At the East Dorset MathsJam Christmas party, @jussumchick (Jo Sibley in real life) posed the following question:

There are two ways to draw a 16-gon with rotational symmetry of order 8 inside a unit circle, as shown. What’s the ratio of their areas?

Screen Shot 2015-01-21 at 10.15.13 Screen Shot 2015-01-21 at 10.12.51 Typically, I look at this sort of question and sigh: it’s going to need a (shudder) calculator to work out the lengths of the various sides, and the areas will be really ugly numbers… but I was assured the answer was ‘something nice’. So, I put down my mince pie and picked up a pen. (For the sake of narrative, I’ve covered up some of my mistakes.)

The splitting up of the shape

My first step was to notice that I could split each of the shapes into 16 congruent triangles – from the centre of the circle (O) to a point on the circle (A), to one of the ‘elbows’ (B) between the circle points. In the 16-gon made of two squares, the angle OAB is 14π, as it’s half of a right-angle. The angle BOA is 18π, because it’s a sixteenth of a circle, and the final angle ABO is 58π. In the other 16-gon, angle BOA is 18π, like before; ABO is easy to find, too – if you continue OB through B, the other side of the angle is 14π – so ABO has to be 34π. OAB turns out to be 18π as well. In each case, the side opposite the largest angle is 1 – the radius of the circle.

An alternative area formula

Every Core 2 student knows what to do from here: work out another side using the sine rule, then use Area=12absin(C) to find the areas, multiply them by 16 and get the ratio. Full marks, next question. And a gaping void in your soul, of course, but you don’t lose marks for that. What a mathematician does is, find a formula for the area of the triangle that works directly from the information we have: a side and three angles. It’s not too tricky: the sine rule says b=sin(B)sin(A)a, so 12absin(C)=12a2sin(B)sin(A)sin(C) or, if we’re tidy-minded, a2sin(B)sin(C)2sin(A), which has a nice symmetry to it. (Incidentally: I had this formula wrong to begin with; my sin(A) and sin(C) were transposed. It smelt funny: there’s no real difference between angles B and C, so it didn’t make sense for them to be divided rather than multiplied).

Now you work out the areas, right?

Nope. Don’t care about the areas. I care about their ratios. If I set up the triangles so that my a is 1 in both cases, I want the ratio: sin(18π)sin(14π)2sin(58π):sin(18π)sin(18π)2sin(34π) How lovely! We can simplify it immediately by dividing both sides by 12sin(18π) to get: sin(14π)sin(58π):sin(18π)sin(34π) Multiply everything through by sin(58π) and sin(34π) to get: sin(14π)sin(34π):sin(18π)sin(58π) – and the left-hand-side works out to be 12. The right-hand-side… well, I’d understand reaching for a calculator… but you really don’t need to! It turns out that cos(AB)cos(A+B)2sin(A)sin(B), which means you can rewrite the right hand side as 12(cos(14π)cos(12π))=12(120), so our ration is: 12:122, or, more neatly, 2:1. What do you know? It was a nice number after all!