Hi Hugh,
The SA 100 should be more linear than the SA200 because of the lower dispersion angle but I think the wavelength errors you are seeing may be more to do with the zero order not quite being in the expected place (ie at 0 Angstroms) ie it is more of an offset error than a non linearity. I have seen this to a greater or lesser extent from time to time and never really managed to nail down the cause. (Something to do with atmospheric dispersion perhaps ? What does a linear fit between the shortest and longest wavelength Balmer lines look like for example ? At the end of the day though wavelength calibration of a slitless system and the transfer of that calibration to other targets is always going to be somewhat approximate.
Cheers
Robin