Earliest Eutherian Mammal Skull, from the Late Cretaceous (Coniacian)Of Uzbekistan

Home , Cimolestes, Gypsonictops, Kennalestes, Leptictida, Leptictis, Palaeoryctes

We dedicate this paper to the memory of Lev A. Nessov (1947 -1995)

Earliest eutherian mammal skull, from the Late Cretaceous (Coniacian)of Uzbekistan

MALCOLM C. MCKENNA, ZOFIA KIELAN-JAWOROWSKA, and JIN MENG

McKenna, M.C., Kielan-Jaworowska,Z., & Meng, J. 2000. Earliest eutherian mammal skull from the Late Cretaceous (Coniacian) of Uzbekistan - Acta Palaeontologica Polonica 45, 1, 1-54.

We describe a partially crushed skull and dentaries of a sub-adult individual of Daulestes nessovi sp. n., from the Coniacian of Uzbekistan. This is the earliest known eutherian skull (about 87 Ma) and the sixth genus of Cretaceous eutherians in which a skull is available. Because the skull of D. nessovi is sub-adult, certain plesiomorphic features may be ontogenetic and should be interpreted with caution. Four upper premolars and five lower premolariform teeth were in use (possibly to become four lowers when fully adult). The upper cheek-teeth have winged conules; M2 has large parastylar and small metastylar projections. Pre- and postcingula are lacking on DP4 and the upper molars. The talonids of dp4-m2 are about 90% as wide as the trigonids, with widely separated entoconid and hypoconulid. The skull has a large sphenorbital fissure, no foramen rotundum, and apparently no pterygoid process of the sphenoid. A large orbital wing of the palatine pre- vents maxilla-frontalcontact within the orbit. The zygomatic arch is slender. The cochlea has one full turn, with an expanded apex, which suggests that a lagena might have been present. A large malleus with a robust anterior process, and a large promontorium may be due to young age of the individual or a primitive retention, as in the platypus. Because of the similarity to Asioryctidae m both cranial structure and dentition, we assign Daulestes tentatively to Asioryctitheria Novacek et al. 1997, family incertae sedis.

Key words: Skull, Eutheria, Daulestes, Asioryctes, Coniacian, Cretaceous, Uzbekistan.

Malcolm C. McKenna [[email protected]], Department of Vertebrate Paleontology, American Museum of Natural History, New York, NY 10024-5192, USA. Zojia Kielan-Jaworowska [zkielan@twarda,pan.pl], Znstytut Paleobiologii PAN, ul. Twarda 51/55, PL-00-818 Warszawa, Poland. Jin Meng [[email protected]], Department of Vertebrate Paleontology, American Mu- seum of Natural History, New York, NY 10024-5192, USA. z Earliest eutherian mammal skal/; MCKENNA er a/.

Introduction

We describeaLate Cretaceousmammal skull from the Coniacianof the Kyzylkum Desert,Uzbekistan (Fig. 1). It is the earliestknown eutherianmammal skull in a rea- sonablyinterpretable state of preservation.It is, however,somewhat distorted and tele- scopedby crushing,notably in the braincaseregion. The skull is designatedDaulestes nessovisp. n., which we tentativelyassign to the orderAsioryctitheriaNovaceket al. 1997,fart1ly incertaesedis. Daulesles Trofimov & Nessov1979,inNessovl & Trofi- mov (1979),was known previouslyonly from a fragmentof a dentary. When the review book on Mesozoic mammalsby Lillegraven et al, (1979) was submittedfor publicationin 1978,only fourteendesignated Cretaceous eutherian gen- erawere known from the entireworld (Kielan-Jaworowskaet al. 1979;Clemens et al. 1979).Since that time, new Cretaceouseutherian taxa from variousparts of the globe havebeen described or discussedby many authors.In their book on the classification of mammals,McKenna & Bell (1991) cited3z eutheriangenera from the Cretaceous. In the sameyear, the posthumousbook by Lev A. Nessov was published(Nessov 1997),in which the author establishedthree new Cretaceouseutherians. These are: EozhelestesNessov, 1997; Ortalesles Nessov, 1997 (possibly congenericwithAspan- lestes,see'Note addedin proof in Nessovet al. 1998); andSazlestes Nesso% 1997, with a single speciesSazlestes /is Nessov,1997 . The latter specieshas, however, not beenfigured andis very poorly known. After the book by McKenna& Bell ( 1997) was published, the following Cretaceouseutherian genera have been erected: Ukhaa- therium Novacek et aI. 1997, EoungulatumNessov, Archibald, & Kielan-Jaworow- ska, 1998, andMontanalestes Cifelb,1999. Sloan& VanValen (1965), Van Valen & Sloan(1965), and Van Valen (1978) listed six eutherian generafrom the Hell Creek Formation of Montana, including taxa from channelsat the top, regardedatthat time as of latestCretaceous age. These channels, the Bug Creek channels,are Paleocenein age (Lofgren 1995; seealso McKenna & Bell1997), and eutheriantaxa from them are accordinglyomitted. McKenna & Bell (I997)questionedCretasorexNessov&Gureev,l98l,describedoriginallyasaCreta- ceoussoricid, which mightbe aLateCenozoic soricoid (with a pocketedcoronoid process of the dentary), possibly wrongly attributed to Cretaceousrocks. Pocketed coronoidprocesses are also found in apternodontids. Among about 40 designatedCretaceous eutherian genera, only five monotypic onesfrom the Late Cretaceousof the Gobi Desert (Kennalestes,Asioryctes, Zalam- bdalestes,Barunlestes, and Ukhaatherium)have beenreported from fairly complete skulls. Daulestesis the sixth Cretaceouseutherian genus for which a skull has been found, and is also the earliest.Originally, Daulesteswas regardedsomewhat uncer- tainly as a zalambdalestid,but in subsequentpapers Nessov (e.g., Nessov et aI. 1994 and referencestherein) assigned it to the Palaeoryctidae.

1 In the paper by Nessov & Trofimov (1979), +hename of Nessov as the author of the paper was spelled 's', with one but with 'ss' asthe authorof a taxon.Nessov used to spellhis namein Englishwith 'ss', although according to rules of Cyrillic transliteration his name should be spelled with just one 's'. Because 's' he changedthe 1979spelling with a single himself in the Nessove/ a/. (1994) papel we follow him and we cite his name in English always as NesSov. ACTA PALAEONTOLOGTCAPOLONICA (45) (1)

.o\ ts ."r\ ( KAZAKHSTAN s"/ )70 ( a: ARAL,,,. ji SEA:]:.

?o\ "o^ \ Le .n. "Lo^

Fig. l. Generalizedtopographic map df south-westemAsia, showing the position (triangle) of Dzharaku- duk locality (modified from Nessover a/. 1994).

Beginning in 1978, the late Dr Lev A. Nessov collected Cretaceousmafirmals (amongother vertebrates)at severallocalities in the Republicof Uzbekistan(Nessov 1982,1984,1985a,1985b, 1987, 1988; Kielan-Jaworowska & Nessov1990, 1992: Nessov& Kielan-Jaworowska1991;Nessov et aL.1994;Nessov1997; Averianov & Kielan-Jaworowska 1999). During field work in 1989 near Dzharakuduk (referred to also as Dzhyrakuduk and,DzharaKuduk)in the central part of the Kyzylkum Desert, a phosphaticnodule containingan unknown amountof the fossitizedskull of a small mammal(Fig.2) wasfound by VyacheslavTrikhin, a memberof Nessov'steam, in the BissektyFormation (Nessov 1990; Nessov et al. 1998;Archibald et al. 1998).In 1990 the nodulewas brought to oslo andthen to New York. It wassubsequently prepared by william Amaral of cambridge,Massachusetts, who skillfully removedthe specimen from the containingmatrix and separatedits tiny jaws from interlockjngocclusion, re- vealingthe dentitionand ventralparts of the sub-adultskull. In the descriptionthat follows we comparethe cranialfe aturesof Daulestesnessovi sp. n. primarily with thoseof Kennalestesgobiensis Kielan-Jaworowska, 1969, Asio- ryctes nemegetensisKielan-Jaworowska, I975a, Zalambdalesteslechei Gregory & Simpson,1926, andBarunlestes butleri Kielan-Jaworowska,1975a (see Gregory & SimpsonI926;Kielan-Jaworowska 1969,I9V5a,198I,19.84a, b, c;Kielan-Jaworow- Earliest eutherian mammal sftull.' MCKENNA er al. ska & Trofimov 1980).The skull of Ukhaatheriumnessovi Novacek et al.,1997 has receiveda preliminary descriptiononly (Novaceket al. 1997)and we refrain from ex- tensivecomparisons with it. Inasmuchas all theseCretaceous genera are monotypic, we refer to them merely by their genericnames. Among these,Kennalestes is especially interestingbecause, among the various specimensfrom Mongolia, there is a sub-adultindividual (ZPAL MgMVI) representedby the skull andjaws, whosedecid- uousdentition is in the processof replacement;almost the samestage of development of the skull andjaws of Daulestesnessovi from Uzbekistan,described here. Among Cretaceouseutherian taxa proposed by Nessov,Dculestes andTaslestes are here subjectively considered congeneric. In spite of the fragmentary nature of the holotypesof the type speciesinvolved, we believe that they differ from Daulestes nessovisp. n. at the specific level (seeTable 1). Another genusproposed by Nessov from the same area and formation, which is of comparable size, is the monotypic KumlestesNessov, 1985a. Only a fragment of a left dentarywith two broken teeth (holotype of Kumlestesolzha Nessov,I985a, CCMGE I/12176) is available.It has beenfigured by Nessov(1985a: pl.2: I5), Nessov& Kielan-Jaworowska(1991: frg. 1),Nessov et al. (1994:p1.2:2)and Nessov (1997: pl. 45: 7). Only talonidsof theteeth (apparentlyml and m2) Ne preserved,the trigonidsbeing virtually unknown.As the dental formula and the trigonids are unknown, the difference betweenit and other taxa of comparablesize from the BissektyFormation cannot be demonstrated.We follow the suggestionof A.O. Averianov (letter of June 14, 1999 to ZKI) and regard Kumlestesolzha as a nomendubium; we shall cite it further in quotationmarks.

Terminology and abbreviations

Anatomicalterms not otherwisespecified follow Gregory(1910), Kielan-Jaworowska (1975a,b), Novacek (1986), Wible (1990), and Fleischer(1973). We denoteupper teeth with capitalletters and lower teethwith lower caseletters: Vi, incisor; Clc, canine;P/p, premolar;.M/m, molar. In the caseof deciduousteeth we useD/d in addition. Abbreviations: Ma, million years;f., foramen;fen. fenesfra;proc., process. Institutional:CCMGE, Czemyshev'sCentral Museum of GeologicalExploration, St. Petersburg;Zfr{,ZoologicalInstitute of the RussianAcademy of Sciences,St. Pe- tersburg;reAl,Institute of Paleobiology,Polish Academyof Sciences,Warsaw.

Description ClassMammalia Linnaeus.1758 Cohort Eutheria Huxley, 1880 Order ?Asioryctitheria Novacek,Rougier, Wible, McKenna, Dashzeveg & HorovitzrL99T Family incertae sedis GenusDazlestesTbofimov & Nessov,1979rin Nessov & Thofimov(1979) Synonym:Tasle ste s Nessov,1 982. ACTA PALAEONTOLOGICAPOLONICA (45) (1)

Typespecies:DaulesteskulbeckenslsTroflmov&Nessov,1979,inNessov&Trofimov(1979:p.952). Otherspecies assigned: Titslestes inobsenabilis Nessov, 1982 and Daulestes nessovi sp. n. Revised diagnosis. upper premola$ 4, upper and lower molars 3, as in eutherians but not primitive metatherians.Shares with Asioryctes, Ukhaatherium, Bobolestes, Cimolestesincisus, C. propalaeoryctes,C. cerberoides,otlestes, Prokennalesle,s, some specimensof Procerberus,andZalambdalestidae the plesiomorphiclack of pre- and postcingula,thus differing fromGypsonictops,Batodon, Kennalestes, Sailestes, Zheles- /e.t, some specimensof Procerberu,r,and severalother speciesreferred to Cimolestes. Shareswith Asioryctesand Ukhaatheriuma stronglyasymmetrical stylar sheHon the uppermolars, especially onM2, with a large parastylarand small metastylarregion, and differs in this respectfrom other Asian Cretaceouseutherians. Differs from Prokenna- lestes and,otlestes in lacking a single labial mandibular foramen, in lacking the preparastyle(Kielan-Jaworowska & Dashzeveg1989), and in having the talonid almost aswide as the trigonid (ratherthan markedly narrower).The entoconidand hypoconulid are not twinned. Differs from Pappotherium,Holoclemensia, Tribotheium, Prokznna- lestes and deltatheroidans(but not fromAsioryctes, Batodon, Bobolestes,Bulaklestes, Gypsonictops,Kennalestes, Otlestes, Sailestes, Zhelestes, hlambdalesfes, and many Cenozoic mammals) in having winged conules.Apparently differs from Kennnlestes, Asioryctes, and Barunlestesin lacking the pterygoid processof the sphenoid.Shmes withAsioryctes generallysimilar proportionsof the alisphenoidand orbitosphenoidand arangement of alisphenoid/orbitosphenoidforamina. The large sphenorbital fissure transmittedthe nervesro, IY vr,Y2, andvL Aforamen dorsalto sphenorbitalfissure and artifactually merged with it is interpreted as the ?sinuscanal foramen. The optic nerve (II) andopthalmic arteryprobably passedthrough an optic foramensituated low in the orbitosphenoid.A large foramenfor carotid distribution (Vidian mery?) is still present within the sphenoid,as in.Asioryctes,metatherians, pantolestids, early Eocene didymoconids(Meng et al. 1994:p. 539), andrhrinaxodon (Estes1961). Differs from metatherians,Zalambdalestes, and Barunlestes, but not from Asioryctes andKenna- lestes,inhaving the foramenrotundum confluent with the sphenorbitalfissure (i.e., there is no foramen rotundum per se; V2 sharesan exit through the sphenorbitalfissure with Vl and other structures).Cochlea with but one full fum. Differs from all known Asian Cretaceouseutherians in being much smaller. Distribution . - Late Cretaceous(late Turonian-Coniacian,Bissekty Formation) of Kyzylkum Desert,Uzbekistan.

