and CE + BE > CD + BD then AB + AC > CD + BD.
The remaining statement about angle inequality is left as an exercise.
2
Euclid’s original version of Proposition 22 deals with the construction
of a triangle from three given segments. It states that “To construct a
triangle from three lengths, it is necessary that when you add any pair
of lengths, the sum is greater than the other length.”
This is a rather odd statement, as Euclid had already showed in
Proposition 20 that if a triangle exists, then the Triangle Inequality
holds. Thus, necessity had already been proven. In looking at Euclid’s
proof of Proposition 22, however, we see that he is really proving suf-
ficiency of the conclusion. That is, if the sum of any two lengths is
greater than the third, than one can construct a triangle with those
three lengths. The next theorem will serve as an equivalent statement
to Euclid’s Proposition 22.
 Foundations of Neutral Geometry  81

Theorem 12.8. Given three segments a, b, and c, if the sum of any

two of these segments is always greater than the third, then there is
a triangle with sides congruent to a, b, and c.

The proof of this theorem requires the circle-circle intersection prop-

erty. We will prove this intersection property in the next section, and
then we will return to the proof of this proposition. We will be careful
not to use this proposition in the remaining proofs of this section.
Continuing in our review of Euclid’s axioms, we come to Proposi-
tion 23. As mentioned earlier in this section, this is part of Universal
geometry, which we have already covered.
Proposition 24 is another theorem dealing with triangle comparison.
It is often called the “Hinge Theorem.”

Theorem 12.9. (Proposition 24) Let ∆ABC and ∆DEF be tri-

angles with AB ∼
= DE and AC ∼= DF . If ∠CAB > ∠F DE, then
BC > EF .

Proof: There are two cases to consider —either AC and AB are congru-
ent or they are not congruent.
Assume that AC and AB are not congruent. We can assume that
AC > AB.

Since ∠CAB > ∠F DE, then

−→
we can find a ray AJ interior
to ∠CAB such that ∠JAB ∼ =
∠F DE (definition of angle order).
Also, from Congruence Axiom III-
−−→
4 we can find a ray DG on the
−−→ −−→
other side of DF from DE such
that ∠GDF ∼ = ∠CAJ. By addi-
tion of angles, ∠CAB ∼ = ∠GDE.
82  Exploring Geometry - Web Chapters

Now, using Congruence Ax-

iom III-1, we can assume that G is
−−→
the point on DG such that DF ∼ =
DG. Then, ∆DF G is an isosceles
triangle and the angles at G and
F in ∆DF G are congruent.

We now show that E, F , and G are non-collinear. By Theorem 12.4

we know ∠BCA < ∠CBA. Suppose E, F , and G were collinear. Then,
by SAS congruence we would have that ∆CAB ∼ = ∆GDE, and thus,
∼ ∼
∠BCA = ∠EGD and ∠CBA = ∠GED. This implies that ∠EGD <
∠GED. But, ∠EGD ∼ = ∠GF D. Thus, we have that the exterior angle
∠GF D would be less than the opposite interior angle ∠GED in ∆DEF
which is impossible. We conclude that E, F , and G are non-collinear.
−−→
Consider ∆EF G. We now show that GE is interior to ∠F GD.
−−→
Suppose that GE was exte-
rior to ∠F GD. By the Exte-
rior Angle Theorem, ∠DF G >
∠DEF . By Theorem 12.4 we
know that ∠DEF > ∠EF D.
Thus, ∠DF G > ∠EF D. But,
∠DF G ∼ = ∠F GD. Thus, ∠F GD >
−−
→
∠EF D. Since GE is exterior to
∠F GD, then ∠EGD > ∠F GD,
and so ∠EGD > ∠EF D. But
this contradicts Theorem 12.7.
−−→
We can now assume that GE is interior to ∠F GD.

Then ∠F GE < ∠F GD. Also,

since ∠DF G ∼ = ∠F GD, then
∠F GE < ∠DF G. But, ∠DF G <
∠EF G. Thus, ∠F GE < ∠EF G.
By Theorem 12.4 we know that
EG > EF . But, by SAS con-
gruence, we know that ∆CAB ∼=
∆GDE and so EG ∼ = BC. Thus,
BC > EF .
 Foundations of Neutral Geometry  83

The proof for the case where AC ∼ = AB is similar and is left as an

exercise. 2
Proposition 25 flips the concluding implication from Proposition 24.

