Configuration Vertices Edges Faces Cells Hypercells Result Remarks
{4, 3} 8 12 {}@1 6 {4}   {4, 3}
2 * 6 {}@2 2 * 4 {3}   2 * {3, 3} total split
{4, 3, 3} 16 32 {}@1 24 {4} 8 {4, 3}   {4, 3, 3}
2 * 24 {}@2 2 * 32 {3} 2 * 8 {3, 3}   2 * {3, 3, 4} total split
2 * 8
2 * 4 {3, 4}   disconnected
2 * 6 {4}   disconnected
32 {}@3   disconnected
{4, 3, 3, 3} 32 80 {}@1 80 {4} 40 {4, 3} 10 {4, 3, 3} {4, 3, 3, 3}
2 * 80 {}@2 2 * 160 {3} 2 * 16 * 5
2 * 10 * 8
{3, 3} 2 * 16 {3, 3, 3} disconnected total split
2 * 10 {3, 3, 4} part connected
2 * 40
2 * 10 * 4 {3, 4}
2 * 10 * 6 {4}   disconnected
160 {}@3   disconnected
2 * 40 {}@4   disconnected total split

The polytopes are getting a bit more interesting. The measure polytope is called the cube in 3D and the tesseract in 4D. This polytope exists also in all dimensions. The number of sections is equal to the dimension (but the last one is not in the database). Even numbered sections totally split the measure polytope in two subsets of vertices and edges, there is no connection from one subset of vertices to the other subsets. Moreover, in some sections the symmetrygroup is not transitive on all elements. In the table above this is the case for the {3, 3}'s in sections 2 of {4, 3, 3} and of {4, 3, 3, 3}.

In {4, 3} and {4, 3, 3} we can find a regular compound in section 2, but you should go to much higher dimensions to find again such regular compounds. There is one in 7D.

Coordinates can easily be generalised to other dimensions, each coordinate tuple entry is either 1 or -1.

To see what regular polygons can be found we have to consider a bit more. It is quite trivial to find that the cosine of the angle between two edges in a section is rational and greater than or equal to 0. It is well known that in order for that to give an angle of n.π for some rational n, the cosine must be either 0 or 0.5, and so the angles in those cases can only be π/2 or π/3, giving only regular squares or triangles. Within a particular section n an edge is formed betweed two vertices that are different in precisely n entries of their coordinates.

Suppose we are to form a regular triangle ABC in section n. Suppose further that A has all entries equal to 1, so both B and C must have n entries equal to -1, and they also must differ in n entries. If n is even, this can be done if the dimension is 3.n/2 - 1: set the first n entries of B to -1 and the first n/2 odd numbered entries of C plus the following n/2 entries. It can not be done when n is odd. So triangles can only be found in even sections. Section 4 in 5D does not qualify (3.4/2 > 5), but it will qualify in 6D, but section 2 already qualifies in 3D.

Next consider squares (ABCD). The edges AB, BC, CD and DA should all have length sqrt(n), so their coordinates should differ in n entries. Also the diagonals AC and BD should have length sqrt(2.n) and so should differ in 2.n entries. This means that the dimension should be at least twice the section number. And so (for instance) squares can be found in section 3 in 6D.

I think similar consideration exist for higher dimensional regular figures. For instance, there are only cubes in a particular section if the dimension is at least three times the section number (again, 6D is the first time that cubes do occur in section 2). I have not yet investigated other higher dimensional regular figures, but I think it may be worthwile to consider.

Especially interesting are those parts where the symmetry group is not transitive on some elements. In the table above there are two such cases, as mentioned. The first one, in section 2 of {4, 3, 3} is relatively easy. The {3, 3}'s split in two subgroups, but each higher dimensional {3, 3, 4} contains exactly 4 {3, 3}'s from each group.

More interesting is section 2 of {4, 3, 3, 3}. The {3, 3}'s split in two groups, one group (A) has 80 entries and the other (B) has 160. The 80 {3, 3}'s from group A are the cell of two {3, 3, 4}'s. The 160 {3, 3}'s from group B are the cell of one {3, 3, 3} and one {3, 3, 4}'s. In fact, no {3, 3, 3}'s share a cell. Each {3, 3, 3} shares a cell with five {3, 3, 4}'s, each {3, 3, 4} shares a cell with eight {3, 3, 3}'s and with eight {3, 3, 4}'s. There is even a quite regular pattern to be found here. In the following the H numbers indicate the hypercells as they are encoded in the database. Also there is (of course) a split between the two different groups due to the total split.

Encode the {3, 3, 3} as follows:

Groupcode
12
H01H0600001
H02H0700010
H03H0800100
H04H0901000
H05H1010000
H14H1100111
H15H1201011
H16H1310011
H19H1701101
H20H1810101
H21H2301110
H22H2410110
H26H2511001
H27H2811010
H29H3011100
H32H3111111

Next the {3, 3, 4} share a face with a {3, 3, 3} with the following pattern in the code ('x' means don't care):

Grouppattern
12
H01H060xxxx
H02H07x0xxx
H03H08xx0xx
H04H09xxx0x
H05H10xxxx0
H13H11xxxx1
H14H12xxx1x
H16H15xx1xx
H18H17x1xxx
H20H191xxxx

Also the first {3, 3, 4} in the table above shares a face with all others (from the same group), except the last. The second shares a face with all others except the last but one, etc.

Finally, all these polytopes can be used to tile the space, and they are the only ones that provide a tiling in 3D or 5D and higher. In 4D you can also tile with either {3, 3, 4} or {3, 4, 3}. The symbols for the tilings are: {4, 3, ..., 3, 4}.