I was recently posed with the following problem:

Let $\Omega$ be a circle passing through vertices $B$ and $C$ in triangle $\Delta ABC$ and let $\omega$ be a circle tangent to segments $AB$ and $AC$ at the points $P$ and $Q$ respectively, and externally tangent to $\Omega$ at point $T$. Let $M$ be the midpoint of arc $BTC$ of $\Omega$. Show that lines $BC$, $PQ$, and $MT$ concur.

Naturally, I first made a quick sketch of the problem:

My Quick Sketch

I then quickly realised that in order for $\Omega$ to pass through both $B$ and $C$, its centre must lie on their perpendicular bisector. This also quickly leads to finding $M$, as it is simply the point where $\Omega$ intersects this perpendicular bisector.

I then managed to loosely show that $T$ has to be on the angular bisector of $\angle BAC$, i.e. the point where $\Omega$ intersects with this angular bisector. This also very simply shows that $P$ and $Q$ must be the same distance away from $A$, or in other words, that the triangle $\Delta PQA$ is isosceles.

With this information, I made a new, more accurate diagram:

My Updated Diagram

I then started trying to prove the statement in the original problem. Since we know $M$ and $T$, and we know $BC$, we can now find the point where the three lines must concur. Then, since we know that $\Delta PQA$ is isosceles, my idea was that we could find the point of concurrence, and draw a line from it to $\Delta ABC$, such that it would make an isosceles triangle at the points where it intersects with the triangle, and check if those points are $P$ and $Q$ (the points of tangency with the circle). If they are, then we have proved it to be true, since $P$ and $Q$ make a line to the point of concurrence, otherwise, it is false.

This is where I got stuck, I don't know how I would go about finding this line from the point of concurrence to $\Delta ABC$, or how I would then go about showing that it intersects at $P$ and $Q$ specifically. I pondered this for a few days but couldn't think of anything, so I decided to ask it here. All help is greatly appreciated!

EDIT: it turns out a claim I made earlier was wrong, $T$ does not have to lie on the angular bisector as @user10354138 pointed out

EDIT: I also had another finding I forgot to include in my original question. Since $\omega$ is tangent to line $AB$ at the point $P$, it follows that the centre of $\omega$ must lie on the line perpendicular to $AB$ at point $P$, and similarly for $AC$ and point $P$, like so:

Tangency of omega and ABC

And since the lines $OP$ and $OQ$ (where $O$ is the centre of $\omega$) must be the same length, $O$ must lie on the angular bisector of $\angle BAC$.

  • 1
    $\begingroup$ $T$ doesn't need to lie on the angle bisector of $A$. If it does, then the centre of $\Omega$ would also lie on this angle bisector but there is absolutely no reason why it has to, unless $AB=AC$. $\endgroup$ Sep 30 '21 at 3:12
  • $\begingroup$ @user10354138 I wasn't exactly sure of it, but I thought I had proved it, I guess I was wrong. Just curious, could I see an example? $\endgroup$ Sep 30 '21 at 3:26
  • $\begingroup$ For every point O on the perpendicular bisector of BC that is sufficiently far away from BC in the opposite direction of A, the circle centred O through BC has a minor arc lying inside triangle ABC and hence you can construct a circle tangent to all three circlelines AB, AC, minor arc BC inside this shape. $\endgroup$ Sep 30 '21 at 3:41
  • $\begingroup$ See this picture for example. $\endgroup$ Sep 30 '21 at 5:28


Let $ω, Ω$ have centers $N, O$ and radii $r, R$, and let $D = BC ∩ MT$. By the law of cosines on $\triangle BON$ and $\triangle BOT$, we have

\begin{multline*} BP^2 = BN^2 - r^2 = (R^2 + (R + r)^2 - 2R(R + r) \cos ∠BON) - r^2 \\ = \frac{R + r}{R}(R^2 + R^2 - 2R^2 \cos ∠BON) = \frac{R + r}{R} BT^2, \end{multline*}

and similarly $CQ^2 = \frac{R + r}{R}CT^2$. Since $∠BTD$ and $∠MTC$ subtend equal arcs of $Ω$, $TD$ is the external bisector of $∠BTC$, so $\frac{BD}{CD} = \frac{BT}{CT} = \frac{BP}{CQ}$. Therefore, $\frac{AP}{PB} · \frac{BD}{DC} · \frac{CQ}{QA} = -1$, which makes $P, Q, D$ collinear by Menelaus’s theorem.

  • $\begingroup$ I almost completely understand what you are saying, I just don't understand the part where you go from an expression with a cosine in it to an expression without it ($\cos \angle BON$ to just $\angle BON$). $\endgroup$ Oct 4 '21 at 8:45
  • $\begingroup$ @TymonMieszkowski That was a typo, fixed. $\endgroup$ Oct 4 '21 at 9:15
  • $\begingroup$ @AndersKaseorg Can you please explain briefly why $TD$ is the external bisector of $\angle BTC$? $\endgroup$
    – YNK
    Oct 4 '21 at 11:59
  • 1
    $\begingroup$ @YNK $∠BTD$ subtends arc $BM$ of $Ω$. $∠CTM$ subtends arc $CM$ of $Ω$. $M$ is the midpoint of arc $BTC$, so arcs $BM$ and $CM$ are equal, so $∠BTD = ∠CTM$ by the inscribed angle theorem. If you’re confused by the unusual configuration of $∠BTD$, you can derive the same result from the more familiar configuration where the supplementary angle $∠BTM$ subtends everything except arc $BM$. If you’re still stuck here, review the inscribed angle theorem. $\endgroup$ Oct 5 '21 at 9:13
  • 1
    $\begingroup$ @TymonMieszkowski It’s not a typo; the signed lengths are important to distinguish the very different configurations of Menelaus’s theorem and Ceva’s theorem. Other than that, this is just simple fraction manipulation combined with $AP = QA$. $\endgroup$ Oct 6 '21 at 2:27

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.