Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
~oologkal3ournaloffhe Linnem So&& (1997), 121: 249-291. With 12 figures An X-radiographic and SEM study of the osseous inner ear of multituberculates and monotremes (Mammalia):implications for mammalian phylogeny and evolution of hearing RICHARD C. FOX' Laboratory for Vertebrate Paleontology, Department o f Biological Sciences, Universib o f Alberta, Edmonton, Alberta, 76G 2E9, Canada JIN MENG2 Department o f Ertebrate Paleontology, American Museum of Natural History, Central Park 10024, U S A . and Institute o f Vertebrate Paleontology West at 79th Street, New Erk, and Paleoanthropology, 20. Box 643, Be$%, China Received FebwaT 1996; accepted for publication S;b 1996 Multituberculate petrosals with well-preserved, three-dimensional internal anatomy from the Late Cretaceous/early Paleocene Bug Creek Anthills, Montana, U.S.A., are described from X-radiographic and SEM images, as well as from conventional visual observations, and are compared with the anatomy of the osseous inner ear in monotremes and in primitive nontherian and therian mammals. Results of this study indicate that: (1) the cochlea of at least some multituberculates retained a lagena, previously known only in monotremes among mammals; (2) an enlarged vestibule evolved in several lineages of multituberculates independently, and hence is not a synapomorphy of the order; (3) the cochlear canal lacks osseous laminae in support of the short, wide basilar membrane, which was probably inefficient in responding to high-frequency airborne vibrations; and (4) consequently, boneconducted hearing in some multituberculate species may have been important in interpretation of their surroundings. Comparisons with the inner ear of monotremes and primitive therians indicate that curvature of the cochlea and cribriform plates for passage of vestibulocochlear nerve branches through the petrosal are unlikely homologues between monotremes and therians. From non-therian to therian mammals, there is a distinct morphological gap in the inner ear transition, characterized by acquisition of a number of neomorphs in the therian inner ear; an intermediate stage has yet to be discovered. 0 1997 The Linnean Societv of London ADDITIONAL KEY WORDS:-anatomy America. - Late Cretaceous - Multituberculata - North ' Corresponding author. * Present address: Department of Biology, University of Massachusetts, Amherst, MA 01003, U.S.A. 0024-4082/97/110249+43 $25.00/0/zj960089 249 0 1997 The Linnean Society of London Introduction . . . . . llaterial and methods . 5Iatrrial . . . . ~lcthods . . . . 'l'rrnmiiioloLT . . . ;ibbrr~~iations . . . . lkscription . . . . . llultituberculates . llonotremes . . . Discussion . . . . . l'hylogrnetic implicatioiis 1;uiictional implicatioIis . .\cknou Irdgeinents . . Krfcrciico . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 251 231 2 2 2 2 264 279 28 1 285 28.5 \Yhile tlie origin of the mammalian ear has long been of interest to comparative anatomists (e.g. Gaupp, 1913; Goodrich, 1958), significant new perspectives concerning it have recently been provided by vertebrate palaeontologists. Important information from the fossil record has been especially relevant to studies focusing on: ( I ) transformation of the reptilian jaw suspensorium into the triossicular hearing systcm of mammals (e.g. Allin, 1975, 1986); (2) reconstruction of vessels and nerves on tlie external surfaces of the petrosal bone as dictated by grooves, foramina, or other external structures preserved there (e.g. MacIntyre, 1972; Wible, 1990; Rougier, JYible & Hopson, 1992); (3) reconstruction of the soft tissue anatomy of the inner ear as inferrcd from the internal bony architecture of the petrosal (e.g. Allin, 1986; Luo & Ketten, 1991). In assessing the significance of such studies, one weds to rcmember that the fossil specimens contain the only direct evidence that is available concerning the course of evolution of the ear. At the same time, these fossils arc the source of novel characters in a temporal context, and can contribute to both phylogenetic reconstruction and functional interpretation on that account. :Ilthough recent gains in knowledge of the early mammalian ear have been important, they have not been uniform, across all parts of the hearing apparatus. AIost is known about the middle ear (see Kermack, 1963; Hopson, 1966, 1969; hlacIntyre, 1972; Allin, 1975, 1986; Archibald, 1979; Kermack et al., 1981; Prothero, 1983: Miao & Lillegravcn, 1986; Novacek & Wyss, 1986; Graybeal et al., 1989; Rougier et a/., 1992; M'ible & Hopson, 1993; Meng, 1992; Meng & n'yss, 1995), ivhile knowledge of inner ear morphology has been more difficult'to obtain, if on1)because of the small number of specimens in collections in which crucial delicate structures arc well preserved. The few investigations of the early mammalian inner car that lia\re been published reveal only general features, such as the shape of the cochlea as obtained by serial sections and computerized tomography (CT) (KielanJaivorowska et al., 1986; Allin, 1986; Graybeal et al., 1989; Luo gL Kctten, 1991; Luo ~t al., 1995). Indeed, for all fossil mammals, detailed inner ear morpholoE has been established in but a few, highly specialized, Cenozoic species (e.g. some Tertiary cetaceans [Fleischer, 1976; Ketten, 19921); comparable details are little kn0u.n in mammals from any time in the Mesozoic. This lack of knowledge is particularly urifortunatc because the inner ear is clearly more important than the middle ear in INNER EAR OF MULTITUBERCULATES 25 I mammalian hearing-and the inner ear contains important phylogenetic information, as well, especially at higher taxonomic levels. It is in this context that we have recently described the osteology of the inner ear and explored its phylogenetic and functional implications among representative early placentals and marsupials (Meng & Fox, 1993, 1995a, b). However, during the course of that research, we also examined the inner ear of Late Cretaceous multituberculates, relying especially on X-radiographic and SEM images but utilizing direct visual inspection, in addition. The results of the multituberculate study are presented here, and are contrasted with aspects of the inner ear in monotremes, as well as in primitive marsupials and placentals as revealed by our earlier work. MATERIAL AND METHODS Material The multituberculate petrosals of this study were collected from the Hell Creek Formation at the Bug Creek Anthills locality, McCone County, Montana (Sloan & Van Valen, 1965)by the first author using dry- and underwater-screening techniques, and are listed in Table 1. It is important to note that these specimens have required no preparation of external or internal features other than removal of loose sand and silt with a camel’s hair brush or, occasionally, with a finely ground needle; hence, no dissolution of delicate structures, always a danger in acid-preparation, nor damage as a consequence of mechanical preparation, has altered the morphology of the specimens after they were collected (see, e.g. Graybeal et al., [1989: 1141; Rosowski & Graybeal [1991: 1381).The specimens are catalogued in the Laboratory for Vertebrate Paleontology, Department of Biological Sciences,University ofAlberta (UALVP). The current uncertainty as to the age of fossils at the Bug Creek locality, whether latest Cretaceous or earliest Palaeocene or both (e.g. Archibald & Lofgren, 1990; Lofgren, 1995), is irrelevant to the purposes of this paper, which are morphologic, phylogenetic and functional, not biostratigraphic. In our Bug Creek collection, multituberculate petrosals of three different morphologies have been recognized, as represented by UALVP 34 144, 26039, and 26037, respectively, and our descriptions are based on these three specimens. Other specimens in this collection are fragmentary; therefore, they are not described here, but all are listed in Table 1. UALVP 34 144 may belong to the same group of multituberculates as the petrosals (MCZ 19177, ZPAL MK-1) studied by KielanJaworowska et al. (1986), while the other two specimens appear different from any Hell Creek petrosals that have been previously described (see, e.g. Kielan-Jaworowska et al., 1986; Luo, 1989; Wible & Hopson, 1995). Taxonomic identification of all such isolated specimens, even to family level, is uncertain (see Kielan-Jaworowska et al., 1986 for similar conclusions), and must await the collection of articulated cranial material in which diagnostic evidence, especially dentitions, is preserved. Luo (1989) recognized three petrosals having different morphologies among the Bug Creek multituberculates, and UALVP 34144 may well correspond to his ‘Ptilodontoidea’ (cf. Luo, 1989: fig. 4, MCZ 21345; fig. 5, MCZ 19177), but we have been unable to assign others of our specimens to his categories. For example, in UALVP 26039, the mastoid region appears to have been fenestrated, a ptilodontoid 252 R. C . FOX AND J. MENG condition according to Luo (1989), but this specimen also has a large vestibule, which Luo (1989) believed to be a taeniolabidoid character. In light of these uncertainties, we have not attempted to assign our specimens to any subgroup of multituberculates (although recognizing that it is virtually certain that they are not plagiaulacoid in their affinities: plagiaulacoids are thought to have become extinct in the Early Cretaceous). The monotreme petrosals included in this study are from specimens in the collections of the departments of Vertebrate Paleontology and Mammalogy, American Museum of Natural History, New York. Methods For X-radiographic examination of internal structures of the petrosal, we used the Radiographic-Fluoroscopic Inspection System and Ready Pack I1 Kodak Industrex M Film. Usually, when the tympanic side of a specimen is placed against the film, a clear image of the cochlea can be obtained. A latex endocast taken from one of our specimens, UALVP 26039, has been useful in revealing fine details of vestibular structure and was constructed according to the following protocol. The petrosal was carefully enclosed in plasticene, blocking all openings except for a broken area in the wall near the fenestra vestibuli (fig. 3); this area allowed access into the interior spaces of the bone. Thin layers of moulding latex were successively painted on to the inner surface of the vestibule with a camel’s hair brush in order to build a self-supporting latex wall. After the latex had dried, talcum powder was sprayed into the vestibule with syringe and needle to coat the exposed surfaces of the latex wall and prevent opposing parts of it from adhering to each other during removal of the endocast. The endocast was then carefully freed from the vestibular walls, extracted from the petrosal, and coated for scanning electron microscopy (Fig. 4). Other SEM photographs were taken directly from uncoated specimens. Measurements of the specimens included in this study are presented in Table 1. All measurements were made using the Reflex Microscope (Reflex Measurement LTD), either directly from the specimens themselves or from radiographs. The 2 x magnification setting and the 5 micron mark on the microscope were used when taking measurements. In our study, the starting point of the cochlear canal is considered to be at the notch produced by the perilymphatic duct in the interior ridge that defines the perilymphatic recess, well anterior to the depression marking the recess (see e.g. UALVP 26037 Fig. 51, a broken right petrosal in which these parts are particularly well preserved). The length of the cochlea was measured along its longitudinal mid-axis; the width of the cochlea was taken at its base. In our descriptions, we refer to the tympanic side of the petrosal as ventral, the cranial side as dorsal, and the apical end of the promontorium as anterior, although the petrosal was probably oriented in the skull such that the tympanic side actually faced ventrolaterally and the cranial side dorsomedially. For anatomical features, we follow the terminology of Gray (1959), Kielan-Jaworowska et a/. (1986) and Rougier et al. (1992), unless otherwise specified. We use the term ‘cochlear canal’ INNER EAR OF MULTITUBERCULATES 253 for the osseous structure that in living monotremes and multituberculates contained the soft cochlear duct (basilar papilla) and lagena; in therians, in which no lagena is developed, the cochlear canal contained the soft cochlear duct alone (see also Luo & Ketten, 1991). In our usage, the term ‘cochlea’ refers, then, to the cochlear canal and cochlear duct together. We use the terms ‘Mammalia’, ‘Theria’, ‘Pelycosauria’,‘Cynodontia’, and ‘Reptilia’ in their traditional, widely understood senses, as in Lillegraven et al. (1979) and Carroll (1988). ABBREVIATIONS aa acs aF a1 aqc asc bm cac cc cn COC er fcn fib fPr fsb fv gcn glsc P d gsu anterior ampulla area cribrosa superior aquaeductus Fallopii anterior lamina of petrosal aquaeductus cochleae anterior semicircular canal basilar membrane cochlear orifice of aquaeductus cochleae crus commune cochlear nerve (branch of eighth cranial nerve) cochlear canal epitympanic recess foramen for cochlear nerve foramen for inferior branch of vestibular nerve fossa for perilymphatic recess foramen for superior branch of vestibular nerve fenestra vestibuli grooves for cochlear nerve gyrus of lateral semicircular canal groove containing the secondary perilymphatic duct groove between sacculus and utriculus la lar lateral ampulla lateral aperture of perilymphatic recess ( = recessus scalae tympani) If lateral flange of petrosal lsc lateral semicircular canal mss mastoid surface that contacts squamosal notch for posterior branch of nPb vestibular nerve promontorium P posterior ampulla Pa perilymphatic recess Per perilymphatic foramen Pf processus recessus P‘ posterior semicircular canal PSC recessus sphericus rs Sf subarcuate fossa groove for stapedial artery sg cone-shaped swelling (‘bulging’) for sPr perilymphatic recess st scala tympani tractus spiralis foraminosus tsf tensor tympani fossa ttf V vestibule vnib inferior branch of vestibular nerve vnpb posterior branch of vestibular nerve vnsb superior branch of vestibular nerve DESCRIPTION Multituberculates (left petrosal) Although several studies have demonstrated that the vestibule in many multituberculates is inflated (Miao, 1988; Luo, 1989; Luo & Ketten, 1991)’ in UALVP UALVP 341 44 2.54 R. C. FOX AND J. MENC 34144 it clearly is not (see below), as evidenced by the retention of a deep, welldefined paroccipital fossa externally and by the configuration of the cavity itself as shown in the X-radiograph (Figs 1, 2). In keeping with the external morpholog of the specimen (Fig. l), the X-radiograph also shows that the lateral semicircular canal and depression for the paroccipital process are located posteroventral to the vestibule, and that the semicircular canals open into bell-shaped ampullae at the two ends of the vestibule. As in the ear of early tribosphenic therians (Meng & Fox, 1995a, b), the ampullae in this specimen are significantly larger in diameter than the canals; in contrast to the therian ear, however, the region between the ampullae and the vestibule proper remains relatively wide, not narrowed to form a neck. In the X-radiograph of UALVP 34144, the spaces for the sacculus and utriculus are not clearly differentiated from one another, but the approximate dimensions of thc two cavities in the specimen can be inferred from two lines of evidence: first, as in most mammals, the semicircular canals join the utriculus through five openings one for the crus coInmune, the others for the medial entrance to the lateral semicircular canal and the three ampullae, respectively-such that the distancc between the junctions of the semicircular canals with the vestibule furnishes at least minimum dimensions for the utriculus. Second, while the sacculus in mammals is usually located at the anteroventral side of the utriculus, in UALVP 34144 the requisite space is very limited and the sac that it contained must have been correspondingly small. In keeping with these proportions, the foramen for the sacculus branch of the vestibular nerve in UALVP 34144 is very much smaller than that for the utriculus branch (see below). In UALVP 34144, the cochlear canal arises anterior to the medial part of the vestibule, closer to the posterior ampulla than to the anterior and lateral ampullae. Thr canal extends anteriorly and horizontally, and bends laterally. Its curved course is suggested by the external topography of the promontorium but is explicitly documented in the X-radiograph. The canal maintains a constant diameter along its length, except at its terminus, which is rounded; unfortunately, its cross-sectional shape cannot be determined, either from the X-radiograph or visually, under magnification. No internal structures within the canal are evident from the Xradiograph. However, the X-radiograph does show four pale, branching lines that cross the cochlear canal nearly normal to its long axis; these