Anda di halaman 1dari 6

Variations on

Steiner’s Porism

teiner’s porism is a remarkable theorem of Euclidean

having the same size and ‘‘sliding’’ along the annulus
geometry from the 19th century. Poristic properties between the circles like ball bearings.
are geometrical results about the existence of geo- A Steiner configuration has amazing properties. For
metrical configurations of figures that cannot exist unless a example, the centers of circles forming a Steiner chain lie
condition is met, but otherwise may exist in infinite on the same conic C whose foci are the centers of the two
number. Another classical example of a poristic property given circles m and l.
is the famous Poncelet’s porism; see [4] and [5]. Apart from Following the classical proof based on inversion that
their profound beauty and elegance, poristic types of transforms two given circles into concentric circles, and
results attract mathematicians because of their deep con- guided by empirical evidence made by experiments in the
nections to other leading research topics in mechanics, software Cinderella, we prove an analogue of the Full
theory of billiards, analysis, etc. A nice treatment of other Poncelet Theorem (see [3]) for the Steiner porism. Before
closure theorems is presented in reference [6, Lecture 29]. we state the theorem, we will briefly remind the reader of
Steiner’s porism is named for the great geometer Jacob some basic definitions and facts from elementary geometry.
Steiner who discovered it in the 19th century, [9]. He studied Let k be a circle with center O and radius R, and let a
a set of a finite number of circles, all of which were tangent point A in the plane be given. Let us consider an arbitrary
to two given nonintersecting circles (see Figure 1), each line p through A intersecting k in two points M and N and
circle in the chain being tangent to the previous and the next let B be the orthogonal projection of O to p (see Figure 2).
circle in the chain, today widely known as a Steiner chain. In Two applications of Pythagoras’s theorem yield
the usual closed Steiner chains, the first and last circles are ƒ! ƒ! ƒ! ƒ! ƒ! ƒ!
also tangent to each other; by contrast, in open Steiner AM  AN ¼ ð AB þ BM Þ  ð AB þ BN Þ
chains they need not be. Several generalizations of Steiner ƒ! ƒ!
¼ ð AB þ BM Þ
chains exist, most notably Soddy’s hexlet and Pappus ƒ! ƒ!
chains, and many of their properties have been discovered. ð AB  BM Þ ¼ AB2  BM 2
A fundamental result is Steiner’s porism, which states that: ¼ ðAO2  OB2 Þ
 ðR2  OBÞ2 ¼ AO2  R2 ;
THEOREM 1 (Classical Steiner porism). If at least one
ƒ! ƒ!
closed Steiner chain of n circles exists for two given circles so the quantity AM  AN is independent of the choice of
m and l, then there is an infinite number of closed Steiner the secant line p and it is called the power of A with
chains of n circles. Moreover, any circle tangent to m and l respect to k.
in the same way is a member of such a chain. The radical line, also called the radical axis, is the locus
of points of equal power with respect to two nonconcentric
Steiner proved his theorem using the method of circle circles. By the chordal theorem, it is perpendicular to the
inversion. Since it preserves tangencies, angles, and circles, line of centers. Coaxal circles are circles whose centers are
inversion transforms one Steiner chain into another of the collinear and that share a common radical line. The col-
same number of circles. One particular choice of inversion lection of all coaxal circles with respect to a fixed radical
transforms the two given circles into concentric circles so axis is called the pencil of coaxal circles. For further reading
all the circles of the Steiner chain are mapped by it to circles we refer reader to [2].

6 THE MATHEMATICAL INTELLIGENCER  2017 Springer Science+Business Media New York

Figure 1. Theorem 1.
Figure 3. Theorem 2 (red circles are coaxal).

Figure 4. The radical line of two circles.