Theorem 12.10. (Proposition 25) Let ∆ABC and ∆DEF be tri-

angles with AB ∼
= DE and AC ∼ = DF . If BC > EF , then
∠CAB > ∠F DE.

Proof: The proof relies on Proposition 24 and is left as an exercise. 2

Proposition 26 consists of two triangle congruence results: ASA and
AAS. ASA triangle congruence has already been proven when we covered
the theorems of Universal geometry. We leave the proof of AAS as an
exercise.

12.1.3 Transversals and Parallels

Before we look at Propositions 27 and 28 we need to review some defi-
nitions concerning angles and parallels:

Definition 12.2. A line t is called transversal to two other lines l

and m if t intersects both lines and the lines are not coincident.

Let t be transversal to l and m, meeting l at A and m at A0 (Fig-

ure 12.1). Axiom I-3 guarantees that there is at least one other point on
lines l and m and axiom II-2 guarantees that we can choose points B
and C on l with B ∗ A ∗ C and B 0 and C 0 on m with B 0 ∗ A0 ∗ C 0 . Also,
we can assume that B and B 0 are on the same side of t, and C, C 0 are
also on the same side.
84  Exploring Geometry - Web Chapters

Figure 12.1

Definition 12.3. Angles ∠CAA0 , ∠C 0 A0 A, ∠BAA0 and ∠B 0 A0 A

are called interior angles. (The angles having AA0 as a side) Also,
∠CAA0 and ∠B 0 A0 A are called alternate interior angles as are
∠C 0 A0 A and ∠BAA0 . All other angles formed are called exterior
angles.

Definition 12.4. Pairs of angles, one interior and one exterior,

on the same side of the transversal, are called corresponding an-
gles. For example, in Figure 12.1, angles ∠A0 AB and ∠DA0 B are
corresponding, as are ∠EAC and ∠AA0 C.

Definition 12.5. Two lines are parallel if they do not intersect.

We can now consider Euclid’s Proposition 27.

Theorem 12.11. (Alternate Interior Angle Theorem) If alternate

interior angles are congruent then the lines are parallel,

Proof: We are given that ∠CAA0 ∼ = ∠B 0 A0 A as shown in Figure 12.2.

Assume that the two lines did meet at some point D. Without loss of
generality, we can assume that D is on the same side of t as B and B 0 .
 Foundations of Neutral Geometry  85

l
t
E C
A
B
m

C’ A’ B’ D

Figure 12.2

−→
There is a point E on AC such that AE ∼= A0 D. In Neutral geometry
we assume that lines cannot double-back on themselves, and so we know
that D 6= E and we have two distinct triangles, ∆A0 AE and ∆AA0 D.
By SAS ∆A0 AE ∼ = ∆AA0 D. Thus, ∠DAA0 ∼ = ∠EA0 A. Now, ∠DAA0
0
and ∠EAA are supplements. By Theorem 11.28 we know that since
∠EAA0 ∼ = ∠DA0 A then ∠DAA0 ∼ = ∠C 0 A0 A. But, the triangle congru-
ence gave ∠DAA = ∠EA A. So, ∠C 0 A0 A ∼
0 ∼ 0
= ∠EA0 A. But axiom III-4 on
the uniqueness of angles would then imply that E must be on m which
is impossible, as then we would have two distinct lines intersecting in
more than one point. 2
The next two theorems make up Proposition 28.

Theorem 12.12. If two lines are cut by a transversal so that cor-

responding angles are congruent then the two lines are parallel.

Proof: Exercise. 2

Theorem 12.13. If two lines are cut by a transversal so that the

interior angles on the same side are supplementary then the two
lines are parallel.

Proof: Exercise. 2
These two theorems have the following extremely useful corollaries.

Corollary 12.14. Two lines that are perpendicular to the same

line are parallel.
86  Exploring Geometry - Web Chapters

Proof: Suppose that lines l and m are both perpendicular to line t.

Considering t as a transversal, then the alternate interior angles made
are both right angles and hence congruent. Then, the lines are parallel
by the Alternate Interior Angle Theorem. 2
Note that this implies that the perpendicular dropped from a point
to a line must be unique, since if there were two lines through the same
point that were both perpendicular to the same line, then they would
have to be parallel.

Corollary 12.15. Through a point there is only one line perpen-

dicular to a given line.

