Might as well throw my hat in here. Forelimb posture is an issue I've been struggling with on my dissertation research, and I presented a hypothesis about the evolution and functional morphology of sauropod forelimbs in a Romer Prize talk in 1999.
Phil Platt, for the record, is an excellent sculptor who has sculpted an unbelievably accurate scale model of the Apatosaurus louisae CM 3018 from the Carnegie Museum of Natural History. I have had many discussions with him, and although I do not quite agree with his interpretation of the evidence, there is a good reason he semi-sprawls the front limbs:
THE RADIUS AND ULNA OF SAUROPOD DINOSAURS DO NOT CROSS EACH OTHER.
To explain my hypothesis and Phil's would be an anatomical treatise onto itself. Let me just give you the highlights.
We know from footprint evidence that the manus (forefoot) of sauropods was pronated during locomotion -- that is to say, the manus was turned forward such that the "palm" surface of the manus faced posteriorly. We pronate our hands (manus) by rotating our radius bone about the proximal (top) end of our ulna bone. This flips our hands (manus) palm-side down. Elephants pronate their manus by permanently crossing their radius bone over their ulna bone.
Sauropods could not have actively rotated their radius about their ulna because it is permanently locked in place at the proximal end of their ulna. Furthermore, their outgroups, like Prosauropods and Theropods, could not actively rotate their radius about their ulna either. In fact, the manus of the theropods and prosauropods are sort of held in a prayer-like position with the palms of the hands held inward. They cannot completely pronate their manus without abducting their humeri (moving their upper arms away from their bodies).
Thus, if we articulate the forelimbs of sauropods together the way they go together in theropods and prosauropods, we have a problem: the radius does not cross over to pronate the manus. The manus is turned sideways -- see Wilson and Sereno (1998) for sauropod reconstructions with this sort of sideways manus. The problem is, we know from footprint evidence that sauropods pronated their manus.
Two ways this could be accomplished. One is Phil Platt's suggestion that we semi-sprawl the limbs -- when this happens, the manus is more or less pronated, and I think Wilson and Sereno (1998) would probably agree with this idea more or less. At the very least, they show sauropods reconstructed with bent elbows -- check it out for yourselves.
My problem with semi-sprawled limbs is: 1) they set up a lot of stress in the limbs of such heavy animals (as per Stratigraphy's suggestion); and 2) the shoulder joint is too angular to really allow this sort of bending to happen.
My hypothesis was that the radius, already ANTERIOR and LATERAL in basal dinosaurs -- have a look for yourselves at any skeleton to confirm this -- shifted to a more ANTERIOR and MEDIAL position as sauropods reverted to a quadrupedal posture, rotating the manus into a pronated position. This is a real tough one to defend without illustrations, so I'll leave it at that until I can get this published -- which I will hopefully do soon, I promise! I was convinced of it by following the morphology of the proximal ulna and radius through theropods and prosauropods, and changes in the manus. If you saw my talk at SVP, you saw some of this evidence, but again, I will not be surprised if many of you heartily disagree. Good for the science either way.
The elbows of sauropods are a pain in the butt to figure out. Phil Platt has done an excellent job by bringing this to everyone's attention, and it is something that needs to be sorted out -- sorry systematists everywhere, you'll have to wait for the morphologists to figure these characters out. =)
In summary, you can't pronate the manus in a sauropod by crossing the radius over the ulna. You either have to bend the forelimbs, or shift the position of the radius with the limb remaining straight. See why sauropods are so cool? =)
----Original Message Follows----