MATERIALS AND METHODS

A dry Sphenodon skull (specimen LDUCZ x036) was subjected to micro-computed tomography (micro-CT) at the University of Hull, UK. Then, using image segmentation and analysis software (AMIRA 4.1, Mercury Computer Systems Inc., USA), the micro-CT dataset was converted into 3D models of the skull and lower jaws. An additional micro-CT dataset (University of Texas, Austin, USA) was used in the construction of the neck (specimen YPM 9194). This additional dataset required some manipulation as the neck was twisted and only part of the 6th vertebra was scanned. With reference to other dry Sphenodon material (LDUCZ x036) the proatlas and the first two vertebrae (atlas and axis) were subsequently realigned during the 3D model construction, and the right-hand portion of the most posterior vertebra and girdle were duplicated and mirrored to form a complete and symmetrical structure. The soft tissue fascia sheets that cover the lower temporal fenestra (onto which part of the adductor musculature attaches) were created manually in AMIRA, and the 7^th and 8^th vertebrae of the neck were simply represented by cylinders and cuboids positioned posterior to the 6^th vertebra. These 3D models were imported into ADAMS multi-body dynamics software (MSC Software Corp, USA), in which a representation of the muscle groups were added.

Each muscle group is represented by several straight cylinders extending between the skull and lower jaw, or skull and neck. The exact origin and attachment point of each cylinder was positioned according to descriptions in Jones et al. (2009), which was, in turn, based on descriptions in the works of previous authors (e.g., Nishi 1916; Byerly 1925; Anderson 1936; Poglayen-Neuwall 1953; Haas 1973; Gorniak et al. 1982; Wu 2003; Al-Hassawi 2004, 2007; Tsuihiji 2005, 2007; Holliday and Witmer 2007) in combination with first-hand observations. When a muscle wraps around other muscles or bone, it is represented by two or three cylinders joined end-to-end. The colour coding and abbreviations used in this paper follow as closely as possible those of Jones et al. (2009).