Let k , k , . . ., k be a finite set of coaxal circles. Steiner Chains section, we provide elementary proofs of
Figure 02. The
1 n
power of a point with respect to a circle.
other interesting results related to generalized Steiner
DEFINITION 1. A sequence of circles cfi0 0 ; j0 g , cfi1 1 ;j1 g , cfi2 2 ;j2 g , chains, which further extend the results from [7]. Finally, in
. . . satisfying the following conditions: the Some Open Questions section, we discuss the question
of criteria on a set of given coaxal circles that guarantee the
fi ;j g
• each circle cp p p is tangent to the circles kip and kjp existence of a generalized Steiner chain of a given pre-
fip1 ;jp1 g fi ;j g scribed type of tangencies.
• neighboring circles cp1 and cp p p are tangent for
every p 2 N
is called a generalized Steiner chain. Proof of Theorem 2
For our generalization of Steiner porism it is of crucial
A finite Steiner chain is said to be closed if the first and importance to understand elementary properties of pencils
the last circle in the chain are tangent to each other. The of coaxial circles. Before we proceed to the proof, we
main theorem of the paper observes the poristic property briefly review some basic facts about the geometry of cir-
of this generalized Steiner chain (see Figure 3). cles and inversion.
The radical line of two given circles is, by the chordal
THEOREM 2 (Generalized Steiner porism). If at least one theorem, a line perpendicular to the line of centers. In the
fi ; j g fi ;j g
closed Steiner chain of m þ 1 circles c0 0 0 , c1 1 1 , . . ., case of intersecting circles it is the line passing through the
fi ;j g
cmm m exists for a given finite set k1 , k2 , . . ., kn of coaxal common points of two circles whereas in the case of tan-
circles, then there is an infinite number of closed Steiner gent circles the radical line is just the common tangent in
chains of m þ 1 circles such that the ordering of tangencies the point of tangency (see Figure 4).
with the given n coaxal circles is preserved. Moreover, any Let k and l be two given nonintersecting circles in the
circle tangent to ki0 and kj0 is the first member of such a plane and p be their radical line. Let O and S be the centers
Steiner chain. of k and l, respectively, and X the intersection point of p
and the line of centers. Let U 2 k and V 2 l be points such
In the Proof of Theorem 2 section, we prove Theorem 2 that the lines UX and UV are tangent to k and l, respectively.
using the same argument as in the standard proof of the By the properties of the radical line it follows that UX ¼ VX
classical Steiner porism. After that, in the Theorems on so the circle x with center in X and radius UX is orthogonal

 2017 Springer Science+Business Media New York, Volume 39, Number 1, 2017 7
Figure 7. An intersecting, a tangent, and nonintersecting

Figure 5. The circle orthogonal to two given circles.

Figure 6. Inversion transforming two nonintersecting circles

into concentric circles.

to both k and l (see Figure 5). Recall that two circles are
orthogonal if the tangent to one circle in a common point
passes through the center of the second circle.
Now let A and B be the intersection points of x with the Figure 8. The picture after inversion.
line of centers of k and l, and C be a circle with center in A.
Let us consider an inversion with respect to C. Then the line
of centers is mapped onto itself, and x is mapped to a line defined by the two point-circles. The two point-circles it
x0 perpendicular to the line of centers. Because k and l are contains are its limiting points and can be used to span the
orthogonal to both the line of centers and x, it follows that pencil. In the case of an intersecting pencil, any two circles
k and l are transformed into the circles orthogonal to the k and l intersect in two fixed but distinct points on the
lines OS and x0 . Thus, their images are k 0 and l 0 —two common radical line and these points are called the base
concentric circles with the center at the intersection point points of the pencil. The pencil is the intersecting pencil of
B0 of the line of centers and x0 (see Figure 6). circles consisting of all the circles through its base points.
Let two nonconcentric circles k and l and their radical The case of a tangent pencil is a limiting case of the two
line p be given. One can consider the set of all the circles c previous ones. All the circles of the pencil are tangent to
such that p is the radical axis of k and c. This is equivalent each other, with a common tangent that is the radical line.
to p being a radical line of l and c. Such a family of circles is
a pencil of coaxal circles, more precisely the pencil of cir- PROOF OF THEOREM 2: We will start the proof of Theo-
cles generated by k and l. The pencil contains two circles k rem 2 by observing that there exists an inversion
and l that span it, but it is also spanned by any pair of transforming coaxal circles into concentric circles. More-
circles it contains (in the same way as a line is spanned by over, this is the same inversion that maps any two circles
any pair of points). With respect to the relative positions of from the set into concentric ones (see Figure 8).
k and l we distinguish three type of pencils: a noninter- It turns out that the circles from the Steiner chain
secting, an intersecting, and a tangent (see Figure 7). touching the same concentric circles are congruent and,
In the case of nonintersecting pencils, any two circles k therefore, the whole figure can be rotated about the com-
and l in the pencil do not intersect and all the points of the mon center of these concentric circles. It is clear that, if
fi ; j g
radical line are outside the circles of the pencil. In this case, starting with c0 0 0 , we get a Steiner Chain; then, starting
fi0 ; j0 g
there are two circles of zero radius (point-circles) and the with d0 we will get the Steiner Chain obtained by
radical axis is the perpendicular bisector of the segment rotation. Clearly, it will contain the same number of circles