have not been seen in other specimens, and their significance is unknown. They may represent small grooves or canals for arterial pathways as occur in living mammals (Axelsson, 1974). Several features of the inner ear can be directly observed through openings in UALVP 34144. As seen through the lateral aperture of the perilymphatic recess (Fig. 1A; sre below and Discussion for terminology), a prominent ridge curves into the cochlear cavity from the aperture and defines an ovate fossa, adjacent to the groove for the secondary perilymphatic duct anteromedially (see Discussion for terminology) and separate from the vestibule more internally. The ridge curves down on to the dorsal side of the bony bar between the lateral aperture and fenestra vestibuli (crista interfenestralis [Wible etal., 1995]),and then swings back more medially, ending slightly internal to the ventral rim of the lateral aperture. Neither ridge nor fossa have been recognized and described in multituberculates previously. From its position and shape, it seems likely that the fossa is the ‘perilymph recess’ of Gray (1 908a: 523) which housed the perilymphatic sac (Romer &Parsons, 1986),and that the ridge formed the boundary between the recess and the cochlear canal; hence, in life the perilymphatic sac was only narrowly confluent with the cochlear duct. More proximally, the perilymphatic ~ INNER EAR OF MULTITUBERCULATES 255 Figure 1. Unidentified multituberculate, incomplete left petrosal (UALVP 34144), Bug Creek Anthills locality, Hell Creek Formation, Montana, USA A, tympanic (ventral) view; B, cranial (dorsal) view. Scale bar = 1 mm. See Abbreviations for key. R. C . FOX AND J. MENG 2.56 A B C D la cc Figure 2. X-radiographs of unidentified multituberculate, incomplete left petrosals (A, UALVP 34144; B, UALVP 26039), Bug Creek Anthills locality, Hell Creek Formation, Montana, USA. C and D outline restorations of the inner ear based on X-radiographs; note that the semicircular canals in A are partly broken. See Abbreviations for key. system communicates with the subarachnoid space via the secondary perilymphatic duct in mammals, and correspondingly in UALVP 34144, as in other multituberculates (see e.g. Luo, 1989),a well-defined sulcus for the secondary perilymphatic duct crosses the dorsal rim of the lateral aperture; there is no separate bony canal (aquaeductus cochleae) developed for the duct as in therians. Hence, the lateral aperture of the perilymphatic recess in multituberculates is equivalent to the fenestra cochleae plus aquaeductus of therians, representing a more primitive pattern in a transformation series (de Beer, 1937; Zeller, 1993). O n the cranial (cerebellar) side of UALVP 34144, four foramina open into the internal acoustic meatus, which is shallow but clearly defined, matching in that respect the internal acoustic meatus in the petrosals of various other Late Cretaceous multituberculates that have been described (see Luo, 1989: 13-14, table 2). Ho\ve\,er, in these other species, the morphological pattern may be different: for example, in IYNER EAR OF MULTITUBERCC'IATES 257 some, only three foramina have been observed (e.g. see Kielan-Jaworowska et al., 1986: 586). In UALVP 34144, the foramina are divided into two groups by a crest or crista falciformis. On the lateral (dorsal) side of the crista are three foramina, arranged in a triangle. The anteriormost is the aquaeductus Fallopii, which transmitted the facial nerve to the tympanic side of the petrosal. Posterior to the aquaeductus, the smallest foramen of the three penetrates the anterior wall of the vestibule; it conveyed the inferior branch of the vestibular nerve to the sacculus and is the functional equivalent of the area cribrosa media in therians. The internal (vestibular) opening of this foramen can not be seen in UALVP 34144 because the vestibular wall here is complete, but its position is confirmed by other specimens (e.g. UALVP 26037, 26039) in which the wall is broken, exposing the internal structure of the vestibule (see below). The most lateral (or dorsal) foramen, not visible in the illustrations, is the largest of the three. It is the entrance to a short channel that runs posterolaterally to the lateral side of the vestibule, and probably carried the superior branch of the vestibular nerve to the utriculus and to the lateral and anterior ampullae; accordingly, this foramen must be functionally equivalent to the area cribosa superior in therians. In Figure lB, only the aquaeductus Fallopii is visible. O n the medial (ventral) side of the crista falciformis only a single, large foramen is present. It opens into the cochlear canal and must have carried at least the cochlear nerve, which was probably housed in a broad groove between the foramen and the canal. However, several lines of evidence indicate that this foramen may also have conveyed the posterior branch of the vestibular nerve, which ran to the posterior ampulla: first, in the most proximal part of the cochlear canal, a welldefined trough, visible through the fenestra vestibuli, leads from the foramen posteromedially to the posterior ampulla. This trough is bounded medially by the ridge bordering the fossa for the perilymphatic recess; while it is best preserved in UALVP 26039, it can be clearly seen in UALVP 34144, as well. Second, other foramina in the internal acoustic meatus are too distant from the posterior ampulla to have contained the posterior branch of the vestibular nerve. This is especially clear in UALVP 26039, in which the vestibule is inflated. Third, the foramen that we believe transmitted the nerve to the sacculus seems too small to have conveyed a second branch of the vestibular nerve as well, nor is it in any way subdivided, as if it carried two nerve rami. Therefore, the large foramen on the medial side of the crista falciformis is equivalent to the foramen singulare for the posterior branch of the vestibular nerve plus the cribriform area for the cochlear nerve in therians (by contrast, Luo [1989: 13-14] and Lillegraven & Hahn [1993: 541 believed that only the cochlear nerve occupied this foramen, but they did not have access to the internal anatomy of the petrosals that they studied). Hence, the arrangement of the foramina for the eighth cranial nerve in multituberculates is different from that of therians, leading to the suspicion that the crista falciformis in multituberculates may not be homologous to the therian crista. It is now known that multituberculates primitively had a common fossa housing facial and vestibulocochlear foramina (in Late Jurassic paulchoffatiids: see Lillegraven & Hahn, 1993: 54-55), contrary to conclusions from other work (Hahn, 1988; Wible & Hopson, 1993). (left petrosal) Although this specimen is larger overall (Fig. 3), the inflated vestibule is the most prominent feature distinguishing it from UALVP 34144. Both the X-radiograph UALVP 26039 Figure 3. Unidcntified multituberculate, incomplete left petrosal (UALVP 26039), Bug Creek ;2iithills localit)-. Hrll Crcek Formation. hlontana, USA: -4, t?mpanic (ventral) vicw; R, cranial (dorsal) view, \\ith thr approximate anterior side upward. Stale bar = 1 mm. See Ilbbreviations for key. INNER EAR OF MULTITUBERCULATES 259 Figure 4. SEM photographs of latex endocast of vestibule, UALVP 26039: A, approximately anterior view; B, dorsal view, with the approximate anterior side downward. Scale bar = 1 mm. See Abbreviations for key. (Fig. 2B) and the endocast (Fig. 4) of UALVP 26039 show a nearly spherical vestibule, which has a maximum diameter of 4mm. Inflation of the vestibule has displaced the petrosal wall ventrally and posteriorly, and has considerably modified the topography of the mastoid part of the bone. In addition to reduction of the fossa for the paroccipital process, the lateral semicircular canal is embedded in the posterior wall of the vestibule, as revealed by the X-radiograph. Nonetheless, the semicircular canals open into the vestibule according to the same pattern as in UALVP 34 144, although the ampullae are significantly smaller relative to the size of the inflated vestibule. As shown by the endocast (Fig. 4), the crus commune is at the dorsal side, the anterior and lateral ampullae at the lateral side, and the posterior ampulla and medial end of the lateral semicircular canal at the medial side of the vestibule. The posterior half of the vestibule is smooth and spherical, while the anterior side is associated with several structures. There is little doubt that both the sacculus and utriculus were enlarged to fill the vestibule in this specimen, but the utriculus was probably the larger of the two, partly for the reasons discussed above in relation to UALVP 34144. However, the endocast of UALVP 26039 provides additional and even more direct evidence in support of this interpretation: it indicates that the superior branch of the vestibular nerve was large, and ran laterally to the anterior and lateral ampullae and to the utriculus; the inferior branch entered at the center of the oval recessus sphericus for the sacculus, which is outlined by a shallow groove; the posterior branch, entering the inner ear together with the cochlear nerve, ran posteromedially to the posterior ampulla. Because of the larger size of UALVP 26039, the cochlear canal is broader in that specimen than in UALVP 34144; in cross section, it is obviously oval, wider than high, but not round, contrary to what has been reported for multituberculates generally (Luo & Ketten, 1991; Wible & Hopson, 1993: 54). The canal in UALVP 26039 is straighter than in 34144; whether or not this difference is size-dependent 260 R. C. FOX AND J. MENG is not clear. The X-radiograph reveals a particularly interesting feature of the canal in UALVP 26039-the inflation of its apical region (Fig. 2B), which strongly suggests the presence of a lagena (see below). Owing to breakage near the fenestra vestibuli in UALVP 26039, a low rounded ridge with surfaces of finished bone can be seen in place ventrolateral to the internal opening for the cochlear nerve. This ridge extends longitudinally along the inner surface of the lateral wall of the cochlear canal and terminates at a level about onethird of the total length of the canal from the cochlear base. UALVP 26037 (below) clearly shows that this ridge marks the entry of the cochlear nerve into the cochlear canal, and because its surfaces are entirely of finished bone, we conclude that it did not support a therian-like bony lamina projecting yet further into the lumen of the canal, nor did it serve as the passageway for the cochlear nerve, as in therians. The more distal parts of the inner surface of the cochlear canal are exposed by breakage anteriorly on the ventral surface of the promontorium (Fig. 3), and can be seen to be entirely smooth. Finally, the fossa for the perilymphatic recess (and its associated ridge), the groove for the secondary perilyrnphatic duct, the trough that may have carried the posterior branch of the vestibular nerve, and the foramina within the internal acoustic meatus are all present in the same positions and have the same relations to one another in UALVP 26039 as in 34144. In UALVP 26039, however, the foramen for the inferior branch of the vestibular nerve is somewhat larger (relatively, as well as absolutely) than in 34144, and a small foramen opens in the bridge of bone that separates the inferior and lateral vestibular foramina. The vestibular opening connecting with this foramen has not been identified, and it may only be a nutritive foramen supplying the bone. However, UALVP 26034, an incomplete specimen from Bug Creek, shows a small foramen in this same position that opens internally, into the vestibule. UALVP 26037 (right petrosal) This specimen is from a multituberculate substantially smaller than those represented by 34144 and 26039. It is incomplete, but owing to fortuitous breakage in the mastoid (Fig. 5), anterior and ventral parts of the vestibule are amply exposed. O n the anteroventral wall of the vestibule, the oval depression marking the recessus sphericus for the sacculus is well defined. The foramen that transmitted the inferior branch of the vestibular nerve to the sacculus penetrates the vestibular wall at the anterodorsal edge of the recessus. Medial to the recessus is the large opening that communicates between the vestibule and the cochlear canal. The dorsal edge of this opening is interrupted by a relatively large, nearly semicircular notch for the trough that presumably carried the posterior branch of the vestibular nerve to the posterior ampulla. Also medial to the opening is a narrow, relatively high ridge that rises from the floor of the vestibule and separates the vestibule proper from the space that contained the posterior ampulla and the orifice of the lateral semicircular canal. The crus commune is exposed at a broken surface laterally on the dorsal side of the vestibulc. The anteromedial wall ofthe promontorium is broken in UALVP 26037, revealing most of the cochlear canal. The canal is slightly bent laterally, and its curvature is mirrored externally by the medial rim of the tensor tympani fossa (M. major fossa of Luo, 1989). As before, the cochlear canal is not round in cross section, but oval, higher medially than laterally, and with its long axis transversely oriented. As in INNER EAR OF MULTITUBERCULATES 261 Figure 5. Unidentified multituberculate, incomplete right petrosal (UALVP 26037), Bug Creek Anthills locality, Hell Creek Formation, Montana, USA A, lateral (slightly dorsal); B, tympanic (ventral) view. Scale bar = 1 mm. See Abbreviations for key. UALVP 26039, a low longitudinal ridge is developed proximally on the inner surface of the lateral wall of the canal marking the most proximal course of the cochlear nerve within the canal. No other structures arise from the walls of the canal, which as far as can be determined from the parts preserved, are entirely smooth, composed of finished bone. No evidence of a shelf or other structure that might indicate the development of a bony lamina that contained the cochlear nerve or supported a basilar membrane is present. The distalmost parts of the canal are partly preserved in this specimen and as a consequence of breakage, are Sully exposed to view. These indicate a slight expansion of the walls of the canal towards its terminus, consistent with a lagena having been developed there. The fossa for the perilymphatic recess and its associated ridge, the groove marking the course of the secondary perilymphatic duct (particularly well developed on this specimen), the trough for the posterior branch of the vestibular nerve, and the foramina within the internal acoustic meatus do not differ significantly from those in UALVP 34144 and UALVP 26039. Monotrmes As shown in Figure 6, the petrosal of Tachyglossus is much larger and more massive than that in Ornithorhynchus. The vestibule in Tachyglossus is roughly oval, with its two ends narrowing to give rise to the ampullae of the semicircular canals. From base to terminus, the cochlear canal curves in the same direction that typifies mammals generally, that is, clockwise in the right ear and counterclockwise in the left; the complete curvature is about 180". In its diameter, the canal remains nearly constant 262 R. C;. FOX AND J. MENG B COC Fig-urc 6. Radiographs of cranial and ear r r p n s of Tuc~~Io.rsus (A and C, AhlNH 42257. teachirig collection of the Departmrnt of Vertebrate PaleontoloLgy)and OmithoihynchuJ (B and D, XP\INH 1 3 15 C:. ,A,, tcachirig collection of DVP). A and R are at the same scale; C and D are not to the samc scale. C is photographically reversed from A. INNER EAR OF MULTITUBERCUL4TES 263 through most of its length, although its apical region appears to be somewhat inflated, housing a lagena (Pritchard, 1881). No osseous structures within the canal are evident from the X-radiographs, nor have such structures been observed through breaks in the walls of the promontorium in the available specimens. Our observations in these respects are in agreement with those previously made on the petrosal of Tuchyglossus by other workers (see e.g. Luo & Ketten, 1991). The cochlear canal in Ornithorhynchus differs somewhat from that in Tuchyglossus: it is relatively straight proximally but abruptly bent distally, or ‘sickle shaped’ (Zeller, 1993: 10I), at its terminus (Fig. 6B, D). As shown in the X-radiographs, the curvature of the canal is certainly less than 270°, contrary to some previous observations (see below), and is even less than the nearly semicircular canal of Tadyglossus. In Ornithorhynchus, as in Tuchyglossus (Gray, 1908a), the apical region of the canal is slightly expanded and in life held the lagena at the tip of the cochlear duct (Pritchard, 1881), and from the X-radiographs, the diameter of the cochlear canal in Ornithorhynchus shows little difference from base to the apical lagena, as already noted by Gates et ul. (1974). A left petrosal of Ornithorhynchus (AMNH 179983 [Mammalogy]) that is broken through the fenestra vestibuli, perilymphatic