I found a generalization of Thomsen theorem and Tucker circle as follows. I'm an electrical engineer, not a mathematician. I don't know how to prove these results.

Let $ABC$ be a triangle, let two points $A_1, A_4$ lie on BC, $A_2, A_5$ lie on $CA$, and two points $A_3, A_6$ lie on $AB$. Let $B_1$ on its edge $BC$ and $B_1 \neq A_1$. A sequence of points and parallel lines is constructed as follows. The parallel line to $A_1A_2$ through $B_1$ intersects $AB$ in $B_2$ and the parallel line to $A_2A_3$ through $B_2$ intersects $AB$ in $B_3$. Continuing in this fashion the parallel line to $A_3A_4$ through $B_3$ intersects $BC$ in $B_4$ and the parallel line to $A_4A_5$ through $B_4$ intersects $BC$ in $B_5$. Finally the parallel line to $A_5A_6$ through $B_5$ intersects AB in $B_6$ and the parallel line to $A_6A_1$ through $B_6$ intersects $BC$ in $B_7$. This problem states that: Six points $A_1, A_2, A_3, A_4, A_5, A_6$ lie on a conic iff $B_7=B_1$

enter image description here

  • When the conic through $A_1, A_2, A_3, A_4, A_5, A_6$ is the Steiner inellipse this problem is Thomsen theorem
  • When the conic through $A_1, A_2, A_3, A_4, A_5, A_6$ is the circumcircle this problem is Tucker circle
up vote 1 down vote accepted

For notational convenience, we'll take the vertices of the triangle to be $P_0$, $P_1$, $P_2$, and define further $P_{i+3} := P_i$, so that we can say that each $A_i$ and $B_i$ lie on $\overleftrightarrow{P_i P_{i+1}}$, for $i =0, 1, 2, 3, 4, 5$.

enter image description here

The configuration of points gives rise to values $a_i$ and $b_i$ ($\neq -1$) defined by $$a_i := \frac{|P_i A_i|}{|A_i P_{i+1}|} \qquad\qquad b_i = \frac{|P_i B_i|}{|B_i P_{i+1}|}$$ (with subscript arithmetic done modulo $6$), where, in the tradition of Ceva's Theorem, we consider these signed ratios of oriented segments: the value is positive when the segments point in the same direction, and negative when they point in opposite directions. Note that we can write, for instance, $$\frac{|P_i A_i|}{|P_i P_{i+1}|} = \frac{|P_i A_i|}{|P_i A_i| + |A_i P_{i+1}|} = \frac{a_i}{a_i+1} \qquad\quad \frac{|A_iP_{i+1}|}{|P_i P_{i+1}|} = \frac{1}{a_i+1}$$

Now, to the problem at hand: The parallelism conditions $$\overrightarrow{A_i A_{i+1}} \parallel \overrightarrow{B_i B_{i+1}}$$ imply that $\triangle A_i P_{i+1} A_{i+1} \sim \triangle B_i P_{i+1} B_{i+1}$ for each $i$. Therefore, $$\frac{|A_i P_{i+1}|}{|P_{i+1} A_{i+1}|} = \frac{|B_i P_{i+1}|}{|P_{i+1} B_{i+1}|} \quad\to\quad \frac{|P_i P_{i+1}|/(a_i+1)}{|P_{i+1}P_{i+2}|\;a_{i+1}/(a_{i+1}+1)} = \frac{|P_i P_{i+1}|/(b_i+1)}{|P_{i+1}P_{i+2}|\;b_{i+1}/(b_{i+1}+1)}$$ so that $$(1+a_{i+1}) b_{i+1}(1+b_i) = (1+b_{i+1})a_{i+1}(1+a_i) \quad\to\quad b_{i+1} = \frac{(1 + a_i) a_{i+1}}{1 - a_i a_{i+1} + b_i + a_{i+1} b_i}$$

That is, we can write $b_1$ in terms of $b_0$; and we can write $b_2$ in terms of $b_1$ (which, in turn, is in terms of $b_0$), and so on, until we arrive at this expression for $b_6$ in terms of $b_0$: $$b_6 = a_0\;\frac{1 + b_0 - ( a_0 - b_0 ) a_1 a_2 a_3 a_4 a_5 \phantom{a_0}}{ 1 + b_0 + (a_0-b_0) a_0 a_1 a_2 a_3 a_4 a_5}$$

Now, point $B_6$ coincides with $B_0$ when (and only when) $b_6 = b_0$. This says $$( 1 + b_0 )( a_0 - b_0 )( 1 - a_0 a_1 a_2 a_3 a_4 a_5 ) = 0$$ Since $b_i\neq -1$, and we assume $A_0 \neq B_0$ so that $a_0 \neq b_0$, the condition reduces to

$$a_0\,a_1\,a_2\,a_3\,a_4\,a_5 \;=\; 1 \tag{$\star$}$$

By Carnot's Theorem for Conics (which I proved in another answer), relation $(\star)$ is satisfied if and only if the $A_i$ lie on a conic. This proves the result. $\square$

  • I thank to Mister @Blue very much, You are a good solution geometer. I have some another problems sine two years ago. The problems are generalization of some classical theorem geometry. Can I post the problems in Math.Stackexchange.com? – Oai Thanh Đào Sep 17 '15 at 17:50
  • @OaiThanhĐào: Thank you for the compliment. (If I have solved your problem to your satisfaction, be sure to click the "√".) I think you should post all of your other problems to this site. (Do not post them all at once, though. One a week might be a good pace.) I find your problems quite interesting; they make me aware of ideas I hadn't seen (like Thebault circles) and remind me of ideas I'd nearly forgotten (like Steiner inellipses). – Blue Sep 18 '15 at 9:21
  • Mister @Blue , Can I using your solution to publish in Cut the Knot – Oai Thanh Đào Sep 18 '15 at 15:42
  • @OaiThanhĐào: So long as I get credit (a link to my profile here on M.SE is sufficient for a Cut-the-Knot post), you may publish my solution with your problem. Thanks for asking! – Blue Sep 18 '15 at 16:26
  • I have about one hundred my problems appear in Cut The Knot. But I need your really name? could you tell me? @Blue ? – Oai Thanh Đào Sep 18 '15 at 17:16

Your Answer


By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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