Two circles of radii $4$ and $16$ units, respectively touch each other externally. Prove that the length of the chord intercepted on the transverse common tangent to the to circles, by a circle of diameter $D$ units which touches the given circle externally, is $\frac{4D}{5}$

enter image description here

  • $\begingroup$ More generally (using slightly different notation): If $\bigcirc A$, $\bigcirc B$, $\bigcirc C$, of radius $a$, $b$, $c$, are pairwise externally tangent, then the line tangent to $\bigcirc A$ and $\bigcirc B$ at their intersection cuts $\bigcirc C$ in a chord of length $4c\sqrt{ab}/(a+b)$. I have a somewhat inelegant coordinate proof of this; I'm investigating a geometric approach. $\endgroup$ – Blue Oct 7 '17 at 5:15
  • $\begingroup$ Yes! Analytic geometry kills it, but it's very ugly. $\endgroup$ – Michael Rozenberg Oct 7 '17 at 5:39

Here's an argument that could probably be streamlined a bit ...

Restating the problem:

Let $\bigcirc A$, $\bigcirc B$, $\bigcirc C$ ---of radii $a$, $b$, $c$--- be pairwise externally tangent, with points of tangency $D$, $E$, $F$ as shown in the image. Let the line tangent to $\bigcirc A$ and $\bigcirc B$ at $F$ meet $\bigcirc C$ in chord $\overline{PQ}$ with midpoint $M$; define $p := |\overline{PM}| = |\overline{QM}|$. Then

$$p = \frac{2c\sqrt{ab}}{a+b} \tag{$\star$}$$

enter image description here

Extend $\overline{FE}$ and $\overline{FD}$ to meet $\bigcirc C$ again at $A^\prime$ and $B^\prime$. Note that $\triangle AEF$ and $\triangle CEA^\prime$ are isosceles, with congruent base angles at $E$ and therefore also at $F$ and $A^\prime$; thus, $\overline{A^\prime C}\parallel\overline{AF}$. Likewise, $\overline{B^\prime C}\parallel\overline{BF}$, making $A^\prime$, $B^\prime$, $C$ collinear on a diameter of $\bigcirc C$.

By Thales' Theorem, $\angle A^\prime PB^\prime$ is a right angle; because $\overline{PM}\perp\overline{A^\prime B^\prime}$, we have the classical construction of a geometric mean: $$|\overline{PM}|^2 = |\overline{A^\prime M}||\overline{B^\prime M}| \tag{1}$$

(This is why I chose $p$ to name the half-chord, instead of the full length of $\overline{PQ}$.) Now our task reduces to a matter of rewriting the right-hand side.

Observe that the perpendiculars at $D$, $E$, $F$ necessarily concur at a point equidistant from them; that is, they meet at center $I$ of the incircle (of radius $r$) of $\triangle ABC$. Of course, $\overline{AI}$ and $\overline{BI}$ are angle bisectors.

Since $\triangle A^\prime M F\sim \triangle IFA$ and $\triangle B^\prime MF\sim \triangle IFB$, we have $$\begin{align} \frac{|\overline{A^\prime M}|}{|\overline{MF}|} = \frac{r}{a} \quad\text{and}\quad \frac{|\overline{B^\prime M}|}{|\overline{MF}|} = \frac{r}{b} &\quad\implies\quad \frac{|\overline{B^\prime M}|}{|\overline{A^\prime M|}} = \frac{a}{b} \tag{2} \\ &\quad\implies\quad |\overline{A^\prime M}| = 2c\;\frac{b}{a+b} \quad\text{and}\quad |\overline{B^\prime M}| = 2c\;\frac{a}{a+b} \end{align}$$

(The above improves on my first pass at this proof. (See the revision history.) There may yet be a better argument that incorporates this proportionality relation: $$\frac{|\overline{A^\prime M}|}{b} = \frac{|\overline{A^\prime B^\prime}|}{|\overline{AB}|} = \frac{|\overline{B^\prime M}|}{a} \tag{2.5}$$ Be that as it may ...) Thus, $(1)$ becomes $$p^2 = \frac{4abc^2}{(a+b)^2} \qquad\to\qquad p = \frac{2c\sqrt{ab}}{a+b} \tag{3}$$ as claimed. $\square$

For the original question, we take $a = 4$, $b = 16$, $c = D/2$, so that the length of the chord is $$2p = \frac{4 \cdot D/2 \cdot \sqrt{4\cdot 16}}{4+16} = \frac{4D}{5}$$

| cite | improve this answer | |
  • $\begingroup$ The "It is easy..." claim need to be PROVEN! $\endgroup$ – Moti Oct 7 '17 at 15:38
  • $\begingroup$ @Moti: $\overline{AB}\parallel\overline{A^\prime B^\prime}$, so isos. $\triangle AEF$ and $\triangle C E A^\prime$ have equal vertex angles, and thus also equal base angles. In particular, the angles at $E$ match, making $\overrightarrow{EF}$ and $\overrightarrow{EA^\prime}$ opposite rays. Done. $\square$ If you like, imagine that I constructed the converse: Extend $\overline{FE}$ to meet $\bigcirc C$ again at $A^\prime$. Then $\triangle AEF$ and $\triangle CEA^\prime$ have equal base angles, and thus also equal vertex angles, so $\overline{AB}\parallel\overline{A^\prime C}$. $\square$ OK? :) $\endgroup$ – Blue Oct 7 '17 at 18:14
  • $\begingroup$ @Moti: Actually, I like the converse construction better. I'm going to edit my answer. :) $\endgroup$ – Blue Oct 7 '17 at 18:16

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.