Cut the knot: learn to enjoy mathematics
A math books store at a unique math study site. Shopping at the store helps maintain the site. Thank you.
Learning Math Online
Sites for teachers
Sites for parents
Terms of use
Awards
Interactive Activities

CTK Exchange
CTK Wiki Math
CTK Insights - a blog
Math Help

III Millennium Olympiad

Games & Puzzles
What Is What
Arithmetic
Algebra
Geometry
Probability
Outline Mathematics
Make an Identity
Book Reviews
Stories for Young
Eye Opener
Analog Gadgets
Inventor's Paradox
Did you know?...
Proofs
Math as Language
Things Impossible
Visual Illusions
My Logo
Math Poll
Cut The Knot!
MSET99 Talk
Other Math sites
Front Page
Movie shortcuts
Personal info
Privacy Policy

Guest book
News sites

Recommend this site

Sites for parents

Education & Parenting

Manifesto  |  Bookstore  |  Contents  |  Amazon store  |  Term index  |  What changed?  |  Contact  |  Recommend
RSS Feed: Recent changes at CTK

9-point Circle as a Locus of Concurrency: What Is This About?
A Mathematical Droodle

 

This applet requires Sun's Java VM 2 which your browser may perceive as a popup. Which it is not. If you want to see the applet work, visit Sun's website at http://www.java.com/en/download/index.jsp, download and install Java VM and enjoy the applet.


Buy this applet
What if applet does not run?

The Simson lines of two diametrically opposite points on a circumcircle of a triangle intersect on the 9-point circle of that triangle. A generalization of this fact is known as Griffiths' theorem: Let the line joining two points P and Q pass through the circumcenter of ΔA1A2A2. Then the pedal circles of points P and Q meet on the 9-point circle of ΔA1A2A2, i.e. the three circles are concurrent. This is known as Griffiths' theorem. In fact the point of concurrency only depends on the line the two points lie on. In other words, the pedal circles of the points on a line through the circumcenter of a triangle all concur at the same point on its 9-point circle. The point is known as Griffiths' point.

(It is worthwhile to note that the 9-point circle is the pedal circle of the circumcenter as well as that of the orthocenter.)

Griffiths' theorem admits a proof based on complex numbers.

Copyright © 1996-2010 Alexander Bogomolny

 

 

 

 

 

 

 

 

 

 

 

 

 

Proof

 

This applet requires Sun's Java VM 2 which your browser may perceive as a popup. Which it is not. If you want to see the applet work, visit Sun's website at http://www.java.com/en/download/index.jsp, download and install Java VM and enjoy the applet.


Buy this applet
What if applet does not run?

Assume the circumcenter of ΔA1A2A3 is at the origin and the circumradius equals 1. Lower case letters denote complex numbers associated with the points denoted by the corresponding upper case so that, for example ai corresponds to Ai, i = 1, 2, 3. We have

  |a1| = |a2| = |a3| = 1.

Let P lie on a straight line l through the origin and C1. We choose l to coincide with the real axis which in particular implies p = p'. Let C2C3 be the feet of perpendiculars from P to the side lines A2A3, A1A3, and A1A2.

Since C1 is collinear A2 and A3

  (c1 - a2) / (a3 - a2) = (c'1 - a'2) / (a'3 - a'2)

From |a2| = 1, a'2 = 1/a2 and a'3 = 1/a3. So we have

 
(c1 - a2) / (a3 - a2)= (c'1 - a'2) / (a'3 - a'2)
 = (c'1 - 1/a2) / (1/a3 - 1/a2)
 = (c'1a2 - 1) a3 / (a2 - a3)

from which

  c1 - a2 = a3 - a2a3c'1

or

  a2 + a3 = c1 + a2a3c'1

If the indices i, j, k are distinct and come from the set {1, 2, 3} then, more generally,

(1) c'k ai aj = ai + aj - ck.

