Mulling the issue over some more, I’d like to take a mulligan on my previous answer to the context question. I feel I punted on the question by offering a critique of some prevalent uses of context, without ever answering the question in a compelling way. One way is to try and add some dimensions to the concept. I guess I think of it along two axes: generic-necessary and tradition-response. I refer to them as axes, because they are probably more descriptive qualities than firm categories.
So, a necessary context is some context necessary to arrive at a proper understanding of something; it provides a motivation. Situating something within a new necessary context can totally change why we see something as having taken place.
A generic context, on the other hand, enriches our understanding of a topic, but will rarely demand a wholesale change in the way we think about something. It describes a gross sort of contingency (i.e., 19th century natural history wouldn’t have existed in anything like the form it did without the imperialist project), or it provides an understanding of why something looks the way it does. Say there’s an operative metaphor or imagery in use that resonates with some other metaphoric tradition that has nothing to do with the history of the subject at hand. Oftentimes, a generic context is something we can safely take for granted, but it doesn’t have to be menial. For example, say we’ve never paid attention to a certain tradition of theory-making, a study of that tradition will tell us both about the theory in question, as well as about the context. However, in subsequent history (provided the original study has achieved canonical status), we can refer to this tradition off-hand, or even ignore it entirely.
A context of tradition speaks to us of a learned behavior. A scientist uses this sort of diagram, which goes back a century, or the tradition of spectroscopic analysis, or the tradition of anthropological characterization in terms of evolutionary principles.
A context of response speaks to us of a more reasoned response to some stimulus. Placing a theory in the context of a certain experiment tells us that to understand the motivation behind the theory, we have to be aware of such and such an experiment. To understand why science funding increased after 1957, we have to be aware of the launch of Sputnik.
So, when is contextual analysis worthwhile? A context of tradition, it seems to me, can move from being a necessary context, if it hasn’t been previously considered, to being a generic context, once the tradition becomes well-understood. Whereas, a context of response is always necessary (or is it?). So, it’s always worthwhile if we can learn about a new context. We can do this through a case study, or, better still, by making the context the subject of investigation itself. One of my favorite history of science books is Andy Warwick’s Masters of Theory. It could place, say, the work of the Maxwellians in the context of the Cambridge Mathematical Tripos, but it’s a much better book than that, because it turns the scenario around and make the Tripos (the ostensible context) the subject instead. Galison’s Einstein’s Clocks is a great book, but it’s maybe a bit awkward, because it’s framed around placing the special theory of relativity in the context of the technical challenges of the late-19th century, but the book ends up being about this context. The effect is to make much of the book seem extraneous to the central point of contextualizing relativity, until you realize that Galison has simply shifted the focus to the context.
Conversely, it’s rarely interesting to contextualize something for the sake of contextualization. To contextualize something significant (like the theory of relativity) is interesting only if the context is necessary. If it’s generic it’s less interesting. So, you could say, Einstein talked a lot about clocks in his 1905 relativity paper, and then say that clocks were everywhere in this period. But Galison goes beyond this, and shows that the problem of simultaneity was a deeply conceptual problem in this period–what could have been generic becomes necessary. However, if the subject is insignificant and is placed within a well-understood context, it’s not interesting. So, say someone placed some other uninfluential paper on time coordination within the context Galison illustrated, it would just come off as a cheap knockoff. That said, there might still be room for a definitive history of time coordination in the late 19th century not as context, but as subject, if there are actors and traditions that need to be made explicit, but not if it’s just a recapitulation of what Galison said.