Hippias and his quadratrix
Hippias of Elis (430 BC) was a sophist who invented the quadratrix curve to trisect an angle. The problem of trisecting a given angle was one of the problems that generated a lot of mathematics during this period, and several mathematicians devised methods for solving this problem. Like many other sophists, Hippias was an itinerant teacher who made his living wowing the locals with his knowledge.
Apparently, he did alright, but didn't leave much of a legacy except for the quadratrix.
Definition of the curve
The curve can be described in a few sentences. Let ABCD denote a square. Over a unit time period, allow the top segment of the square to fall at a uniform speed to the bottom of the square. During the same time, allow the left side of the square to rotate clockwise at a uniform speed to the bottom of the square. At each time, the two segments will intersect in a point P. The totality of all these points P is defined as the quadratrix.
Drawing the quadratrix
One can imagine how Hippias might have sketched the quadratrix in the sand, but one can hardly image how he would have made an accurate sketch of it.
First make a square. Color it tan and label the vertices A, B, C, and D.
>
sqr :=plots[display](
[plots[textplot]({[0,-.05,`A`],[1,-.05,`B`],[1,1.05,`C`],[0,1.05,`D`]}),
plots[polygonplot]([[0,0],[1,0],[1,1],[0,1]],color=tan)],scaling=constrained,axes=none):
sqr;
As time t goes from 0 to 1, we want the top of the square to fall to the bottom at a constant speed. That means that the y coordinate will be 1-t at time t.
An animation of the top falling to the bottom
a fleeting image of the quadratrix
Great. We can see the point of intersection P of the top and right sides as they fall to the bottom, but it is a fleeting image. We need parametric equations for the curve so we can trace it out.
What do we have? At time t, the point P = [x,y] is on the segment , so . But also it is on the segment , say
,
for some w. Equating and solving for w, we get that . Hence x = = . These are our parametric equations.
the quadratrix being traced out
>
Using the quadratrix to trisect an angle
How did Hippias use the curve to trisect an angle? Very simply. He placed the square containing the curve over the angle with the vertex at the lower left hand corner and the initial ray on the bottom of the square. He then marked the point P on the curve where the terminal ray crossed it and found the point Q at the same height on the left side of the square. Then he constructed the point R on the segment AQ such that AR/AQ = 1/3 and found the point S on the curve whose height is the same as R. By the definition of the curve, the measure of angle SAB is 1/3 the measure of the original angle.
>
Problems with the quadratrix
1. Find the x-coordinate of the lower end of the quadratrix.
> p(1);
Error, (in p) division by zero
> p(t);
> limit(p(t)[1],t=1)=evalf(limit(p(t)[1],t=1));
This can also be obtained easily with L'Hospitals rule.
2. Find the length of the quadratrix.
> F := int(sqrt(diff(p(t)[1],t)^2+diff(p(t)[2],t)^2),t=0..1);
> evalf(F);
3. Find the area under the quadratrix.
> Int(p(1-t)[1],t=0..1)=int(p(1-t)[1],t=0..1);
> rhs(%) = evalc(rhs(%));
> (rhs(%))=evalf(rhs(%));
>
Using the quadratrix to square the circle.
The quadratrix can be used to square the circle, in fact the name, quadratrix, means 'square making'. Dinostratus (circa 350 BC) was the first to use it for this purpose, according to Pappus (circa 300 AD).
>
What Pappus did was prove that the length of the blue arc was the reciprocal of the length of the magenta segment AG, when the segment AB is taken as the unit. We have already done this using trig and L'hospital's rule above, because we showed that AG = . Since the blue arc has length , that proves it.
Once you have a segment of length , you can construct a segment of reciprocal length . From there it is easy to 'rectify the circle', that is construct a rectangle whose area is . Once you have done that you can 'square the rectangle'. Voila!