Jeph said:
I use option 2.
Have you ever wondered why they don't make d3s? It'd be two 4-sided pyramids stuck together, with one pyramid having triangles with heights equal to 1/4th that of the other side.
Actually no. It's a simple geometrical law.
2 points (minimum) are required to define any line.
3 points (mimimum) are required to define any plane, and they cannot be part of the same line. They will form three lines and this is minimum number of sides to a polygon (2D object).
4 points (minimum) are required to define any 3D space so long as no more than 2 of them are part of any given line and no more than 3 of them are part of the same given plane. 6 lines will be formed between them, this is the minimum number of edges to a polyhedron. 4 planes will confine the space so defined, and this is the minimum number of planes (or sides) for any polyhedron.
If all the angles between the lines are equal, the polygons are said to be perfect or equilateral. There are an infinite number of equilateral polygons. Polyhedrons are a different matter. A "perfect polyhedron" is composed of polygons that are identical in size and shape and the angle of their edges are all equal. So far as is known, only 5 exist and they form the shapes you see in a d4, d6, d8, d12 and a d20. The common diamond shaped d10 is not a perfect polyhedron because not all the angles between the polygon edges are equal.
All this said, it is perfectly possible to make "dice" that can only come to rest on X number of sides, but making a die like this that has an equal chance of coming up on any given side is tricky. A recent trick has been to shape the die like a cylinder and point the tips so that it cannot come to rest on the tips. These cylinder dice can, theoretically, be made to any number of sides, though the greater the number of sides becomes the less distinct each comes (making the die hard to read). Remember that as the number of edges a perfect polygon has, the closer it comes to becoming a circle (which is a polygon with infinite edges). The same principle applies to a sphere for polyhedrons, though if there are any true perfect polyhedrons between a 20 sided and a sphere none have been found (to my limited knowledge).
As a footnote, a Zodoctahedron (sp?), or d100 is simply a golfball sized sphere with 100 indentations cut into the surface so that it, in theory, has an equal chance of coming to rest on any given indentation. My experience with such dice has shown them to have a marked tendancy to come to rest along their molding seams.
Sorry for the long pointless post, but I have a love of geometry (I hate all other forms of math though).