Comments on age

The holotypesof Daulesteskulbekensls and Taslestesinobservabills were recovered from the lower part of the Bissekty Formation,whereas the holotype of 'Kumlestes olzha' is from the middle part of the BissektyFormation. The holotype of D. nessovi wasfound in the upperpart of the BissektyFormation and is stratigraphicallyyounger 'Kumlestes than two other speciesof Daulestes and olzha'. The lower part of the BissektyFormation, previously known as the TaykarshiBeds of Sochava(1968), site cDZtl-|1a, wasrefened to by Nessov& Mertiniene(1986) and Nessov (1993, 1997) Earliest eutherian mammal skall: MCKENNA et al. aslate Turonian. Nessov et al. (1994)and Nessov et al. (1998)regarded the lower part of the BissektyFormation as possibly late Turonian, and the middle andupper parts as Coniacianin age,an idea acceptedby Averianov (1999).We follow this assessment. The total amountof time representedby the BissektyFormation is perhapsseveral million years at most and the skull of D. nessoviis about 87 million years old (see Gradsteinet al. 1994,1995). Daulestesnessovi sp. n. Figs2-19, Tables 1-4. Skull of a proteutherian;Nessov & Kielan-Jaworowska1991: fig. 1. Skullwithtwo lowerjawsofEutheriaof the new genus and species,lO7ll2455; Nessov 1993,frg.l:4. Skull of a smalleutherian under study; Nessov et al. 1994:.p. 19. Skullofaplacentalmammal, probably Daulesles sp. nov.,ZINC.79066, siteCBI-5;Nessov 1997l.pl. 48:l. Holotypeand only knownspecimen: ZIN C.79066',a nearlycomplete but posteriorlycrushed skull with articulateddentaries of a sub-adultindividual, including a mix of deciduousand replacement teeth,but lacking premaxillaeand upper incisorsbecause of breakage,associated with an incompleteatlas and separatetympanic bone. The specimenwas preserved in a phosphaticnodule. Originally the skull wasassociated with a slightly damagedspinous process of a mammalian vertebra(possibly an axis), adhering to theposteriorpart ofthe skull.It is seenin Figs5 and6, but hasnow beenseparated from the skull. The vertebrais relatively largein proportionto the skull, and we arenot surewhether it belongedto the sameanimal as the skull. Typehorizon and locality: Channel in theupper part ofthe Coniacianpart ofthe BissektyFormation (Nessov1990: p. 59;Nessov e/ al. 1994:p. 73),locality CBI-5a (Nessov's field locality,unpub- lishedfield notes;CBI = centralKyzylkum, Bissekty), about 10 m higherin thestratigraphic section thanlocality CBI-14 (Kielan-Jaworowska& Nessov1992).East of Dzharakuduk,central part of the Ky2ylkum Desert,Uzbekistan.

Diagnosis.-Smallestspeciesof Daulestes(basedonthesize of m2,seeTable 1);differs from the holotypes of D. inobservabilis and D. kulbeckensis in having m2 about 85Vo of their size. D. nessovi and D. kulbeckensis differ in structure of p2, which is double-roote{ in the new species(unless it is dp2) and single-rooted inD. kulbeckensis (see also discussion on dental formula of D. kulbeckensis in Nessov et al. 1994: p. 56). Differs from D. inobservabills in having three talonid cusps symmetrically arranged on m2, while in D. inobservabilis the hypoconid is set off, and the hypoconulid and entoconid are slightly approximated (see Nessov et al.1994: pI. I: 4a).The dentary of D. nessovi is much less robust (shallower below m2) than in D. inobservabilis, which, however, is in part due to the youth at death of ZIN C.79066. The only other described taxon from Uzbekistan that falls within the same size category as species of Daulestes is Otlestes meimanNessou 1985a, from the early Cenomanian. D. nessovi differs from O. meiman in having four lower premolars and a temporarily retained dp2 instead of five permanent lower premolars, less molarized p3, and lack of labial mandibular foramen, which do not indicate a close relationship.

2 According to A. dverianov (personalcommunication 2February, 1999), 107/12455(cited by Nessov 1993)was a numberintended to be,but that wasnot actuallylisted in the cataloguefor the collectionsof Chernyshev'sGeological Museum, St. Petersburg.After Nessov'sdeath all of Nessov'sunpublished specimenswere transferred to the Zoological Institute of the RussianAcademy of Sciences.The proper numberfor the soecimendescribed here is thereforeZII:I C.79066. ACTA PALAEONTOLOGICAPOLOMCA (45) (I)

Table l. Measurements(in mm) of holotypeof Taslestesinobservabilis (= Daulestesinobservabilis);igltt 'Kumlestes (r) and left (I) m2s of Daulestesnessovi; m2 of holotype of olTha' (nomendubium); ml of holotype of Daulestes kulbeckensis,and m2 of holotype of Otlestesmeiman. Measurementsand explanatory drawing are taken frorn Nes sov et al. (1994: table 1 and fig. 1), augmentedby optical micrometer measurementsof theDaulestes nessovi,Zft{ C.79O66,takenwithShopscope- equipment. See drawing for ex- planation of the symbols: a, anteroposterior length of trigonid; b, remaining length of tooth to rear of hypoconulid;c, width of trigonid; d, width of talonid; x, direct distancebetween apexes of entoconidand hypoconulid;y, directdistance between apexes ofhypoconid andhypoconulid; x', anteroposteriordistance fromentoconidapex to level of hypoconulid;y', anteroposteriordistancefrom apexof hypoconidto level of hypoconulid; a+b, total anteroposteriorlength of tooth. Anteroposterior = parallel to tooth row.

T i-t'- I t) I il c.rI l I I \ _+_ \ \..r

Species a b d r x v a+b Taslestes,m2 0.65 0.67 0.90 0.84 0.26 0.37 0.22 0.16 r.32 D. nessovi,rm2 0.53 0.59 0.84 0.68 0.31 0.33 0.11 0.23 1.13 D. nessovi,lm2 0.51 0.61 0.84 0.76 0.33 0.32 0.15 o.2l r.t2 'Kumlestes',m2 0.55? 0.56 0.70 0.64 0.27 0.32 0.20 0.25 l.l 1? D. kulbeckensis,ml 1.33 Otkstes,m2 0.61 0.69 r.o2 0.76 0.34 0.35 0.21 0.32 1.30

Skull as a whole (Figs2-:t,Table2)

The anterioraspect of the snout(the most anteriorpart of which is missing)is narrow and tubular. Its posterior end widens abruptly in front of the anterior op;ning of the infraorbital canal, atthe level of P3. The anterior margin of the orbit is abovethe anterior margin of M1. The zygomatic arch,preserved on the left side,is very slenderand only slightly expandedlaterully, which may be ontogeneticandlor partly causedby distortion. There is a slight interorbital constriction,but no postorbitalprocess. Al- thoughZIN C. 79066was sub-adultat death,we havehad difficulty in delimiting su- turesbetween various bones, especially in the orbit. We interpretthis asthe resultof at leastpartial fusion. Also, the bonesof the sphenoidcomplex (presphenoid,basisphenoid, orbitosphenoid,and alisphenoid)are fused, and our descriptionofthem as individual is conjectural.Fusion of skull suturesis also true of Solinodon,Nesophontes, many soricoids(McDowell 1958),and somebats. The posteriorend ofthe skull is broken,variously displaced, and distorted. In palatal view, the left sideof the snouthas been displaced at the midline about0.9 mm poste- Earliest eutherian mammal sftzll.' MCKENNA e/ al.

Fig. 2. Lateral (A) anddorsal (B) views of the skull of D aulestesnessovr sp. n. (ZIN C. 79066)in its originally preservedcondition. Arrows in A point to the mentalforamina. riorly with respectto the right side, so that homologousteeth are no longer opposite oneanother (Fig. a). On the dorsalaspect of the anteriorpart of the skull the left sideis not displaced,although fracturing resemblingdisplaced sutures has taken place.The basicraniumhas been bent upward toward the rear,such that the right petrosalnow lies higher than the parietalon the skull roof. In additionto the upwarddisplacement, the right squamosaland otic region have beentelescoped forward and laterally with respectto the right alisphenoid.The left petrosal,although separate now from adjoining bone except the squamosalanteriorly, is more nearly in place with respectto the alisphenoidthan is ihe right petrosal.The left basicranialregion is largelybroken, with only ttre basisphenoidand the promontoriumpreserved. The breakagegoes through the left fenestravestibuli andthe fenestracochleae; the left vestibuleand semicircular canalsposterior to this breakageare all missing.The promontoriumis nonethelesspre- servedlngood condition and is approximatelyin its original positionwith regardto the basisphenoid,although it is now stronglyrotated outward (as seenfrom the rear).

Fig. 3. A. Stereophotographshowing dorsal view ofthe skullof Daulestesnessovi sp. n. (ZIN C. 79066),x 7. B. Explanatorydrawing for the same. ACTA PALAEONTOLOGTCAPOLONTCA (45) (1)

postparietalvascular f.

parietal

squamosal

B 10 Earliest eutherian mammal sftzl/.' MCKENNA et al.

Table 2. Optical micrometer measurements(mm) of skull and dentary of Daulestes nessovi sp. n., ZIN C. 7 9066, taken with Shopscope* equipment.

Length of preservedskull 11.55

Length ofhypothetically undamaged sub-adult skull (Fig. 7) 15.6

Length of adult skull (estimated by analogy with Kennalestes) 19JO

Length from condyle to anteriormost preserved part of dentary 12.36

Depth ofjaw bucally, beneath dp4 1.2

Depth ofjaw bucally, beneath m2 l.l

Length from front of DC to the rear of M2, projected on sagittal plane 5.86

Length from front of DC to rear of M2, not projected on sagittal plane 6.17

On the right side, the skull was fractured ventrally, approximately at the level of the pterygoid; the fracture runs dorsally along the posterodorsaledge of the orbit. The entire half of the cranium posterior to the fracture was rotated counterclockwise about 25" (as seenfrom the rear) and was pusheddorsally and slightly anteriorly; consequently,the preservedskull is now appreciablyshorter than it was in life. Becauseof distortion and displacementsat the rear of the skull, the position of the occipital plate with respectto the plane of the teeth is conjectural. Despite the bending and fracturing of the posterior part of the cranium, the ear region in regardto its surroundingelements remains roughly with many of its original topological relations and is the best pre- servedand most informativepart of the skull. In Fig. 7 we presenta tentativereconstruction of the skull of Daulestes,according to which the length of a skull of an undamagedsub-adult individual would havebeen 15.6 mm. By analogy with Kennalestes,we estimatethe length of an adult skull to havebeen between 19 and 20 mm Oable 2).

Bones of the skull (Figs 2-9)

The nasals (Figs 2, 3) are not completeanteriorly. Anteriorly narrow, they broaden sharplyon the dorsalside of the snout,above the infraorbital canal.Each nasal contacts the maxilla laterally, and the lacrimal posterolaterally, as in more plesiomorphic synapsids(Gregory t920). Abroad contactwith the frontal is presentposteriorly. The suturewith the frontal is faintly discernible,and is convexposteriorly as it crossesthe roof of the skull from side to side abovethe anterior part of the orbit. The anterior ends of the nasalsare archedin cross-section,and eachnasal is overlappedby the maxilla throughout the entire length of the contact between them. Anteriorly the nasals are completelyunfused; posteriorly the sagittalsuture between them becomesless separate but is still distinct.

Fig. 4. A. Stereophotographshowing ventral view ofthe skullof Daulestesnessovl sp. n. (ZIN C. 79066),x 7. B. Explanatorydrawing for the same.Vidian foraminaare not seenin this view. ACTA PALAEONTOLOGTCAPOLONTCA (45) (1) ll

DC DP1

\t.

palatine postpalatinef. postpalatine vomer torus frontal palatine pterygoid parietal choanae ?opticf. presphenoid orbitosphenoid displacedalisphenoid ?sinuscanal f. carotidf. alisphenoid basisphenoid sphenorbitalfissure f. ovale squamosal rostraltympanic proc. postglenoidproc a fen. vestibuli

B promontorium fen. cochleae 12 Earliest eutherian mammal sftul/.' MCKENNA et a/.

The premaxilla and its incisorsare missing;thus the position of the suturebetween the premaxilla and maxilla is unknown other than that it must have beencompletely in front of, rather than at, the alveolus of the upper caniniform tooth (interpretedas DC). The maxilla (Figs4-7) makesup the side-wallof the snout,but contributeslittle to the anterior end of the zygomatic arch. We do not know how far the maxilla extendedin front of the alveolus for DC, but the latter lies wholly within the maxilla. The infraorbital foramen is 0.6 mm high. Leading anteriorly from the infraorbital foramen is a short,groove-like depression,in front of which the lateral wall of the maxilla is slightly concaveand then flares slightly at the level of the tooth that we identify as either DP2 or P2. On the palate(ventral view) the maxilla is not fenestrated(Fig. 4). No tracesof incisive foramina are visible at the anterior edgeof the palate;therefore, incisive foramina must have been in the premaxilla, anterior to the level of the canine tooth. The anterior end of the preservedpart of the skull hasbeen slightly twisted, such that the left maxilla is displacedsomewhat posteriorly with regard to the right one and also rotated slightly ventrally. At the rear of the palate the right and left maxillae have also beentelescoped medially, so that the left maxilla and the horizontal part of the palatine have come to dorsally overlap the right maxilla and palatine. M3 was not yet erupted. Each maxilla of an adult would thus bear a canine and (presumably) seven postcanineteeth. The posterior margin of the palatal wing of the maxilla, the medial part of which hasbeen preserved on the left side,is thickened(Frg. 4). Its lateralpart (wherethe incompletelydeveloped M3 was contained),and the whole margin on the right side,are broken. On the left side,to the rearof M2, we recognizethe lingual part of an alveolusfor M3, and tentativelythe broken lingual root of the developingbut now missingM3. Afragment of bonethat (in ventralview) obscuresthe left sideof the sphenoid(presphenoid) is possiblya brokenpart of the right maxilla. Betweenthe tri- gonsof the uppercheek-teeth there is someindication of shallowpitting of the maxilla for receptionof cusp apicesof the trigonids of lower cheek-teeth. Within the orbit there is alarge, horizontal, triangular maxillary plate that ventrally floors the anterior part of the orbit. We interpret the contactof the maxilla with the palatine'sorbital wing to occurin the anglebetween the horizontalmaxillary plateand the vertical orbital wall. The maxilla doesnot contribute to the orbital wall abovethe medial side of the horizontal maxillary plate. Rather,an extensiveorbital wing of the palatine lying aboveand posterior to the maxilla bars contactof the maxilla with the frontal within the orbit, as in Potamogalebut not other lipotyphlans(MacPhee & Novacek 1993:p.18). The anteriorend ofthe orbital floor continuesanteriorly as the floor ofthe infraorbital canal.The dorsalends of roots of Ml and M2 areexposed on the dorsal surfaceof the maxillary plate, on the orbital floor. The posterior end of the infraorbital canal (maxillary foramen)opens above the level of the M1 paracone.The boundary betweenthe horizontalmaxillary plateand the verticalorbital processof the palatineis marked by a groove for nervesand vessels('groove' in Fig. 5), which leadsto the venfromedial corner of the maxillary foramen and further to the infraorbital canal. Within the left orbit, the areabearing this groove is broken but could have houseda

Fig. 5. A. Stereophotographshowing right view of the sksllof Daulestesnessovi sp. n. (ZIN C. 79066).8. Explanatory drawing for the same.?Sinous canal foramen and sphenorbitalfissure are not seenin this view. Spinous processof a vertebra (possibly an axis), found with the skull, is preserved,x 7. ACTA PALAEONTOLOGICAPOLOMCA (45) (I) 13

DP1 maxilla DP2or P2 infraorbital f.

nasal lacrimalf. groove M1 maxillaryf. frontal palatine ?opticf jugal vomer parietal choana postparietal orbitosphenoid vascularf. alisphenoid

sprnousproc. B of a vertebra l4 Earliest eutherian mammal skal/: MCKENNA et al.

sphenopalatineforamen; in the right orbit this foramenis only weakly suggestedand would havebeen very small,if correctlyidentified.