Corollary 12.16. If l is a line and P is a point not on l, then there

is at least one parallel to l through P .

Proof: By the theorem above there is a unique perpendicular t through

P to l. There is also a unique perpendicular m through P to t. Then, t
is perpendicular to both l and m and so l and m are parallel. 2
Note that this result is Proposition 31. Also note that the parallel
through P is not necessarily unique. In fact, Hyperbolic geometry satis-
fies all of the axioms discussed so far in this chapter, and in this geometry
there are multiple parallels through P .
Exercise 12.1.1. Prove Theorem 12.2. Also, show that this theorem is equiv-
alent to Euclid’s version of Proposition 17.

Exercise 12.1.2. Prove Theorem 12.4. That is, show that in any triangle,
the greater angle lies opposite the greater side. [Hint: Figure 12.3]

A D B

Figure 12.3

Exercise 12.1.3. Prove Theorem 12.3. That is, show that in any triangle
 Foundations of Neutral Geometry  87

the greater side lies opposite the greater angle. [Hint: Assume triangle ABC
has ∠BCA > ∠BAC. Then, either AB > BC or AB ∼ = BC or AB < BC.
Show that two of these are impossible.]

Exercise 12.1.4. Prove the AAS triangle congruence result. That is, given
triangles ∆ABC and ∆DEF if ∠ABC ∼ ∼ ∠EF D and
= ∠DEF , ∠BCA =
∼ ∼
AC = DF , then ∆ABC = ∆DEF (Figure 12.4).

C F

A D
B E

Figure 12.4

[Hint: Suppose that BC and EF are not congruent. If EF < BC then

there is a point G between B and C with GC ∼ = EF . Show that this leads to
a contradiction of the exterior angle theorem.]

Exercise 12.1.5. Finish the proof of Theorem 12.7. [Hint: use the Exterior
Angle Theorem.]

Exercise 12.1.6. Finish the proof of Theorem 12.9. [Hint: There are only
two paragraphs in the proof that depend on the two segments AC and AB
being non-congruent. Only a small modification in the argument is neded for
each paragraph.]

Exercise 12.1.7. Prove Theorem 12.10. [Hint: There are three cases to
consider: either ∠CAB < ∠F DE, ∠CAB ∼
= ∠F DE, or ∠CAB > ∠F DE.
Also, use Proposition 24.]

Exercise 12.1.8. Prove Theorem 12.12. Hint: Suppose the lines intersect
and find a contradiction.

Exercise 12.1.9. Prove Theorem 12.13.

Exercise 12.1.10. Let ∆ABC be a triangle with right angle ∠ABC. Call
the side opposite the right angle the hypotenuse and the other sides legs of the
triangle. Show that both legs are less than the hypotenuse.

Exercise 12.1.11. Let ∆ABC be a triangle with right angle ∠ABC. On

←→
AB let D be a point with D ∗ A ∗ B. Show that BC < AC < DC.
88  Exploring Geometry - Web Chapters

12.2 CONTINUITY REDUX

In Chapter 11 we discussed (without proof) the Circle-Circle continuity
principle:

• (Circle-Circle Continuity) Given two circles c1 and c2 , if

c1 contains a point inside of c2 and also contains a point
outside of c2 , then there are exactly two distinct points
of c1 that are also on c2 . ( We say they intersect in two
points)

There is also a Line-Circle continuity principle:

• (Line-Circle Continuity) Given a circle c and a line l, if

l contains a point inside of c and also contains a point
outside of c, then there are exactly two distinct points of
c that are also on l.

In this section we will show that in Neutral geometry, both of these

continuity principles can be proven from Dedekind’s axiom and the other
results we have shown so far. The material in this section is technically
dense, much like the material in section 11.8. But, it is worth wading
through, as it is crucial to a complete understanding of continuity prin-
ciples in geometry.
We will start with Line-Circle continuity. We first review some defi-
nitions regarding a circle.

Definition 12.6. The circle of radius AB and center O is the set of

all points X such that OX ∼= AB. A point P is said to be an interior
point (or said to be inside the circle) if P = O or OP < AB. If
OP > AB the point P is said to be an exterior point (or outside
the circle).

Theorem 12.17. (Line-Circle Continuity) Given a circle c and a

line l, if l contains a point inside of c and also contains a point
outside of c, then there are exactly two distinct points of c that are
also on l.