Figure 11. Proposition 1.
Let Yi be the second intersection point of the line UI
and the circumcircle of MA0 i B0 i I . The power of the point U
Figure 9. Lemma 1. with respect to the circumcircle of MA0 i B0 i I yields
UA0 i  UB0 i ¼ UYi  UI for all i:

Also, the power of the point U with respect to c0i gives

UA0 i  UB0 i ¼ UA0 iþ1  UB0 iþ1 for all i:

Thus, the last two equalities imply that Yi  Yiþ1 for every Y
so the lines Ai Bi are concurrent in the point X ¼ wðY Þ:
Similarly, the line Pi Qi is mapped onto the circumcircle
of MP 0 i Q0 i I under w for all i. Let Zi be the second inter-
section point of the line UI and the circumcircle of MP 0 i Q0 i I .
The power of the point U with respect to the circumcircle of
MP 0 i Q0 i I yields
UP 0 i  UQ0 i ¼ UZi  UI for all i:
Figure 10. Proof of Lemma 1.
Also, the power of the point U with respect to c0i gives

and the ordering of tangencies with the given n coaxal UA0 i  UB0 i ¼ UP 0 i  UQ0 i for all i:
circles will be the same. (
Now it easily follows that Zi  Yi for all i and therefore the
lines Pi Qi pass through the point X.
Theorems on Steiner Chains The points I, O, S, and U are collinear by standard
In this section we study various types of collinearities
properties of inversion maps, so the point X clearly lies on
among points in the configuration of Steiner chains.
this line. (
We will start with the following lemma:
From the proof of Lemma 1 we can easily deduce that
UI  UY ¼ UA0 i  UB0 i ¼ UP 0 i  UQ0 i for all i, which clearly
LEMMA 1. Let k and l be two nonintersecting circles and
implies the following results.
O and S their centers, respectively. Let c0 , c1 , c2 , . . . be
circles touching both circles k and l. Suppose that Ai and Bi
are the intersection points of circles ci1 and ci , and Pi and
Qi the points of tangencies of ci with k and l, respectively, • For every i and j, the points Ai , Bi , Aj , and Bj lie on the
for every i 2 N: Then all of the lines Ai Bi and Pi Qi are same circle.
concurrent in the same point X, and the points O, S, and X • For every i and j, the points Pi , Qi , Pj , and Qj lie on the
are collinear (see Figure 11). same circle.
• For every i and j, the points Pi , Qi , Aj , and Bj lie on the
PROOF: Let us consider an inversion w, which transforms same circle.
circles k and l into two concentric circles k 0 and l 0 whose
common center is the point U, and let I be the center of As an immediate consequence of Lemma 1 we have:
w. Let c0 i be the image of ci under w for every i. The line
Ai Bi is mapped to the circumcircle of MA0 i B0 i I (see Fig- COROLLARY 1. Common tangents of neighboring circles
ure 10). in a classical Steiner chain are concurrent.

 2017 Springer Science+Business Media New York, Volume 39, Number 1, 2017 9
Figure 13. Money-Coutts theorem.

q ¼ tan2 np. In this case we have a Steiner n-cycle, and such

an n-cycle can be constructed starting with any circle tan-
Figure 12. Theorem 3 (the lines in shades of green color are gent to both k and l [8, pp.98–100].
of type Pj
fp;qg fp;qg
Qj , the lines in shades of red color are of Experimenting with Cinderella, we constructed various
type tfe;f ;g;hg whereas the black line is the line of centers of examples of generalized Steiner chains. The easiest way to
coaxial circles.
obtain such configurations is to start with a set of concentric
circles and to construct a chain of circles inscribed in some
of the rings formed by those circles and then apply an
inversion to it. However, this method does not provide any
Now we turn our attention to the generalized Steiner reasonable conjecture for a starting condition guaranteeing
chains in the light of Lemma 1 and Corollary 1. Suppose the closure of the whole configuration.
fp ;q g
that a circle ci i i from the generalized Steiner chain is
fp ;q g
tangent to the circles kpi and kqi in the points Pi i i and PROBLEM 1. Establish a condition on m given coaxial
fpi ;qi g i
Qi . Let tfpi ;piþ1 ;qi ;qiþ1 g be the common internal tangent of circles that implies the existence of a generalized Steiner
fp ;q g fp ;q g
two neighboring circles ci i i and ciþ1iþ1 iþ1 . By similar chain of prescribed order of tangencies as in Definition 1.
reasoning, as in the proof of Lemma 1, we obtain the fol-
lowing theorem. Our Theorem 2 is a Steiner analogue of the Great Pon-
celet theorem. There are a number of other closure
THEOREM 3. theorems resembling Poncelet’s and Steiner’s porisms. One
fp;qg fp;qg
interesting theorem concerns mutually tangent circles
• All of the lines Pj Qj (through the points of inscribed in polygons and its variations, the Money-Coutts
tangencies of all circles touching kp and kq ) are Theorem (see [10] and [11]).
concurrent in the point Xfp;qg
• All of the tangents tfe;f ;g;hg between neighboring circles THEOREM 4 (Money-Coutts theorem). Let ABC be a tri-
such that the first touches ke and kg , and the second angle. Let k1 be a circle inscribed in the angle ABC, k2 the
touches kf and kh intersect in the common points
circle tangent to k1 and inscribed in the angle BCA, k3 the
Yfe;f ;g;hg .
circle tangent to k2 and inscribed in the angle CAB and so
• All of the points Xfp;qg and Yfe;f ;g;hg lie on the line through
on cyclically. Then the sequence of circles is 6-periodic:
the centers of coaxial circles k1 , k2 , . . ., kn (see Figure 12).
k1 ¼ k7 (see Figure 13).