foramen, internal acoustic meatus and cochlear canal reveals aspects of otic morphology that have not been clearly described before (Fig. 7). On the cerebellar side of the petrosal within the internal acoustic meatus, the passageway for the facial nerve is a single oval foramen, as in therians. Also as in therians, the foramen for the vestibular nerve and the foramen for the cochlear nerve (the tractus spiralis foraminosus in therians) are closed by cribriform plates. In contrast to the pattern in therians, the cribriform area for the cochlear nerve does not form a spiral belt. No secondary lamina is developed on the inner surface of the bony bridge between the fenestra vestibuli and the lateral aperture of the perilymphatic recess, nor is this lamina present on the inner surface of the cochlear canal, resembling multituberculates but not therians in this respect. However, on the dorsal surface of the canal in this specimen (Fig. 7), many (approximately a dozen) small grooves fan out from the internal acoustic meatus toward the canal at an angle nearly perpendicular to its course. This indicates that the cochlear nerve branches into small rami as it approaches the basilar membrane in Ornithohynchus. It also demonstrates that the pathway for the cochlear nerve can leave bony traces on the inner surface of the cochlear canal, a phenomenon that has not been observed in any fossil forms other than therians (Meng & Fox, 1995a, h). This pattern differs from Pritchard’s (1881) report that the cochlear nerve runs parallel to the ‘lamina’ (see Discussion). It also differs from the therian condition in that the cochlear nerve is not supported by an osseous lamina projecting into the lumen of the cochlear canal and dividing the canal into a scala tympani and scala vestibuli, in agreement with previous observations by Zeller (1989) and Luo & Ketten (1991). Although possessing a cribriform area for the entrance of the cochlear nerve, a modiolus is not developed within the petrosal, which differs from the therian pattern. Finally, the perilymphatic foramen opens directly into the cochlear canal with no fossa and ridge intervening within the petrosal internal to the opening. This pattern contrasts with that in therians (Meng & Fox, 1995a, b), as well as multituberculates, as described above. The space housing the lagena in AMNH 179983 is smoothly rounded, as in the multituberculate petrosals already discussed. R. C. FOX AND J. MENG F i p r c 7. SEM photographs of broken left petrosal of Omithor/ynchus, (AMNH 179983 FIammalogy]): A. cranial view, showing internal acoustic meatus and related structures; B, lateral view (of counterpart of A), showing inside of cochlea. Scale bar = 1 mm. See Abbreviations for key. DISCUSSIOK Peribmphatic foramen and fenestra cochleae As we pointed out in previous work (Meng & Fox, 1995b), Gray (1 908a) undertook the first thorough discussion of the relationships of the perilymphatic recess, fenestra cochleae and aquaeductus cochleae among reptiles, birds and mammals, although de Beer (1937: 401-402) has also contributed importantly to understanding of the relationships of these structures in living vertebrates. From their results, those from later (especially palaeontological) studies, and our own data on the inner ear in primitive mammals, we have summarized several features of this region that we INNER EAR OF MULTITUBERCULA’IES 265 believe to be synapomorphies of therians. These include: (1) merger of the perilymphatic recess with the basal part of the scala tympani within the petrosal bone; (2) formation of the fenestra cochleae and aquaeductus cochleae by development of the processus recessus, a caudal outgrowth of the pars cochlearis of the auditory capsule (de Beer, 1937; Kermack et al., 1981; Zeller, 1985a, b; Wible, 1990); and (3) separation of the fenestra cochleae and fenestra vestibuli by a wide bony bridge that supports the basal part of the secondary lamina on its inner surface; hence, the basilar membrane in therians was elongated not only apically but also proximally, extending to the area between the fenestra cochleae and fenestra vestibuli. In nontherian mammals, the bony bridge between the fenestra vestibuli and the lateral aperture of the perilymphatic recess is usually narrow, and a bony lamina along the outer wall within the cochlear canal on the inner surface of the bridge is not devefoped. Therians are characterized additionally by a spiral cochlea, a primary osseous spiral lamina, a radial pattern of the cochlear nerve and a modiolus that forms the central axis or pillar of the cochlear spiral (Gray, 1959; Meng & Fox, 1995b). Kncelestes@ougier etal., 1992; Hopson & Rougier, 1993),ofLower Cretaceous age and the only pre-tribosphenic therian in which the basicranial region of the skull is known, may display intermediate conditions in this region, but its inner ear morphology has yet to be described. Here we discuss the homologies and distribution of several of these features, starting with the perilymphatic foramen, the aquaeductus cochleae, and the fenestra cochleae in therian and non-therian mammals (Fig.8). To understand these structures in fossils, reference must be made to the perilymphatic system in living taxa (see e.g. Gray, 1908a; de Burlet, 1934; de Beer, 1937; Kuhn, 1971; Wever, 1978, 1985; Zeller, 1985a, b, 1989, 1993), but the discussion below attempts to extend these works to the history of early mammals, and within a broader comparative context than in the literature at present. In return, the fossil record can contribute to understanding of the evolution of the mammalian ear, for which data from living forms alone are not sufficient. In our consideration of the perilymphatic system, we first explore two issues: (1) the perilymphatic duct in relation to the perilymphatic foramen and the aquaeductus cochleae, and (2) the secondary tympanic membrane in relation to the lateral aperture of the perilymphatic recess ( = recessus scalae tympani of authors; see Gray, 1908a) and the fenestra cochleae. The anatomy of these structures is well displayed in lizards and crocodilians (e.g. Wever, 1978),and their sequence in the perilymphatic system-scala leading to foramen leading to sac-appears to be primitive for tetrapods (de Beer, 1937; see Wever, [1985] for Recent amphibians as further confirmation of the primitive attributes of this pattern). According to a broad spectrum of vertebrate morphologists (Gray, 1980a; de Beer, 1937; Goodrich, 1958; Wever, 1978; Bellairs & Kamal, 1981; Rieppel, 1985; Zeller, 1985a, b, 1993; Manley, 1990), the perilymphatic foramen in living nonmammalian amniotes carries the perilymphatic duct between the perilymphatic sac and the scala tympani of the cochlea, with the sac held within the bony perilymphatic recess. For example, although Gray (1908a) did not use the term ‘perilymphatic foramen’, his equivalent, ‘oval opening’, refers to this passage between the perilymphatic recess and the cochlea in birds and mammals. According to de Beer (1937: 40 1 4 0 2 ) , “the term foramen perilymphaticum [refers to] the posterior opening in the wall of the auditory capsule of Tetrapoda, through which the ductus perilymphaticus leaves the cavity of the capsule and enters the recessus scalae Figure 8. SEA1 photographs showing perilymphatic foramen and related structures in Rug Creek manimals and extant Omitlzorh3,nc-hus. A, unidcntified placental, incomplete left pctrosal (left is approxiniatcly anterior) (UA1,VP 26043 [see LZeng & Fox, 199SbI); B, unidentified multituberculate: iriconiplcte right petrosal (UA4L\’P 26037); C, Onzit/zor/ync/iu.c,left ear region (AIlNH 13 15 C. A . ~ teaching collection, Department of Vertebrate Paleontology); D, unidentified multituherculatr. leli p r t r o d (L;AL\‘P 26039). B and 1) are approximately posteromedial views of the promontorium. The \.c.ntralpart of thc promontorium in the placental petrosal (A) is broken. exposing thc internal structure5 of thc petrosal. In A, the solid sqiiarc indicates the approximate position of the perilymphatic fbrarnrn \vithin the cochlea. bet\vccn the hasc of the scala tympani (st) and the perilymphatic recess (qx.‘conch a p e d bulging’ for the perilymphatic recess of Gray); the dashed circle indicates thc position of thr feriestra rochlcac (hoken) and the arrotv marks the lateral rdge of the fcnestra \-rstibuli. In B arid 1). the lateral aperture of the perilymphatic recess is in a plane that bounds the per, pf. and most of gpd mcdially; therefore per, pf, and gpd are within the petrosal bone in B and 1); b u t more cxtcrriall\located in C:. Not to scale. See Xbbreviations for kc). tympani”, while for Wever (1978: 72), “from the scala tympani a short tube, often not xvell defined. runs as the perilymphatic duct through the perilymphatic foramen into the rccesus [scalae tympani], where it expands to form the perilymphatic sac”. In mammals, however, the term ‘perilymphatic duct’ is conventionally applicd to the tube that communicates between the perilymphatic space of the cochlea arid the subarachnoid space of the braincase (Gray, 1959; Schuknecht, 1970; Zeller, 1985a, b). As it leaves the petrosal, this duct is contained in a groove or sulcus in non-therian mammals and in the bony canal termed aquaeductus cochleae in t herians (Fig. 8).Because the sulcus or aquaeductus cochlea contains a ‘perilymphatic duct’ extending between the perilymphatic recess and the subarachnoid space iii mammals, but the prrilymphatic foramen in non-mammalian amniotes hold5 a INNER EAR OF hfULTITUBERCULATES 267 ‘perilymphatic duct’ extending between the scala tympani and the perilymphatic recess, it is necessary to distinguish terminologically between the two structures. Therefore, we term the duct between the scala tympani and perilymphatic sac the ‘primary perilymphatic duct’ and the more elongate tube between the perilymphatic sac and the subarachnoid space in mammals the ‘secondary perilymphatic duct’. According to Gray (1908a: 525, fig. I), the structure equivalent to the mammalian perilymphatic duct in non-mammalian amniotes is a “dehiscence on roof of perilymphatic recess through which the perilymph is in direct communication with the cerebro-spinal fluid. This dehiscence becomes drawn out into the form of a short tube in the bird and a long tube in the mammal, and in these two divisions is termed the aqueduct of the perilymph”. Apparently, the secondary duct is a neomorph that grows from the perilymphatic sac; it is not simply an elongation of the primary duct. An example of these differences in living mammals is provided by Tachyglossus,in which a definite perilymphatic recess holding a perilymphatic sac is developed (Gray, 1908a): the sac communicates with the scala tympani by the primary perilymphatic duct passing through the perilymphatic foramen (the ‘oval opening’ of Gray), and the secondary perilymphatic duct is a long tube that extends from the posterior end of the perilymphatic sac (ibid.: pl. 20, fig. 10). In Omithorhynchus, Zeller (1985a, 1993) observed a similar pattern: the perilymphatic duct passes into the perilymphatic recess, where it widens, and on leaving the recess, it runs medially in a groove to the cranial cavity. It should be pointed out that although Zeller (1985a, 1993) considered the elongation of the perilymphatic duct in mammals to be secondary, he concluded that the elongation took place independently in Zchyglossus and therian mammals, because in Tachyglossus the glossopharyngeal nerve runs together with the perilymphatic duct within the aquaeductus cochleae (Kuhn, 197 1). Zeller’s (1985a, 1993) description includes the primary and the secondary perilymphatic ducts differentiated by us. The second issue of interest concerns the fenestra cochleae, the lateral aperture of the perilymphatic recess and the secondary tympanic membrane. In lizards and crocodilians, the fluid-filled perilymphatic sac faces onto the air-filled tympanic cavity via an opening conventionally termed ‘round window’ or ‘fenestra rotunda’ (Wever, 1978). For purposes of discussion here, however, a better term for the ‘round window’ in lizards and crocodilians is ‘lateral aperture of the perilymphatic recess’ or ‘lateral aperture of the recessus scala tympani’ (de Beer, 1937; Bellairs & Kamal, 1981; Zeller, 1993): it is doubtful that this opening is homologous to the ‘round window’ (termed, henceforth, fenestra cochleae) of living therians, even though the latter has a comparable position and function to the opening in lizards and crocodilians (see below), and in fact, de Beer (1937) coined the term ‘fenestra pseudorotunda’ for the crocodilian ‘round window’. The part of the sac wall that closes the lateral aperture off from the tympanic cavity in lizards and crocodilians is generally termed the ‘secondary tympanic membrane’, a pressure-releasing device of the perilymphatic system (Weber, 1978). Unlike de Beer (1937), Zeller (1985a, b, 1993) believed that the secondary tympanic membrane is homologous throughout the Amniota, including mammals, and that the fenestra cochleae (rotunda)of therians is homologous with the lateral aperture of the recessus scalae tympani of reptiles, although Bellairs & Kamal (1981) and Rieppel (1985) were less certain. By incorporating evidence from early mammals and other amniotes into the data from 268 R C FOX AND J MENG living forms, as discussed below, the homology of these structures may be considered fi-om a palaeontological perspective, in which multituberculates play a key role. As described above, we have identified the perilymphatic recess in all of the multituberculate petrosals in our Bug Creek collection that have this region preserved and accessible to inspection. The recess is marked by a clearly defined depression in the roof of the petrosal between the cochlear canal and the enlarged external opening posteromedial to the fenestra vestibuli (Fig. 8B, D). The recess communicates with the scala tympani via an inner opening, the perilymphatic foramen, that is defined by a distinct crest. In multituberculates, the inner and external openings have not been clearly distinguished and described before (e.g. Kielan-Jaworowska et al., 1986; Luo, 1989; Lillegraven & Hahn, 1993; Wible & Hopson, 1993). If our identification of the perilymphatic foramen in multituberculates as a structure that is located wzthin the interior of the petrosal is correct, that is, internal to the perilymphatic recess, this external opening obviously cannot itself be the perilymphatic foramen. We believe the external opening is simply a lateral aperture of the perilymphatic recess over which the secondary tympanic membrane was stretched in life. As noted above, the rim of the aperture is cut by a narrow but well-defined groove for the secondary perilymphatic duct in its passage to the subarachnoid space. The configuration of this groove agrees with studies of the multituberculate petrosal by previous workers (Kielan-Jaworowska et al., 1986; Luo, 1989; Lillegraven & Hahn, 1993; \/t’ible & Hopson, 1993, 1995). We have sought to extend our observations on the proximal parts of the perilymphatic system in multituberculates to other mammals, and have found that the basic multituberculate pattern may have been already established in the early stages of mammalian history, as is presently best exemplified by the Early Jurassic non-therian mammal, Morganucodon. Kermack et al. (1981) have shown that in .Vlorgunucodon, the perilymphatic foramen (which they termed ‘fenestra cochleae’) is kvell developed, equal in size to or even larger than the fenestra vestibuli. The rim of the opening in Morganucodon is cut by a narrow groove for the secondary perilymphatic duct. Although some aspects of the internal anatomy of the petrosal in Morganucodon are known from serial sections and C T scans (Grayheal et al., 1989; Luo et al., 1995), fine details comparable to those of the Bug Creek multituberculate pctrosals described here have yet to be determined: for example, whether or not a perilymphatic recess is present in Morganucodon is not clear. Of published material, however, some specimens (Kermack et al., 1981: figs 75B, 83A; Luo, 1994: fig. 6.3A) display a recess of some kind medial to the perilymphatic foramen, whereas in other specimens (e.g. Kermack et ul., 1981: fig. 82A), this recess is less distinct. The location of the recess suggests that it is the perilymphatic recess, resembling that in multituberculates and monotremes, although it seems less internally placed than in multituberculates. The variable conditions of the recess in actual specimens in ,tlorgunucodon suggest that the perilymphatic sac may have been at least partly held in some individuals by a distinct osseous depression medial to the perilymphatic foramen. Detailed description and illustration of this region in Morganucodon, comparable to that which we have provided for multituberculates and monotremes (fig. 8), are needed to clarify the features in question, but whenever further evidence becomes available, we hold that Morganucodon will be in accord with the