The palatine is nearlycomplete on eachside but, becauseof distortion,the left oneis partly hidden from ventral view on the palate by the ventrally overlapping right one. On the palatethe horizontal part of the palatine extendsanteriorly up to the level of the anteriorpart of M1 (Fig. 4). An anteriorpalatine foramen has not beenfound. How- ever,a partly hidden openingoccurs at the anterioredge ofthe palatalprocess ofthe left palatinebone. This openingseems to be too largeto be only the anteriorpalatine foramen; it may represent a foramen expandedby breakage and filled with matrix. Similar but evenmore extensivebreakage filled with matrix occurson the right side. Leading from these openings are very shallow grooves that extend anteriorly on the maxilla, bolsteringthe conclusionthat foramina were present.The choanalmmgin, placedopposite the future site of an eruptedM3, forms a gentlearch, possibly without a medial postpalatinespine. The curved postpalatinetorus is more prominent and wider laterally than medially. It forms the posterior edgeof the palate.The torus delim- its the floor of the choanal opening. At its buccal left end, in the suture with the maxilla, a shortfurrow is tentativelyrecognized as a fissure-likeposterior palatine foramen(Fig. a). On the orbital wall (Figs 5, 6), the exposureof the palatineis extensive, but its boundariesare indistinct. The palatinecontacts the maxilla anteroventrally,the frontal and lacrimal anterodorsally,and the orbitosphenoidposteriorly, but its boundary with the orbitosphenoidis conjectural.

The lacrimal is large, and has on the facial region a prominenttriangular antorbital wing that wedgesbetween the maxilla and the nasal.There is an extensivecontact of the lacrimal with the broad rear part of the nasal, a plesiomorphic feature (Gregory 1920). Posterodorsally,the lacrimal contacts the frontal. The extent of the lacrimal within the orbit is alsolarge; it contactsthe frontal, maxillary andpalatine bones there. There is no lacrimal tubercle.The single lacrimal foramen is prominent, lying just within the orbit and facing rearward. Beneaththe lacrimal foramen, the lacrimal has a smallzygomatic wing that extendsa shortdistance back along the rim of the orbit, dorsally overlappingthe anteriormostpart of thejugal. Within the orbit andmedial to this overlapping, the lacrimal forms the dorsal roof of the posterior opening of the infraorbital canal(maxillary foramen).

The orbitosphenoid (Figs5-9) is.preservedon both sidesof the skull. The boundaries of this bonewith othersare mainly conjectural,especially so on the animal's left side. The orbitosphenoidis either mostly or totally fused to the palatine anterovenfro- medially, the frontal anterodorsally,and the parietal dorsally. The posterior margin is betterpreserved. It is markedby two large foramina (sphenorbitalfissure and ?sinus canalforamen) at the boundarywith the alisphenoid,which on the right side slightly overlapsthe orbitosphenoid.At the postero-ventralpart ofthe orbitosphenoidthere is a largeforamen, possibly enlarged on the left sideby damage,but intacton the right side. This foramen facesanteroventrolaterally, and within the wall of the orbit it hasramify- ing passages.We identify it as the ?opticforamen as inAsioryctes (Kielan-Jaworow- ska 1981: fig. 3). A smallopening (or breakage),connected with the ?opticforamen by a tiny groove,lies posteroventrallyto it, at the presumedboundary with the presphe- ACTA PALAEONTOLOGICAPOLONICA (45) (1) 15

Fig. 6. Stereophotographshowing left view of the sku,llofDaulestes nessovi sp.n (ZIN C. 79066).Arrows point to the spinousprocess of a vertebra(possibly an axis), found with the skull, x 7. noid. we cannot identify it. A small ethmoid foramen for the recuffent, ethmoid branchof the opthalmicdivision (V1) of the trigeminalnerve is seenon the left sideof 'sinus the skull anterodorsalto the canalforamen' at what appearsto be the boundary of the orbitosphenoidand frontal.

The jugal forms the ventrolateralmargin of the orbit, and at its anterior end lies above the maxilla (Figs 5-7), whereits pointedanterior tip is dorsallyoverlapped for a short distanceby the posterolaterallyprojecting zygomaticwing of the lacrimal. Theserela- tionshipscan be seenon the right sideof the skull, althoughthe moreposterior parts of the zygomaticarch aremissing on that side.Above the maxilla the height of thejugat graduallyincreases posteriorly to reachits deepestpoint oppositethe ectoflexusof M2 andthen tapers away beyond that point. The body ofthe slenderleftjugal extendsposte- riorly to reachapproximately the glenoid areaof the left squamosal,but that areais dam- agedand telescoped. The posterior end ofthe left zygomaticarch has been displaced; the expectedsuture between the jugal and a dorsally overlapping zygomaticprocess of the left squamosalhas not beenseen. Part of the squamosalcontribution to the zygomatic arch,if any,may be missingor evencovered by the rotatedleft petrosal.In its currentdis- tortedcondition, the rearend ofthe preservedpart ofthejugal now unnaturallycontacts 16 Earliest eutherian mammal sftall.' MCKENNA er al.

?sinuscanalf. ethmoid f. alisphenoid palatinelac1m4f. parietal infraorbitalf. I \ squamosat\ I ,/ \. ,

subsquamosalf. - -1 , :.,....!.../ 2mm lacrimal orbitosphenoid ?oPticf. groove maxillaryf.

Fig. 7. Tentativereconstruction ofthe skull of Daulestesnessovi sp. n. (ZIN C.79066) in lateral view. the anteriormargin of the petrosalas well as the alisphenoidnear the foramen ovale, in- steadof reachingan anterior projection of the squamosalnear the glenoid fossa. The pterygoid probably forms the posterior part of the lateral wall of the choanal channel, but the suture with the anterior part (built from the palatine) is conjectural to dislo- 1Fig.a). The roofs and sidesof both the right andleft choanaeare broken due cati,onof the basicraniumwith regard to the anterior part of the skull. Also as the result of distortion andbreakage, the lateral marginsof the choanaeprobably convergeposte- riorly more stronglythan in life. The vomer (Figs 4, 8) is possibly representedby a longitudinal centralridge, as re- vealedby breakageof the roofs of the choanae.The ridge may lie too far back to be a part of the uo*e.. At presentthe possiblevomer is separatedfrom the presphenoidby postmortemdamage. The presphenoid(Figs 4, 8) hasbeen shifted posthumously because of translationand rotation of the basicraniumwith respectto the more anterior part of the skull. The anterior endof the presphenoidnow touchesthe ventralfloor of the right choana(out of the midline) and his beenseparated from the vomer.Its anteriormargin is possiblymiss- ing. Therewas a medianridge, only the baseof which hasbeen preserved. There is no suture(as opposedto breakage)between the presphenoidand basisphenoid. The basisphenoid(Figs 4, 8) is estimatedtobe2.6 mm long at the midline in ventral view, continuouslaterally with the alisphenoid,asinAsioryctes andKennalestes'This medial part of the basisphenoidis roughly trapezoidal,strongly wideningposteriorly' Aprominent ridge extendsalong its midline for the anterior third and gradually fades posteriorly.The ridge alsopersists onto the presphenoidas a club-like process.Several

Fig. 8. A. SEM stereomicrograph showing ventral view of the right basicranial region of the skull of Daulestesnessoul sp. n. (ZN C. 79066)8. Explanatory drawing for the same' ACTA PALAEONTOLOGICAPOLONICA (45) (1) t7

-+frontal+ palatine

vomer ?sinuscanal f. (below) cnoana B ?opticf. alisphenoid sphenorbitalfissure presphenoid Vidianf. basisphenoid glenoidfossa medialridge anteriorproc. postglenoidproc. f. ovale subsquamosalf. carotidf. manubrium postglenoidf. bonefragment ?incus rostraltympanic proc. fen. vestibuli malleus

promontorium fen.cochleae cristainterfenestralis groovefor promontoryartery 18 Earliest eutherian mammal skal/; MCKENNA et al. pits of varioussizes on the surfaceof the medialpart possiblyrepresent nutrient foramina that arefilled with matrix. Anteriorly, an opening on eachside of the skull occursat the presumedtransition between the presphenoidand basisphenoid.If they represent original foramina, these foramina are very large and occur at the entrance of the choanae.By position,they areunlikely to be anteriorcarotid foramina, because the latter aremuch moreposteriorly positioned in Asioryctes(Kielan-Jaworowska 1981: fig. 3) andin many fossil andliving marsupials(MarshalI et al.1995; entocarotidforamen of Gregory 1910:fig. 16). Theseforamina are tentativelyidentified aspaired Vidian foraminathat in life would haveconveyed the Vidian artery(McDowell 1958),but the largesize of theseforamina, even allowing for damage,is puzzling.On the right sideof the basisphenoid,a faint groove startsanterior to the ear region and extendsto the Vidian foramen.Closer to the foramen,the groovebecomes increasingly broader. It appearsthat the main groove leadsto the foramen,while a smallerone branchesoff and coursesby lateral to the foramen. This groove compareswith the pharyngeal sulcusthat containsthe pharyngealnerve and artery, as in leptictids(Novacek 1986). Posteriorly,the medial part of the basisphenoidappears completely preserved but the basioccipitalis entirely brokenoff beyondthe suturebetween them.

The alisphenoid (Figs 4, 5, 7 -9) is nearly completeon the right sideof the skull and brokeninto two partson the left. The alisphenoidsform broadly roundedridges, projecting ventrally,anterior to the tympanicregion (inAsioryctesa similar structureex- tends along the medial border of the alisphenoid).Jhe ridges define a valley that is wider posteriorly than anteriorly.The ectopterygoidprocess (flange) of the alisphenoid, which occurs at the boundary of the basisphenoidwith the alisphenoidin Kennalestesand Asioryctes (Kielan-Jaworowska 1981), is apparentlyabsent in Dau- lestes.Onthe right sidethe alisphenoidhas been pushed somewhat anteriorly over the orbitosphenoid,while the squamosalhas been pushed over the alisphenoidposteriorly. As preserved,an extensivegap is present at the anterior and lateral sides of the alisphenoid.Sutures delimiting the alisphenoidare not clear.We assumethe broken edgesof the boneto approximatethe anteriorand lateral sutures. Given theseassump- tions, the alisphenoidappea$ to be a very extensive,roughly fan-shapedextension of the sphenoidin the lateralwall of the braincase,with a roundedand thickened anterior margin that is convex anteriorly and that forms the posterior boundary of the sphenorbital fissure.The breakagegap anteriorto the alisphenoidhas left two large openings, which arepartly separatedby a bony bar.The ventralopening is identified asthe sphenorbitalfissure (foramen lacerum anterius) for nervesIII, IV Y r,Y z, andVI. This foramenis positionedlateral to the suturebetween the basisphenoidand presphenoid. The secondopening, lying dorsolateralto the sphenorbitalfissure, is probably en- largedas a result of breakage.We identify it as the ?sinuscanal foramen. On the right sidethe ?sinuscanal foramen lies at the mutualcontact of the threebones (alisphenoid, orbitosphenoidand parietal) (Figs 8, 9). On the left side of the skull the sphenorbital fissureis alsobroken, and the anteriormargin ofthe detachedalisphenoid projection builds the posterolateralmargin of the fissure.There is no alisphenoidcanal. It seems that the posterodorsalmargin (obscuredby the squamosal)and the ventral margin of the alisphenoidconverge posteriorly toward the foramenovale. On the right sideof the skull the foramen ovale is blocked from view ventrally by the anteriorly displaced ACTA PALAEONTOLOGICAPOLONICA (45) (1) t9

Fig. 9. SEM stereomicrographshowing ventro-lateral view of the right basicranial region of the skull of Daulestesnessovi sp. n. (ZIN C. 79066); al, atsphenoid; i, ?incus;m, malleus; o, orbitosphenoid;p, promontorium; the arrows point to: 1, ?optic foramen;2, ?sinuscanal foramen; 3, sphenorbitalfissure (ob- scured by overthrust margin of the alisphenoid, only ventral part is partly seen); 4, carotid foramen; 5, postglenoidforamen; 6, subsquamosalforamen; 7, 8, additionalvascular foramina. The postglenoidforamen looks double on this micrograph, which might be an artifact. malleus.Between the posteromedialpart of the alisphenoidand the anteromedialcor- ner of the petrosalthere is a large carotid foramen,well preservedon the right side (Fig. 8). On the left sideof the skull only the anteriormargin of this openinghas been preserved,because the petrosalis broken acrossit. On the left side the medioventral part of the alisphenoidhas beenbroken off and extendsposterolaterally toward the promontoriumas a process-like,laterally concave projection (Fig. a). The anteriorpart of this projectionis partly obscuredby two displacedbone fragments. On the left side of the skull, the dorsalpart of the alisphenoidhas been pushed medially, resultingin a large gap in the middle of this part of the alisphenoid.

