A set of 1982 speckle interferometry observations led to a simple preliminary model of Herculina as a three-axis object, perhaps 260 by 220 by 215 km.

1985 analysis of this data concluded there was a nonspherical shape with one bright spot, whilst a 1987 photometric astrometry study concluded the object was spherical with two dark spots (and rotated around a completely different pole), which was in turn negated by a 1988 thermal study which showed the object could not be spherical.

By the late 1980s, the generally accepted model was a three-axis object with major albedo or topographical features.

[7] Recent (2002) modelling of photometric data indicates that Herculina is not spherical, but a blocky shape not unlike a battered cuboid - or, as the analysis described it, it "resembles a toaster".

This analysis indicates the presence of multiple largish craters, similar to 253 Mathilde, but no major variation in albedo.