In "common parlance" we throw around chemistry, biology, physics, and all, sort of throwing off the diversity within these disciplines. Gosh, in my comps I answered (or attempted to answer) a question about how useful it was to talk about "scientists" and non-scientists. Going the other way, I'll frequently discuss "research areas" or "invisible colleges" (Price [a] and of course Crane[b]*) or even some of the other groupings of scientists:
- social circle [c]
- paradigm (Kuhn) [d]
- epistemic culture (Knorr Cetina)[e]
- thought collective (Fleck)[f]
- core set (Collins - includes people who disagree on a scientific controversy - people on both sides) [g]
These sub-disciplinary groupings gather scientists who study the same phenomena using the same research methods. They publish in the same places, they attend the same meetings, and they cite each other. There are lots of ways to find these and it's usually pretty easy to make the argument at this level, but what about the discipline level?
When we talk about disciplines, is there enough that chemists have in common to make talking about them as a discipline useful? In LIS, when we're training science librarians, we sure do talk about the disciplines, but is this simplification or 30,000 feet view ok? Sometimes we go even more abstract talking about physical sciences or life sciences (where does chemistry go?)
Researchers using the domain-analytic view say that there are practices and epistemic cultures that define a discipline. The primary researchers in this in the past few years** are Fry and Talja (together and apart [1-3]). They have studied how the various ways scientists communicate online and accept and use e/digital/online scholarly materials are shaped by the features of their scientific discipline. The two main dimensions of discipline that they get from Whitley (2000, cited in all of their stuff) are mutual dependence and task uncertainty. These sort of map to resource concentration and agreement on what's good that I mentioned when talking about Birnholtz' work, but he's most definitely talking sub-discipline - after all, HEP is NOT like anything else in or outside of physics. Turns out, though, that the actual units of analysis in the case studies by Fry and Talja are sub-disciplinary fields, too. So what about disciplines?
As Meyer & Schroeder point out, various researchers have found that despite widespread concerns, collaborating across disciplines is actually easier than distributed work or even collaborating across institutional boundaries . Disciplines in these cases, are often organizational schemes of universities: departments and schools. If you're in the chemistry department or in the physics department changes whether it's physical chemistry or chemical physics. There's also the idea that where you are trained is not necessarily where you work, and that researchers will and do use work from other disciplines - if they can find it. (there are issues in finding information outside of your research area because the language is different and you'd have to look different places). In research areas with a lot of scatter, researchers have to look lots of different places to find what they need 
So I've been wandering around in this post***. I guess the point is that for most purposes, it is useful to replace "discipline" with the largest cohesive sub-group (research area, etc), that works for the purposes of the discussion. The time when it makes sense to keep using "discipline" is when you're talking about universities and their organizational structure and providing services based on money from, coordination with, tailored to elements of university organizational structure.
 Fry, J. (2006). Scholarly research and information practices: a domain analytic approach. Information Processing & Management, 42(1), 299-316. doi: 10.1016/j.ipm.2004.09.004
 Fry, J., & Talja, S. (2007). The intellectual and social organization of academic fields and the shaping of digital resources. Journal of Information Science, 33(2), 115-133. doi:10.1177/0165551506068153
 Talja, S. (2002). Information sharing in academic communities: Types and levels of collaboration in information seeking and use. New Review of Information Behaviour Research, 3, 143-160. Retrieved from http://www.uta.fi/~lisaka/Taljaisic2002_konv.pdf
 Meyer, E. T., & Schroeder, R. (2009). The world wide web of research and access to knowledge. Knowledge Management Research and Practice, 7(3), 218-233. doi:10.1057/kmrp.2009.13
 Vakkari, P., & Talja, S. (2005). The influence of the scatter of literature on the use of electronic resources across disciplines: A case study of FinELib. Research and Advanced Technology for Digital Libraries, 3652, 207-217.
* do you all want full citations for all of these? I could provide them, but since I'm just mentioning them in passing... lemme know
** for maybe a century or so there has been talk of "tribes" and such... but I'll just stick to the recent bits
*** what's new.
Update - I was asked for the other references so here they are
[a] Price,D.J.D. (1986) Little Science, Big Science...and Beyond. New York:
Columbia University Press.
[b] Crane,D. (1972) Invisible Colleges: Diffusion of Knowledge in Scientific
Communities. Chicago: University of Chicago Press.
[c] Polanyi, M. (2000) The Republic of Science: Its Political and Economic
Theory. Minerva, 38, 1-21.
[d] Kuhn,T. S. (1996) The Structure of Scientific Revolutions. 3rd ed.
Chicago, IL: University of Chicago Press.
[e] Knorr-Cetina,K. (1999) Epistemic Cultures: How the Sciences make
Knowledge. Cambridge, MA: Harvard University Press.
[f] Fleck,L. (1979) Genesis and Development of a Scientific Fact. Chicago: University of Chicago Press. Originally published in 1935.
[g] Collins,H. M. (1985) The scientist in the network. In Changing Order:
Replication and Induction in Scientific Practice (pp. 129-157). Beverly Hills, CA:
Domain, kingdom, phylum, class, order, family.... division levels resulting from prior speciation branch points; admittedly, there's more obvious ongoing lateral transfer in the realm of knowledge than in biological genetics, but the similarity strikes me nonetheless.
Yes, it's an old problem. Western philosophers were arguing about it in the early nineteenth century and after. Those arguments can be instructive, and they can be limiting.
Pasteur drew much opposition because he was not a doctor, but a chemist, and spontaneous generation was out of his discipline...he did not know how to think, they said. One of Andre Gide's close relatives opposed Pasteur.
Disciplines tend to have their preferred ways of thinking. One, unnamed here, starts by naming everything in sight, as if Adam. Then what? On the ground these preferred ways are the real differences between the disciplines. And yes, "for most purposes," you are most likely to net a like group of preferred ways of thinking by using sub-groups.