Frontal and parietat (Figs 2,3, 5-7). The sufurebetween the frontal and parietalis unclear;most likely it runs transversely,anterior to the breakageand posteriorto the narrowestregion of the skull, on the left side of the skull roof, as shownin Fig. 3. No frontal foramina have been seen.The frontal has a large extent in the orbit, but a postorbital processis not developed.The parietal is broad but does not form a lambdoidalcrest (ontogenetic feature?). Given the shapeof thepreserved skull andthe largeparietal, the generaloutline ofthe skull is broadin the basicranialregion. Dorsal to the fracture that separatesthe parietal and the alisphenoid,the transparentbone dis- 20 Earliest eutherian mammal skull; MCKENNA et aI plays a light coloredline that most likely reprcsentsacanal within, or a grooveon the medial side of the parietal.By position, this canal or groove is similar to that of the ophthalmic sulcus for the ophthalmic artery and vein in the skull of leptictids (Nova- cek 1986).A largeopening on the parietal(Figs 3, 5) representsa vascularpostparietal foramen(Cope 1880).For taxathat havea broadtemporal region, on which muscula- ture must have aIlached,a vascularforamen, or severalforamina, is frequently present aspassageway for arteriesandlor veins, as inEoryctes (Thewissen& Gingerich 1989) and didymoconids(Meng et al. 1994: p. 536). Theseare temporalbranches of the orbitotemporalsystem. The squamosal(Figs 4,5,7-9) is well shownon theright side.Although displaced forward and rotated clockwise along with the right pefrosal (in lateral view), the right squamosalis largely preserved.The left squamosalis now almosttotally missing. Ante- riorly, the right squamosalis separatedfrom the alisphenoid by a break, where the squamosalis thrust forward and venfially over the alisphenoid.Because the alisphenoid is thus overthrust,the anterior edgeof the squamosaloverlaps the alisphenoidfor some distanceventrally. However,the relationsof the squamosalto the parietaldorsally, and to the petrosal posteriorly, remain undistorted.The undistorted (and reconstructed)relationshipsare generallysimilar to thoseof Asioryctes,Didelphis, and leptictids,but in contrastto the latter, the squamosaltn Daulestesapparently does not send an anterior processto intrude deeplybetween the alisphenoidand parietal. The squamosalis moderate in size, particulady so comparedwith the extensivepefrosal; therefore, its contribution to the sidewall of the braincaseis probably limited. Venffally, the confribution of the squamosalto the zygomaticarch is not known. On the right side,however, the displaced and rotated squamosalbears an almost completeglenoid fossa.The fossais moderately concave,delimited posterior$ by a fransverseshort postglenoid process, the medial part of which is narrowerand bent anteriorly.There aretwo vascularforamina in this region. By comparisonwithKennalestesandAsioryctes, we identiff (Figs7-9) the fissure-like; longitudinally directed opening, situatedposterolateral to the postglenoidprocess, as a subsquamosalforamen (Cope 1880;Gregory 1910;Butler 1956;in Leptictis called a suprameatalforamen by Novacek 1986). In the squamosal,above and behind the glenoid process,small openingsin the squamosalcorrespond to additional vascularfo- raminadiscussed by theseauthors, e.g., a postsquamosalforamen ('7' and'8' in Fig. 9). Aprominent circular opening,slightly posteromedialto the postglenoidprocess, is identified as the postglenoid foramen (Figs 8 and 9). It opens ventromedially from a posterodorsallylocated sinus in the squamosal(to which other foramina in the squamosal also connect)for primarily venousdrainage (sphenoparietal vein). Abony projection (entoglenoidprocess) limits the foramenimmediately posteromedially. The projection would actually lie more on the posterior side of the foramen if the distorted basicraniumwere restoredto its original shape.This would be a similarity to asiory- ctitherians.The medial contribution of the squamosalto the earregion is obscuredby the somewhatdisplaced malleus.

Ear region (Figs8-11) The petrosalsare relatively well preserved(Figs 4, 8-10). Both promontoriaare present. The right one is in anatomicalposition with respectto the basisphenoidand ACTA PALAEONTOLOGTCAPOLONTCA (45) (l) 2l

cochlearcanal

spacefor lagena?

Fig. 10. Radiographic photographrevealing the shapeof the cochlear canal in the left petrosal of D aulestes nessovi sp. n. (ZIN C. 79066), approx. x 7. The arrow indicates a narrow areabetween the cochlear canal and the spaceprobably for the lagena. Seetext for further discussion. alisphenoid,and more or less with the squamosal(shifted anteriorly and rotated).In Fig. 88 we drawthe positionof the promontoriumwith its long axis directedobliquely anteromedially-posterolaterallyas preserved,which is less longitudinally oriented than in certain other primitive eutherianmammals (e.g,, Asioryctes, Barunlestes, Palaeoryctes,and others).The left promontoriumhas been displacedventrally and tilted laterally in an unnaturalway; however,it is still connectedwith the basicranial region by the displacedmedial part of the alisphenoid.The promontorywall is very thin (possiblybecause of the youth of the individual). The shapeof the cochleais partly visible. Along the medial and anteriormargins of the promontoriumthere is a flat bony shelf (rostraltympanic process of MacPhee1981), which beginssome dis- tancein front of the fenestracochleae and extends up to the carotidtbramen. This shelf is well preservedon the left sideof the skull, but partly obscuredby unidentifiedmin- ute bone fragmentson the right side. The promontoriumis largerelative to the skull size(Figs 4,8,9), possiblydue in part to the animal's immaturity. The fenesfravestibuli is round-oval,about 0.5 mm in diame- ter, similar in dimensionto the samestructure in marsupialsbut significantly lesselongate than that of known Late Cretaceouseutherians (Maclntyre 1972; Archtbald1979; Wible 1990;Meng & Fox 1995a,b). The fenesfa is slightly inclined toward a horizontal position. Separatedfrom the fenesffavestibuli by a broadbony bridge (cristainterfenesfralis), the fenesfracochleae is more posteromediallypositioned. The fenestracochleae, still blocked with matrix, is seenat the baseof a pocket that is medially boundedby a bony ridge. The ridge, which marks the medial boundaryof the posteriorhalf of the pefosal, runsbackward to end at the gyrus of the posteriorsemicircular canal. It is unclearwhether an aqueductuscochleae (canaliculus cochleae) is presentor whetherthe bony ridge is the processusrecessus that divides the fenestracochleae and the aqueductusin therians.On the medial side of the ridge, a small artifactual hole is present,penefrating the petrosal wall into the cochlearcanal at the medial end of the promontorium. zz Earliest eutherian mammal skall: MCKENNA et al.

At first sight, judged from the external shapeof the promontorium only, the cochlear canal appearsto have a broad basalsection and then graduallytaper apically. However,the X-ray photographof the left earregion (Fig. 10),which is brokenposte- rior to the vestibule,shows that the cochlearcanal is actually not tapering;its apical part is asbroad as its basalsection. At the medialbase ofthe canal,a narrowtunnel rep- resentsthe connectionbetween the fenestracochleae and the cochlear canal. The X-ray photographreveals a nrurowregion ofthe cochlen canalat abouta turn of 180' from the base.This morphologydiffers from the osseouscanal of the cochleain either non-therian mammals, such as monotremes and multituberculates (Luo & Ketten 1991;Fox & Meng 1997),or in early therians(Meng & Fox 1995a,b). Two possibili- ties may be considered.First, the centerwardarea of the narrowedregion is an artifact representingunderexposure resulting from blockageof the X-rays by bone (for instance,the modiolus)and X-ray opaquematrix filling in the internal auditorymeatus. Second,the narrowregion is a true structureand is the boundarybetween the spaceof the cochlearcanal and the spaceof the lagena.The latter implies thatDaulestes pos- sesseda large Iagena,which would be unique among therian mammals.Among mammals, monotremesand, probably, multituberculates have alagena,which is indicated osteologicallyby an expandedapex ofthe cochlearcanal (Pritchard 1881; Gray 1908; Griffiths 1968,1978; Fox & Meng 1997and referencestherein). The cochlearcanal curves laterally, as in other mammals,and has just one full coil as has beenreported for Zalambdalestes(KieIan-Jaworowska 1984b: p. 162 and fig. 2). The coiling is slightly lessextensive than that of a theriantaxon from the Late Cte- taceousMilk River Formation of Canada(Meng & Fox 1995a).In the Milk River specimen,the apex of the cochlearcanal overlapsthe basal canal,but in Daulestes thereis no overlap.At the centeror vertical axis of the coiling, which is on the medial sideof the broadbasal cochlear canal, a dimple is formedon the extemalsurface of the promontorium of the Daulestespetrosal. This morphology differs from that of all known Cretaceousmammals in which the cochlearcanal is fully coiled (Maclntyre 1972; Archlbald 1979;Kielan-Jaworowska 1984b; Wible 1990;Meng & Fox 1995a, b; Wible et al. 1995) or partially coiled, such as Vncelestes(Bonaparte & Rougier 1987;Rougier et al. 1992).In thoseforms, the surfaceof the promontoriumis univer- sally convex and 'almond-shaped'.In non-therians,such as multituberculatesand fficonodonts(Kermack 1963; Kermack et al. 1981;Kielan-Jaworowska et al. 1986; Graybealet al.1989; Luo 1989;Luo & Ketten 1991;Fox & Meng 1997)in which the cochlearcanal is slightly curved,the promontoriumis usually elongate,although its medial edge(which is usually the lateral boundaryof the tensortympani fossa)may suggestthe curvatureof the cochlearcanal in somemultituberculates (Fox & Meng 1991).ThedegreeofcurvatureofthecochlearcanaLinDaulestes (Fig. 10)cannotbe determined with precision, but it seemsmore curved than that of monotremes and Vncelestes,but lessso thanin any otherknown therianmammals. This may represent a primitive conditionoccurring in the commonancestor of marsupialsand placentals. The threesemicircular canals are well-preserved in the left petrosal(Figs 8, 9). As in other mammals (Meng & Fox 1995b),the lateral canal surroundsthe fossa for the stapediusmuscle. The posteriorpart of the lateralcanal is broken,exposing the empty canal within. Also as in other mammals,the gyrus of the anterior canal forms the rim of the subarcuatefossa on the intracranialside of the petrosal.The subarcuatefossa is ACTA PALAEONTOLOGICAPOLONICA (45) (1) LJ

very deep;its roof is broken,forming a large,artifactual opening between the posterior and lateral semicircularcanals on the posteriorsurface of the petrosal.The crus com- mune, formed by the posterior and anterior canals,bounds the posteriorrim of the subarcuatefossa, near which the intracranialorifice ofthe aqueductusvestibuli is not observed.Among the three canals,the anteriorone is the largestin diameterof the loop, but measurementsof the canalsare difficult to make,given the orientationand incompleteexposure of the structures. Becauseof breakageat the occipital region, the intracranial side of the right petrosalis partly visible, but not well enoughto be illustrated.Detailed morphology of the internal acousticmeatus is obscurebecause of matrix filling, but it is clearthat on the ventral side of the subarcuatefossa a depression,the internal acousticmeatus, is present.The meatus is further divided by a bony ridge into a more anterior opening (presumablyfor the facial nerve,VII) anda posteriorfossa in which no foramenfor auditory nerve VIII can be seenbecause of matrix filling. The areaanterior to the petrosal is coveredby matrix, furtherremoval of which would be likely destructiveto the bone. Therefore, although it can be said that the petrosal contributes a great deal to the posterolateralwall of the braincase,it is impossibleto determinehow largethe petrosal actuallyis and whetherit has an anteriorprocess. The intracranialside ofthe displacedleft petrosalis betterexposed but poorly preserved.It showsa somewhatdifferent morphology from the right one,probably owing to breakage,and doesnot merit description. On the ventralsurface of the rightpromontorium,two shallowgrooves are visible, sug- gestingpresence of the internal cmotid artery (promontory artery) and stapedialartery Gigs 8, 9). The starting(medial) points of the two grcovesare anteromedial to the fenestra cochleae.Judged from their tendencies,the branching point of thesegrooves, not im- printed on the bone,is closeto the proximal endsof the two grooves.This situationdiffers from the condition of Vncelestes,in which the branchingpoint of the promontory artery and stapedialartery is very closeto the ventralrim of the fenesfa vestibuk @ougSeret al. 1992).T\e groove presumablyfor the promontory artery extendsanterolaterally to the concaveregion of thepromontorium; further anteriorly,the courseof the grooveis blocked from venffal view by the displacedmalleus. The groovepresumably for the stapedialartery travels laterally toward the fenestravestibuli; its destiny on the lateral side of the fenestravestibuli is indeterminable,owing to poor preservation.On the right promontorium €ig. 4), the groovefor the prcmontory arteryis presentbut is lessdistinctive than the one on the left, whereasbecause of breakagethe stapedialgrcove is not preserved. The ectotympanicis preservedas an isolatedelement found in the matrix attherear of the skull (Fig. 11). The size of the ectotympanicmatches the skull. It is roughly U-shaped,consisting of a transversepart that arcsventrally and two limbs. Becauseit is not in its original anatomicalposition, the sideof the skull to which the bonebelongs is in doubt.However, we believethe boneis the left ectotympanicfor two reasons:first, the longer limb - the anterior one accordingto our identification - bearsa trough on the medial (if the bonewas placed vertically) or the dorsal(placed horizontally) side of the limb. This trough is undoubtedlyfor articulation of the anterior processof the malleus; it compareswell with thoseof horseshoe-shapedectotympanics in extantmammals, suchas Didelphis.In extantmammals, a similar troughoccurs at the sameposition on the left ectotympanic,although it varies in size and shapeamong various taxa. In any 24 Earliest eutherian mammal sftall: MCKENNA et al.