Proof: Let O be the center of circle c, let AB be the radius, and let P be
 Foundations of Neutral Geometry  89

the point inside c. There are two cases to consider. First, suppose that
O is on l. Then, on the two rays on either side of O on l we know by
congruence axiom III-1 that we can find unique points I1 and I2 with
OI1 ∼= AB ∼ = OI2 .
Now, suppose that O is not on l. We can drop a perpendicular from
O to l which intersects at some point N . Then, N must be an interior
point of the circle. This is obvious if N = P . If N 6= P then ∠ON P is a
right triangle and thus by exercise 12.1.10, we have that ON < OP and
thus N is interior.
−−→
We will create a partition for Dedekind’s axiom l as follows. Let N T
−→
be a ray on l and let N S be the opposite ray (Figure 12.5).

T
N

P
S O
l

Figure 12.5
−−→
Let Σ1 consist of N T together with all of the points W on l that are
inside c (i.e., OW < AB). Let Σ2 consist of the remaining points of l.
−→
Note that Σ2 consists of points solely on N S. Clearly, Σ1 is not empty.
−→
Also, by congruence axiom III-1, we know there is a point R on N S with
NR ∼ = AB. Since ∆ON R is a right triangle we have, by exercise 12.1.10,
that OR > N R and R is in Σ2 .
For the betweenness condition of Dedekind’s axiom let Q1 , R1 be
two points of Σ1 and Q2 , R2 be points of Σ2 . Suppose that Q1 ∗ Q2 ∗ R1 .
−−→
There are three cases to consider. First, if Q1 and R1 are both on N T ,
then Q2 cannot be between these two points as it is on the opposite
−−→
ray. Second, suppose only one of Q1 and R1 is on N T . We may assume
that Q1 is. Then, OR1 < AB and R1 , Q2 are both on the same side of
N on l. Now, since Q1 ∗ N ∗ Q2 and Q1 ∗ Q2 ∗ R1 then by four-point
betweenness we have N ∗ Q2 ∗ R1 . Since ∆ON Q2 is a right triangle
and N ∗ Q2 ∗ R1 then, by exercise 12.1.11, we have OQ2 < OR1 . But,
OR1 < AB, and so OQ2 < AB which contradicts the fact that Q2 is in
90  Exploring Geometry - Web Chapters

−→
Σ2 . Lastly, suppose both Q1 and R1 are on the ray N S with OQ1 < AB
and OR1 < AB. Then, since Q1 ∗ Q2 ∗ R1 , we can use exercise 12.1.11 to
show that either OQ2 < OQ1 or OQ2 < OR1 . In either case, OQ2 < AB
which contradicts Q2 being in Σ2 .
−−→
Now, suppose that Q2 ∗ Q1 ∗ R2 . This is impossible if Q1 is on N T ,
−→
as Q2 and R2 are on the opposite ray. So, OQ1 < AB and Q1 is on N S.
Since Q2 ∗ Q1 ∗ R2 , then one of Q2 and R2 is on the same side of Q1 as
N . We can assume that Q2 is on the same side of Q1 as N . Then, either
N ∗ Q2 ∗ Q1 or N ∗ Q1 ∗ Q2 . We cannot have N ∗ Q2 ∗ Q1 , as N is in Σ1
and we have already shown that a point of Σ2 cannot be between two
points of Σ1 . Thus, we must have N ∗ Q1 ∗ Q2 (or Q2 ∗ Q1 ∗ N ). From
Q2 ∗ Q1 ∗ R2 , we conclude that N and R2 are on the same side of Q1 .
Thus, either Q1 ∗ R2 ∗ N or Q1 ∗ N ∗ R2 . Again, we have shown above
−→
that Q1 ∗ R2 ∗ N is impossible. Since Q1 and R2 are both on N S, we
cannot have Q1 ∗ N ∗ R2 .
Thus, by Dedekind’s axiom there is a unique point X separating Σ1
and Σ2 . We claim that OX ∼ = AB, and thus X is an intersection point
of l with c.
For suppose that OX < AB. By segment addition and congruence
−−→
axiom II-1, we can find a point V on the vector opposite to XN with
XV + OX ∼ = AB. (see figure 12.6) By the triangle inequality we have
OV < XV + OX = AB.