primiti1.e amniote pattern in which the cochlear canal is partly separated from the more proximal perilymphatic recess (which held a perilymphatic sac), and the perilymphatic foramen opens between the two spaces. From these considerations, we believe that INNER EAR OF MULTITUBERCUWTES 269 the lateral aperture of the perilymphatic recess ofMorganucodonshould be distinguished from the perilymphatic foramen, in keeping with conclusions that de Beer (1 937) and Zeller (1987, 1993) had already determined from living mammals, and in agreement with our own conclusions regarding multituberculates. Where, then, was the secondary tympanic membrane attached in Morganucodon and multituberculates? Kermack et al. (1 98 1) noted that preservation of some petrosals of Molganucodon was complete enough to show that at least part of the rim of what they called the ‘fenestra cochleae’ was developed as a thin flange of bone; they interpreted this flange to be a support for a secondary tympanic membrane. Instead, we believe that this flange is the counterpart of that in the multituberculate petrosals described above that defines the perilymphatic foramen and separates the scala tympani from the perilymphatic recess. If so, the secondary tympanic membrane in Morganucodon would have been medial to the perilymphatic recess and sac, just as in other amniotes, and would have attached to the poorly defined external edges of the recess found there, not to the flange. In Ornithorhynchus, it is the lateral aperture of the perilymphatic recess, not the perilymphatic foramen, that is closed by the secondary tympanic membrane (Gray, 1908a; Kuhn, 197 1; Zeller, 1991; see below), and the osseous rim that supports the membrane is poorly developed (Zeller, 1993: 10 1)-just as we hypothesize for Morganucodon. In multituberculates, the lateral aperture (traditionally called the ‘perilymphatic foramen’, ‘fenestra cochleae’ or ‘round window’ by authors, but incorrectly) is better defined, encircled by a complete rim of bone (Fig. 8B, D). According to. our determination of homologies, this opening must have been closed by the secondary tympanic membrane, with the perilymphatic foramen and perilymphatic recess located more internally in the petrosal than in Mo?ganucodon and Ornithorhynchus. In contrast to the assumption that Kermack et al. (1 98 1) made in respect to Morganucodon, attachment of the secondary tympanic membrane may or may not leave distinctive marks on the bone. Two of the Bug Creek petrosals described above, UALVP 26037 and 26039, show the development of a narrow groove along the ventral part of the rim of the lateral aperture that splits the rim into two parallel flanges, perhaps evidence of a secondary tympanic membrane attaching there. However, other parts of the rim in these specimens are smoothly rounded, and the ventral rim itself is smoothly rounded in UALVP 34 144, also a well-preserved specimen: no flanges are developed. In Didebhis (UALVP unnumbered specimens), the rim of the fenestra cochleae is smoothly rounded in its entirety, and the bone lacks a distinctive configuration that might be taken as the site of attachment of the secondary tympanic membrane, even though a membrane is present in Didebhir, as in other living therians. As noted above, the same is true for the lateral aperture of the perilymphatic recess in Ornithorhynchus: no distinct osseous crest or rim for attachment of the membrane has been observed (Fig. 8C). Consequently, the lack of a flange along parts of the rim of what we identify as the lateral aperture can not qualify as evidence that a secondary tympanic membrane did not attach to this rim in Morganucodon nor in multituberculates. Conversely, the three-dimensional curvature of the internal ridge that we believe defines the perilymphatic foramen and recess internally in the petrosal of multituberculates precludes it from supporting the secondary tympanic membrane, which need have been held in only a single plane, as by the rim of the lateral aperture. If our interpretation of the attachment of the secondary tympanic membrane in Morganucodon and multituberculates is correct, the primary perilymphatic duct was “70 R. C. FOX KYD J. MENG short in these forms, as it is in monotremes. The path for the secondary perilymphatic duct as it leaves the petrosal is clear in Morganucodon and multituberculates, following the groove that arises at the antermedial extremity of the perilymphatic recess; its location as it leaves the recess is comparable to that of the cochlear (internal) orifice of the aquaeductus cochleae in Late Cretaceous therians (Meng & Fox, 1995b, fig. 1; Fig. 8A), a point of significance in reconstructing the evolution of the therian pattern. The therian oriface is at the medial end of a ridge that reflects the position of the perilymphatic foramen within the cochlea (marked by a black square in fig. 812) hetween the scala tympani and the perilymphatic recess. According to Gra). ( I 908a: 523), there is little development of a clear-cut boundary between cochlea (scala tympani) and perilymphatic recess in living therians, and the recess is indistinct, represented only by what he called a ‘cone-shaped bulging’ (Gray, 1908a); we havc extended Gray’s observations to Late Cretaceous therians, in which we have identified a cone-shaped swelling, as well (Meng & Fox, 1995b: figs 1, 4). Because of their comparable positions in relation to the perilymphatic foramen, the therian fenestra cochleae and the aqueductus cochleae are readily derivable evolutionarily from division of the lateral aperture of the perilymphatic recess (as the recess is seen, for example, in multituberculates) by formation of a processus recessus, the latter documented in the ontogeny of the therian petrosal (e.g. de Beer, 1929, 1937; Zeller, 1993). This transformation series from the fossil evidence gains support from at least two forms that may show morphologically intermediate stages: Encelestes and an unnamed taxon represented by a newly described, isolated petrosal from the Khoobur locality, Lower Cretaceous, Mongolia (Rougier et al., 1992; Wible et al., 1995; Rougier et al., 1996); in both, a fenestra cgchleae and aquaeductus cochleae arc developed (we note, however, that the fenestra cochleae in the Khoobur petrosal was described as the perilymphatic foramen [Wible et al., 19951)but an aquaeductus cochleae has since been identified in this specimen [Rougier et al., 19961). In the Khoobur petrosal, the perilymphatic recess is completely contained within the petrosal, but a spiral lamina accommodating the cochlear nerve had not yet developed (personal observation, JM) and the cochlea is straight (Wible et al., 1995: 3 ) . Presence of the secondary spiral lamina has been cited in Hincelestes (Wible et al., 1995: appendices 1 and 2), but evidence for it and the morphology of the inner ear more generally are undescribed. Hence, the fossil evidence of early mammals is consistent with de Beer’s (1937) conclusion from ontogenetic studies that part of the latcral aperture of the perilymphatic recess (not the perilymphatic foramen) in nonthcrian mammals becomes the fenestra cochleae of therians, and that the secondar)tympanic membrane covering the aperture or its derivative fenestra cochleae is homologous in these forms. Our choice of lizards and crocodilians as an out-group to establish the primitive amniote and primitive mammalian sequence of perilymphatic recess-perilymphatic foramen-cochlear canal leads to an additional question, however. Is the lateral apcrture of the perilymphatic recess in primitive mammals homologous to the ‘round window’ of non-synapsid amniotes and, hence, is the secondary tympanic membrane homologous among all amniotes, as Zeller (1985a, b, 1993) has maintained? M’e helie1.e not, with the substance of our conclusion based on evidence from the fossil record of early amniotes. That record indicates, it is now generally agreed, that the most primitive synapsids, the Palaeozoic pelycosaurs, probably lacked a tympanic mcmbrane, and if capable of hearing airborne sounds at all, these animals could have transmitted only low frequencies to the otic capsule, via the intervention of INNER EAR OF MULTITUBEKCUIA’rES 25 1 unspecialized surface tissues activating the massive stapes (Allin, 1975, 1986; Allin & Hopson, 1992). Indeed, the bones intervening between any superficial soft tissues in the future tympanic area and the stapes seem themselves incapable of independent movement, thereby further limiting the potential for stapedial activation in sound transmission (Allin & Hopson, 1992). Pelycosaurs lacked a lateral aperture that would have supported a secondary tympanic membrane and lacked the perilymphatic foramen (Romer & Price, 1940; Olson, 1944; Allin & Hopson, 1992), possessing only a jugular foramen for the exit of cranial nerves IX, X, and XI and the internal jugular vein (Romer & Price, 1940; Romer, 1956). In primitive cynodonts (e.g. 7hrinaxodon and Probainognathus; see Allin & Hopson, 1992: fig. 28.8; Luo et al., 1995: fig. lA), the pelycosaur pattern in the proximal perilymphatic system appears to have been maintained: a separate perilymphatic foramen is not developed. In a young specimen of lhrinaxodon liorhinus, Estes (1961: fig. 3) restored the major parts of the inner ear as if they were in place, with the ‘fenestra rotunda’ at the end of a ‘distinct channel’ opening within the walls of the jugular foramen (p. 172). While cynodonts at the Thinaxadon stage of evolution probably were capable of hearing air-borne sounds via a specialized tympanum and moveable connecting bones (AUin, 1975, 1986; Allin & Hopson, 1992), Estes’ (1961) restoration suggests that the secondary tympanic membrane was probably not developed as no part of the perilymphatic system seems to have abutted against the air-filled tympanic cavity. Among pre-mammalian synapsids, tritheledonts and tritylodonts are the first to resemble mammals explicitly in the history of the proximal perilymphatic system. They display an opening adjacent to but not confluent with the jugular foramen, although the two openings are closer to one another than in early mammals (Allin & Hopson, 1992; Crompton & Luo, 1993; Lucas & Luo, 1993). Olkobphus well represents this stage: both the new opening and the jugular foramen are completely encircled by bone, but they are housed in a common recess with only a slender osseous bar between them (Crompton, 1964: fig. 8; Luo, 1994: fig. 6.3D). Th‘is new opening is the perilymphatic foramen, opening in the wall of the auditory capsule (de Beer, 1937), although it has been mistakenly termed ‘round window’ by some workers (Luo, 1994; Luo et al., 1995). Relying on the pattern in lizards as a model (e.g. Wever, 1978: 72-74; fig. 3-13), one can envisage in Oligokyphus a perilymphatic sac held in the common recess that communicates with the scala tympani through the perilymphatic foramen laterally, expands into the cranial cavity through the jugular foramen medially, and closes the opening for the common recess ventrally. Because the perilymphatic and jugular foramina are so near to each other in Oligokyphus, development of a secondary perilymphatic duct seems not necessary (see e.g. Luo, 1994: fig. 6.3D). Continued expansion of the otic capsule and basicranium with refinement of hearing and increasing size of the brain would have resulted in further spatial separation of the perilymphatic foramen from the jugular foramen, as seen, for instance, in Morganucodon and other early mammals (Kermack et al., 1981; Crompton & Luo, 1993; Lucas & Luo, 1993; Luo et al., 1995). In those groups having the two foramina more distantly separated than does Oligokyihus (perhaps accompanied by reduction of the perilymphatic sac), the secondary perilymphatic duct was developed and a groove holding it was formed. These definitely include morganucodontids (Kermack et al., 198l), Haldanodon (Lillegraven & Krusat, 1991: fig. 8), triconodontids (Wible & Hopson, 1993: fig. 5.3B), multituberculates (KielanJaworowska et al., 1986; Luo, 1989; this paper), and monotremes (Wible & Hopson, 272 K C:. FOX AND J RfENG 1993);whether or not a groove for the duct is present in tritheledontids, tritylodontids. Adelobasileus, or Sznoconodon has been disputed (sec Rougier et al. [1996]), but does not altcr the sequence of morphological events established here in our attempt to determine the homologies of the secondary tympanic membrane among amniotes. In summary, the major features of the evolution of the proximal perilymphatic qystcm from pelycosaurs to early mammals when interpreted in a context of comparative information from living amniotes seem straightforward and explain the nature of thc perilymphatic foramen, the lateral aperture of the perilymphatic recess, the fenestra cochleae, and (implied) secondary tympanic membrane: we conclude that the ‘mammal-like’ aspects of the perilymphatic system in non-synapsid amniotes are homoplasious, but how satisfactorily does this account apply to living mammals? The only problematic living mammals in this respect are monotremes. In TactyglossuJ, a superficially therian-like, partial ‘aquaeductus cochleae’ is developed that contains the secondary perilymphatic duct, but it also contains the glossopharyngeal newe. This pattern has clearly evolved independently of that in therians, in which the true aquaeductus cochleae contains only the secondary perilymphatic duct, not the nerve (Kuhn, 1971; Gauthier et al., 1988; Zeller, 1993). Quite different is thc arrangement in Ornithorhynchus, in which the secondary perilymphatic duct is not yet enclosed in a bony aquaeductus (resembling in that respect multituberculates and other non-therian mammals), and the perilymphatic recess is little ossified (in contrast to that in multituberculates and presumably more primitive). We agree with Zeller (1 993)) however, that the secondary tympanic membrane is homologous in living monotremes, irrespective of ontogenetic differences in the development of the lateral aperture of the perilymphatic recess and position of the secondary perilymphatic duct in Tuchyglossus and Ornithorhynchus. At present, the fossil record can make no contribution to understanding of these patterns. Szzt) ofthe vestibule It is nobv well known that in several multituberculate genera the vestibule of the labyrinth is enlarged (Miao, 1988; Luo, 1989). It is significantly larger relative to petrosal size and inferred skull size than that in monotremes, therians and lMorganucodon (Luo & Ketten, 1991: fig. 4)) and enlargement of the vestibule has been interpreted dS a possible multituberculate synapomorphy (Luo, 1989; Luo & Krtten, 1991; Simmons, 1993). An extreme expression of this specialization is seen in the earl) Tertiary multituberculate Lambdopsalzs (Miao, 1988; Meng & Wyss, 1995). Indeed, the vestibule in Lambdopsalis is so inflated that it was originally identified as the tvmpanic bulla, from which the specific name, L. bulla, is derived (Chow & Qi, 1978). Lillegraven & Hahn (1 993) have shown, however, that the vestibule is not inflated in the Late Jurassic Paulchoffatiidae, the most primitive multituberculates known. As noted above, the vestibule is not inflated in UALVP 34144, either, and in this respect the specimen resembles the petrosal of paulchoffatiids and the Paleocene ptilodontoid Ptzlodus montanus (Simpson, 1937). In sum, the distribution of an inflated vestibule among multituberculates suggests that inflation is a derived condition within the order, and that it probably originated more than once, independently, given the phylogeny of multituberculates (Simmons, 1993). As a consequence of inflation, the lateral semicircular canal can be embedded in the poyterior wall of the vestibule, as in UALVP 26039, or can even become confluent with the vestibule, as in LambdopsaliJ (Miao, 1988; Meng & Wyss, 1995). INNER EAR OF MULTITUBERCULATES 273 Inflation of the vestibule mostly reflects inflation of the sacculus and utriculus. These are the two primary sacs that occupied the vestibule in life, and the available evidence indicates that they filled the entire vestibule in multituberculates, leaving no significant parts of it free. Luo & Ketten (1991: 225) were not certain of the relative sizes of the two sacs in multituberculates, stating: “It remains unclear from our CT scans of the petrosal and previous studies (Kielan-Jaworowska et al., 1986; Miao, 1988; Luo, 1989) whether the vestibular inflation of multituberculates relates to the enlargement of the saccule, the utricle, or both”. It is indeed difficult to make this assessment from only osseous remains, but it is not impossible given the opportunity for direct visual inspection of the interior of the petrosal, as we have already shown. Based on our own observations, it is certain that the utriculus was the major occupant of the inflated vestibule in multituberculates, as it appears to be in at least Tachyglossus among monotremes (Griffiths, 1968; Jmgensen & Locket, 1995; although see Gray [1908a: 5 191 for an opposing view). However, because