Fig. 11.SEM stereomicrographoftheleftectotympanic ofDaulestes nessovisp. n. (ZIN C.79066). case,the trough on the anterior limb helps to confirm its identity as the ectotympanic. Second,given our identification,the boneconverges laterally (whenvertically placed) so that the lateral opening is smaller in diameterand irregular in shape,suggesting formation of an extemal auditory meatus;in contrast, the medial side is more regularly oval-shapedwith a largerdiameter, in keepingwith an oval-shapedear drum. The bestpreserved ectotympanic yet describedin a Cretaceouseutherian is that of Asioryctes(see Kielan-Jaworowska 1981). The ectotympanicof ZIN C.79066 is similar to that of Asioryctesin that its ventral(if vertically placed)or anteromedial(if horizontally positioned)transverse portion is slightly expanded,more so than the ring- shapedbone of Didelphis. However, the expansionin Daulestesis less than that in Asioryctes.The extemalsurface of this expandedpart is smoothand convex, whereas its medial side is concave,although in ZIN C.79066 the medial side is still covered with somemaffix andglue. Adistinctive groovefor attachmentof the tympanumis ab- sent,probably damaged during preservation. The anteriorlimb of the ectotympanicap- pearslonger and slimmerthan that of Asioryctes.The mostnotable morphology of the ectotympanicof Dauleste,l assumingthat it is complete,is its U-shapedoutline. In severalextant placentals,the ectotympanicat an early stageof developmenttakes a U-shapeand progressesto becomemore horseshoe-shapedin adult forms. Although the U-shapedectotympani c in Daulestesmay be attributedto its immaturity,or possibly to the lossof a part,two otherpossible interpretations come to mind. First, the posterior limb may havebeen displaced and pushed closer to the anteriorlimb thanin its original condition.This is likely becausea breakoccurs at its midpoint,where the posterior limb assumesthe unusualshape. A secondpossibility is that the supposedposte- rior limb is not a part of the ectotympanicat all. Possibly,this little pieceof bonemight representa part of anotherear ossicle,secondarily attached to the ectotympanicat the break. Mentally removing this part makes the general shapeof the remaining ecto- 'normal' tympaniclook more but, on the otherhand, enlarges the gapbetween the two endsof the bone.That gap,however, can be filled by the relatively large malleus. The malleusis nearly completeon the right side of the skull (Figs 4, 8, 9) and is partly preservedon the left side.In fact, ttreright malleusis the bestpreserved malleus amongknown Cretaceousmafitmals. The right malleushas been moved from its original position anteriorly and is rotated slightly within the horizontal plane, with its anterior end more laterally shifted whereasthe posterior end is more medially shifted. Con- sequently,the anteriorprocess (goniale) now points anteriorly.In life, it would have ACTA PALAEONTOLOGTCAPOLONICA (45) (t) 25 pointedmore anteromedially.The anteriorprocess is long, althoughits tip may be in- complete.The processappea$ to be robust,but its thicknessis likely exaggeratedby glue andby attachedmatrix on its dorsalside. However, removal of thesealien materi- als might prove to be destructiveto the bone.Owing to the indeterminatesurface condition, we are unableto determinewhether the foramen for the chorda tympani is present, as in monotremes(Doran 1878;Fleischer 1973) and multituberculates(Meng & Wyss 1995),but the homology of the anteriorprocess (goniale) with the nonmammalian synapsidprearticular is assumed.The width of the anteriorprocess matches the trough on the medial (or dorsal) side of the ectotympanic,as describedabove. The lamina (body) of the malleusis in all aspectstypically therian(Doran 1878;Fleischer 1973).Its ventral surfaceis concave,and laterally the body is boundedby a distinct ridge. This ridge representspart of thicknessformed at what appearsto be the incus- malleusarticulation. Although the acfualarticulation facet is not well seen,the ridged structuresuggests that the incusis placedposterior to the malleus,as in living therians. In monotremes,in which the incusoccurs on the dorsalside of the malleus,the ridge, if any, is very low andblunt (personalobservations). In multituberculatesthe incus also is dorsalto the malleus(Meng & Wyss 1995;Hurum et al. I996;Rougier et al.1996b). The neck of the malleus is slightly nrurower than the lamina and, on its medial end, thickens to form a somewhatrounded basefor the manubrium. This morphology is almost identical to that of living therians. Surprisingly, the manubrium, or at least the mostbasal portion of the manubrium,is preservednearly in the original position.It is a very thin process,extending from its roundedbase anteriorly (in original position it would be directedanteromedially) about half the length of the anteriorprocess. The anteriorend of the manubrium,at leastin the position in which it is preserved,pres- ently abutsagainst the carotidforamen. The manubriumparallels the anteriorprocess in ventralview; however,these two processesdo not extendin the sameplane. As in living mammals (monotremesand therians alike), the manubrium extends in a plane that is more dorsallytilted thanthat of the anteriorprocess; the anglebetween the two planesvaries among taxa. In ZIN C. 79066, the angle is about 30'. The tip of the manubriumis probablybroken: if present,the tip shouldbe wider thanthe body of the manubrium and it should bear a flat ventral surfacefor attachmentof the tympanum. The ?incus,interpreted here to be preservedon the right side ofthe skull (Fig. 9), adheresto and ventrally partly coversthe posterolateralborder ofthe malleus.Ifthis boneis indeedan incus,it hasapparently been overturned posterolaterally. On the left sideof the skull an incus cannotbe identified with any certainty.There is, however,a small bit of bone that adheresto the lateral margin of the transversepart of the left malleus,which might possiblybe an incus. The ?stapescannot be identified with any certainty. On the left side of the skull a roundedpiece of bone that may be a part of the stapesis closely associatedwith and may actuallyenter the fenestravestibuli (Fig. a). We cannotsay anythingmeaningful aboutthe shapeof the stapesor abouta stapedialforamen.

Atlas and occipital condyle (Fig. 12) Separatedfrom the skull are two fragmentary bonesthat are stuck together.One piece is a left half of the atlas vertebra,with the intercentrum missing (transferredto become the odontoidprocess of the missing axis vertebra?).The preservedatlas is similar to 26 Earliest eutherian mammal skull: MCKENNA et al.

cranialafticular fovea

occipitalcondyle transverseproc.

Fig. 12.A. SEM stereomicrographof two fragmentarybones of Daulestesnessovi sp. n. (ZIN C.79066); the leftpiece is a left part ofthe atlas;the right oneis apartialright exoccipitalwiththe occipitalcondyle.B. Explanatory drawing for the same. thoseofAsioryctes andKennalestes (Kielan-Jaworowskaet al.1979) in having a short transverseprocess and lacking a transverseforamen. The articularfacet for the occipital condyleof the skull is oval and gently concave.The medial (distal)end of the arch is wider than the lateral (proximal) end.Unlike A sioryctesor Kennalestes,the sulcus arteriaevertebralis is shallowerinDaulestes.InAsioryctes the right andleft archesare fused, althoughno ridge or processis formed at the junction.In Dauleslesthe two archesappear separate, because the preservedarch has a straight,smooth medial (distal) edge;breakage is unlikely to have producedsuch a shape,had the archesfused. This morphologyreflects either unfused structure in animmature individual, suchas in juvenile dogs,or a more primitive condition than that of Asioryctes,or both. The other piece of bone is alarge part of the basioccipitaland exoccipitalof the skull, the latterbearingan occiptalcondyle. The condyleis flat andrough-surfaced.At its anteriormargin, a laterallydirected condyloid foramen occurs. On the dorsalside of the basioccipital(within the braincasein life), the intracranialopening of the condyloid foramen appearsfilled with matrix that is homogenizedwith surroundingbone by glue; applying acetoneto this region revealsit.

Mandible (Figs2A, 13-15) Both dentariesare present.The right dentarybears a probably deciduouscanine and eight postcanineteeth (including dp2 - wholly behindp2, anda developingm3, Fig. ACTA PALAEONTOLOGICAPOLONICA (45) (I) 2',7

Fig. 13.Stereophotographs ofthe left dentaryof Daulestesnessovl sp. n. (ZIN C.79066). A. Lateral view. B. Medial view. x 8.5.

15). The left dentary(Fig. 13) bearsthe alveolusof at leastone now-missinganterior incisor. At leasttwo additionalincisors lie behindit, the anteriorof which is largeand possessesa slightly expandedcrown and the posteriorof which is minute and closely appressedto the posteriorpart of its anteriorneighbor (Figs 13, 14). Following these single-rootedteeth is a double-rootedcaniniform tooth that we interpretto be a deciduouscanine. Following the caniniformtooth arefour premolariformteeth that we identify as the following linear sequence:dpl, p2, dp2, and the eruptingp3. Following p3 arefour molariform teeth:dp4 not yet shed,two eruptedmolars, and a developingm3 in its alveolus.Both right and left dentariesare quite shallow, surely an ontogenetic feature.The dentarycondyle, preserved on the left dentary,is best seenin posterior (slightly posterodorsal)view, in which it appearsroughly circular.In posteriorview it consistsof two roundedridges. The dorsomedialridge is larger,the smallventrolateral 28 Earliest eutherion mamnrulsftal/: MCKENNA e/ a/

Fig. 14A. SEM stereomicrographof the left lower dentition,occlusal vtew, of DauLestesnessovi sp. n. (ZIN C. 79066).B. SEM micrographof themedial view of the same.Scale bar is for A andB. SeeFig. 16for identification of the dentition.C. Stereophotographof the left dentaryof the samein occlusalview. x 8.5. oneprotrudes more stronglyposteriorly. In lateraland medial views, the condyleis directed anterodorsally-posteromediallybut the situationis the reverseof that seenin shrews,in which the lower capitularfacet is placedmore internallythan the upperone. Two condylesoccur in advancedSoricidae, but thereis little reasonto believethat the ACTA PALAEONTOLOGICAPOLONICA (45) (I) 29

Fig. 15.SEM micrographof the right lower dentitionof Daulestesnessovl sp. n. (ZIN C .'79066),inlateral view. The mental foramen beneaththe anterior root of do4 is obscuredbv SEM artifact. doublecondyle inDaulestes is homologousto that of the non-heterosoricineSoricidae or Nesophontesor Solenodon(Feanthead et al. 1955; McDowell 1958: figs 3, 30; Mills 1966:ftg. IZ).II seemsmore probable that it is dueto the youth of ZIN C.79066, in which the condylemay be interpretedas not having beenentirely ossified. In buccal view, the coronoid processarises at about a 60" angle from the horizontal ramusffig. 13). It is not pocketedmedially. Thereis a deeplypocketed masseteric fossa, which is floored by a shelf running forward from the condyle.Three and possibly more tiny vascularforamina lie at the anteroventralend of the massetericfossa, in place of a single buccal foramen(abial mandibularforamen) at this site in someCretaceous mammals,e.g., Kielantherium, Otlestes,andProkercnalestes (Marshall & Kielan-Jaworowska 1992:p.365). Thesetiny foraminaare visible on the right dentary;those on the left are still obscuredby matrix and a preservativecoating. TWo mental foramina occur beneath the anteriorpart of right dp1, and a slit-like depressioncontaining one or more mentalforamina occursunder the anteriorroot of dp4 in both dentaries(Figs 2A, 7, 15). The position of suchforamina in a frrlly adult individual may well havebeen somewhat different. In lingual view the symphysisextends from the front of thejaw to a point approximately beneathdp1 (Fig. 13B). At the anterodorsalbase of the somewhatinflected angularpro- cessof the dentarya mandibularforamen is seen.Extending anteriorly from the mandibular foramenis a very shallow groove (presenton both dentaries)that becomesnalrower and more distinct anteriorly and disappearsat a point below the posteriormostalveolus. This groove is less obvious than a groove referred to as a remnant of the Meckelian groove in Prokennalestes(Kielan-Jaworowska & Dashzeveg1989: figs 17, 20,23). Above the groove the dentaryis slightly swollen.

Dentition (Figs 13-19) ZIN C. 79066 appearsto be only slightly more mature than the juvenile skull of Kennalestesgobiensis, ZPAL MgM-V1, describedby Kielan-Jaworowska(1981). All the teethof ZIN C.79066 are similar in color (in juvenile Kennalestesthe deciduous teeth are lighter in color), which cannothelp to distinguishbetween deciduous and successionalteeth. However, comparison with both the dentalformula andthe premolar morphologyof Kennalesteshelpedusto hypothesizethe dentalformulaof Daules- tes at the ontogeneticstage represented by ZIN C. 79066 as follows. Upper incisor 30 Earliest eutherian tnemmal skal/.' MCKENNA et al.

countunknown, DC, (D)P1,either DP2 or a replacementPz,eruptingP3 (DP3 already shed),DP4 still in place and presumablycovering a replacementP4if that tooth had formedyet, M1, M2, andfragments of the developingM3. In the lowerjaw the following sequenceofteeth is present:dil or il (in the left dentaryin front ofdi2 thereis a root identified as di1 or iI), d12or i2, di3 or 13,di4 or i4 (or somecombination of at leastthree of these),dc, (d)pl, eruptingp2 , dp2 still presentbut lying wholly behind p2 ratherthan aboveit (resulting in a premolar count of 5 in use at one time), erupting p3, dp4 still in position over the site of a presumablydeveloping p4, eruptedml and m2, andm3developing in its crypt but unerupted.We postulatethat an adult specimen of Daulesteswould havehad an unknownupper incisor count but possiblymore than three,C, (D)P1,P2,P3,P4, Ml, M2, M3 /lowerincisorcountofat leastthree,c, (d)p1, p2,p3,p4,mI,m2,m3.

Lower dentition (Figs 13-16, Table 3). Incisors.- A partial root of an apparently large anteriorincisor, designateddil (or i1) and the last two incisorsare preserved in the left dentary.Because of ambiguityconcerning the numberof lower incisors,we as- signdi1 (or i1) to the first tooth for which the root is preserved.The next incisor (di2 or i2), which is the first tooth with a completelypreserved crown (penultimateincisor) projectsanterobuccally from its alveolusat about45", bears a simple,blunt apex,and hasa somewhatoval cross-sectionwith anteriorand posterior cresting. It is wider posteriorly than anteriorly. Lingual to its alveolus is a second,empty alveolus that either housedan additionalincisor or possiblya deciduouspredecessor of the penultimateincisor.The ultimateincisor (di3 ori3) is a very small,simple, peg-like tooth, projecting also at about 45' and lying somewhatposterolingual to the penultimateincisor, to which it is closelyappressed. Although thereare three or possiblymore lower incisors, the retentionof deciduousteeth among replacement incisors cannot be discounted. The lower deciduouscanine is double-rootedas seenfrom the buccalside, the anterior root being the smaller of the two. There is an anteriody located,transversely appressedmain cusp and a low heel at the rear, giving the tooth a premolarifoffn as- pect.In view of the depthof thejaw supportingit, it is difficult to imaginea largertooth at this locusat the stageofontogeny represented, so we regardthis tooth asa deciduous rather than replacement lower canine. It occludes in front of the much larger, single-rootedupper caniniform tooth, which we alsointerpret to be deciduouson thebasis of its similarity to DC of Kennalestes. The first lower premolar(dpl or p1), which usually is called dp1 in placentalsexcept tapirs (seeSimpson 1945), is a moderate-sized,completely erupted premolariform tooth; it is double-rootedand hasa blade-like apex.A distinct, small heel cusp lies at the posterior end of the tooth. A sharpcrest runs from the heel cusp all the way to the apexof the main cusp.Compared to its neighbors,the tooth leanssomewhat lingually. The secondlower premolar(interpreted here as ap2 that did not displacethe dp2 behind it during eruption;Luckett 1993)is a small, double-rootedtooth almostfully eruptedon the animal's left side and incompletelyerupted on the right side.Its morphology is generally similar to that of its neighbors,which tightly confine it. Full eruption ofp2 andp3 would doubtlesspush dp2 out ofthe dentary.Posterior to p2 is a double-rooted,antepenultimate, lower premolarthat we interpretas a non-sheddp2. This tooth formed and eruptedbehind, rather than above,p2. The posteriorroot of dp2 is ACTA PALAEONTOLOGICAPOLONICA (45) (1) 31

Table 3. Optical micrometer measurements(mm) of lower dentition of Daulestes nessovi sp. n., ZIN C. 79066, taken with Shopscope*equipment. (AP, anterior-posterior).

