Proof of Morley's Theorem
Date: 08/09/2000 at 00:31:34 From: Karamalizadeh Subject: Trisecting angles If you take any triangle ABC and trisect each angle, ABK = KBL = LBC ACJ = JCL = LCB BAK = KAJ = JAC Each pair of trisectors meet a point, forming an equilateral triangle JKL. Can you prove it? Thanks. B /^\ / \ / L \ / K* * \ / * J \ A /_ _ _ _ _ _ _ _\C
Date: 08/09/2000 at 07:54:38 From: Doctor Floor Subject: Re: Trisecting angles Hi, Karamalizadeh, Thanks for writing. You are referring to the famous "Morley trisector theorem," by Frank Morley (1904). Note that I use DEF for your JKL. We will use the Law of Sines, which states that a b c ----- = ----- = ----- = 2R sin(A) sin(B) sin(C) and thus, for instance, also a = 2R sin(A) where a = BC, b = AC, c = AB, and R is the radius of the circumcircle of triangle ABC. We apply this to triangle BDC and find BD a a -------- = -------------------- = ------------- sin(C/3) sin(180 - B/3 - C/3) sin(120 + A/3) where we use the fact that A+B+C = 180 degrees. We rewrite this to a sin(C/3) 2R sin(A) sin(C/3) BD = -------------- = ------------------ .................. sin(120 + A/3) sin(120 + A/3) Now we will use the following trigonometric identities: 4 sin(t)sin(t+60)sin(t+120) = 4 sin(t)(sin(t)cos(60) + cos(t)sin(60))*(sin(t)cos(120) + cos(t)sin(120)) = 4 sin(t)(1/2 sin(t) + 1/2 sqrt(3)cos(t))*(-1/2 sin(t) + 1/2 sqrt(3)cos(t)) = 4 sin(t)(3/4 cos^2(t) - 1/4 sin^2(t)) = 4 sin(t)(1/2 cos^2(t) + 1/4(cos^2(t) - sin^2(t))) = 2 sin(t)cos^2(t) + sin(t)cos(2t) = sin(2t)cos(t) + sin(t)cos(2t) = sin(3t) So  can be rewritten as BD = 8R sin(A/3)sin(C/3)sin(60 + A/3) In the same way we find BF = 8R sin(A/3)sin(C/3)sin(60 + C/3) and we can apply the Law of Cosines to triangle BDF: DF^2 = BD^2 + BF^2 - 2*BD*BF*cos(B/3) = 64R^2 sin^2(A/3)sin^2(C/3)*(sin^2(60+A/3) + sin^2(60+C/3) - 2 sin(60+A/3)sin(60+C/3)cos(B/3)) ... Now again we will have to apply a trigonometric identity, using the fact that the sum of 60+A/3, 60+C/3 and B/3 is 180 degrees. I will show below that if u+v+w = 180 degrees, then we find: sin^2(u) + sin^2(v) - sin^2(w) = 2 sin(u)sin(v)cos(w) ..... Applying  to  we find that DF^2 = 64R^2 sin^2(A/3)sin^2(B/3)sin^2(C/3) and thus DF = 8R sin(A/3)sin(B/3)sin(C/3) Since this expression is independent of permutations of A,B and C, we will find the same expression for DE and EF, so that indeed DEF is an equilateral triangle. ******** Proof of identity . I recall from the following message from the Dr. Math archives, "Equilateral Triangles: Area Formula and Proof," http://mathforum.org/dr.math/problems/andy6.16.98.html that if x+y+z = 2*pi = 360 degrees then cos^2(x) + cos^2(y) + cos^2(z) = 1 + 2cos(x)cos(y)cos(z) We apply this to u+90, v+90 and w, and find, using that cos(t+90)= -sin(t) for any angle t. sin^2(u) + sin^2(v) + cos^2(w) = 1 + 2 sin(u)sin(v)cos(w) sin^2(u) + sin^2(v) + cos^2(w) = sin^2(w) + cos^2(w) + 2 sin(u)sin(v)cos(w) sin^2(u) + sin^2(v) - sin^2(w)= 2 sin(u)sin(v)cos(w) which proves that identity  holds indeed. ******** There are several other proofs of Morley's theorem. One was given by Professor John H. Conway of Princeton in the geometry-puzzles newsgroup, hosted by the Math Forum. Note that he starts with a triangle with angles 3A, 3B and 3C, so that A, B and C are the trisected angles. The message can be found at: http://mathforum.org/kb/message.jspa?messageID=1084185 (Quoting) This is "Morley's Trisector Theorem", which has long had a justly deserved reputation as being difficult to prove. So I was very pleased when I found a very simple proof some years ago. Here it is: Let the angles be 3A, 3B, 3C, and let "X+" mean "X + 60 degrees". Then there certainly exist 7 abstract triangles having the angles: A++,B,C; A,B++,C; A,B,C++; A,B+,C+; A+,B,C+; A+,B+,C; 0+,0+,0+ 1 2 3 4 5 6 7 since in every case the triple of angles adds to 180 degrees. Now these triangles are only determined up to scale, and I determine the scale by saying that certain lines are all to have the same length. I draw the situation below, but unfortunately can't draw all the lines and can't put the vertices exactly where I want them: B B B 5 3 C++ A+_____C+ A++ 1 0+_______0+ \ / B+ \ 7 / B+ A \ \ / / C \ 0+ / 4 \ / 6 C+ A+ A B++ C / \ / 2 \ / \ A--------X-------Y----------C Triangle number 7, with angles 0+,0+,0+, is clearly equilateral, and can take all its edges to have some fixed length L. Then I arrange that the edges B+ C+ C+ A+ A+ B+ of triangles 4, 5, 6 should also have length L. The way we choose the scale of the other three triangles I illustrate only with triangle number 2. Namely, I draw the isosceles triangle B++ X Y whose base angles have the value B+, and make the two slanting sides of this have length L. Then it's easy to see that these all fit together to make up a triangle whose angles are 3A, 3B, 3C, and which is therefore similar to the original one, so proving Morley's theorem. To see this, you just have to check that any two sides that come together have the same length, and that the angles around any internal vertex add to 360 degrees. The latter is easy, and the former is proved using congruences such as that that takes the vertices A, C+, B+ of triangle number 4 to the points A, B++, Y of triangle number 2. I know of no other proof that's less than twice as long! John Conway ******** It can be shown that by a good choice of external angle trisectors instead of the internal ones we used above, there can be found 17 more equilateral trisector triangles. If you need more help, just write back. Best regards, - Doctor Floor, The Math Forum http://mathforum.org/dr.math/
Date: 08/14/2000 at 05:27:00 From: Karamalizadeh Subject: Thanks for your answer Hi, Dr. Floor, Thanks for your answer about trisecting angles. Could this problem be solved without reference to trigonometry? If so, I would appreciated any help you could give me. Thanks a lot.