the condition in potential outgroups within the Mammalia, including Adelobasileus, Sinoconodon and Morganucodon, is unknown (Luo & Ketten, 1991; Luo et al., 1995), the polarity of this character cannot be established at present. Curvature of cochlear canal In most reptiles, the cochlear canal (also termed cochleolagenar canal, cochleolagenar recess, or cochlear recess) is short and is ventral to the vestibule (Miller, 1966, 1968; Baird, 1970, 1974; Wever, 1978; Lewis et al., 1985). The canal in crocodilians is somewhat elongate and bends anteroventrally (Wever, 1967, 1978; Baird, 1970; Manley, 1970). In birds, it is also variably elongate but bends medially, in a direction opposite to that in monotremes and therians (Pritchard, 1881). In cynodonts, the cochlear canal is usually short and is located anteroventral to the vestibule (Estes, 1961; Quiroga, 1979; Allin, 1986; Allin & Hopson, 1992; Luo et al., 1995), although Simpson (1933: 288-9) reconstructed the cynodont ear with a more elongate cochlea that extends “downward and forward from the floor of the vestibule, very much as in Ornithorhynchus”. The cochlear canal of primitive mammals has been most extensively studied in Morganucodon. In Morganucodon it resembles that of multituberculates in several respects: Graybeal et al. (1989) found the canal to be “a slightly curved tube” (p. 112), ‘%-tually straight tube” (p. 113), “straight” (p. 114), or “had a very slight curve” (p. 115), while Kermack et al. (1981) described it as “almost straight”. Rosowski & Graybeal (1991) portrayed the Morganucodon canal as similar to that in birds and crocodilians, and Graybeal et al. (1989: 1 14) characterized it in much the same way. The special resemblances that these authors noted are in its shortness and relative straightness, and in its lack of bony lamina (Rosowski & Graybeal, 1991: 138)features, however, that are also seen in multituberculates. Although none of the above authors have explicitly stated the direction in which the cochlear canal curves in Molganucodon, it seems to be laterad (see especially Rosowski & Graybeal, 1991: fig. 3B, D), as in other primitive mammals but not reptiles or birds. Knowledge of inner ear morphology in the most primitive mammals has been amplified significantly by recent description of the petrosal in Sznoco~odon(Luo et al., 1995). In this genus, the cochlear canal is still more primitive in certain respects than in Morganucodon. The Sinoconodon canal is about half the length of that in Morganucodon, and is straight proximally but is slightly curved laterally towards the R. C . FOX AND J. MENG 2i-I TARI,E I . Comparative measurements of multituberculate petrosals, Bug Creek Anthills locality. :1, length of cochlea; B, long axis of fenestra cochleae; C, short axis of fenestra cochleae; D; long axis of fenrstra vestihuli; E, short axis of fcnestra vestibuli; F, diameter of crus commune; G. diametrr of semicircular canal; H, radius of lateral semicircular canal; I, diameter of cochlear canal. Aatrrisks indicate measurements from X-radiographs A UALYP 26032 LALVP 2603-1 L.ILVP 26035 UALYP 26036 UALVP 2603 7 L-AI,vP 26039 L-~AI,\’P31111 H c: u c F 0.62 3.3* +.9* 0.78 4.5* 0.99 1.311 0.54 1.2 0.88 0.53 0.87 H 1 4.0* 2.j* 1.-17* 0.22 0.53 0.86 G 0.66 0.55 0.22 0.33 0.3 1.1:3* terminus, more so than in Murganucodon (Luo et al., 1995: 118, table 1). Thc configuration of the cochlear canal is unknown in Adelobasileus, purportedly the earlicst known mammal (Lucas & Hunt, 1990; Lucas & Luo, 1993; Luo et al., 1995). O u r demonstration that in multituberculate petrosals having otherwise different morphologies (see e.g. Fig. 2), the cochlear canal extends anteriorly in a nearly horizontal plane and curves slightly laterally clearly differs from obsemations reported by Sloan (1979), Mia0 (1988) and Luo & Ketten (1991). According to Sloan (1979), the cochlear canal is ‘hooked’ in the early Tertiary neoplagiaulacid Ecppudus (but see Miao [1988] for reservations), whereas a ‘rod-like and straight’ cochlea has been cited for Catopsalis, ?Meniscoessus and Lumbdopsalis (Miao, 1988; Luo & Ketten, 1991 ; and see Luo et al., 1995). Our observations are more in keeping with those of Simpson (1937: 751) on Ptiludus, in which a slightly curved cochlear canal appears to have been developed. In all known multituberculates in which the cochlea is definitely curved, the direction of curvature matches that in other mammals, i.e. clockwise in the right ear and counterclockwise in the left, but the degree of curvature in multituberculates appears to differ among different taxa. From what is known so far, we suspect that lateral cun7ature of the cochlea may be a synapomorphy of thc Mammalia (see e.g. Lillegraven & Krusat [1991] for ~ u ~ d u n o ~but u ~ )any , final conclusion along these lines must await clarification of cochlear morpholo9 in Ade&basileus and in a broader array of primitive mammals, including primiti\.e multituberculates, than is presently available. The degree of curvature of the cochlea in monotremes has been uncertain, with conflicting accounts recorded in the literature, as we have discussed elsewhere (Meng & Fox, 1995b). Kermack et al. (198 I), citing Denker (1 90 l), reported that the cochlea curves 180” in Tacachygossus and 270” in Ornithorhynchus, and this conclusion has been widely accepted since (e.g. Kerinack & Mussett, 1983; Wible, 1990; Luo & Ketten, 1991; Allin & Hopson, 1992). However, in describing the cochlea in Ornithor/ynchuJ, Pritchard (1881: 267) had long before observed: “This tube is directed almost horizontally forwards and is slightly curved outward, its apex forming a little projection on the outer border of the petrous bone close to its pointed end”, and, as illustrated (Pritchard, 1881: pl. 45I), the cochlea is only slightly cuwed, about 180”. Gray (1908a: 5 18)made much the same observation: “In reality, the labyrinth of the echidna presents in this, as in other aspects, a very considerable advance upon that of the platypus, in which there is only the slightest degrec of curnature INNER EAR OF MULTITUBERCULATES 275 in the cochlea, with a bulbous knob on the tip of the organ”. But according to Griffiths (1978: 189), in Ornithorhynchus “the cochlea exhibits a three-quarter spiral turn and terminates in a lagena with lagenar macula” (and see J~rgensen& Locket, 1995: fig. l), while Zeller (1993: 101) reported that “the monotreme cochlear duct is only slightly curved; its rostra1 end, containing the papilla lagenae, is sickle shaped”. The disagreement about curvature in the monotreme cochlea may well have its source at least partly in what have been different meanings for the term ‘cochlea’ by different authors. The term sometimes seems to refer to the membranous structure that consists of the endolymphatic and perilymphatic ducts (the soft tissue cochlear duct and scala vestibuli and tympani), at other times to the cochlear canal (the bony housing for the ducts) or even to both, in different publications. Here we attempt to clarify these differences so that the nature and significance of our own observations are clear. In mammals, the cochlear duct is in indivisible association with the scalae as parts of the membranous labyrinth, and is contained within the osseous cochlear canal, a part of the bony labyrinth. In monotremes, two lines of evidence indicate that the soft membranous ducts may have a greater curvature than does the osseous canal in which they are housed. First, Luo & Ketten (1991: 226) have shown that in the monotreme inner ear “the bony cochlear cavity does not coil in correspondence to the coiled membranous labyrinth”, an observation which obviously contradicts that of Pritchard (188 1). Second, with decalcification of the petrosal in preparing a specimen for examination of the soft tissues within, it is possible that the membranous ducts come to display a variable degree of curvature, even greater than that of the cochlear canal itself, but only as a consequence of tissue distortion during the decalcification technique; in specimens that have undergone this treatment to enhance the visibility of the membranous ducts, it is impossible to tell with certainty what the original curvature of the ducts within the canal might have been. In any case, no evidence exists to indicate what the degree of curvature of the membranous ducts within the osseous canal actually was in multituberculates or any other early mammals (cf. Wible & Hopson, 1993: 54): curvature of the ducts may have matched that of the canal or it may have exceeded it, given the observations of Luo & Ketten (199 1) for monotremes. This leads us to conclude that any meaningful comparison in such studies must be between the same structures, and that can only be a comparison of the osseous canal between extant and fossil forms. During the evolution of therians, the cochlear canal probably achieved its first full turn and then subsequent turns only gradually, with early steps in this transformation being documented by the Early Cretaceous pre-tribosphenic therian Encelestes (270” curvature: Rougier, 1990; Rougier et al., 1992) and therian petrosals from Late Cretaceous species (slightly more than 360” curvature: Meng & Fox, 1993, 1995a, b). In all extant therians, the membranous ducts spiral through at least one and a half turns (Fernandez, 1952; Fernandez & Schmidt, 1963; von BCkksy, 1960; Pye, 1966a, b; Wever, 1974; Wever et al., 197 1, 1972; Fleischer, 1976; Bruns, 1976; Aitkin et al., 1979; Lewis et al., 1985), and the only living therians that are known to have as few as one and a half turns are the marsupial mole, hedgehog and sea-cow (Gray, 1980b; Lewis et al., 1985). In Mesozoic therians for which the internal cochlear structure is known, there is probably a closer correspondence between curvature of the membranous ducts and curvature of the osseous canal than in non-therians owing to the support of the ducts in therians by the osseous laminae that arise from the walls of the canal. The origin of these laminae within 276 R. C. FOX AVD J. MENG the Theria has, however, yet to be determined, with observations and descriptions of them from no species older than Late Cretaceous (Meng & Fox, 1995a, b). hngth of cochlea Obviously, the length of the cochlear canal is an important aspect of cochlear size in a functional context, but it is of significance phylogenetically, as well. Because the cochlear cavity is proportionally short and straight in extant diapsid reptiles (Oelrich, 1956; Wever, 1978; Bellairs & Kamal, 1981) and in therapsids (Olson, 1944; Estes, 1961; Allin, 1986; Allin & Hopson, 1992), Luo & Ketten (1991) and Luo et al. (1995: table 1) concluded that the cochlear canal in multituberculates and Morganucodon is longer (relative to skull length) and thus more derived than the canal in therapsid outgroups. However, we have found that comparisons of the length of the cochlear canal in previous work (e.g. Graybeal et al., 1989; Luo & Ketten, 1991; Rosowski & Graybeal, 1991) are sometimes oflimited usefulness because of differences in identifying the point at which the canal begins (and see Luo et al. [1995: table 11). In our studies of the inner ear in Late Cretaceous therians, this point was taken to be at the inflection of the osseous spiral lamina anterior to the fenestra cochleae, where the cochlear duct begins its spiral course (Meng & Fox, 1995a, b). Because the cochlear canal in multituberculates has no osseous lamina, we have objectively defined the starting point of the cochlear duct as at the point where the secondary perilymphatic duct notches the rim of the perilymphatic foramen: this point most closely corresponds to the position of the aquaeductus cochleae at the base of the scala tympani in therians. For Sinoconodon, Luo et al. (1995: 1 18) measured the length of the canal from the anterior rim of the fenestra vestibuli to the terminus, a somewhat different dimension from ours in that it includes part of the vestibule. A potentially more important statistic than the raw measurement of the length of the cochlear canal is its length relative to skull length (Luo et al., 1995: table 1)-although in the present context this is of limited use owing to the lack of association of petrosals with complete skulls in our sample and, indeed, for most early mammals. The length of the cochlear canal in modern therians corresponds directly to the length of the basilar membrane because the lagena is no longer present as a distal extension of the cochlear duct in these mammals. This relationship between canal and membrane is assumed to hold as well in the Late Cretaceous therians that we investigated previously (Meng & Fox, 1995a, b). In other early mammals, such as Morganucodon (Graybeal et al., 1989) and multituberculates, the length of the cochlea must also have approximated the length of the basilar membrane, but this approximation was probably less precise than that in therians because of the presence of a lagena (see below). Pattern of innervation of the cochlea Previous work has indicated that in primitive mammals, such as Morganucodon (Kermack et al., 198 l), Haldanodon (Lillegraven & Krusat, 199 1) and multituberculates (Kielan-Jaworowska et al., 1986; Luo, 1989; Lillegraven & Hahn, 1993; Figs 1, 3, 9B), the cochlear division of the auditory nerve passes through the petrosal wall via a single foramen, thereby contrasting sharply with the condition in extant therians (Lorente de NO, 1937; Bast & Anson, 1949; Gray, 1959; Bredberg, 1968; Spoendlin, 1972, 1974; Sando, 1965). In even the earliest tribosphenic therians for which evidence is available (Meng & Fox, 1995a, b, c), the cochlear nerve branches into INNER EAR OF MULTITUBERCULATES 277 Figure 9. SEM photographs showing the cribriform plate of a therian (A) and the single foramen of a multituberculate (B) for the cochlear nerve in the internal acoustic meatus. Not to scale. See Abbreviations for key. COC bm C A B c Figure 10. Diagrams showing cochlear innervation patterns. A, a hypothetical parallel innervation pattern for a coiled cochlea; B, a parallel innervation pattern presumably present in early mammals, including multituberculates; C, a radial pattern in therians. See text for discussion. See Abbreviations for key. many small rami that pass through the numerous tubules (the tractus spiralis foraminosis) of the modiolus (Fig. 9A), forming a radial pattern matching that in extant therians. The cochlear nerve in adult monotremes resembles its homologue in therians in that it passes through a cribriform plate (the area cribosa anterior of Alexander [ 19041 and Gaupp [1908a, b]) in the floor of the internal acoustic meatus (Simpson, 1938; Kermack et al., 1981; Fig. 7). Once within the canal, the newe differs in its orientation from that in therians. As Pritchard (1881: 275) noted, the nerve in the therian spiral cochlea “necessarily runs at right angles to the lamina spiralis, [whereas] in this cochlea [of platypus] it runs parallel to the corresponding lamina” (see our description of the apparent nerve pattern in a specimen of Ornithorhynchus -above, however). We term the latter configuration of the nerve the ‘parallel pattern’ or ‘fanned pattern’ (Fig. 10B), and given the short, relatively straight cochlear canal of multituberculates, the fibres of the cochlear nerve must also have followed a similar pattern in that group. This pattern may well be less efficient in certain respects than 278 R C FOX XYDJ MENG the radial pattern of therians, especially in innervating an elongate organ of Corti. It is likely that the radial pattern of the cochlear nerve in therians functions to minimize differences in length required of individual fibres that pass to different parts along the basilar membrane (West, 1985; Meng & Fox, 1995a, b, c). In therians in which the basilar membrane is elongate, a parallel pattern of the cochlear nerve fibres, as in monotremes (Pritchard, 1881)-and presumably, from the shape of the cochlear canal, in Morganucodon, Haldanodon and multituberculates as well--would require that fibres innervating hair cells distally would need to be significantly longer than those passing to proximal cells, nearer to the base of the membrane (Fig. 1OA). Therefore, for those mammals in which the cochlear canal is elongate and has several spirals (over four spirals in the guinea-pig, for example [Manley, 1972; Lewis et al., 1985]), this pattern would result in a significant difference in conduction time of the nerve impulse from the two ends of the basilar membrane to the auditory cortex of the brain. Such a pattern, however, is not present in reality. A radial pattern of the cochlear nerve in all tribosphenic therians minimizes this difference, d hypothesis to be further tested. The blood Lressels that supply the cochlear ganglion and its cell bodies housed in the spiral canal of the modiolus adopt a different pattern in therians. As shown by Axelsson (1 974), the main trunk of the spiral modiolar artery spirals along with thc cochlea, between the acoustic nerve centrally and the spiral ganglion peripherally. The