Length (front dc - m2) projectedon sagittalplane (estimated) 6.5 AP length right dc at alveolus, projected on sagittal plane 0.53 AP length left dc at alveolus, projected on sagittal plane 0.52 Width rieht dc at alveolus 0.33 Width left dc at alveolus 0.26 AP length right dpl at alveolus, projected on sagittal plane 0.57 AP length left dpl at alveolus, projected on sagittal plane 0.58 Width right dpl at alveolus 0.28 Width left dpl at alveolus 0.31 AP lenglh right p2 at alveolus, projected on sagittal plane 0.47 AP length left p2 at alveolus, proiected on sagittal plane 0.47 Width right p2 at alveolus o.29 Width left p2 at alveolus 0.21 AP length right dpZ at alveolus,projected on sagittalplane 0.59 AP length left dp2 at alveolus,projected on sagittalplane 0.58 Width risht dp2 at alveolus 0.31 Width left dp2 at alveolus 0.31 AP length incompletely erupted right p3, on angle, projected on sagittal plane 0.82 AP length left p3, on angle,projected on sagittalplane 0.72 Width, right p3 o.4l Width,leftp3 0.45 AP max. length right dp4, paraconid to hypoconulid, projected on sagittal plane 1.06 AP max. lensth left dp4. paraconid to hypoconulid, proiected on sagittal plane 1.09 Width hisonid risht dp4 0.68 Width talonid rieht dp4 0.58 Width trigonid left dp4 0.65 Width talonid left dp4 0.58 AP max. Iength right ml, paraconid to hypoconulid, projected on sagittal plane 1.20 AP max. length left ml, paraconid to hypoconulid, projected on sagittal plane 1.21 Width trigonid right ml 0.79 Width talonid rieht ml 0.68 Width trieonid left m1 0.79 Width talonid left ml 0.68 AP max. length right m2, paraconid to hypoconulid, projected on sagittal plane 1.16 AP max. length left m2, paraconid to hypoconulid, projected on sagittal plane t.r4 Width hisonid rieht m2 0.80 Width talonid rieht m2 0.71 Width trieonid left m2 0.83 Width talonid left m2 0.68 Earliest eutherian mammal skal/: MCKENNA et al.

DP1 DP2 P3 (or P2) B

brokenroot ol di1or i1

di2 di3 dc dp1 p2 dp2 p3 dp4 m2 alveolus for m3 alveolus tor ?i2

lmm E

Fig. 16. Slightly diagrammatic drawing of the left upper and lower dentitions of Daulestes nessovi sp. n. (ZIN C. 79066). A, B. Upper dentition in lateral and occlusal views. C-E. Lower dentition in medial, lateral, and occlusalviews. noticeablylarger than its anteriorone. The crown of dpZ has a steeplyrising anterior edgeand a slopingposterior edge somewhat more blunt than is found in neighboring premolars.There is only a small posterobasalheel cusp(seen in buccalview). On the right side,pI,p2 anddp2appear heavily worn or broken;the latterinterpre- tation is favoredbecause the eruptingp2 shouldnot haveextensive wear. The penulti- ACTA PALAEONTOLOGICAPOLONICA (45) (1) JJ mate lower premolar(p3) is incompletelyerupted in both lower jaws. The talonid is below the alveolarmargin in both rami. Had the animal lived longer,the tooth probably would haverotated somewhat more during the last stageof its emplacement.How- ever,it is nearlyin the positionit would havehad in an adult,but not yet so highly projecting. Thereis a single,high trigonid cusp,presumably the protoconid,without trace of either paraconidor metaconid.The crest leadsfrom the protoconidapex postero- lingually, ending at the trigonid base.A large, single, blade-like heel cusp projects stronglyto the rear. Four molariform lower teethare indicated: dp4, ml,m2, andm3. Of these,the first three have erupted,whereas the last molar is representedonly by a large dentary alveolusand fragmentary remains within the left alveolus.The sizeof the eruptedteeth increasesfrom dp4 throughm2. If our assumptionthat M3 wasreduced (see above) is true, onecan presume that m3 might alsohave been relatively small.At leastdp4, ml, ardm2 aredouble-rooted, but the numberof rootsof the developingm3 is unknown. The first molariform lower tooth is believedto be dp4. Although its weardoes not show any significantdifference from the molars,X-ray photography(not figured) re- vealsthat the toothhas thinner enamel and more open roots than the molars.Moreover, a p4 germis visible betweenthe roots of dp4 within the dentaryin a radiographicpho- tograph.The dp4 hasa moreopen trigonid thanthe succeedingteeth. Its paraconidjuts forward as a short blade ending anterolingualto the protoconid,reaching approxi- matelythe mid-point of the anteriorface of the trigonid. The protoconidis higherthan the metaconidand thosetwo cuspstogether form a high, nearly straight,posteriorly facing wall. On all molariform teeth,the threecusps of the talonid are symmetrically arranged(hypoconulid and entoconidare not twinned). On dp4 the entoconidis low andsomewhat crested, the hypoconulidcentrally located, the hypoconidthe highestof the threecusps. A crestfrom the entoconid(entocristid) runs forward to the posterolingual margin of the rear wall of the trigonid. A crest from the hypoconid (cristid obliqua) reachesthe centralpart of the baseof the rear wall of the trigonid. Both of thesecrests, especially the cristid obliqua (betterpreserved in the left lower jaw) are roundedand wide rather than sharp,which is due to the youth of the individual and consequentlack of wear.Between the entocristidand cristid obliquathe talonid basin is a deeprounded depression; to therear the talonidbasin gradually raises. There are no cingulids,but a faint swelling may be seenalong the buccalbase of the paraconid. The first lower molar hasa more compressednigonid than the precedingmolariform tooth (dp4),with the metaconidless protruding posterodorsally. The talonid is wider, with alatger hypoconulid that points posterodorsally.A strong anterior cingulid rises steeply on the anteriorslope of the protoconid,terminating at the boundarybetween the basesof the protoconid and the paraconid.Other details are essentiallythe sameas in dp4. The secondlower molar hasa somewhatcompressed and recurved trigonid, with a blade-like paraconidmore lingually placed than in the precedingmolariform teeth, sharplydelimiting a deeptrigonid basin.The talonid is similar to that of ml, but the hypoconulidpoints dorsallyrather than posterodorsally. An anteriorcingulid like that of ml is present.There is afainttraceof a posterobuccalcingulid below the hypoconid and hypoconulid. Only a traceof the developingcrown of m3 is preservedin the incompletelyformed alveolus at the rear of the cheek-toothrow of the left dentary. 34 Earliest eutherian mammal skal/.' MCKENNA et a/.

Upper dentition (Figs 4-6, 16,4.,B, 17, 19l.Table 4). - Upper incisorsare not preserved.However, they probablywere numerous, judged from the lower dentition.The upper caniniform tooth (DC) is the most anterior tooth present,partly broken on the left sideand originally completelypreserved on the right sidebut now missing,only an alveolusbeing preserved. It is a projectingand canine-liketooth, oval in crosssection andwith a simpleapex. We interpretit to be a fully eruptedbut not yet sheddeciduous caninelikethatof ajuvenile skullof Kennalestes,ZPALMgMVl (Kielan-Jaworowska 1981),not an incompletelyerupted replacement tooth. There apparentlywas a single root, as in ZPAL MgMU1. The dorsalend of the root is borderedposteriorly by the maxilla, but it may have been bordered anteriorly either by the maxillary or by the missingpremaxilla. If the caniniformtooth shouldprove to be a replacementtooth, its single root would be a profound difference from the double-rooted and shorter- crowned replacementupper canine of Kennalestes,Asioryctes, Zalambdalestes and other taxa with double-rootedpemanent uppercanine teeth. At the sametime, a single-rootedupper permanent canine would be a similarity to Hyotheridium,Ukhaathe- rium, carnivores,creodonts, and cimolestids.However, the single-rootedcondition observedin Barunlestes(Kielan-Jaworowska 1975a) is obviouslyconvergent, associated with reduction of the anterior teeth other than the enlargedincisors. Projectingless than the deciduousupper canine(or possiblyits partly eruptedre- placement),the first tooth of the upperpremolar series is double-rootedand somewhat elongateand laniary. It bearsa high paracone,a faint indicationof an anterobasalcusp, and a weak posteriorbasal swelling suggestiveof a heel. The anteriorsurface of this premolar is slightly convex and oriented nearly vertically, whereasthe posterior slope of the main cuspis more sloped.There is no indicationof a replacementtooth within the maxilla dorsal to the roots. We identify the preservedtooth as the tooth usually called DP1 in placentalsother than tapirs (Simpson1945). A double-rootedsecond premolar (DP2 or P2) is preservedon the right sideof the skull but is missing on the left side,making a gap therethat separatesDPl from P3. The right tooth is much lower and apparently more worn than the first premolar, and immediatelyfollows it. The secondpremolar is essentiallya small replica of the first, as in Ukhaatherium andAsioryctes but not Kennalestes(Novacek et al. 1997 Kielan- Jaworowska1981). P3 had almost completedits eruption in the right maxilla, but is somewhatless erupted on the left side of the skull. At this stagein ontogeny P3 lies anteroventralto the anterior opening of the infraorbitalcanal; its position when fully eruptedmay have shifted. Although primarily an anteroposteriorly trenchant tooth, P3 appearsto have threeroots: anterior,posterolingual, and posterobuccal. P3 is the highestupper cheek tooth.The crown is a very high, bladeJikestrucfure consisting of a paraconeand its as- sociatedcutting edges.The anterior cutting edge is vertically orientedas a convex curve thatjoins a basalcusp anteriorly.A concavecrest runs from the paraconeapex posterodorsallyto a pointjust lingual to a posteriorbasal heel cuspthat appearsto be suppliedwith short cingula extendinga short distanceforward along the lingual and buccalsides of the crown base.The posteriorend of the lingual cingulumis slightly enlarged.There is no traceof a metaconeon this tooth.P3 resemblesthe penultimateupper premolar of Prokennalestesin its stage of molarization, although in Proken- nalestesthe posterolingualcingulum is possiblyslightly moreprominent. It is lessad- ACTA PALAEONTOLOGICAPOLOMCA (45) (1) 35

Table 4. Optical micrometer measurements(mm) of upper dentition of Daulestes nessovi sp. n., ZIN C. 79066, taken with Shopscope* equipment. (AP, anterior-posterior).

AP length DC at alveolus,projected on sagittalplane 0.45 Width right DC at alveolus 0.38 AP length right DPl, projected on sagittal plane 0.59 AP length left DP1, projected on sagittal plane 0.51 widthrishtDPl 0.27 widttr left DPI 0.29 AP lengthright DY2 orP2, projectedon sagittalplane 0.69 AP length left DP2 or P2, projected on sagittal plane 0.63 Width right DP2 orP2 0.24 Width leftDP2 orP2 0.2'1 AP length right P3, parastyleto metastyle,on angle,projected on sagittalplane 1.05 AP length left P3, parastyle to metastyle, on angle, projected on sagittal plane t.t4 Width right P3, normal to sagittalplane 0.62 Width left P3, normal to sagittal plane 0.73 AP length right DP4, parastyleto metastyle,projected on sagittalplane 0.89 AP length left DP4, parastyle to metastyle, projected on sagittal plane 0.93 Width right DP4, normal to sagittalplane t.l4 Width left DP4, normal to sagittalplane 1.15 AP length right Ml, parastyle to metastyle, projected on sagittal plane r.t7 AP length left Ml, parastyle to metastyle, projected on sagittal plane 1.16 Width right M1, normal to sagittal plane r.42 Width left Ml, normal to sagittal plane 1.49 AP length right M2, parastyleto metastyle,projected on sagittalplane l. l5 AP length leftM2, parastyle to metastyle, projected on sagittal plane l.l7 Width right M2, normal to sagittal plane t.62 Width left M2, normal to sagittal plane 1.68 vancedin this respectthan that of Kennalestes,in which P3 has a distinct heel. We havenot seenany tracesof root remnantsof the shedDP3. DP4 is the smallestof the eruptedsequence of molariform teeth.Atooth germrep- resentingP4 is presentin the maxilla aboveit. DP4 is lesstransverse than the preserved molarsbut is by no meansequilateral. As a molariform tooth,it bearsa fully developed protocone,paracone and metacone,of which the high, somewhatforward-leaning paraconeis the largestcusp. The metaconeis presentposterior to the paraconeand is about half its height. The basesof the paraconeand metaconeare not conjoined,in contrastto M1. Extendingalong the anteriormargin of the paraconeis a niurow groove for reception of the protoconid of dp4. Anterior to the groove the tooth margin is 36 Earliest eutherian mammal sftall; MCKENNA et al.

