The shape of the crystals is of petrological interest and is also required for stereological correction of 2D observations. Overall shape is usually specified as the Short (S), Intermediate (I) and Long (L) dimensions of a bounding parallelopiped. These parameters may be estimated from separated crystals, by careful examination of crystals in thin sections, or from the distribution of width/length ratios of intersections.

For a populations of uniform rectagular blocks or ellipsoids the mode (not the mean) of the intersection width/intersection length is equal to the ratio of S/I (Higgins, 2000). The ratio I/L is more difficult to determine from 2D observations and cannot be measured with the same precision.

Higgins (2000) suggested using the 'skewness' of the intersection width/intersection length distributions. Morgan and Jerram (2006) developed an Excel spreadsheet, CSDSlice, that fits measured intersection aspect ratios distibutions to a library of ideal distibutions. The method does not estimate the error in determinations of S, I and L. There are also errors in some of the library of shapes.

The most recent approach, an Excel sheet called ShapeCalc, also uses a larger library of shapes and estimates the error in both S/I and I/L (Mangler et al., 2022) . ShapeCalc is available here. You can tranfer data from CSDCorrections to ShapeCalc easily. In CSDCorrections click right on the data table (Raw data DS1) and select 'Copy Length, Width'. Paste into ShapeCalc, page 'Shapecalc', columns labelled 'Input'.

In natural materials, the fit between the 3D models and the actual data is not always good. This can be caused by non-regular, non-ellipsoid shapes and/or by a variation in shape between small and large crystals. The latter effect can be seen partially by selecting different size ranges or by inspecting the graph of intersection width/intersection length versus intersection length.