It is impossible to reconstruct the step cycle of an extinct animal beyond a certain level of accuracy. Even if the bony joints are undistorted, joint cartilages are not preserved, and the underlying joint surfaces can infer only approximate range of movement. This is a particular problem in most dinosaurs, in which the articular surfaces appear to be generally less well ossified than in either living crocodiles or squamates. Other data are completely missing. For example, there was almost certainly significant movement of the scapulocoracoid during locomotion (Paul and Christiansen 2000), but since there is no bony articulation with the rib cage, it is impossible to estimate the excursion. Although the presence of ossified tendons, as well as the close association between the pelvic girdle and posterior ribs suggests that, unlike most primitive tetrapods, the vertebral column underwent little or no lateral flexion during locomotion (Ford 1997; Paul and Christiansen 2000), even modest movement, if it did occur, could affect stride length. In any limb with multiple joints, many variants in specific stances are potentially consistent with evidence derived from data such as total joint excursion inferred from osteological features and trackways. In our simulation, we have tried to be conservative in estimating range of movements at joints by keeping well within the physical boundaries of the preserved joint surfaces. Specifically, we would only move a joint until the edges of the opposing surfaces meet, and further excursion could only be accomplished by separation of the joint surfaces. In cases where it was not possible to even estimate range of movement, such as scapulocoracoid excursion on the rib cage and lateral undulation of the trunk, the joints involved were held immobile.

Although complementary adjustments in angles of some of the joints in the limb can produce multiple postures at each stage of the step cycle (limb segment redundancy–see Hutchinson and Gatesy 2006), we would argue that there are still enough constraints imposed by the manus print and glenoid orientation to restrict the possibilities to a relatively small subset of (for our purposes) very similar possibilities. Thus, even within the limitations of the following admittedly restricted data set, we believe that is possible to draw some useful conclusions regarding limb stance and step cycle in Chasmosaurus, and by extension, neoceratopsids in general.

Trackway data are essential if the orientation of the manus is to be estimated. Although numerous trackways attributable to sauropods, ornithopods, and theropods are known (Fastovsky and Smith 2004), neoceratopsid trackways (attributable to Triceratops) have been identified unequivocally from only one locality (Lockley and Hunt 1995). Thus, we are obliged to use trackways belonging to another (albeit closely related) neoceratopsid in the current study. However, postcranial anatomy differs little within ceratopsid neoceratopsians (Forster and Sereno 1997), and it is reasonable to assume that forelimb kinematics was closely comparable in all members of this group.

The specific orientations reconstructed for the various elements of the limb of a particular model at each point of a plausible step cycle will no doubt depend on such factors as the preservational quality and extent of ossification that are particular to the specimen upon which it is based. Thus, the angles presented in this study should be taken as representative of this specimen of Chasmosaurus only. We ran this simulation several times, and obtained very consistent results. We had considered rounding the measurements taken from the final (and most meticulously executed) simulation presented in this study to avoid giving the impression that these were intended to represent a precise description of the step cycle of Chasmosaurus, but decided to leave this to the reader. Rather, these numbers are intended to represent, literally, the results of a simulated step cycle based on a particular specimen of Chasmosaurus under the constraints described above. To the extent that the anatomy of the limb elements of other specimens of Chasmosaurus, and for that matter, other neoceratopsids, resembles that of our specimen, we predict that their step cycle would approximate ours. We will leave it to others to test that prediction.