In Table 1 are the original data (2 sheets) provided by John Howe but measured by a student who apparently made some mistakes. Some data appeared redundant, either because they originated from the right and left side of the same animal or because the same specimen was measured twice. In Table 2 are the data I used. Unfortunately two important measures were not taken: the proximal maximal and the distal supra-articular widths.

Scatter diagrams (Fig.1-4) show some strange specimens and even inside the "bulk" the variation is very large. Most Natural Trap radii are bigger than in extant E. hemionus onager.

Middle-sized Radii
Statistics of the main group are in Table 3; Simpson’s diagrams, bone by bone, in Fig.5.
The proportions resemble those of the extant E. grevyi but the size is smaller. I have not found similar proportions among the fossil American species; the closest is A. leoni but it is more robust (Fig. 6).
As in other limb bones, a few specimens seem larger or smaller, but not otherwise different:

Large Radii
There are only two of them (Fig.7).

Small Radii
Four specimens were separated from the "bulk" because they plotted a little apart on the scatter diagrams (Fig.7).

Caballine Radii
One Radius could belong to a Caballine (Fig.8) but the proximal articular depth is too small.

E. semiplicatus-like Radii
Specimen 42309 is rather like those of Channing (Fig.8). If the diaphysis depth is wrong, specimen 35104 could belong to the same form.

Undetermined Radii
I have no idea of what species they could belong to (Fig.9).