Too many what ifs. In Euclidean geometry it is the locus of a set of points at a ficed distance from a fixed point. That is how I always define it.
I actually prefer Daleks' definition to this one. All of the problems with his definition can be fixed by just being a little more careful. The "fixed distance" definition, however...
So, What do you mean by distance? There are actually many (infinitely many, to be exact) ways to define distance. There is daleks' way, where the distance betwen two points is the square root of the sum of their squares.
Ie the distance between Y,X is (Y^2 + X^2)^1/2
In his equation, he sets this equal to a constant, R, and then squares both sides. Note that in this case, x+y is denoting a single point (x,y), so this equation is giving the distance from this point to the origin. So, he is basically saying a circle is the set of all points R away from the center. This comes basically from the Pythagorean theorem.
But, what if I used a different way to define distance? What if I did something like this:
If d(X,Y) is the distance from the the point (x1,x2) to (y1,y2),
then let d(X,Y)= |y1-x1| + |y2-x2|
Where |z| is the absolute value of z.
Then, this definition has all the properties of distance you would want: its always positive, the distance from Y to X is equal o the distance from X to Y, and the distance from X to Z plus Z to Y is great than or equal to the distance going from X straight to Y.
In this picture, the distance between the two points is equal to the length of the yellow, blue, or red lines.
BUT (and some careful thought will prove this to you) a circle using this definition is actually a square!
This distance is informally called the taxicab distance, since the distance is measured as you travel between the two point in a zig zag, like taxi would in a large city.
Here is what a "circle" looks like:
It has a radius of two.