Manuscript to be reviewed I.2.2 Cancellous bone in brief One aspect of osteology that has remained understudied by dinosaur palaeontologists is the threedimensional (3-D) architecture of cancellous ('spongy') bone, the other main type of bone tissue found in limb bones. Cancellous bone is found throughout the vertebrate skeleton, including in the ends of long bones, vertebrae, throughout short bones (e.g., those of the wrist and ankle) and between the opposing cortices of many flat bones, such as those of the skull (Carter & Beaupré 2001; Currey 2002; Martin et al. 1998). This work will only consider cancellous bone in the endochondral bones of the appendicular skeleton. Furthermore, it will not consider medullary bone, the loosely packed bone that is periodically formed in birds (Dacke et al. 1993) and at least some dinosaurs (Hübner 2012; Lee & Werning 2008; Schweitzer et al. 2005), despite its superficial similarity to cancellous bone. Medullary bone is rapidly laid down to act as a calcium reservoir for the production of eggshells before they are laid, and consequently its tissue is not as mechanically competent as that of other, permanent bone tissues: its primary function is metabolic, rather than mechanical (Currey 2002). The macroscopic architecture of cancellous bone is characterized by a complex, 3-D lattice-like array of interlinking bony struts called trabeculae, from the Latin trabecula, meaning 'small beam' (Fig. 1). The shape of individual trabeculae may be rod-like, plate-like or some variant in between (Singh 1978). Despite being not as mechanically competent as cortical bone, cancellous bone forms a key component of the skeleton; in humans, it comprises some 70% of the whole skeleton by volume (Huiskes 2000). The highly complex macrostructure of cancellous bone gives it an exceptionally high ratio of surface area to volume, which makes it a useful reservoir for calcium homeostasis (Clarke 2008; Swartz et al. 1998). More importantly, this high surface area also leads to a rate of remodelling that is an order of magnitude greater than that of cortical bone; in humans, some 25% by volume is remodelled per year, compared to 2-3% for cortical bone (Clarke 2008; Huiskes et al. 2000;