The subStrings command walks back when reading a .txt or .html. file that has some UTF8 characters apparently. The File List tool does not show any strange characters in the files which subStrings errors on. Thus I hope the dichotomy of behaviors may be resolved by a preference setting.

Or is it just a permanent glitch I must write my own subStrings command to get around?

Dan Ingalls called Squeak a sandbox. I think the sandbox needs cleaning.