Let $\Delta ABC$ be a triangle. If we place points $D,\ E,\ F$ arbitrarily on the sides $\overline{AB},\ \overline{BC}$ and $\overline{CA}$ respectively, then the circumcircles of the triangles $\Delta ADF,\ \Delta BDE$ and $\Delta CEF$ will all pass through a common center via Miquel's theorem. Let $M$ be this common point, the Miquel point.

If we locate the centers of the circumcircles and name them $P,\ Q,\ R$ respectively, then the resulting triangle $\Delta PQR$ will be similar to triangle $\Delta ABC$. I want to prove that the Miquel point $M$ is the center of the spiral similarity which carries $\Delta PQR$ to $\Delta ABC$         miquel

If somebody could provide a proof or a reference to a proof that would be greatly appreciated. Thank you.

If someone could just provide a proof for the regular similarity between the two triangles, that would also be appreciated. I have looked around online but could not find a proof.

  • $\begingroup$ So does a "spiral similarity with center M" mean that you can perform a combination of rotation about that point and dilation to match the figures? $\endgroup$ – rschwieb Oct 4 '12 at 16:15
  • $\begingroup$ @rschwieb Yes, that's what I mean. A dilation and rotation with center $M$ bringing $\Delta PQR$ to $\Delta ABC$. $\endgroup$ – EuYu Oct 4 '12 at 16:44

Let's consider only two different settings in our proof, as shown in figure 1 and 2. You can complete the proof for other settings.

(1) Case ($\angle ERM$ and $\angle MRF$ are adjacent angles).

See figure 1.


We know that $\triangle MRE$ is isosceles and $QR$ is bisector of $\angle MRE$. We also know that $\triangle MRF$ is isosceles and $PR$ is bisector of $\angle MRF$.

Let $m(\angle MRE) = 2 \alpha$ and $m(\angle MRF) = 2 \beta$, then we can conclude that $m(\angle ECF)= \alpha + \beta$, since $\angle ECF$ is a inscribed angle and its measure is half the measure of the central angle $\angle ERF$. ($m(\angle ERF) = 2\alpha + 2 \beta$).

So it follows that $$\angle ECF = \angle QRP.$$

By a similar approach we can conclude that

$$\angle DAF = \angle QPR.$$

Therefore $\triangle ABC \sim \triangle QPR$.

(2) Case ($\angle ERM$ and $\angle MRF$ are not adjacent angles).

See figure 2.

In that setting there is a superposition of $\triangle MRE$ and $\triangle MRF$, so $m(\angle ERF) = 2 \beta - 2 \alpha$, and no more $2 \alpha + 2 \beta$.

Note that $m(\angle QRF)= \beta - \alpha$ and that $m(\angle ECF)= \beta - \alpha$, since $m(\angle ECF)$ is still half the measure of $\angle ERF$.

Therefore $$\angle ECF = \angle QFR.$$

As in case 1 (by a similar approach), we can conclude that

$$\angle DAF = \angle QPR.$$

Therefore $\triangle ABC \sim \triangle QPR$, as in case 1.

  • $\begingroup$ Excellent. Thank you. $\endgroup$ – EuYu Jan 9 '13 at 20:24
  • $\begingroup$ @EuYu. You're welcome:) $\endgroup$ – RicardoCruz Jan 9 '13 at 20:28

Fact. If $x$, $y$, $z$ are points on the complex plane, then the circumcenter of $\triangle xyz$ is the point

$$ i \frac{x(z\overline{z} - y\overline{y})+ y(x\overline{x}-z\overline{z})+z(y\overline{y}-x\overline{x})}{x(\overline{z} - \overline{y})+ y(\overline{x}-\overline{z})+z(\overline{y}-\overline{x})}$$

where "$\overline{x}$" indicates the complex conjugate of $x$.

Let us consider $\triangle abc$, defining $d$, $e$, $f$ on sides $bc$, $ca$, $ab$, respectively:

$$d := b + \alpha(c-b) \qquad e := c+\beta(a-c) \qquad f := a+\gamma(b-a)$$

for real scalars $\alpha, \beta, \gamma$.

Let $p$, $q$, $r$ be the respective circumcenters of $\triangle aef$, $\triangle dbf$, $\triangle dec$; for simplicity, we take $a$, $b$, $c$ on the unit circle so that $\overline{a}=1/a$, etc. Then,

$$\begin{align} p &= i\;\frac{c(b-a)+ \beta b ( a-c) + \gamma c( a - b )}{b-c} \\[6pt] q &= i\;\frac{a(c-b)+ \gamma c ( b-a) + \alpha a( b - c )}{c-a} \\[6pt] r &= i\;\frac{b(a-c)+ \alpha a ( c-b) + \beta b( c - a )}{a-b} \end{align}$$

We deduce

$$\frac{p-q}{a-b} = \frac{q-r}{b-c} = \frac{r-p}{c-a} =: u$$

which, by taking the modulus, proves $\triangle abc \sim \triangle pqr$. A little algebra reveals that $$u + \overline{u} = 1$$

Now ... A point, $m$, (which may or may not be the Miquel point) is the center of spiral similarity from $\triangle abc$ and $\triangle pqr$, if and only if

$$\frac{a-m}{p-m} = \frac{b-m}{q-m} = \frac{c-m}{r-m}$$

(This characterization is what inspired me to approach the problem in the complex plane.) Solving for $m$ via the first equality gives

$$m = \frac{aq-bp}{(a-b)-(p-q)} = \frac{r - c u}{1-u}$$

whence $$m-r = \frac{u(r-c)}{1-u} = \frac{u}{\overline{u}}(r-c) \qquad\qquad \overline{m}-\overline{r} = \frac{\overline{u}}{u}(\overline{r}-\overline{c})$$ so that $$|m-r|^2 = (m-r)(\overline{m}-\overline{r}) = (r-c)(\overline{r}-\overline{c}) = |c-r|^2$$

indicating that $c$ and $m$ are equidistant from $r$: thus, $m$ is on the circumcircle of $\triangle dec$. By symmetry, it is also on the circumcircles of $\triangle aef$ and $\triangle dbf$; the center of spiral similarity is in fact the Miquel point.

  • $\begingroup$ This result has been a curiosity at the back of my mind for a very long time. Thank you for providing such a nice proof. $\endgroup$ – EuYu Jan 10 '13 at 16:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.