T
N
P
V X
O
l

Figure 12.6

Thus, V is in Σ1 . But, this would imply that X is between two points

of Σ1 , namely V and N , which is impossible. A similar argument shows
that OX > AB is not possible.
Thus, OX ∼ = AB. It is left as an exercise to show that X must be
−→
on N S.
 Foundations of Neutral Geometry  91

Will this intersection point be unique? Assume that there is another

−→
point of intersection X 0 of ray N S with the circle. Let M be the midpoint
of XX 0 . Then, triangles ∆OXM and ∆OX 0 M are congruent by SSS
congruence and thus ∠OM X ∼ = ∠OM X 0 and both must be right angles.
But, we know by theorem 12.15 and its corollaries that the perpendicular
from O to l must be unique, which implies that N = M . But, this is
impossible since if X and X 0 are both on the same side of N , then all
points on XX 0 are on that side of N . Thus, X is unique.
To finish the proof we need to find an intersection point along the
−−→
ray N T . We leave this as an exercise. 2
It is clear from the proof that we only used the point P that was
inside the circle in our argument for the first intersection. We thus have
the following:

Corollary 12.18. Given a circle c and a line l, if l contains a

point P inside of c, then there are exactly two distinct points of c
that are also on l. Also, the intersection points occur on each of the
two opposite rays on l defined by P .

The proof of the second statement of this corollary is left as an

exercise. We also have the following result about segments and circles:

Corollary 12.19. Given a circle c and a segment P Q, if P is inside

c and Q is outside c, then there is a point X interior to P Q that is
also on c.

Proof: Exercise. 2
Before we prove the circle-circle continuity property, we will take a
little side-trip to show that Dedekind’s axiom can be extended to arcs
of circles.

Definition 12.7. Let c be a circle with center O. If A 6= B are two

points on c then we call the segment AB a chord of the circle. If
a chord passes through the center O we say it is a diameter of the
circle.
92  Exploring Geometry - Web Chapters

Definition 12.8. A chord AB of a circle c will divide the points of

←→
a circle c into two parts, those points of c on one side of AB and
those on the other side. Each of these two parts is called an open
arc of the circle. An open arc determined by a diameter is called a
semi-circle. If we include the endpoints A and B, we call the arc or
semi-circle closed.

Definition 12.9. One of the two open arcs defined by a (non-

diameter) chord AB will be within the angle ∠AOB. This will be
called a minor arc. The arc that is exterior to this angle is called a
major arc.

In order to extend Dedekind’s axiom to arcs of circles we need to

have a notion of betweenness for points on arcs.

Definition 12.10. Let AB be a chord of a circle c with center O

and let σ be the minor arc of the chord. Let P, Q, R be points on σ.
−→
Then we say that R is between P and Q if the ray OR is between
−→ −→
OP and OQ. (see figure 12.7)

R Q
P
A

B
O

Figure 12.7

Theorem 12.20. Dedekind’s axiom and the Archimedean property

can both be extended to minor arcs of circles.
 Foundations of Neutral Geometry  93

Proof: The definition of betweenness for points on a minor arc is that

of betweenness for rays in an angle. There is a direct correspondence be-
tween points on a minor arc and rays interior to the angle defined by
the chord of the arc. What this correspondence allows us to do is to
substitute any statement about points on an arc, including betweenness
properties of such points, to statements about rays in an angle and be-
tweenness properties of rays. Since we have already proven extensions
of Dedekind’s Axiom and the Archimedean property for rays within an
angle, then we automatically have these same properties for points on a
minor arc. 2
We now return to the task of proving circle-circle continuity. We will
need the following result for the proof.

←→
Lemma 12.21. Let P be a point on circle c with center O. Let OR
←→
be perpendicular to OP at O. Then, for any number s > 0 there are
points S1 and S2 on c such that P S1 < s and P S2 < s. Also, S1
←→ ←→
and S2 are on opposite sides of OP and the same side of OR.

Proof: By Theorem 11.55 (Segment measure), there is a segment AB

←→
having length s. At P construct the perpendicular to OP .
By congruence axiom III-1, we
know there are points U and V
on opposite sides of P on this per-
pendicular such that P U ∼= AB ∼ =
P V . By congruence axiom III-1,
−→
there are points S1 and S2 on OU
−−→
and OV , respectively, such that
S1 and S2 are on c.

