For many years one of the arguments against birds being dinosaurs was that birds have a furcula and dinosaurs do not. Now clearly the avian furcula (or wishbone) is a highly specialised bone and it is formed by the fusion of the clavicles (the collar bones) but the absence of a furcual in theropods hardly rules out theropod ancestry. After all, basal theropods don’t have feathers, or beaks, or could fly. One rather key concept of evolution is that things change over time, so an absence of a furcula hardly rules them out of contention – nothing else had a furcula either so unless birds somehow sprang out of the ground fully formed, then the furcula had to have a genesis in one lineage or another.
However, despite this obvious flaw in the argument, there is a rather better argument – the presence of a furcula in numerous theropods. A recent survey of the literature and specimens shows that they are now known in coelophysoids, spinosauroids, allosauroids, tyrannosauroids, compsognathids, oviraptorosaurs, torrdontids and dromaeosaurs. Outside of the theropods, there’s clavicles known in a prosauropod and several in some ornithischians. Dinosaurs have clavicles and furculae.
And just to show off what they look like, here is the furcula of Sinornithosaurus. It’s the boomerang-shaped element in the centre of the picture and is quite typical in morphology, though many of them also have a short spar coming out of the middle.
Dave Hone's Archosaur Musings 
Huh! I wasn’t aware that ornithischians had furculae. For whatever reason, I was under the impression that the group had simply lost their clavicles.
“birds have a furcula and birds do not”
I assume that second “birds” should be “dinosaurs,” right?
Yes, thanks for the spot. Fixed it.
Here’s how I thought the argument against a dinosaurian ancestry for birds went. No dinosaurs at one point were known to have possessed a furcula, while some “thecodonts” or “pseudosuchians” allegedly did. It was thought that this bone could not have re-evolved, and therefore it was thought that birds could not be descended from dinosaurs. That’s my understanding, please correct me if I’m wrong.
Ah yes true. Though of course that still presumes a similar problem that some things can’t evolve twice – which is a pretty basic issue given how rampant convergence is and how well documented it is.
The idea was that theropods had lost their clavicles (Oviraptor’s furcula was misidentified as an interclavicle), so could not evolve into birds which had clavicles. More basal taxa like Euparkeria still had clavicles though, so could be ancestors of birds. The fusion of lack of it was not an issue, since as you say, birds’ ancestors had to have unfused clavicles at some point.
Also, no sauropodomorphs or ornithischians are known to have furculae, though Massospondylus’ clavicles are articulated in the same configuration.
Well Nesbitt et al. in their furculae review paper of last year or so talk about them (saurops. and orniths.) and provide citations. That was my source for them, and I’ve not seen anything to contradict that.
Nesbitt et al. talk about clavicles in ornithischians, not furculae. Furculae are fused clavicles (as you said yourself above), but unfused clavicles are not furculae and to refer to them as such as just sloppy terminology that Nesbitt et al. do not commit. For proof of that, see pages 856, 862 and 872– the authors never refer to the clavicles as ‘furculae’ when talking about ornithischians.
You’re right I misread their sentence sorry. I’ll edit the text.
No ornithischians have furculae, only unfused clavicles (see Bryant and Russell 1993), nor have they been reported as having wishbones/furculae, but only as clavicles (I verified this by checking Osborn 1924, Brown and Schlaikjer 1940, and Sternberg 1951). The recorded presences of clavicles in ornithischians are strictly limited to some basal ceratopsians (Psittacosaurus, Leptoceratops and Protoceratops– as one might guess based on the other refs I’ve cited above). The clavicles in these ornithischians aren’t articulated in a bracing manner as they do in Massospondylus and theropods (Yates and Vasconcellos 2005).
Nick is correct – the clavicles in Massospondylus are “furcula-like”, but still unfused – and that is what Nesbitt et al say.