Date: 08/15/2000 at 06:46:39 From: Doctor Floor Subject: Re: Thanks for your answer Hi, again, Karamalizadeh, Thanks for your reaction. I thought that perhaps the "puzzle proof" by John Conway was such a proof without trigonometry. Here is another proof that, not unlike Conway's proof, works somewhat the other way around: Let DEF be an equilateral triangle. Let P be a point on the extended median from D (the line connecting D and the midpoint of EF), beyond the midpoint of EF. Points Q and R are chosen likewise on the medians from E and F respectively. Let A be the point of intersection of RE and QF, B of PF and RD and C of PE and QD. Now let l be the reflection of the line BP through BR, and m the reflection of CP through CQ; and let X be the intersection of l and m. We can define the points Y and Z likewise. Note that AR and AQ are the trisectors of <YAZ. We will name the magnitude of <RAQ = a. Similarly <PBR = b and <QCP = c. Let x be the base angle of (isosceles) triangle PEF, y of QFD and z of RDE. Then <BDC = < QDR = y + z + 60 deg and in the same way <CEA = x + z + 60 deg <AFB = x + y + 60 deg Now note that <AED = <AER - <RED = 180 deg - z and in the same way <AFD = 180 deg - y Since the angles of a quadrilateral add to 360 deg, we find in AFDE a + 60 deg + (180 deg - y) + (180 deg - z) = 360 deg a + 60 deg = y + z a = y + z - 60 deg .................................... With this we can compute <AZB = 360 deg - a - b - <AFB = 360 deg - a - b - (x + y + 60 deg) = 240 deg - (a + b + c) ........................... From the symmetry of this result, we see that <BXC and <CYA must have the same value. Why all this reasoning? Well, if we let a', b', c' be given so that a' + b' + c' = 60 deg, so that a', b', and c' are the trisected angles of a given triangle. Then we build the above construction with x = 60 deg - a', y = 60 deg - b' and z = 60 deg - c'. The result we get is that: * From  we see a = y + z - 60 deg = 60 deg - b' - c' = a' and in the same way b = b' c = c' * From  we see that <AZB = 240 deg - 60 deg = 180 deg so Z is a straight angle (and so are angles CYA and BXC). And we conclude that ABC is a triangle with the same angles as the given triangle, and that the angle trisectors intercept an equilateral triangle. As desired, to prove Morley's theorem. I learned this proof from a booklet by O. Bottema ("Hoofdstukken uit de Elementaire Meetkunde"). I hope this is a proof you like even better. Best regards, - Doctor Floor, The Math Forum http://mathforum.org/dr.math/
Date: 08/22/2000 at 05:54:57 From: Karamalizadeh Subject: Thanks for your answer Hi again, Dr. Floor Thank you very much for your answer about trisecting angles without reference to trigonometry. Please explain to me why x = 60 deg - a' y = 60 deg - b' z = 60 deg - c' and identify a', b', and c' in the figure. Best regards, Karamalizadeh
Date: 08/22/2000 at 09:35:01 From: Doctor Floor Subject: Re: Thanks for your answer Hi, again, Karamalizadeh, Thanks for your reaction. In my message, until the line "Why all this reasoning?" I did preliminary work. The figure is just for this first part. After this part a new episode begins, in which I start of course with a triangle. That triangle has angles, say, A', B', and C' (I already used A, B, and C). After that I take a' = A/3, b' = B/3, c' = C/3, and since A+B+C = 180 deg, we have a'+b'+c'= 60 deg. I then TAKE x = 60 deg - a', y = 60 deg - b' and z = 60 deg - c' and use these specific x, y, and z in the more general case of the first part. You should think that all the work of the first part is redone. So x, y, and z are not calculated, they are taken such that we get the results we want (that is, these x, y, and z are chosen such that the A, B, and C from the first part form a triangle similar to A'B'C', and thus proving Morley's theorem). And indeed that works out great. I hope this clears it up. Kind regards, - Doctor Floor, The Math Forum http://mathforum.org/dr.math/
Search the Dr. Math Library:
Ask Dr. MathTM
© 1994- The Math Forum at NCTM. All rights reserved.