Also, ∠OP U and ∠OP V are right angles, so ∠P OU and ∠P OV

are acute (Proposition 17). Then, ∠P OS1 < ∠P OR. Let ∠P OT be the
supplementary angle to ∠P OR. Then, ∠P OS2 < ∠P OT (angle order
properties with supplementary angles). Thus, S1 and S2 are on opposite
←→ ←→
sides of OP and the same side of OR.
Now, ∆OP S1 is an isosceles triangle. Thus, ∠OS1 P must be acute
(Proposition 17). Then, ∠P S1 U is obtuse, by Theorem 11.35. In trian-
gle ∆P S1 U we than have ∠P U S1 is acute (Proposition 17). Thus, by
94  Exploring Geometry - Web Chapters

Theorem 12.3, we have that P S1 < P U . Since the length of P U is s, we

have P S1 < s. A similar argument shows that P S2 < s. 2
We will also use the following result to put circle-circle continuity
into a “standard” configuration.

Theorem 12.22. Let c1 , c2 be two circles with centers O1 , O2 . Sup-

pose there is a point P on c1 that is inside c2 and there is a point Q
on c1 that is outside c2 . Then, the line through the centers O1 , O2
must meet circle c1 in two points I1 and I2 with one of these points
inside c2 and the other outside c2 .

Proof: For our analysis, we will use the segment measure properties
we covered in Chapter 11. Specifically, we will let r1 be the measure
of the radius for c1 , and r2 the measure for the radius of c2 . Since O1
←−→
is inside of c1 then by exercise 12.2.4 we have that O1 O2 intersects c1
in two points I1 and I2 . There are three cases to consider for point I1 .
Either I1 ∗ O1 ∗ O2 or O1 ∗ I1 ∗ O2 or O1 ∗ O2 ∗ I1 .

Case (1): I1 ∗ O1 ∗ O2 . Then,

I1 O2 ∼= I1 O1 + O1 O2 , or I1 O2 = P
O1 O2 + r1 . Also, by the triangle Q

inequality for ∆O1 QO2 we have

O2 Q < O1 O2 + r1 . Since Q is out- I1 O1 O2 I
2
side c2 , we have O2 Q > r2 . Then,
r2 < (O1 O2 + r1 ) (= I1 O2 ). Thus, c2
I1 is outside c2 . c1

Case (2): O1 ∗ I1 ∗ O2 . Then,

O1 O2 = O1 I1 + I1 O2 = r1 + I1 O2 . P
Q
Also, by triangle inequality for
∆O2 P O1 we have O1 O2 < r1 +
O2 P . Thus, r1 + I1 O2 < r1 + O2 P I O1 I1 O2
2
and so I1 O2 < O2 P . Since O2 P <
r2 we have that I1 O2 < r2 and I1 c2
c1
must lie inside c2 .

Case (3): O1 ∗ O2 ∗ I1 . A similar argument to case (2) shows that I1

must lie inside of c2 (exercise).
What about I2 ? In Case (1), we have I1 ∗ O1 ∗ O2 . Also, I1 ∗ O1 ∗ I2 ,
 Foundations of Neutral Geometry  95

as I1 and I2 must be on opposite sides of O1 (by the work we did in the

proof of Theorem 12.17). Thus, O2 and I2 are on the same side of O1
and we have either O1 ∗ I2 ∗ O2 or O1 ∗ O2 ∗ I2 , In both cases, we can
replace I1 by I2 in the work above. By doing so, we would either be in
Case (2) or (3) for I2 , and so I2 must be inside c2 .
Now consider Case (2) or (3) for I1 . We have O1 ∗I1 ∗O2 or O1 ∗O2 ∗I1 .
Again we have I1 ∗ O1 ∗ I2 . By four-point betweenness, in either case we
get that I2 ∗ O1 ∗ O2 . Thus, for I2 , we would be in Case (1) and I2 must
be outside c2 . 2
We are now ready to prove circle-circle continuity.

Theorem 12.23. (Circle-Circle Continuity) Let c1 , c2 be two cir-

cles with centers O1 , O2 . Suppose that c1 has one point P inside
and one point Q outside of c2 . Then, the two circles intersect in
two points.

Proof: By Theorem 12.22 we can assume there is a diameter I1 I2 of c1

with I1 outside c2 and I2 inside c2 (Figure 12.8).