spiral modiolar vein follows a similar route. The size of the spiral modiolar artery gradually diminishes as it gives off branches on its way from the base to the apex of the cochlea. Ikgmi Monotremes are the only living mammals known that possess a lagena at the distal end of the cochlear duct, a specialization comparable to that in reptiles and birds (Pritchard, 1881; Denker, 1901; Alexander, 1904; Gray, 1908a, b; Fernandez 8L Schmidt, 1963; Griffiths, 1968, 1978;Jorgensen & Locket, 1995). Manley (1972: 61 1) thought that “the cochlear canals of early mammals, which probably also contained a lagena macula, were very short e.g. 3-4mm in Triconodon mordux (Kermack, 1963)”.Although this statement seems likely to be a valid characterization of the cochlea in early mammals-at least for non-therians-no evidence of a lagena in Triconodon is actually known, nor has any structure that suggests its presence been previously found in Morganucodon (Graybeal et al., 1989), Sinoconodon (Luo et al., 1995), or any other early mammals. Accordingly, it is of great interest that the X-radiograph of at least one of our multituberculate specimens (UALVP 26039) unambiguously displays an inflated terminus of the cochlear canal, comparable to that of monotremes. \Ye believe that this structure provides the first evidence of the lagena in mammals other than monotremes; no other realistic interpretation of the structure suggests itself. The other specimens (UALVP 34144, 26037) do not show a terminus that is expanded to the same degree, but while the distal end of the canal ends abruptly in these specimens, at a rounded tip, it is nonetheless slightly expanded relative to thc proximally adjacent parts of the canal. Previous studies (Luo & Ketten, 1991: fig. 3) have indicated that distally the cochlear canal narrows, rather than expands, in certain multituberculates as well as in monotremes; therefore, if a lagena was actually present in these multituberculates, it is either not reflected osseously by an inflated cochlear terminus or the method of reconstruction of the canal based on serial sections has been too coarse to capture details of its actual shape. INNER EAR OF MULTITUBERCULATES 279 PHYLOGENETIC IMPLICATIONS At present, no consensus concerning the phylogenetic position of multituberculates exists among specialists who have recently concerned themselves with the question (Kemp, 1983; Miao, 1988, 1991; Rowe, 1988, 1993; Wible, 1991; Kielan-Jaworowska, 1992; Sereno & McKenna, 1995). Postcranial resemblances have been cited in support of a multituberculate-therian sister-group relationship on the one hand (but see Kielan-Jaworowska & Gambaryan [1994] for an alternative), whereas dental features have been invoked in support of a monotreme-therian sister-group relationship to the exclusion of multituberculates on the other (Archer et al., 1985). As a further complicating factor, the monotreme-multituberculate affinities originally suggested by Broom (1914), and then elaborated on by Kermack & KielanJaworowska (1971) from resemblances in structure of the braincase, have gained renewed support from recent work (Hopson & Rougier, 1993; Wible & Hopson, 1993; Meng & Wyss, 1995). Given this current diversity of opinion, which stems ultimately from different speculations about which morphological similarities are homologous and which are homoplasious in the absence of a definitive fossil record connecting multituberculates to other mammals, the relationships for multituberculates remain unresolved (Rougier et al., 1992; Wible et al., 1995). It is in this context that we discuss the phylogenetic implications of two features of the multituberculate ear examined in our study, the curvature of the cochlea and the inflation of the vestibule, that previous authors have concluded were informative in determining the relationships of multituberculates. The configuration of the cochlea is a character sometimes thought to be important in determining the interrelationships of the major mammalian lineages. A cochlea having a curvature of greater than 180” has been cited as support for the hypothesis that monotremes and therians are sister-groups (Luo & Ketten, 199l), or as a trait that is at least consistent with a monotreme-multituberculate-therianclade (Rowe, 1988). However, either a reversal in multituberculates to a short, uncoiled cochlea is required by Rowe’s analysis, or coiling of the cochlea in monotremes and higher therians is not homologous, the latter evidently being Rowe’s preference p.Rowe, pers. comm. in Luo & Ketten, 1991: 2251; and see below). Unexpectedly, and in contrast to the conclusions of previous workers (e.g. Luo & Ketten, 1991), our study shows that the osseous cochlear canal of Ornithorhynchus is in fact less curved than that of Tachyglussus (Fig. G), while at the same time it is not significantly more curved than that in at least some multituberculates (compare e.g. Figs 2 and 6). Given that the inner ear of early monotremes (Archer et al., 1985, 1992; Pascual et al., 1992; Flannery et al., 1995) and that of therians more primitive than Encelestes is unknown, we caution that the curvature of the cochlear canal may be an unreliable character for determination of phylogenetic relationships between these groups. Furthermore, from the evidence of a lagena-like expansion of the cochlear canal in the multituberculate petrosals described above, the absence ofa lagena can not be interpreted either as a derived feature uniting all multituberculates or linking multituberculates with other mammals that lack a lagena. By contrast, the cribriform passage for the cochlear nerve in monotremes is a condition anatomically more similar to the therian pattern than to that in multituberculates, although it is preceded by a “more united foramen” ontogenetically in monotremes (Simpson, 1938: 1 l), and hence, may not be homologous to the pattern in therians. The peculiar curvature of the cochlear duct in Ornithurhynchusas discussed by Luo 280 R. C. FOX Ah’D J. MENG & Ketten (1991) raises the question as to whether the curvature is homologous to the coiling in therians (Zeller, 1989), especially since the duct in Tuc’achyglossusis not curved the same way as has been purported for that in Ornithorhynchus, while the duct in therians always coils in correspondence with the osseous cochlear canal (Gray, 1908a, b; Fernandez & Schmidt, 1963; Lewis et al., 1985). From these differences, we conclude that curvature of the cochlear duct in Omithorhynchus is unlikely a precursor of the coiled duct in therians, and that, instead, the monotreme condition is apomorphous, and hence not validly invoked as support of monotremetherian relationships. In addition, our X-radiographs indicate that the diameter of the cochlear canal in neither Ornithorhynchus nor Tachyglossus changes substantially along its length proximal to the lagenar expansion, contrary to Luo & Ketten (1 99 1: fig. 3). The 270” bend of the cochlear duct within the narrow osseous canal in Ornithorhynchus, as observed by Luo & Ketten, is abrupt, an unexpected configuration for a hearing function that is dependent on vibrations of the basilar membrane. The question of the propagation of travelling waves along a basilar membrane having these properties remains unanswered; no comparable pattern is known in other vertebrates (Wever, 1978, 1985; Lewis et al., 1985; Manley, 1990). The second feature of the petrosal sometimes thought to be significant in determining the relationships of multituberculates-inflation of the vestibuleappears from our study to be a derived state, occurring only in some advanced species, and hence not indicative of monophyly of the order. This conclusion is strengthened by its agreement with the size distribution of the vestibule in more primitive mammals and in their therapsid predecessors. Although Luo & Ketten (1991: 226), citing earlier work (Olson, 1944; Quiroga, 1979; Bellairs & Kamal, 1981; Allin, 1986), observed that the vestibule in therapsids and diapsid reptiles is relatively small, this comparison is valid only when it is made with those advanced multituberculates in which the vestibule is clearly inflated. Furthermore, it need be remembered that the size of the vestibule in therapsids relative to that in mammals (including Morganucodon, early multituberculates, monotremes and therians) is unknown; no broad-spectrum comparisons of this kind have yet been undertaken. The vestibule in primitive synapsids has been characterized as ‘very large’, but its apparent size in these groups may be misleading, at least in part as the vestibule was probably lined with ‘substantial amounts’ of cartilage in life, a tissue that is not normally preserved by the processes of fossilization (Allin & Hopson, 1992: 606). Similarly, the apparent confluence of the cavity of the horizontal semicircular canal with the vestibule in these animals may be only an artifact of the failure of cartilage lining the vestibular walls to fossilize (Allin & Hopson, 1992: 606). A different view of the significance of vestibular size to the analysis of early mammalian interrelationships was furnished by Gray (1 908a: 520), in comparing monotremes and extant reptiles: “In one respect, however, the vestibule of the echidna’s labyrinth is distinctly more like that of mammals than that of at least the majority of reptiles. Its size, relative to the rest of the organ, is small, whereas in reptiles the vestibule is frequently the bulkiest portion of the whole labyrinth and occupies almost all the space contained by the planes of the three canals”. In this perspective, the vestibule of Morgunucodon is mammal-like in its small size (Graybeal ef d., 1989; Luo & Ketten, 1991) and in its freedom from the semicircular canals except at their terminal openings. A recent reconstruction of the endocast of the inner ear from serial sections indicates that the vestibule in Sinoconodon is similar to that in Morgunucodon in these regards (Luo et al., 1995). Consequently, if a small INNER EAR OF MULTITUBERCULATES 28 1 vestibule and free semicircular canals are mammalian characters and Sinoconodon plus Morganucodon comprise the outgroup of multituberculates, monotremes and therians, two distributional patterns of vestibular morphology suggest themselves: first, the vestibule in multituberculates evolved from a small, uninflated ancestral state, as exemplified by Sinoconodon and Morganucodon, implying that the common ancestry of multituberculates, monotremes and therians had a small, uninflated vestibule, as well. Alternatively, the small, uninflated vestibule of monotremes and therians is the consequence of a reversal from an inflated vestibule in the predecessors of these two clades-either in their immediate common ancestry, if monotremes and therians are sister-groups (a relationship that from present' evidence seems unlikely; see e.g. Flannery et al., 1995), or at a more remote 'node' in the central mammalian stem, or in the therapsid ancestors of mammals. We conclude that an uninflated, Morganucodon-like vestibule marked the earliest mammals and was retained in the earliest multituberculates-this is the more realistic hypothesis biologically (functionally), and is in keeping with our earlier argument from different evidence that an inflated vestibule is a uniquely derived feature only for one or more subgroups within the Multituberculata. Figure 11 summarizes the distributions of inner ear features, based on this and our previous studies. We have found no derived characters of the inner ear that indicate a preferred position of multituberculates among mammals nor that ally monotremes with therians. In light of the evidence presented in this paper, several derived conditions of monotremes, such as a curved cochlear canal of up to 180°, a cribriform plate for the passage of the cochlear nerve and grooves for the nerve fibres within the cochlea, are autopomorphies of monotremes. An array of neomorphs of the inner ear diagnoses the clade of tribosphenic therians, and these document a distinct morphological gap in the transition from the non-therian to therian ear. From presently known forms, an intermediate stage is lacking, even if Encelestes (Rougier et al., 1992) is included. FUNCTIONAL IMPLICATIONS Air-conducted hearing in multituberculates We have already speculated on hearing abilities in Late Cretaceous therians (Meng & Fox, 199513) using regression equations derived from acoustical data in living mammals (Manley, 1971, 1972; West, 1985; Rosowski & Graybeal, 1991; Rosowski, 1992). We applied these equations to data from fossils, following the rationale that inner ear structures in the extinct therians that we studied are virtually identical to those of their extant relatives and would have imposed similar constraints on function in Late Cretaceous species. The inner ear structures that are of greatest interest in these comparisons are discussed elsewhere (Meng & Fox, 1993, 1995a, b) and listed in Figure 11. Owing to differences in anatomy, there is no basis for applying the equations that are applicable to the inner ear in therians to the numeric data from the petrosals of multituberculates and other early non-therian mammals. It is certain that at least some of the unique features of the therian ear play important roles in the high frequency hearing that is universal in therians and that is already indicated in Late Cretaceous species. For example, the basilar membrane in Late Cretaceous R. C. FOX AND J. hlENG Figure 1 1. Cladogram depicting distribution of inner ear features in selected mammalian p u p s . 1. S o inner ear feature diagnoses this node given the position of multituberculates; threr ear ossicles may he characteristic. 2, Independent osseous aqueductus cochleae (Rougier et al., 1992). 3, A lullp coiled cochlrar canal, reduction of the lagena, development of primary and secondary osseous spiral laminae, development of the modiolus, a perilymphatic recess that merges with the scala tympani of the cochlea, a radial pattern of the cochlear nerve, and a narrow, elongate basilar nienihranc that is supported by the osseous spiral laminae and that extends from the inner surface of the bony hridgc between the fencstra cochleae and fenestra vestibuli (5leng & Fox, 1993, 1995a, h). 4, A nearly 180" cull-rd cochlea, a cribriform plate for the passage of the cochlear nerve in the internal acoustic meatus. and grooves for the branched cochlear nerve within the cochlea. 5, S o inner ear synapomorphp: an inflatcd vestibule occurs in subgroups of this clade. Sce text for discussion. marsupials and placentals was supported by rigid osseous spiral laminae and was narrow (Meng & Fox, 1995a, b). The presence of the secondary osseous spiral lamina in extant terrestrial therians implies a high frequency ear (Reysenbach de Haan, 1956; Hinchclfie & Pye, 1969; Sales & Pye, 1974; Ketten, 1992), while a narrow basilar membrane is also a likely correlate of high-frequency hearing (hlanlq . 