Let me define the penumbra of a paper as the data, hypotheses, speculations, argumentation, theory, etc. that do not make it into the final version of the paper, even though they are in some sense relevant and played a role in your thinking when writing.
When starting to write a paper, you hardly knows what it will contain at the end. Of course, there is a main idea, a central argument, some core data, but the way that the paper actually turns out in the end is largely undetermined by what you start with at the beginning of the process. As you write the paper, you write a paragraph or page, and then delete most of it, you group some scattered paragraphs into a section, you cobble things together, find a new argument, fill in the gaps and add paradigms. You may realize that somebody has already worked quite a bit on one of your generalizations or hypotheses, so then you rewrite to incorporate that work.
You may also get feedback, causing to you cut entire sections that seemed like a core part of the paper at the beginning, or to add entire sections that did not seem very relevant at first. Ultimately, you produce an unchangeable final, published paper.
But in the end, a lot gets left out. This is the penumbra of the paper. The more you have thought about your paper, and the more time you put into it, the larger and more dense the penumbra grows.
The penumbra can contain lots of references and connections to the content of those references. Some of this stuff comes up at talks, where audience members ask questions trying to go deeper into the topic. Later the penumbra can split off into other papers, or projects, or to prevent you from pursuing a reckless path, so nothing was a waste of time.