Figure 12.8
←−→ ←→
By Theorem 11.47, we can construct a perpendicular O1 R to I1 I2 at
−−→
O1 . Let O1 R be one of the two rays defined. Then, by congruence axiom
III-1 we can assume R is a point such that the measure of O1 R is r1 ,
where r1 is the measure of the radius of c1 . That is, R is on c1 . Now, R
is either outside c2 , inside c2 , or on c2 (exercise). If it is on c2 , we have
96  Exploring Geometry - Web Chapters

found an intersection point for the two circles. Let this point be M for
←→
reference later in this proof. We note that M is on one side of I1 I2 .
If R is inside c2 , then the points I1 and R form a pair of inside-outside
points to c2 . If R is outside c2 then the points R and I2 form a pair of
inside-outside points. In either case, we have a well-defined angle, with
vertex at O1 , with sides being two rays intersecting c1 at points inside
and outside c2 . Without loss of generality we assume that R is outside
c2 and consider angle ∠I2 O1 R.
Since I2 is inside c2 , then O2 I2 < r2 . Let s = r2 − O2 I2 .

Using Lemma 12.21 for c1 and

s
the number , we have that there
2
is a point S interior to arc σ =
s
arc(I2 R) such that I2 S < . Like-
2
wise, since R is outside c2 then
O2 R > r2 . Let s0 = O2 R − r2 .
Then, there is a point S 0 on this
s0
arc such that RS 0 < .
2
By the triangle inequality, we have that O2 S < I2 S + O2 I2 . Since
s
I2 S < < s < r2 − O2 I2 , we have O2 S < r2 − O2 I2 + O2 I2 = r2 .
2
Thus, S is inside c2 . Again, by the triangle inequality, we have that
s0
O2 R < RS 0 + O2 S 0 . Since RS 0 < < s0 , then O2 S 0 > O2 R − RS 0 >
2
s0
O2 R − > O2 R − s0 = r2 . Thus, S 0 is outside c2 .
2
Define a Dedekind cut for the rays within ∠I2 O1 R as follows. Let Σ1
consist of rays with points on σ inside circle c2 . Let Σ2 consist of rays
with points on σ that are on or outside circle c2 . Clearly, these two sets
are non-empty. We now show these sets satisfy the betweenness property
for Dedekind’s axiom for arcs.
−−−→ −−−→
Suppose we had a ray O1 Q1 of Σ1 that was between two rays O1 Q2
−−−→
and O1 R2 of Σ2 .
 Foundations of Neutral Geometry  97

Then, ∠I2 O1 Q1 is greater than

∠I2 O1 R2 (or ∠I2 O1 Q2 ). In either
case, we can show, using Theo-
rem 12.9 that this implies that
O2 Q1 > O2 R2 (or O2 Q1 > O2 Q2 )
(exercise). But, this would mean
that Q1 is outside c2 , which con-
tradicts Q1 being in Σ1 .

Similarly, we cannot have a point of Σ2 between two points of Σ1 .

Dedekind’s property for angles (Theorem 11.56) requires a partition
of rays (which we have), a betweenness condition (shown above), and the
condition that both subsets of the partition contain at least one interior
−−→ −−→
ray to the angle. These are O1 S and O1 S 0 defined above. Thus, there is
−−−→
a unique ray O1 M interior to ∠I2 O1 R separating Σ1 and Σ2 . We need
to show that O2 M = r2 , where r2 is the radius of c2 .
Suppose O2 M > r2 . Let t =
←−→
O2 M − r2 . Let O1 U be perpendic-
←−→
ular to OM at O. We can choose
U so that U is on the other side
←−→
of O1 O2 as M . By Lemma 12.21,
there is a point T on c1 such
that T is interior to ∠U O1 M and
t
MT < .
2

Now, since ∠I2 O1 M is within and less than ∠U O1 M , it may be that T is

exterior to ∠I2 O1 M . If this is the case, then we replace T by the point T 0
that is on c1 and on the bisector of ∠I2 O1 M . Since ∠T 0 O1 M < ∠T O1 M
we can show, using Theorem 12.9 that this implies that T 0 M < T M
t
(exercise). Then, for T 0 we have M T 0 < M T < .
2
We conclude that if O2 M > r2 , then for t = O2 M − r2 , there is
t
a point T interior to ∠I2 O1 M such that M T < . By the triangle
2
t
inequality, we have that O2 M < M T + O2 T . Since M T < < t, then
2
t
O2 T > O2 M − M T > O2 M − > O2 M − t = r2 . Thus, T is outside c2 .
2
But, this implies that M is between two points, R and T , that are both
outside c2 . This contradicts Dedekind’s property for rays.
98  Exploring Geometry - Web Chapters