1971, 1972; Ketten, 1992). Unfortunately, the width of the basilar membrane in rnultituberculates and other early non-therians cannot be determined from direct evidmce, because neither a rigid osseous frame nor evidence of cartilaginous structures arising from the walls of the cochlear canal as support for the membrane arc preserved in these species. However, the lack of such structures in multituberculates suggests that the basilar membrane was wider than in therians having cochleae otherwise of the same diameter and/or supported by a relatively flexible franc of cartilage; hence the multituberculate ear was less well adapted to high-frequency hearing. Other than in its lack of a cribriform plate in the internal acoustic meatus and grooves within the cochlea for the cochlear nerve, the inner ear of multitubcrculates in its overall anatomy is most comparable of that of monotremes. Unfortunately, little is known about the relationship between the configuration of the inner ear and hearing ability in either the platypus or echidnas. What is known from previous work on monotreme hearing function implies that the ear in these animals is relatively inefficient in sensing airborne vibrations, and that consequently monotremes occupy an intermediate level in hearing ability between extant reptiles and other mammals (Aitkin &Johnstone, 1972; Gates et al., 1974). From present knowledge of petrosal anatomy, it seems likely that the ability to hear airborne sounds in INNER EAR OF MULTITUBERCUL4TES 283 multituberculates and other early mammals having a straight or nearly straight cochlea was no better than that of living monotremes. In addition to the lack of osseous cochlear laminae, two other features may have influenced the hearing ability of multituberculates: the inflated vestibule and the length of the cochlear canal. Based on comparison with certain extant amphibians and turtles (Patterson, 1960; Bramble, 1982; Lewis & Narins, 1985; Wever, 1985) that have an enlarged vestibule and are sensitive to low frequency sounds, Luo & Ketten (1991: 226) concluded that the enlargement of the vestibular space in multituberculates was probably indicative of adaptation to low frequency hearing. We agree with this conclusion, but we approach the subject not from a comparison with non-mammalian tetrapods but from the perspective of functional morphology. In this context, the acoustical phenomena demonstrated by the experiments and models of Bkktsy (1 960) are crucial: these show that since high frequency vibrations dissipate rapidly with distance, they are sensed at the proximal part of the basilar membrane, near the footplate of the stapes, whereas low frequency vibrations are sensed more distally, towards the apex. This tonotopic organization is found not only in mammals but is also generally applicable to birds and reptiles, excepting only squamates (Manley, 1990). The distribution of frequencies along the basilar membrane is also reflected in an orderly projection of the auditory nerve fibres on the brain nuclei, as demonstrated in mammals (Sando, 1965). Apparently, the distance from the stapes to the basilar membrane is one of the factors affecting frequency limits sensed by mammals (Btktsy, 1960). In multituberculates, and probably in other early mammals that lacked osseous laminae, the basilar membrane failed to extend proximally to the inner surface of the bony bridge between the fenestra vestibuli and fenestra cochleae, thereby failing to approach the footplate of the stapes closely. This pattern contrasts with the condition in Late Cretaceous marsupials and placentals in which the basilar membrane extended to the bridge, as indicated by the presence of the secondary osseous spiral lamina there (Meng & Fox, 1995a, b). In those multituberculates in which the vestibule is inflated, the gap between the fenestra vestibuli and the most proximal part of the basilar membrane may be further increased as a consequence of vestibular inflation. An extreme example of this pattern is seen in Lumbdopsalis (Meng, 1988; Meng & Wyss, 1995): the fenestra vestibuli is positioned relatively far from the entrance to the cochlear canal simply as a consequence of the greatly inflated vestibule. In fact, the distance from the fenestra to the canal in Lambdopsalis is equal to the total length of the canal itself (pers. observ.). In this case, it seems likely that high-frequency vibrations could easily have been dissipated before they were able to activate the basilar membrane. In living mammals, the cochlea is the major acoustic organ, whereas the vestibule (containing the sacculus and the utriculus) is primarily concerned with balance sensitivity (Wilson & Melvill Jones, 1979; Stebbins, 1983: 55; Lewis et al., 1985: 82). In the evolution of tetrapods from amphibians to mammals, the auditory papilla (precursor to the basilar membrane) increased substantially in length, with a corresponding increase in the number of hair cells (Wever, 1974: 450). Manley (1972: 616) recognized the adaptive significance of these changes, noting that “animals with short membranes are less sensitive to sound, a phenomenon which may be partly due to their possessing only a small number of hair cells”. If elongation of the cochlea approximates elongation of the basilar membrane (and we assume it does), a longer membrane would obviously support more hair cells within it, with 28.1 R. C. FOX ANDJ. MEN(; the effect of both increasing the range in frequency of sounds that an organism can receive (Manley, 197 1, 1972; Wever, 1974) and enhancing the ability to discriminate between frequencies within that greater range (Manley, 1972; Allin & Hopson, 1992). It must be pointed out, however, that a short cochlea does not in all cases imply low frequency hearing: some investigations have revealed an inverse relationship between basilar-membrane length and high-frequency hearing, i.e. certain eutherian mammals with the shortest basilar membrane are most sensitive to high frequency sounds (West, 1985). The cochlear canal in multituberculates is among the longest in early mammals (Luo & Ketten, 1991; Luo et al., 1995), but it is still considerably shorter than that in therians of the sarhe body size (Keen, 1940; Bkkksy, 1944, 1960; Schuknecht, 1953; Manley, 1971, 1972; West, 1985; Rosowski & Graybeal, 1991; Fay, 1992: Ketten, 1992; Meng & Fox, 1995a, b). In retaining a relatively short, straight. cochlear canal, the ear of multituberculates not only resembles that in other known primitive mammals (Sznoconodon, Morganucodon, monotremes), but it also resembles those of living reptiles and birds; consequently, it is a reasonable conclusion that multituberculates were capable of hearing only lower frequency airborne sounds in a narrower frequency range than therian mammals, resembling in that respect living monotremes and non-mammalian amniotes. The ear in multituberculates could have been as sensitive to airborne sounds as that of monotremes but less so than that of therians. Bone-conducted hearing in multituberculates The inflated vestibule is such a conspicuous aspect of petrosal structure in those multituberculates in which it occurs, it must have had some special function. We are in agreement with other authors who have concluded that inflation of the vestibule is likely an adaptation for low frequency and bone-conducted hearing (Miao, 1988; Luo & Ketten, 1991). This hypothesis is consistent with the assumed fossorial life of some multituberculates (Miao, 1988; Kielan-Jaworowska, 1 989: Kielan-Jaworowska & Qi, 1990; Luo & Ketten, 1991), as well as with the relatively massive ear bones of Lambdopsalis (Miao & Lillegraven, 1986; Meng, 1992; Meng & Wyss, 1995). Most mammals, including man, are capable of bone-conducted hearing (see e.g. Wever & Lawrence, 1954; Bkkesy, 1960). According to Bkkksy (1960), vibrations picked by the bones of the ear can be roughly resolved into parallel, compressional and rotational components affecting the cochlea and middle ear. If an asymmetry is present between the volume of the scala tympani and scala vestibuli. i.e. the space of the scala vestibuli is larger than that of the scala tympani, parallel and compressional vibrations can produce a significant displacement of fluid near the stapes, which causes disturbance of the basilar membrane (Fig. 12). If inflation of the vestibule, as in some multituberculates, greatly increases the space between the footplate of the stapes and the scala vestibuli, then bone-conducted hearing may become necessary. It may well have been the case that some multituberculates, namely those in which the vestibule was greatly enlarged, were more highly adapted to bone-conducted hearing than for the reception of airborne sounds. It is well known that bone can transmit high frequency vibrations effecti\Tely (Bkkesy, 1960). For example, the human ear responds to airborne sounds only up to 20kc./s. in frequency, but the inner ear can sense sounds up to lOOkc./s., although different frequencies higher than 15 kc./s. cannot be readily distinguished INNER EAR OF MULTITUBERCULATES 285 c .Stap h Round .- A B Figure 12. Diagrams showing two patterns of bone-conducted hearing. A, the cochlea under parallel vibrations; and B, the cochlea under compressional vibrations (after Wever & Lawrence, 1954; Btkksy, 1960). See text for discussion. (Pumphrey, 1950). Whether high frequency bone-conducted vibrations are audible as sound depends in part on the structure and physiology of sound-receiving and transmitting soft tissues of the inner ear, as well as those of the brain, none of which are preserved as fossils. Hence, it is uncertain whether high frequency vibrations transmitted by bone-conduction could have been sensed as sound in multituberculates if only because the inner ear of multituberculates is significantly different in structure from that of therians, and of course nothing is known of multituberculate auditory physiology. ACKNOWLEDGEMENTS We thank the field parties from the University of Alberta who assisted in collecting the samples that contained the multituberculate specimens described here. Mrs A. Voss, UALVP, sorted the screened concentrates from Bug Creek; L.A. Lindoe constructed the latex endocast of the vestibule of UALVP 26039; G.D. Braybrook prepared the SEM photographs. We thank R.D.E. MacPhee, Department of Mammalogy, and R.H. Tedford, M.C. McKenna, MJ. Novacek and G.W. Rougier, Department of Vertebrate Paleontology, the American Museum of Natural History, for permission to study monotreme specimens and for access to the Khoobur petrosal specimens under their care. The manuscript was improved with comments from two anonymous reviewers, to whom we extend our thanks. Approximately 60% of financial support for research by RCF on Cretaceous and early Tertiary mammals is provided by Natural Science and Engineering Research Council (NSERC) of Canada operating grants to him. During the course of this study, Meng has been supported by a felldwship from NSERC, the Frick Postdoctoral Fellowship from the AMNH and an NSF grant (DEB-9508685). REFERENCES Aitkin LM, Gates GR, Kenyon DE. 1979. Some peripheral auditory characteristics of the marsupial brush-tailed possum, Trichosums uulpecula. Journal of Expimental <vvloQ 209: 3 17-322. K.C. FOX L\KD,J. hlENG ?8(i Aitkin LM, Johnstone BM. 1972. Rliddlr-ear fiinctioii in a nionotrrme: the echidna ( ~ ? I ~ / ~ J & \ . \ u \ czix1rnt~t.c). Joiirnul pf Eupn-zrnmtril ~ o o l 180: o ~ 245-230. Alexander G. 1904. Entwicklung und Bau dcs innerens Gehororgans \on Echidna ac~tlecila. Smori.! <ooligicrhr Forsrhung.nreisen in dutralien 3: 1-1 18. Allin EF. 1975. E\rolution of the mammalian middle ear. Journal ofMo$holoo 147: 403--438. Allin EF. 1986. Auditor). apparatus of advanced niammal-like reptiles and early mammals. In: Hotton N. XIacLean PD, Roth JJ, Roth EC:, eds. Tlie E(,ologand Biolo<gof,Zlanininl-like Keptilfs. \\'ashi~l,qton and London: Sinithsoniari Institution Press. 283-294. Allin EF, Hopson JA. 1992. Evolution of the auditor). system in S>-napsida('Inammal-like reptiles' and primiti1.c mammals) as seen in the fossil rrcord. In: IVebster DB, Fay KR. Popprr .YY. tds. The EL*ohtionayBioloB of Hearing. N e u York: Springer-\'erlag. 587 6 14. Archer M, Flannery TF, Ritchie A, Molnar RE. 1985. First hlesozoic mammal from .\ustralia~-an Early Cretaceous monotreme. :%&ire 318: 363-366. Archer M, Jenkins FA Jr, Hand SJ, Murray P, Godthelp H. 1992. Description of the skull and non-vestigial dentitioii of a Miocene plat~pus(Obdurodon dicksoni n. sp.) from Rivcrsleigh. Australia and the prohleni of nioiiotremc origin. In: ;-\ugee M, rd. I'la~~pi.\and Erhidnm. Sydne) : KO) a1 Zoological Socict). of New South \Vales. 15-27. Archibald JD. 1979. Oldest known cutherian staprs a n d a marsupial petrosal boiir from rhr 1,atc Cretaceous of North America. . M u r e 281: (569-670. Archibald JD, Lofgren DL. 1990. Mammalian zonation near the Cretaceous-Tertiary Ixiundar).. In: Rose KD, Broxvn 'I, eds. Da\vn of thr Xgc of klammals in thc northern part of thc Rock>Xlountain Interior, North America. Geolog Sorie(y ?f'.-lnierim Special Paper 243: 3 1 50. Axelsson A. 1974. The blood supply of the inner ear of mammals. In: Keidcl LVD. Neff \I'D. eds. Huridbook o j S e n q Pl~,:siolop, Lbl. PI:.i?idit(q System, .4mtniy, Ph_linoloa, ( E 4 . NPLCYork: Springer\.erlag. '213 260. Baird IL. 1970. L sul n e l - of the periotic labyrinth in somc rrpresentative Kecent reptiles. T?ir I 7ni7,rr.s~!~ ncp Bulletin 41: 8 9 1 ~981. . 'I'hc anatomy of the rrptilian car. In: Gans C, rd. Bioloa. qf the Reptilia. 1 2 . 2. I,ondon and i%cw York: ilcadrinic Press. 193--275. Baird IL. 1974. Anatomical featurcs of thr inncr ear in sulxnammalian \wtrtirntcs. In: Kirdrl \VI>. Yeff \I'D: rds. Handbook of Smsoy P/~i:\inlog. 161. 1'1: .iuditoT $'.rtPln, .inatom)., PhrJioiou fEm). Ncxv York: Springer-Verlag, 159-212. Bast TH, Anson BJ. 1949. ' R i p 'Ternpol-a1 Bone and t h Ear, ~ Springficld, Illinois: Charles C. 'l'hornas. BCkesy G von. 1944. n b r r die mechanische Frequcnzanalysc in der Schneckc versclliedcncr Tirrr. z4kziificc~hrZcitschnft 9: 3- 1 I . er (trans). Nebv Uork: hfcGraw-Hill. Bekesy G von. 1960. E,r,berirnents in Hpuing. EG \ Bellairs A D'A, Kamal AM. 1981. The chondrocranium and the development of the skull in Kecent reptiles. In: Gans C:, cd. Biolog)' ofthe Rrptilza. LX 2. N e w York: Academic Press, 1-363. Bramble A. 1982. Smptochrbs: generic rwision and evolution of gopher tortoises. Copria 1982: 8.52-867. Bredberg G. 1968. Cellular pattern and n e n e supply of the human organ of Corti. .Irtn Otolrigrgol(giia S ~ ~ p ~ ~ p ~236: f u rI n135. Broom R. 1914. O n the origin of mammals. Zuruartions of thp Ryal Socieg; London B206: 1 4 9 . Bruns V. 1976. Peripheral auditory tuning for fine frrquency analysis by the CF-FLI bat, ~ < h i n o k o / h ~ j?n-urnequinum, 11. Frequency mapping in the cochlea. Journal of Conzpai-atioePhysiolog 106: 87 97. Carroll RL. 1988. lirtebratf Paleonto/o<gand E~~ohtion. Nrxv York: Frccman. Chow M, Qi T. 1978. Paleocene mammalian fossils from Nomogcii Formation of Inner 1\Ion,yolia. Ihtpbrntn P(i1iisiatira 16: 77-85. Crompton AW. 1964. O n the skull of Ol&+k4pt/uJ. Bullhi afthe Briti.th r\luseum (w~hturcrl Histog'). G e o l q g 9: 69 82. Crompton AW, Luo Z. 1993. Relationships of the Liassic mammals Sinoconodoiz, L\lopnzlrodorz orhlel-i, and Dinrrrthrriurn. In: Szalay FS, Sovacek hIJ, XlcKenna hIC, eds. .2rlrmzmal P/ylogeiy, [hi,I : .\leso:oir IXffcrmtinii(in, i\fullit~ib~rc7ilate,pr. Jlonotrme.,. Earli, Tlzpi7an.r. and Mai-supiuls. New York: Sprin~rr-\Terlaq;, ~ :x--t+. d e Beer GR. 1929. The development of the skull of the shrew. I'hilosojihid 7?ai~snrfiontuj /he K g n l ,\brieo ofLondon 217: 41 1-480. d e Beer GR. 1937. 'The Deoelopzent qf the Tirtehrak Skiill. Oxford: Clarendon Press. d e Burlet HM. 1934. Zur vergleichcnden hnatomir und Physiolo,gie des perilymphatischrn Kaunicc. .l(ki o k ~ l n ~ ~ ~1 ~3 ~153-187. ~ l o g i ~ ~ ~ INNER EAR OF MULTITUBERCULATES 287 Denker A. 1901. Zur Anatomie des Gehororganes der Monotremata. Denkschniten der Medizinisch~atuatulwissenschaftlichenGesellschaft zu Jena 6: 635-662. Estes R. 1961. Cranial anatomy of the cynodont reptile Thrinaxodun liorhinus. Bulletin ofthe Museum of Comparative ~ o o l o g y125: 165-180. Fay RR. 1992. Evolution, perception, and the comparative method. In: Webster DB, Fay RR, Popper AN, eds. The Evulutionaly Biology of Hearing. New York: Springer-Verlag, 2 11-263. F e r n b d e z C. 1952. Dimensions of the cochlea (guinea pig). Journal ofthe Acoustical Sociep ofAmerica 24: 519-523. Fernandez C, Schmidt RS. 1963. The opossum ear and evolution of the coiled cochlea. TheJuurnal $Comparative Neurulogy 121: 151-159. Flannery T, Archer MyRich THYJones R. 1995. A new family of monotremes from the Cretaceous of Australia. Nature 377: 418420. Fleischer G. 1976. Hearing in extinct cetaceans as determined by cochlear structure. Journal of Paleontology 50: 133-152. Frick H. 1952. Uber die Aufteilung des Foramen perilymphaticum in der Ontogenese der Sauger. xeitschnftjir Anatomie und Entwicklungsgeschichte 116: 523-55 1. Gates GR, Saunders JC, Bock GR, Aitkin LM, Elliott MA. 1974. Peripheral auditory function in the platypus, Omithorhynchus anatinus. Journal ofthe Acoustical Sociep ofAmerica 56: 152-156. Gaupp E. 1900. Das Chondrocranium von Lucerta agilis: Ein Beitrag zum Verstandnis des Amniotenschadels. Anatumische H@e 15: 164-433. Gaupp E. 1908a. Ueber die Entwicklung und Bau der ersten Wirbel und der Kopfgelenke von Echidna aculeata, nebst allgemeine Bemerkungen uber die Kopfgelenke der Amnioten. Semon, Zoologische Forschungsreisen in Australien. Denkschnten der medizinzrchen-natur~senschaftlichen Gesellschaft zu Jena, pt. 2: 481-538. Gaupp E. 1908b. Zur Entwicklungsgeschichte und vergleichenden Morphologie des Schadels von Echidna aculeata var. typica. Semon, Zoologische Forschungsreisen in Australien. Denkschnten der medizinirchen-naturwissenschaftlichen Gesellschaft zu Jena, pt. 2: 539-788. Gaupp E. 1913. Die Reichertsche Theorie (Hammer-, Amboss-, und Kieferfrage). Archivf i r Anatomie und E n ~ ~ k l u n ~ s ~ e s c1912: h i c h ~1-41 6. Gauthier J, Kluge AG, Rowe T. 1988. Amniote phylogeny and the importance of fossils. Cladistics 4: 105-209. Goodrich ES. 1958. Studies on the Structure and Development of Krtebrates. New York: Dover. Gray AA. 1908a. An investigation on the anatomical structure and relationships of the labyrinth in the reptile, the bird, and the mammal. Proceedings ofthe Royal Suciety, ser B80: 507-528. Gray AA. 1908b. The Labyrinth ofAnimals, Wl. II. London: J. & A. Churchill. Gray H. 1959. Anatomy ofthe Human Body, 27th edn. Philadelphia: Lea and Febiger. Graybeal A, Rosowski J, Ketten DRYCrompton AW. 1989. Inner ear structure in Morganucodon, an early Jurassic mammal. ~oologicalJournal ofthe Linnean Society 96: 107-1 17. Griffiths M. 1968. Echidna. Oxford: Pergamon Press. Griffiths M. 1978. The Biology ofthe Monotremes. New York: Academic Press. Hahn G. 1988. Die Ohr-Region der Paulchoffatiidae (Multituberculata, Oberjura). Palaeuvertebrata 18: 155-185. Hinchcliffe R, Pye A. 1969. Variations in the middle ear of the Mammalia. Journal 4<ooluQ-The xoolugical SocieQ o f London 157: 277-288. Hopson JA. 1966. The origin of the mammalian middle ear. American <uologist 6: 437-450. Hopson JA. 1969. The origin and adaptive radiation of mammal-like reptiles and nontherian mammals. In: Petras JM, Noback CR, eds. Comparative and Evolutionaly Aspects ofthe Ertebrate and Cent?