Fig. lT.SEMstereomicrographoftherightupperdentition,occlusalview,of Daulestesnessovisp.n.(ZIN c.79066).

incurved.A strongprotocone lies lingually. The tooth has weakly developedwinged conules,of which the paraconuleis largerand projects somewhat anterolingually. The paraconuleis connectedby a short anterior cingulum to the parastylararea. The metaconuleis merely a swelling along the crest from the protocone apex to the posterolingualbase of the metacone.The prominentparastylar projection incorporates a poorly developedparastyle and stylocone.The labial edgeof the stylar shelf posterior to the styloconeis raisedas a ridge and appearsto beartwo or more tiny cuspules, the mostposterior of which would be the metastyle.A ridge runsfrom the metastyleto the posterolabialside ofthe metacone,extending up to its apex.There are three roots. Ml is triangular and is more transversethan DP4 but less so than M2. The high paraconedominates the much lower metaconeand is basally conjoinedwith it. The paraconeleans slightly forward and has a convex anterior surfaceand a concave crest-likeposterior one continuouswith the crest-like,transversely appressed metacone.The metaconeis abouthalf the heightof the paracone.At the tooth's lingual apex the protoconeis narrowand juts slightly forward,aligning with theparacone. A strong, winged paraconulelies at the anterolingualbase of the paraconeand juts anteriorly. The metaconuleis much smallerthan the paraconule,being a swelling along the crest ACTA PALAEONTOLOGICAPOLONICA (45) (I) JI

Fig.18. SEM stereomicrograph of the right upper dentition, lateral view, of Daulestes nessovi sp. n. (ZIN c.79066). from the protoconeto the posterior base of the metacone.The prominent anterolingually placedparaconule and somewhatmore posteriorly locatedmetaconule are situatedon the anteriorand posterior edges of the trigon, respectively,with the deepest concavity of the trigon basin betweenthem, in contrastto a more lingually placed trigon basin in many mammals.The parastylarwing juts sharplyanteriorly from the anterolabialbase of the paraconeand bearsa low parastyleand a low stylocone.In occlusalview theparastylar wing is partially hiddenby the metastylararea of DP4, and thereforeappears smaller than if viewed laterally.In lateral view, the parastylepro- trudesslightly ventrally dorsalto the ventral margin of the parastylarwing, whereas the styloconehas an appearanceofarounded cusp.A crest(preparacrista) connects the paraconeand stylocone.The groovefor receivingthe protoconidis wider thanthat of DP4. The ectoflexusis gently curved,being slightly shallowerthan on DP4 andmuch shallowerthan on M2. The metastylarwing is more prominentthan on DP4 andbears severaltiny cuspulesat its raised buccal margin. A swelling extendsbetween the parastylarand metastylarregions along the ectoflexus,at the end of which in the metastylarregion small cuspulesprotrude slightly ventrally. Posteriorand anterior cingulaare not present.There are three roots. 38 Earliest eutherian mammal sftzl/: MCKENNA et al.

Fig. 19. SEM micrographsof the left upper dentition of Daulestesnessovi sp. n. (ZIN C.'79066). A. Occlusalview B. Lateral view.

M2 is the widest of the eruptedmolariform series.It differs from Ml not only in proportions(see dimensions), but especiallyin having a strongly asymmetricalstylar shelf (seebelow). As on Ml, the paraconeand metaconeare conjoinedat their bases but each is slightly smaller than the coffespondingcusp on Ml. As on Ml, the paraconeleans anteriorly, whereas the smallermetacone points ventrally.The protoconeis a high and piercing cusp,connected anterior$ by a low crestto a prominent paraconule,from which a preparaconulecrista continuesas the anterioredge of the tooth to the baseof the parastyle.A postparaconulecrista runs from the paraconule apex posterolabiallytoward the posterolingualbase of the paracone.The winged paraconuleis larger than the metaconuleand, as on M1, the paraconuleis more lingually placed,being closerto the protoconethan to the paracone.Although occupying the anterioredge of the tooth, the paraconuleprojects less anteriorly than that of M 1. The metaconuleis a tiny cuspulethat lies alongthe postprotocristafrom the protocone to the posterior base of the metacone;it is, however,V-shaped, with short conule crests.The parastylarregion is very prominentand protrudes more labially thanthat of M1 so that the ectoflexusis deeperthan on Ml. In occlusalview the parastylarregion is largerthan that of Ml, but appearsslightly smallerin lateralview. It bearstwo cusps: ACTA PALAEONTOLOGICAPOLONICA (45) (1) 39

a more anterior and prominent parastyle and a barely developed sfylocone, situated closeto the ectoflexus.In occlusalview, the anteriormargin of the parasfyleis slightly obscuredby the posteriormargin of the metastylarareaof Ml, which eruptedearlier. In lateral view, the styloconeis more roundedthan on M1. The groovefor the reception of the protoconidof m2 is similar to that of Ml. Lingual to the parastyle,the anterior margin of the tooth is strongly incurved. Extending anterolabuccally from the anterobuccalside of the paraconeto about the middle of the parastylar lobe, there is a weak crest. In contrastto the very prominentparasfylar wing, the metasfylarone is much less developed.The labial margin of the metastylar wing is gently rounded and only weakly protrudeslabially. Thereis no metastylarcusp, but a ridge extendsalong the buccal margin of the metastylar shelf, bearing several poorly defined cuspules, betterseen on the left M2 thanon theright one.The metacristaextends along the posterior margin of the tooth to the baseof the metacone.In lateralview, the metacristahas the appearanceof a tiny cuspule. The crown of M3, normally the last tooth erupted in the molariform sequencein placentals,is not preserved.Because of the immaturestage represented by zN c. 79066, M3 simply was not yet fully formed and is poorly fossilized. There are, however,some clues about its size.Large parastylarand reducedmetastylar wings of M2 occur in ani- mals in which M3 is reducedor lost, so that M2 becomesM3-like (e.g.n Deltatheroides, see Kielan-Jaworowska 1975b: fig. 1). The difference in size of the parastylar and metastylarwings of M2 in Daulestessuggests reduction of M3. The preservedalveolus also suggestsa reduced but still three-rootedM3. Traces of an apparently reduced, three-rootedtooth can be seenat the rem of the left maxillary plate @gs 168 and 19).

Anatomical comparisons

Groove for internal carotid and stapedial arteries. - Early ideas of Gregory (1910),Matthew (1909),Maclntyre (1972),McKenna (1966), and orhershave been 'medial replaced by a concept that the promontory and branch' of the carotid artery (entocarotidartery) are one and the same(Presley 1979;MacPhee 1981; wible lgsT). 'measurements conroy & wible (1978:p. 84) cautionedthat of the promontoryca- nal in mammalianpaleontological studies do not provide unequivocalevidence of the size or, indeed,the presenceof the promontory artery in fossil forms'. In addition, presenceof a stapedialforamen in the stapesdoes not necessarilyimply presenceof the stapedial artery,because the artery is absentin marsupialsand some placentalsin which a stapedialforamen persists (Bugge 1974;wible 1987).Nonetheless, arteries are commonly reconstructedfor fossils that have,or evenlack sulci or grooves,as in Probainognathus,Morganucodon, multituberculates, Vincelesfes, various eutherians, and the Khoobur (Khovboor) pefrosalfrom the Early Cretaceousof Mongolia (Kielan- Jaworowskaet al. 1986; Miao 1988; Luo 1989; Wible 1990; Rougier et al. 1992; Wible & Hopson 1995;Wible et aI.1995). Without a grooveto dictatethe courseof a possible artery,reconstruction of the blood vesselsbecomes highly arbitrary, as shown, for instance, by three different patterns of the internal-stapedialarteries pro- posedforMorganucodon (Wible 1990:fig. 7A; Rougieret al.1992: fig. 88; Wible & Hopson 1995: fig.68). Given inferred soft structureto be unreliableas charactersin 40 Earliest eutherian mammal skllll: MCKENNA et al

fossils,we adoptCifelli's (1982)criterion that only grooveson the promontoriumare countedas characters. Taxa bearing grooves on the promontorium,possibly for the internal carotid artery and/orstapedial artery,include multituberculates,Vincelestes, eutherians,and Daulestes. One petrosalof an unidentifiedtribosphenic therian pos- sessesagroovesimilartothatofVincelestes (Meng&Fox I995a).Thepositionofthe groove for the promontory artery in Daulestesdemonstrates, again, that the trans- promontorialcondition is primitive for placentals(Wible 1987).

Fenestracochleae and aqueductuscochleae. - The fenestracochleae ('fenestra rotunda', or'round window') and aqueductuscochleae are structuresrelated to the perilymphaticsystem of the inner ear,which functionsas the route that channelsfluid vibrationsand releases pressure provided by soundsconducted to the petrosalby way of the stapes(Gray 1908;de Beer 1937;Wever I97B);therefore, they areimportant for hearingfunction. Meng & Fox (1995b)have summarizedseveral features of this region typical for therian mammals:(1) merging of the perilymphaticrecess with the basal part of the scala tympani within the petrosal (Gray 1908); (2) ttue fenesffa cochleaeand aquaeductuscochleae resulting from developmentof the processus recessus(de Beer 1937;Zeller 1985a,b; Wible 1990);(3) distantseparation of the fenestracochleae and fenestra vestibuli by a bony bridge; and (a) the basalpart ofthe secondaryspiral lamina extendson the inner surfaceof the bony bridge betweenthe two fenestrae,implying a basalelongation of the basilarmembrane to this area. In most nontherian mammaliamorphs,such as morganucodontids,multituberculates,and monotremes,the perilymphaticduct is not enclosedby a bony canal,but held by a narrow sulcusconfluent with the lateral apertureof the perilymphatic recels (recessusscalae tympani; seeFox & Meng 1997,for review). The perilymphaticre- cessis a pouchmedial (proximal) to the perilymphaticforamen, holding the perilymphatic sac,in nonmammalianamiotes and monotremes(Gray 1908; de Beer 1937; Wever 1978;Kuhn I97I; Zeller 1985a,1985b, 1993). Such a pouchis presentin multituberculates,illustrated by many (e.g., Kielan-Jaworowskaet al. 1986; Luo 1989;Liltegraven & Hahn 1993; Wible & Hopson 1993, t995), which is explicitly identified asthe perilymphaticrecess by Fox & Meng (1997).External to the recessin multituberculatesis the opening,the lateralaperture of the perilymphaticrecess, over which the secondarytympanic membrane is presumablystretched. Previously, the lateral aperturein nontherianmammaliaforms, except monofremes, was not recognized 'fenestra andthe perilymphaticforamen was variably andincorrectly called cochleae', 'round 'fenestrarotunda',or window' (e.g.,Kermacket aL.1981; Kielan-Jaworowska et al.1986; Wible 1990,1991; Lillegraven & Hahn 1993;Luo I994;Luo et aI.1995; but seeWible & Hopson1993).Neither the perilymphatic foramen nor the lateralaper- 'fenestra ture of the perilymphatic recess should be equatedwith the cochleae', 'fenesffarotunda', or 'round window', which are only correctly appliedto the condi- 'processus tion in which the aqueductuscochleae is formed. Coining the term recessus' to substituteVoit's (1909) 'processusintraperilymphaticus', de Beer (1929) rejected Gaupp's(1900) hypothesis that the fenestracochleae (rotunda) and aqueductus cochleaeare formed as subdivisionsof the perilymphaticforamen; instead, de Beer (1937:p. 402)proposed that the secondarytympanic membrane stretches over the lateral apertureof the recessusscalae tympani, part of which becomesthe fenestraco- ACTA PALAEONTOLOGTCAPOLONTCA (45) (1) 4l

chleae.The lateral aperture,therefore, can be viewed as the progenitor for the fenestra cochleae. In other than tribosphenic therians, a true fenestra cochleae (round window, fenestrarotunda) has been recorded inVincelestes and the Khoobur petrosalfrom the Early Cretaceousof Mongolia (Rougieret al. 1992,1996a;originally, a perilymphatic foramen was reported from the latter by Wible et al. 1995). Neither of these have a fully coiled cochlearcanal (= a full turn of 360'). The fossil recordshows that development of the fenestracochleae and aqueductuscochleae occurred prior to the full coiling of the cochlear canal. Whether the secondaryspiral lamina is developedin Daulestesisunknown,butitwascodedforVincelestes (Wible etal.1995),of which the inner ear hasyet to be described.

Curvature of cochlea.- In most reptiles,the cochlea(or its equivalent)is short and venfralto thevestibule (Miller 1966,1968BardI970a,b,1974; Wever1978; Lewis et al. 1985).The cochlearduct of a crocodile, amongreptiles, is elongate(Manley 1970;Wever 1967 , 1978)and bends anteroventrally (Baird l970a,b: p.260, fig.27; Wever 1978).In birds,the cochlearduct is alsoelongate and bends medially, in a direc- tion oppositeto that of monotremesand therians(Pritchard 1881). In nonmammalian cynodonts,the cochleais generallyshort and is anteroventralto the vestibule(Estes 1961;Quiroga 1979;Allin & Hopson 1992),with the exceptionof the reconsffuction of a cynodont cochleathat is elongateand extendsanterovenffally and then anteriorly (Simpson1933). The cochlearcanal of multituberculatesis in a somewhathorizontal position and bent laterally in variable degrees(Simpson 1937;Luo & Ketten 1991; Fox & Meng 1997).Recent study (Fox & Meng 1997)demonstrates that the cochlear canalsin monotremescurve around180', differing from the conventionthat the platypus hasa cochleawith a 2T0 ctrlivahne;consequently, the monotremecondition is not significantlydistinctive from that of multituberculates.InVincelestes the cochlearcanal curvesaboat270" (Rougier et al.1992; Wible & Hopson 1993).The cochlearcur- vaturein Daulestesis greaterthan that of monotremes,close to a full circle of coiling. Like the cochlearcanal with a constantdiameter throughout its entirelength in monotremes(Gates et al. 1974),and like a petrosalof an unidentifiedtribosphenic therian (Meng & Fox I995a), the cochlear canal of Daulestes doesnot taper, and has an expandedapex as in monotremes.

Ectotympanic. - Besidesmonotremes and therians, a fully suspendedectotympanic has been reported in multituberculates(Miao & Lillegraven 1986; Meng & Wyss 1995;Rougier et aL.1996b).The ectotympanicin multituberculatesdoes not expand ventrally, differing from the condition in Asioryctes(Kielan-Jaworowska 1981) and Daulestes; however, it bearsan internal groove for attachmentof the tympanic membrane,as in extantmammals. The conditionsin Asioryctesand Daulesles appear more derivedthan in multituberculatesand didelphidmarsupials in the ventralexpansion of the ectotympanic,but a slender,ring-shaped ectotympanic is often found within eutherians,such as leptictids(Novacek 1986).

Malleus. - The malleusis nearly complete,the bestpreserved among known Creta- ceous marnmals.As in the case of the ectotympanic,the only non-monofreme, nontherian mammalsin which the malleus is recorded are multituberculates (Miao & /a Earliest eutherian mammal skzll; MCKENNA et al.

Lillegraven 1986; Meng & Wyss 1995; Hurum et al. 1996; Rougier et al. I996b). Again, the bestmorphology of the multituberculatemalleus is from Lambdopsalis,in which the compositenature of the malleus- a combinationof the articular,goniale, and prearticular(de Beer 1937;Allin & Hopson 1992)- is shownby presenceof the foramenfor the chordatympani nervein the massiveanterior process (Meng & Wyss 1995).Although the condition of preservationimpedes identification of this foramen in the malleusof DaulesteJ,we assumea compositemalleus in Daulestes.One of the featuresthat distinguish monotremesfrom theriansis the dorsal placementof the incus: the articulationfacet is on the dorsalside, at the anteriorpart of the body, of the malleus.A fossaat the sameposition in the malleusof Lambdopsallssuggested the dorsoventralincudo-malleus relationship in multituberculates(Meng & Wyss 1995, 1996),although this was questionedby Rougier et al. (I996b), basedon a specimen (Miao & Lillegraven 1986)in which displacementwas suspected.A detaileddiscus- sion concemingthe shapeand position of the malleusand incus in multituberculates was providedby Rougieret al. (I996b).In Daulestesthe articulationof the incus and malleusis therian,demonstrated by a thickenedarticular area posterior to the body of the malleus.