On the other hand, PCk is perpendicular to AiAj, so that

  (p - ck)/(ai - aj) + (p' - c'k)/(a'i - a'j) = 0,

which, using |ai| = |aj| = 1, we modify as follows:

  (p - ck)/(ai - aj) + (p' - c'k) ai aj /(aj - ai) = 0,

or, with p = p',

  p - ck - (p - c'k) ai aj = 0.

Finally, we obtain

(2) c'k ai aj = p ai aj - p + ck.

A comparison of (2) and (1) gives

(3) ck = (ai + aj + p - p ai aj) / 2.

Introduce point

(4) w = (a1 + a2 + a3 - a1 a2 a3) / 2.

We shall show that four points C1, C2, C3, W are concyclic, i.e. W lies on the circumcircle of ΔC1C2C3. Note W is independent of P. In particular, W is the same when P is at the origin, i.e. when P coincides with the circumcenter in which case the circumcircle of ΔC1C2C3 is the nine point circle of ΔA1A2A3. Thus we shall have proved Griffiths' theorem and showed that W is exactly the Griffiths point.

For the concyclicity of C1, C2, C3, W we have to check that

  (w - c2)/(c1 - c2) : (w - c3)/(c1 - c3) = (w' - c'2)/(c'1 - c'2) : (w' - c'3)/(c'1 - c'3).

which we shall prove in another form

(5) (w - c2)/(c1 - c2) : (w' - c'2)/(c'1 - c'2) = (w - c3)/(c1 - c3) : (w' - c'3)/(c'1 - c'3).

First of all, from (3) and (4),

 
w - c2= (a1 + a2 + a3 - a1 a2 a3) / 2 - (a1 + a3 + p - p a1 a3) / 2
 = (p - a2)(a1 a3 - 1) / 2.

From here, using |ai| = 1,

 
w' - c'2= (p - a'2)(a'1 a'3 - 1) / 2
 = (p - 1/a2)(1/a1 1/a3 - 1) / 2
 = -(p a2 - 1)(a1 a3 - 1) / 2a,

where a = a1a2a3. Also

  c1 - c2 = (a1 - a2)(p a3 - 1) / 2

and

  c'1 - c'2 = -(a1 - a2)(p - a3) / 2a.

Similar expressions are obtained for the differences in (5) that include c3. The left hand side in (5) becomes

  (p - a2)(a1 a3 - 1)(a1 - a2)(p - a3) / (a1 - a2)(p a3 - 1)(p a2 - 1)(a1 a3 - 1)

or, after reduction,

  (p - a2)(p - a3) / (p a3 - 1)(p a2 - 1),

which is symmetric in a2 and a3 showing that the right hand side in (5) equals the same expression, thus proving (5) and, with it, Griffiths' theorem.

(Line l passing through the circumcenter intersects the circumcircle at two points. For each of these, the three projections C1, C2, C3 are collinear and lie on the simsons of ΔA1A2A3. It follows that the two simsons meet at the Griffiths point associated with l.)

Assume now there are four concyclic points A1, A2, A3, and A4. Taken by three, they form four triangles Tm = ΔAiAjAk, where i, j, k, m are distinct elements from {1, 2, 3, 4}. To each Tm and the given line l there correspond Griffiths point Wm. Using the same technique as above we can show (after J. Tabov) that the four Griffiths points W1, W2, W3, W4 are collinear. It is sufficient to show that

  (w1 - w2) / (w'1 - w'2) = (w1 - w3) / (w'1 - w'3) = (w1 - w4) / (w'1 - w'4).

This and an additional feature of the configuration is considered on a separate page.

A purely geometric proof can be found in [Advanced Euclidean Geometry, pp. 245-246].

References

  1. R. A. Johnson, Advanced Euclidean Geometry, Dover, 2007
  2. J. Tabov, Four Collinear Griffiths Points, Mathematics Magazine, v. 68, n 1, February 1995, pp. 61-64

Copyright © 1996-2010 Alexander Bogomolny

35683325Page copy protected against web site content infringement by Copyscape

Search:
Keywords:

Google
Web CTK