-a1Nmous $stem. Annals New rOrk Academy o f Sciences 167: 199-2 16. Hopson JAYRougier GW. 1993. Braincase structure in the oldest known skull of a therian mammal: implications for mammalian systematics and cranial evolution. In: Dodson P, Gingerich P, eds. Functional Murpholugy and Evolution. American Journal o f Science 293-A: 268-299. Jergensen JM, Locket NA. 1995. The inner ear of the echidna Tachyglossus aculeatus: the vestibular sensory organs. Proceedings ofthe Royal Society ofLondon B260: 183-1 89. Keen JA. 1940. A note on the length of the basilar membrane in man and in various mammals. Journal ofAnatomy 74: 524-527. Kemp TS. 1983. The relationships of mammals. ~oologicalJuurna1ofthe Linnean Sociep 77: 353-384. Kermack KA. 1963. The cranial structure of the triconodonts. Philosophical Transactions of the Royal Sucieg ofLondon B246: 83-103. 288 R. C. FOX AND J. MENG Kermack KA, Kielan-Jaworowska Z. 1971. Therian and non-therian mammals. In: Kermack DM, Kermack KA, eds. Earb Mammals. <oologiralJournal of the Linnean Sociep 50 (suppl. 1): 103-1 15. London: Academic Press. Kermack KA, Mussett F. 1958. The jaw articulation of the Docodonta and the classification of Mesozoic mammals. Proceedings .f the Royal SocieQ $London B148: 20W2 15. Kermack KA, Mussett F. 1983. The ear in mammal-like reptiles and early mammals. .-lck~ Palaeontologica Polonica 28: 147-1 58. Kermack KA, Mussett F, Ringney HW. 1981. The skull of Moqanucodon. <oologiral Joui-nal Oj’fhc Linnenn Socieg 71: 1- 158. Ketten DR. 1992. The marine mammal ear: specializations for aquatic audition and echolocation. In: Webster DB, Fay RR, Popper AN, eds. l 7 i e Evolutionary Biology $Hearing. New York: SpringerVerlag, 7 17-750. Kielan-Jaworowska Z. 1989. Postcranial skeleton of a Cretaceous multitubcrculate mammal. J r f a Palaeontologica Polonica 34: 75- 85. Kielan-Jaworowska Z. 1992. Interrelationships of Mesozoic mammals. Historical Hzolog)l 6: 185-201. Kielan-Jaworowska Z, Gambaryan PP. 1994. Postcranial anatomy and habits of Asian multituberculate mammals. Fossils 6’ Strata 36: 1-92. Kielan-Jaworowska Z, Presley R, Poplin C. 1986. The cranial vascular system in taeniolahidoid multituberculate mammals. Philosophical Eansactions of the R p a l Sociep ofLondon B313: 525-~~602. Kielan-Jaworowska Z, Qi T. 1990. Fossorial adaptations of a taeniolabidoid multituberculate mammal from the Eocene of China. ktebrata PaOlsiatica 28: 81-94. Kuhn HJ. 1971. Die Entwicklung und Morphologie des Schadels von Tachglossus aculeatur. Abhandlungpn der Senckenbergischen Natuforschenden Gesellschaft 528: 1-1 92. Lewis ER, Leverenz EL, Bialek WS. 1985. 771e I4rtebrate Inner Ear Boca Raton, Florida: CRC: Press, Inc. Lewis ER, Narins PM. 1985. Do frogs communicate with seismic signals? Scienre 227: 187-189. Lillegraven JA, Kielan-Jaworowska Z, Clemens WA [eds]. 1979. Mesozoic rClammals: The First Tiwthirds ofillammalian HistoT. Berkeley: University of California Press. Lillegraven JA, Hahn G. 1993. Evolutionary analysis of the middle and inner ear of Late Jurassic multituberculates. Journal $Mammalian Evolution 1: 47-74. Lillegraven JA, Krusat G. 1991. Cranio-mandibular anatomy of Haldanodon exspectutu.! (Docodonta; Mammalia) from the Late Jurassic of Portugal and its implications to the evolution of mammalian characters, Contributions to Geology, Universig o j I;ljioming 28: 39-1 38. Lofgren D. 1995. The Bug Creek problem and the Cretaceous-Tertiary transition at McGuire Creek, Montana. Publications in Geological Sciences, Universig of Calfornia 140: 1-1 85. Lorente de NO R. 1937. The sensory endings in the cochlea. Laryngoscope 47: 373-377. Lucas SG, Hunt AP. 1990. The oldest mammal. ,%re, M e x i c o 3oumal $Science 30: 41-49, Lucas SG, Luo Z. 1993. Adelobasileus from the Upper Triassic of West Texas: the earliest mammal. Journal of ktebrate Paleontolopy 13: 309-334. Luo Z. 1989. Structure of the petrosals of Multituberculata (Mammalia) and the molar morph010~q of the early arctocyonids (Condylarthra, Mammalia). Unpublished Ph.D. thesis, Universit). of California at Berkeley. Luo Z. 1994. Sister-group relationships of mammals and transformations of diagnostic mamnialian characters. In: Fraser NC, Sues H-D, eds. In tlie Shadore, ofthe Dinosaurs Ear4 iVfeAozoir ‘Gtrapods. London: Cambridge University Press. Luo Z, Ketten DR. 1991. C T scanning and computerized reconstructions of the inner ear of multituberculate mammals. Journal of Vertebrate Paleontolopy 11: 220-228. Luo Z, Crompton AW, Lucas SG. 1995. Evolutionary origins of the mammalian promontorium and cochlea. Journal .f Vertebrate Paleontologj 15: 1 13- 121. MacIntyre GT. 1972. The trisulcate petrosal pattern of mammals. In: Dobzhansky T, Hecht MK, Stearc WC, eds. EvolutionaZy Biology, El. 6. New York: Appleton-Century-Crofts, 275-302. Manley GA. 1970. Comparative studies of auditory physiology in reptiles. <eitschnft fur zwgleichunde Plysiologie 67: 363-381. Manley GA. 1971. Some aspects of the evolution of hearing in vertebrates. h u t u w 230: 506--509. Manley GA. 1972. A review of some current concepts of the functional evolution of the ear in terrestrial vertebrates. Evolution 26: 608-62 I . Manley GA. 1990. Peripheral Hearing lMechanisms in Reptiles and bird^. New York: Springer-Verlag. Meng J. 1992. The stapes of Lurnbdopsalis (Multituberculata, Mammalia) and its implications for the stapes transformation of Mammaliaformes. Journal of Webrate Paleontologj 12: 459-37 1 , - INNER EAR OF hIULTITUBERCULATES 289 Meng J, Fox RC. 1993. Inner ear structures from Late Cretaceous mammals and their systematic and functional implications. Journal of Ertebrate Paleontology 13 (suppl. 3): 50A. Meng J, Fox RC. 1995a. Therian petrosals from the Oldman and Milk River formations (Late Cretaceous), Alberta, Canada. Journal of Vertebrate Paleontology 15: 122-1 30. Meng J, Fox RC. 1995b. Osseous inner ear structures and hearing in early marsupials and placentals. <oologicalJournal ofthe Linnean Sociep 115: 47-7 1 . Meng J, Fox RC. 1995c. Evolution of the inner ear from non-therians to therians during the Mesozoic: implications for mammalian phylogeny and hearing. In: Sun A, Wang Y, eds. Sixth Symposium on Mesozoic Tiestrial Ecosystems and Biota, Short Papers. Beijing: China Ocean Press, 235-242. Meng J, Wyss AR. 1995. Monotreme affinities and low-frequency hearing suggested by multituberculate ear. Nature 377: 141-144. Miao D. 1988. Skull morphology of Lambdopsalis bulla (Mammalia, Multituberculata) and its implications to mammalian evolution. Contributions to Geology, University of Wyoming, Special Paper 4: 1-104. Miao D. 1991. O n the origins of mammals. In: Schultze H-P, Trueb L, eds. Orpns ofMajor &ups of Etrapods: Controuersies and Consensus. Ithaca, NY: Cornell University Press, 579-597. Miao D, Lillegraven JA. 1986. Discovery of three ear ossicles in a multituherculate mammal. National Geogaphic Research 2: 500-507. Miller MR. 1966. The cochlear duct of lizards. Proceedings of the Calijornia Academj of Sciences 33: 255-359. Miller MR. 1968. The cochlear duct of snakes. Proceedings of the Calijirnia Academy of Sciences 35: 425475. Novacek MJ, Wyss AR. 1986. Origin and transformation of the mammalian stapes. In: Flanagan KM, Lillegraven JA, eds. Vhtebrates, Pblogmy, and Philosophy. Contributions to Geology, University of Wyoming, Special Paper 3, 35-53. Oelrich TM. 1956. The anatomy of the head of Ctenosaura pectinata (Iguanidae). Miscellaneous Publications ofthe Museum of <oology, Uniuersip of Michigan 94: 1-122. Olson EC. 1944. Origin of mammals based upon cranial morphology of the therapsid suborders. Special Pape-rs ofthe Geological Socieg ofAmerica 55: 1-136. Parrington FR. 1979. The evolution of the mammalian middle and outer ears: a personal review. Biological Reviews 54: 369-387. Pascual R, Archer My Ortiz Jaureguizar E, Prado JL, Godthelp H, Hand SJ. 1992. The first non-Australian monotreme: an early Paleocene South American platypus (Monotremata, Ornithorhynchidae). In: Augee M, ed. Plagpus and Echidnas. Sydney: Royal Zoological Society of New South Wales, 1-14. Patterson NF. 1960. The inner ear of some members of the Pipidae (Amphibia). Proceedings of<oological Sociep ofLondon 4: 509-546. Presley R. 1980. The braincase in Recent and Mesozoic therapsids. Memoires Sociiti Geologique de France N.S. 139: 159-162. Pritchard U. 1881. The cochlea of the Ornithorhynchus plappus compared with that of ordinary mammals and birds. Philosophical Transactions ofthe Royal Socieg of London 172: 267-282. Prothero DR. 1983. The oldest mammalian petrosals from North America. Journal OfPaleontoLogy 57: 1040- 1046. Pumphrey RJ. 1950. Upper limit of frequency for human hearing. Nature 166: 571. Pye A. 1966a. The structure of the cochlea in Chiroptera I. Microchiroptera: Emballonuroidea and Rhindolphoidea. Journal ofMorphology 118: 495-5 10. Pye A. 196613. The structure of the cochlea in Chiroptera 11. Megachiroptera and Vespertillionoidea of the Microchiroptera. Journal ofMo$hology 119: 101-120. Quiroga JC. 1979. The inner ear of two cynodonts (Reptilia-Therapsida) and some comments on the evolution of the inner ear from pelycosaurs to mammals. Gegenbaun Morphologisches Jahrbuch &$I@ 125: 178-190. Reysenbach de Haan FW. 1956. Hearing in whales. Acta Otolayngologica Supplementurn 1 3 4 1-1 14. Rieppel 0. 1985. The recessus scalae tympani and its bearing on the classification of reptiles. Journal ofHe-rpetolopy 19: 373-384. Romer AS. 1956. 7 h e Osteology ofthe Reptihs. Chicago: University of Chicago Press. Romer AS, Parsons TS. 1986. 7 h e Vutebrate Body, 6th edn. Philadelphia: Saunders. Romer AS, Price LW. 1940. Review of the Pelycosauria. Geological Society ofdmen'ca Special Papers 28: 1-538. R. C. FOX A S D J. MENG 4'.10 Rosowski JJ. 1992. Hearing in transitional mammals: predictions from the middle-ear anatomy and hearing capabilities of extant mammals. In: JVebstcr DB, Fay RR, Popper hi,eds. ?he fi>o:i:ohtzona~~~ Hiolugl, of Hearing. New York: Springer-Vcrlag, 6 15-63 1. Rosowski JJ, Graybeal A. 1991. h'hat did :\loTanucodon hear? ~oologicalJournal of'the Linnenn Sorip!? 101: 13-168. Rougier GW. 1990. Primeras cvidencias sobrc la morfologa del oido interno en un trrio no tribosfknico. Resumenes VIIJornadas Agntinas de Paleontologia de Ertebrados, Ameghiniana 26: 249. Rougier GW, Wible JR, Hopson JA. 1992. Reconstruction of the cranial vessels in the Early Cretaceous mammal T4ncelestes neuquenianus: implications for the evolution of the mammalian cranial vascular system. Journal of Vertebrate Paleontolug, 12: 188-2 16. Rougier GW, Wible JR, Hopson JA. 1996. Basicranial anatomy of Priacodon fTuztaen~~s(Triconodontidae, Mammalia) from the Late Jurassic of Colorado, and a reappraisal of mammalian interrelationships. American A21useum.Vor'itates 3183: 1--38. Rowe T. 1988. Definition, diagnosis, and origin of hlammalia. Journal of' L'trtpbmte Paleontolog 8: 24 1-264. Rowe T. 1993. Phylogcnetic systematics and the early history of mammals. In: Szalay FS, Sovacck AiJ, McKenna hIC, eds. Mammal Ptylogey, L61, 1: i\ilesozoZc Differentiation, rlhdtituberrulates, LZhonohrmes, Earb Therians, and Marsupials. New York: Springer-Verlag, 129-1 45. Sales G , Pye D. 1974. LTltrasonic Communication By Animals. New York John Wile); and Sons. Sando I. 1965. The anatomical interrelationships of the cochlear nerve fibres. Acta Otoluryngologira 59: 41 7--436. Schuknecht HF. 1953. Techniques for study of cochlear function and pathology in experimental animals. Acta Otolarq'ngologica 58: 377-397. Schuknecht HF. 1970. Pathophysiology of the fluid systems of the inner ear. In: Neff JVD, ed. Contribufions to Smsoy P/ysioloD, Kl. 4. New York Academic Press, 15 -93. Sereno PC, McKenna MC. 1995. Cretaceous multituberculate skeleton and the carly evolution of the mammalian shoulder girdle. "Vature 377: 144-1 47. Simmons NB. 1993. Phylogeny of hlultituberculata. In: Szalay FS, Novacek MJ, McKenna X f C > eds. L21ammal Phylogey, El. 1: IMesoZoic Dzfferentiation Multituberculates, .Zlonotremes, Ear!? Tierran.,. and ,Ilarsufiial.i. New York: Springer-Verlag, 146- 164. Simpson GG. 1933. The ear region and the foramina of the cynodont skull. AmericanJournal of'Sciencr 26: 285-294. Simpson GG. 1937. Skull structure of the Multituberculata. Bulletin ofthe American A21u.ieum ofAatural Hijtoy 73: 727 -763. Simpson GG. 1938. Osteography of the ear region in monotremes. American ,Museum .Vouztnte.t 978: 1-15, Sloan RE. 1979. Multituberculata, pp. 492--498. In: Fairbridge RW, Jablonski D, eds. The Encyclopedia o j Paleontolog. Stroudsberg: Dowden, Hutchinson & Ross, Inc., 492-498. Sloan R, Van Valen L. 1965. Cretaceous mammals from Montana. Science 148: 220-22'7. Spoendlin H. 1972. Innervation densities of the cochlea. Acta Otolayngolugica 73: 235--248. Spoendlin H. 1974. Neuroanatomy of the cochlea. In: Zwicher E, Terhardt E: eds. Facts and L\Jode/c in Hearing. New York Springer-Verlag, 18-32. Stebbins WC. 1983. 7 h e Acoustir Sense uf' Animals. Cambridge, Massachusetts: H a n a r d University Press. Voit M. 1909. Das Priinordialcraniuin des Kaninchens unter Berucksichticgungder Deckknochcn. Anatomische Hefie 38: 425-61 6 . West CD. 1985. The relationships of the spiral turns of the cochlea and the length of the hasilar membrane to the range of audible frequencies in ground dwelling mammals. Jo21m~lo f t h e Acoustical Socie!y uj'rimerica 77: 109 1 1 10 1. Wever EG. 1967. Tonal differentiation in the lizard ear. Larq'ngoscupe 77: 1962-1973. Wever EG. 1974. Thc evolution of vertebrate hearing. In: Keidel WD, Neff WD, eds. Handbook qf SmsoT Plyiulogy, Rl. T'J: Audit09 $ystem, Anatomy, Ptysiolog (Ear). New York Springer-Verlag. 423-454. Wever EG. 1978. The Reptile EaI: It1 Structure and Function. Princeton, New Jersey: Princeton University Press. Wever EG. 1985. ?he ilmphibian Ear Princeton, New Jersey: Princeton University Press. Wever EG, Lawrence M. 1954. Ptysiological Acoustics. Princeton: Princeton University Press. Wever EG, McCormick JG, Palin J, Ridgway SH. 1971. The cochlea of the dolphin, Tur&ps hlincatits: general morphology. Proceedings o f t h e .latianal Academy of'Sciencrs. U.S.A. 68: 238 1 238.5. ~ INKER EAR OF hIULTITUBERCCWTES 291 Wever EG, McCormick JG, Palin J, Ridgway SH. 1972. Cochlear structure in the dolphin hgenortpnchus obliquidens. proceedings of the &ational Academj of Sciences, U.S.A. 69: 65 7-66 1, Wible JR. 1990. Petrosals of Late Cretaceous marsupials from North America, and a cladistic analysis of the petrosal in therian mammals. Journal of Vertebrate Paleontology 10: 183-205. Wible JR. 1991. Origin of Mammalia: the craniodental evidence reexamined. Journal of Vertebrate Paleontology 11: 1-28. Wible JR, Hopson JA. 1993. Basicranial evidence for early mammal phylogeny. In: Szalay FS, Novacek MJ, McKenna MC, eds. ibfamnal Phylogey, El. 1: Mesozaic ~~erentiatzon, Mu~titub~culates~ Monotremes, Early Thm'anr, and Marsupials. New York Springer-Verlag, 45-62. Wible JR, Hopson JA. 1995. Homologies of the prootic canal in mammals and non-mammalian cynodonts. Journal of Ertebrate Paleontology 15: 33 1-356. Wible JR, Rougier GW, Novacek MJ, McKenna MC, Dashzeveg D. 1995. A mammalian petrosal from the Early Cretaceous of Mongolia: implications for the evolution of the ear region and mammaliamorph interrelationships. Amm'can Museum Novitates 3149: 1-19. Wilson VJ, Melvill Jones G. 1979. Mammalian Estibular Physiology. New York Plenum Press. Zeller U. 1985a. The morphogenesis of the fenestra rotunda in mammals. In: Duncker H-R, Fleischer G, eds. Vertebrate Morphologv, Fortschritte der Zoologie 30. Stuttgart: Gustav Fischer Verlag, 153-157. Zeller U. 1985b. Die Ontogenese und Morphologie der Fenestra rotunda und des Aquaeductus cochlea von Tupaia und anderen Saugern. Gegenbaurs ibforphologisches Jahrbuch Lipzip 131: 179-204. Zeller U. 1987. Morphogenesis of the mammalian skull with special reference to Tupaia. In: Kuhn H-J, Zeller U, eds. Morph0genesi.r ofthe Mammalian Skull. Hamburg and Berlin: Verlag Paul Parey, 17-50. Zeller U. 1989. Die Entwicklung und Morphologie des Schadels von Ornithorhynchus anatinw (Mammaha: Prototheria: Monotremata). Abhandlungen der Senckenbep'schen JVaturforschenden Gesellschaft 545: 1-1 88. Zeller U. 1991. Foramen perilymphaticum und Recessus scalae tympani von Ornithorhynchus anatinus (Monotremata) und anderen Saugern. E'erhandlung der anatomischen Gesellschaft 84: 44 1-443. Zeller U. 1993. Ontogenetic evidence for cranial homologies in monotremes and therians, with special reference to Ornithorhynchus. In: Szalay FS, Novacek MJ, McKenna MC, eds. Mammal Ptylogmy, El. I : Mesozoic Dgerentiation, Multituberculates, Monotremes, Earb Thenam, and Marsupials. New York: Springer-Verlag, 95-1 07.