A similar argument shows that O2 M < r2 is impossible. Thus,

O2 M = r2 and M is on c2 .
To find the second intersection point, we can copy angle ∠I2 O1 M
←−→
to the angle on the other side of O1 O2 , by congruence axiom III-4,
yielding an angle ∠I2 O1 M 0 . By congruence axiom III-1 we can assume
that O1 M ∼ = O1 M 0 . This implies that M 0 is on c1 . Then, ∆O2 O1 M ∼=
∆O2 O1 M by SAS triangle congruence. Thus, O2 M ∼
0
= O2 M 0 , which
means that M 0 is on c2 . 2
We can now complete the proof Euclid’s Proposition 22, which we
had to leave “dangling” in the last section.

Theorem 12.24. Given three segments a, b, and c, if the sum of

any two of these segments is always greater than the third, then
there is a triangle with sides congruent to a, b, and c.

Proof: By repeated pair-wise comparison of the three segments, using

segment ordering, we can assume that either 1) all segments are pair-
wise non-congruent, 2) two of the three are congruent, or 3) all three are
pair-wise congruent.
If all three segments are pair-wise non-congruent, then we can assume
that a > b > c. Let l be a line and A a point on l. By Axiom III-1 in
section 11.5 there is a point A0 on l such that segment a is congruent to
AA0 . Since a > b then there is a point B between A and A0 such that
AB is congruent to b. Also, there is a point B 0 on the other side of A
from B such that AB 0 is congruent to b. Since a > c there is a point C
between A and A0 such that A0 C is congruent to c. Also, there is a point
C 0 on the other side of A0 from C such that A0 C 0 is congruent to c.

Since b + c > a, then AB + A0 C >

AA0 . Since B and C are between
A and A0 , then, by exercise 11.8.4
we know that C is between A and
B and B is between C and A0 .
 Foundations of Neutral Geometry  99

Construct the circle c centered

at A of radius AB and the circle
c0 centered at A0 of radius A0 C.
Since C is on AB, then AC < AB
and C is interior to circle c.

In Neutral geometry, lines do not “return” upon themselves, and so C 0

is not on AB. Thus, AC 0 > AB and C 0 is exterior to circle c. So, by the
circle-circle continuity principle, circles c and c0 must intersect at two
points D and D0 . Then, triangle ∆ADA0 will have sides congruent to a,
b, and c.
The case where two of a, b, and c are congruent can be argued in a
similar fashion. The proof is left as an exercise.
The case where a, b, and c are all congruent is equivalent to the
construction of an equilateral triangle, which was proven in Chapter!11.
2
This finishes our tour of Neutral geometry. The observant (and pa-
tient) reader will see that we have now proven all of the first 28 Propo-
sitions of Book I of Elements, plus Proposition 31. Euclid’s Proposition
29 deals with the properties of a line crossing two parallel lines. The
proof of this proposition relies on the uniqueness of parallels, and thus
is outside of Neutral geometry. This is the first of Euclid’s results that
is uniquely a property of Euclidean geometry.
Exercise 12.2.1. In the proof of Theorem 12.17 (Line-Circle Continuity)
−−→
show that the point X given by Dedekind’s axiom must be on N P .

Exercise 12.2.2. In the proof of Theorem 12.17 (Line-Circle Continuity)

−−→
show that there is a second intersection point of l with c on N T .

Exercise 12.2.3. Finish the proof of Corollary 12.18. [Hint: In the proof of
Theorem 12.17 we showed that the intersection points were on opposite rays
−−→ −−→
N S and N T , where N was interior to the circle and both of these rays were
incident on l. There were two cases for N . Use these as the basis for your
proof.]

Exercise 12.2.4. Prove Corollary 12.19. [Hint: See the hint for exer-
cise 12.2.3. Also, make ample use of four-point betweenness.]
100  Exploring Geometry - Web Chapters

Exercise 12.2.5. Finish the proof of Theorem 12.24. That is, given three
segments a, b, and c, where exactly two of the three are congruent, show there
is a triangle with sides congruent to these three segments. [Hint: Follow the
outline of the first part of the proof of Theorem 12.24, but use circles based on
the two segments of differing lengths.]

Exercise 12.2.6. Let C be a circle and P a point in the plane. Show that
P is either on the circle, outside the circle, or inside the circle.