This package has evolved since 1998 from my tinkerings in python. There are huge improvements that could be made (search for TODO in the source files; you may even find some FIXME comments, too) but it's hobby work, so don't expect it to be made "finished" any time soon. Patches would be welcome if anyone finds this fun enough to play with to make it worth patching.
Most documentation is in doc-strings, where it belongs; you should get useful information by printing the .__doc__ of whatever object you're playing with, be it Planck's constant, the Polynomial class or the top-level study package object itself - a good place to start, if you want to learn more about this package. There's also a partial description of the package on my web-site.
PEP 8 contains ample sound advice. I even agree with most of it; and the rest is eminently sensible, too. Some of my deviations from it are deliberate -outlined in Style.rst - but this package contains plenty that are mere accidents of history. Some simply predate the advice, or the language feature that makes the advice sound (or my hearing of either). Most of these are slated for fixing - see the "PEP 8 conformance" item in study.TODO - but shall take a lot of work to fix, simply by virtue of there being quite a lot of code and, especially, doc-strings to revise.
So, if you're thinking of submitting a patch to cure such ugliness, don't be shy - but it's probably prudent to ask me about it before you expend large amounts of effort, only to discover I preferred it the old way. In any case, please make such clean-up changes in commits separate from any more meaningful code changes.
The first toy (April 1998) was an early version of study.maths.ratio's Rational and approximate(). The next (Jan 1999) was a first stab at what has since become the Quantity class, initially just supporting a numeric value and its units; it soon grew the ability to track error bars on its numeric values. Right at the outset, Quantity included the SI units, some derived units, and a few silly archaic units (that now live in study.value.archaea). The next innovation was the first incarnation of Lazy (now in study.snake.lazy and deprecated in favour of study.cache.property). This sets the pattern for the three main areas of development: mathematical toys, scientific quantities and pythonic support infrastructure.
All of this happened in RCS; later, I migrated it to CVS. The way I'd been using RCS made this easy, as it happened. I had all my RCS files in a directory tree mirroring the source tree, whose top-level directory had a symlink named RCS pointing at the top of the RCS tree; each sub-directory then had an RCS -> ../RCS/name symlink using its own name. The resulting RCS tree was thus a ready-made CVS module, that I merely had to move into my CVSROOT. There's consequently no obvious sign of when I did that, but I can guess it was at the same time (2001/Autumn) as I converted my web-site to CVS.
In the RCS/CVS era of history, I could re-arrange the directory tree at whim, without making any changes visible in the commit history, aside from the consequent changes to how modules import one another. I think the last major re-structuring was 614f6ac in 2007/March, although a few relatively minor moves may still be hidden from the version-control history between then and the summer of 2010, when I imported the whole lot from CVS into git. So checking out historical revisions before summer 2010 may, and before spring 2007 certainly shall, give you a source tree that probably doesn't even import cleanly.