An intriguing question is to obtain a nested version of

Some Open Questions
this theorem with several polygons included in the
A natural question related to porisms is what is a necessary
and sufficient condition for the existence of a poristic
property. For the Poncelet porism it is the remarkable
Cayley condition; see [4] and [3]. Such a condition is also
PROBLEM 2. Define and find a nested set of polygons
known for the case of the classical Steiner porism. For two having a poristic property that in some sense generalizes the
given nonintersecting circles k (O, R) and l (S, r) such that l Money-Coutts theorem.
lies in the interior of k, let d be the distance between their
centers. Let
ðR  rÞ2  d 2 The authors were supported by Grant 174020 of the Min-
q :¼ :
4Rr istry for Educationand Science of the Republic of Serbia
Then there is a closed chain of n mutually tangent circles and Grant 19/6-020/961-120/14 of the Ministry of Science
each tangent internally to k and externally to l if and only if and Technology of the Republic of Srpska.


The authors are greatly indebted to the anonymous REFERENCES
reviewers for their valuable comments and remarks. The [1] Marcel Berger, Geometry Revealed—A Jacob’s Ladder to
authors are grateful to Sergei Tabachnikov and Djordje Modern Higher Geometry, Springer-Verlag, Heidelberg (2010).
Žikelić for their interest, useful suggestions, and fruitful [2] H. S. M. Coxeter, Introduction to Geometry, John Wiley & Sons
discussions, which significantly improved the presentation (1969).
of the article. [3] V. Dragović and M. Radnović, Bicentennial of the Great Poncelet
Theorem (1813–2013): Current advances, Bulletin of the Amer-
-Dord¯e Baralić ican Mathematical Society 51:3 July (2014), pp. 373–445.
Mathematical Institute SASA [4] V. Dragović and M. Radnović, Poncelet Porisms and Beyond,
Kneza Mihaila 36, p.p. 367 Integrable Billiards, Hyperelliptic Jacobians and Pencil of
11001 Belgrade Quadrics, Springer, Basel (2011).
Serbia [5] L. Flatto, Poncelet’s Theorem, AMS (2009).
e-mail: [6] D. Fuchs and S. Tabachnikov, Mathematical Omnibus—Thirty
Lectures on Classic Mathematics, AMS (2007).
Dušan Jokanović [7] C. H. Lu, Exploring Steiner’s Porism with Cabri Geometry, Ph.D.
Production and Management Faculty thesis, National Chiao Tung University (2007).
University of East Sarajevo [8] D. Pedoe, Geometry, A Comprehensive Course, Dover Reprint,
Stepe Stepanovića b.b. (1988).
89101 Trebinje [9] J. Steiner, Einige geometrische Betrachtungen, J. Reine Angew.
Republic of Srpska Math. (Crelle) 1 (1826) 161–184.
e-mail: [10] S. Tabachnikov, Going in Circles: Variations on the Money-
Coutts Theorem, Geometriae Dedicata 80 (2000), 289-296.
Marina Milićević [11] D. Ivanov and S. Tabachnikov, The Six Circles Theorem Revisited,
Production and Management Faculty to appear in Amer. Math. Monthly., arXiv:1312.5260.
University of East Sarajevo
Stepe Stepanovića b.b.
89101 Trebinje
Republic of Srpska

 2017 Springer Science+Business Media New York, Volume 39, Number 1, 2017 11