Dentition. - Kielan-Jaworowska(1984c: p. 178) statedthat: '...theLare Cretaceous Asian eutheriangenera, although differing from each other in the coronal structureof the molars, are similar in the general appearanceand orgarization of the dentition, when seenin lateral view. All have a strongdouble-rooted upper canine(secondarily single-rootedin Barunlestes),situated some distance to the rear of the premaxillary- maxillary suture,followed, after a short space,by small Pl and P2 that are double-rootedand do not comeinto occlusionwith the lower premolars.In all generaP3 is the strongesttooth of the whole series,with a high, piercing paraconeand P4 is semi-molariform.' Daulestesshares with Mongolian Late Cretaceouseutherian genera Kennalestes, Asioryctes,Zalambdalestes and Barunlestes the above-described pattern of the premolar andmolar sequence,but differs from adultsof the first threegenera in having a single-rootedupper canine asinHyotheridium andUkhaatherium.We believe,however, that the tooth at the caninelocus of thejuvenile skrtllof Daulestesis deciduousrather thanpermanent, and thus it is not excludedthat the uppercanine of adultscould have beendouble-rooted. The situationwould thenbe ashas been demonstrated for juvenile vs. adult specimensof Kennalesles(Kielan Jaworowska1981). Because the permanent uppercanine in Daulestesremains to be found, for the time being we tentatively acceptthat it was single-rooted.As in the other Mongolian Cretaceousgenera men- tioned,the upperdeciduous canine of Daulestescould be someunknown distance behind the premaxillary-maxillary suture,but it could as well be bordered anteriorly by the now-missingpremaxilla. Comparison with the Early CretaceousMongolian Pro- kennalestes(Kielan-Jaworowska & Dashzeveg1989; Sigogneau-Russell e/ al.1992) cannotbe made,because in Prokennaleslesneither the uppercanine nor its deciduous predecessorhave yet beenfound. The dentitionof Daulestesshows a mixture of primitive and advancedcharacters. The lack of pre- and postcingula(shared with Otlestes,Asioryctes, Ukhaatherium, Zalarrbdalestidaeand somePalaeorvctidae. Cimolestidae. and miscellaneouszalarrrb- ACTA PALAEONTOLOGICAPOLONICA (45) (1) 43

dodont mammals),the sequenceof premolarsand molars mentionedabove (shared with Prokennalestesas well as the Mongolian Late Cretaceousgenera), the possible (but uncertain)presence offour lower incisors,and perhapsthe apparentpresence of both dp2 andp2 in place at the sametime (of indefinite duration),one in front of the other, are all primitive characters. DPl and the secondupper premolar (DP2 or P2) in Daulestesare very similar to thoseof Kennalestesand Asioryctes, and do not come in contactwith the lower premolars;the differenceis that,in Daulestes,there is no diastemain front of DP1, or be- tweenDPl andthe secondupper premolar, or betweenit andP3, althoughthis may be the resultof the relativeyouth of the new specimen,ZIN C.79066.P3 of Daulestesre- semblesthe penultimateupper premolar of Prokennalestesin its stageof molarization, although in Prokennalestesthe posterolingualcingulum is possibly slightly more prominent.Daulestes is lessadvanced in this respectthan Kennalestes or Asioryctes, wherein P3 is provided with a distinct heel. In ZalambdalestidaeP3 is widely trans- verseand hasdeveloped a true protocone. P4 in MongolianLate Cretaceousgenera and its equivalent,the last upper premolar in Prokennalestes,are semimolariform and transversely wide, but lack developmentof more than affaceof a metacone.In juvenile DaulestesDP4 is completelymolariform. It is shorterthan M1 or M2 andits metaconeis well developed,although lower thanthe paracone.As we identify the last premolar of Daulestesas DP4, we are ignorant of whetherits permanentsuccessor is molariform. Featuresof Daulestesthat are advancedinclude the winged molar conulesand a very specialstructure of the stylarregion, in which the metastylarprojection is reduced as tn Asioryctes (see crompton & Kielan-Jaworowska1978) and ukhaatherium (Novaceket aI. 1997),indicating an apparentreduction of M3.

Allocation to and within Eutheria

Allocation of Daulestesto Eutheriais supportedby its dentalformula andthe general similarity to variousCretaceous basal eutherians. We do not compareDaulestes with 'tribotheres'. metatheriansor with the so-called Butler (1978)proposed the infraclass Tribotheriafor Albian tribosphenicmammals from Texas,but subsequentlywithdrew his infraclass (Butler 1990) and the group is now used informally. some of the 'tribothere' genera,such as Pappotherium,Holoclemensia, Kermackia andslaugh- teria, have been regardedby some authorsas possibleeutherians (see Marshall & Kielan-Jaworowska1992, for summary),but we regardthem as Theria incertae sedis. Novacek (1986) madean attemptat the classificationof higher categoriesof Re- cent and selectedfossil eutherians,in which Cretaceouseutherians were not included. This wasunforfunate, as his classificationwas subsequentlyused in cuffent textbooks of vertebratepaleontology (e.g., canoll 1988;Benton 1990).Moreover, the diagnosis 'Eutheria' of given by Novacek(1986, see also Novacek 1982)was basedalmost ex- clusively on soft anatomyand physiology appropriatefor extantplacentals. The only 'relatively osteologicalcharacters given by him (1986: p. 81) were narrow stylar 'epipubic shelveson uppermolars' (a vaguediagnostic character) and bonesabsent'. Given that the majority of purportedearly placentals are known exclusivelyfrom teeth 44 Earliest eutherian mammal skull: MCKENNA er al. or fragmentsofjaws with teeth,identification of themas belonging to eutheriansusing Novacek's(1986) definition is virtually impossible.The presenceof epipubicbones is plesiomorphicfor mammals(Lillegraven 1969).Epipubic bonesare absentin all ex- tant and post-Cretaceouseutherians, but do occur in zalambdalestidsand Ukhaathe- rium (Novaceket al. 1997).Kielan-Jaworowska (I975c) first provided indirect evi- dencethat they might havebeen present in the Late CretaceousBarunlestes. They have beenlost in laterplacentals, possibly in relationto the prolongationof the gestationpe- riod (Lillegrav en 1969,197 5 ; Lillegraven et al. 1987). An assignmentof Daulestesto a known eutherianhigher systematic category poses difficulties, as is the casewith most Cretaceouseutherians. Nessov et al, (1994) as- signedDaulestes, poorly known atthattime, to Palaeoryctidae,while Nessov(1997: explanationtoplate48),identifiedCCMGEIll2I76as'Skullof aplacentalmammal, probablyDaulestes sp. nov.' Kielan-Jaworowska(1981) assignedAsioryctes(placed in her new subfamily Asioryctinae)to Palaeoryctidae.This assignment,accepted by Thewissen& Gingerich(1989), was rejected by McKenna& Bell (1997).McKenna & Bell placed Palaeoryctidaein magnorder Epitheria, grandorder Ferae, order Cimo- lesta.The only Late Cretaceouscimolestan recognized by theseauthors is Cimolestes, assignedto the Didelphodontidae,classified earlier as a didelphodontinepalaeoryctid (e.g.,Lillegraven 1969; Clemens I973;Fox 1984,1989). McKenna & Bell (1997)did not useAsioryctinae and assignedAsioryctes to magnorderEpitheria, without further qualification.Novacek et al. (1997)erected Asioryctitheria (no systematicrank), into which they placedAsioryctidae Kielan-Jaworowska, 1981 (assigningfamily rank to it). Charactersused to diagnoseasioryctitherians include (1) a tall, curvedcrista inter- fenestralisthat links the paroccipitalprocess to the promontorium;(2) a strongento- glenoid process;and (3) a postglenoidforamen that opensinside the glenoid area. Novaceke/ al. (1997)attributed to Asioryctidaetwo genera,Asl oryctes and Ukhaathe- rium, the latter genusonly preliminarily described. Daulestesis similar to the asioryctidgenera A sioryctesand Ukhaatheriuminboth the dentitionand skull structure(the latterdescribed fully only inAsioryctes).Daules- tes shareswith asioryctidgenera postcanine dental formula (the numberof incisorsis unknown) and reducedmetastylar projection on the molars. It shareswith Ukhaa- therium a single-rootedupper canine (as here interpretated;double-rooted in Asio- ryctes),lack ofpre- and postcingula,winged conules,and an enlargedparastylar region. It differs from both generain having upper molars wider antero-posterior$ and lesselongated transversely, with a well-developedmetaconule, which disappearedin Asioryctes(Crompton & Kielan-Jaworowska1978); Asioryctes is highly specialized in this respect.Daulestes differs further from both generain having a lessmolarized P3, without evenan incipient protocone,and an apparentlyless molarized p3. The snoutis similarly shapedin Daulestesand Asioryctes, and the arrangementof bones,as far as can be reconstructed,is similar. The lacrimal is poorly preservedin 'The Asioryctes,but, as statedby Kielan-Jaworowska(1984: p. 30) lacrimal is provided by a facial wing, probably pointed anteriorly...', which is also the case in Daulestes,There is a postpalatinetorus in both, anda postpalatineforamen, developed inAsioryctes as a notch.The braincase(incompletein Daulestes)is generallysimilar to that of Asioryctes,although the mesocranialregion, very long inAsioryctes,is relatively shorter in Daulestes.The shorteningconcerns in particular the size of the ACTA PALAEONTOLOGICAPOLONICA (45) (I) 45 presphenoidand orbitosphenoid,which arerelatively smallerin Dqulestes(a juvenile character?).The alisphenoidhas a thickenedanterior margin in both generaand over- hangsventrally the smallerorbitosphenoid. There are four foraminain this region in both genera,the differencebeing that the sinuscanal foramenin Asioryctes(Kielan- Jaworowska1984) is placedmore dorsally, and is clearly separatedfrom the sphenorbital fissure,whileinDaulestes theseforamina are hardly separated(possibly due to damage).The optic foramenis placedlow in the orbitosphenoidin both genera.The ethmoidforamen is more obviousinAsioryctes, but is recognizableon the left sideof the skull in Daulestes.The differencebetween Asioryctes and Daulestes concerns the apparentlack of the pterygoidprocess of the sphenoidin Daulesfes,which, however, may be easilylost in fossilization,and its lack may be due to the stateof preservation. The structure of the postglenoid processand arrangementof vascular foramina in the squamosalis similar,but not identical,in both genera.The promontoriumis very large in both, but possibly more horizontally arrangedin Daulestes.The ectotympanicis horseshoe-shapedinAsiorycresand U-shapedinDaulestes,which, however, may be a juvenile characterin the latter. Therelationshipsbetween Daulestes,Asioryctes, andUkhaatheriumonthe onehand, and Palaeoryctidae,on the other, cannot be demonstrated(contra Kielan-Jaworowska 1981). Asioryctidae and Palaeoryctidaeresemble one another in having upper molars very narrow and strongly elongated transversely (the transverseelongation is less strongly pronouncedin Daulestes);all have an enlargedparastylar region. They differ in having differently shapedlowermolars andpremolarsand differin dentalformulae (four premolars in Asioryctidae, three in Palaeoryctidae).The similarity of Daulestes and Palaeoryctidaeis that, in both, i2 is larger than i3 (tn Daulestespossibly dr2 and dl3), which is not the casein Asioryctidae.The most dramatic differenceconcerns the skull structure(see Thewissen& Gingerich 1989 and referencestherein, for description of palaeoryctidtaxa). In Palaeoryctidae,in confrastto Asioryctidae,the mesocranialregion is very short, there is no Vidian foramen, no sphenorbitalfissure, the optic foramen is placed far anteriorly, and there is an alar canal. In Palaeoryctidaethe glenoid fossa is placed much more anteriorly with respectto the promontorium than in Asioryctes and Daulestes,and the whole basicranialregion hasa very different structure.In view of these differencesit seemsreasonable to presumethat the similaritiesin uppermolar structureof Asioryctidae and Palaeoryctidaeare due to convergence,rather than relationship. Another eady eutherianthat invites comparisonwith Daulestesis the Early Creta- ceousProkennalestes Kielan-Jaworowska & Dashzeveg, 1989 (see also Sigogneau- Russelle/ a/. 1992).Prokennalesteshasbeen assigned by Kielan-Jaworowska& Dashze- veg (1989) to the paraphyleticorder Proteutheria,superfamily Kennalestoidea,family Otlestidae,but by Sigogneauet al. (1992) to Eutheria incertaesedis.Tl'rc order Proteu- theriawas erected by Romer (1966)to include incertaesedis eutheians then known from the Early Cretaceous,as well as non-lipotyphlanCretaceous and Tenary insectivorous mammalsgrouped into the Leptictoideaand Apatemyoidea. Butler (1972)recommended useof Proteutheria,but pointedout its waste-basketnature; he stated(p. 264): 'Furtherrc- searchshould reduce the sizeof the Proteutheria,by discoveringthe relationshipsof some of its groups to other orders, and possibly by making new orders out of other groups.' Many authors assignedCretaceous or early Tertiary genera to Proteutheria(see e.g., Lillegraven et al. 1979 and referencestherein). McKenna & Bell (1997) mbitrarily as- 46 Earliest eutherian ruamrual skzll; MCKENNA et al.

signedProkennalestes and Kennalesles to Gypsonictopidaewithin the superorderLeptic- tida. However, Kennalestesdiffers from other Leptictida (I\4cKenna 1975; Novacek 1986)in the lack of an entotympanicbulla and alisphenoidcanal. The skull andcomplete dental formula of Prokennalestesarc unknown.Daulestes shareswith Prokennalestesa plesiomorphiclack of pre- andpostcingula, but differs in having four premolars(five in Prokennalestes),different structureof the uppermolars with reducedmetastylar region (not reducedin Prokennalestes),winged conules,di2 or i2 larger than di3 or i3 (subequalin Prokennalestes),paraconid and metaconid higher with respectto protoconid,and lack of singlelabial mandibularforamen. Among Cretaceouseutherians, the skull is alsoknownin KennalestesandZalarrtb- dalestidae.Kennalestes is known from an almostcomplete dentition, in which only the number of incisors is uncertain(Kielan-Jaworowska 1969, I98I). Daulestesdiffers from Kennalestesin having uppermolars without pre- and postcingula,with reduced metastylarregion, and possibly a single-rootedupper canine. The basicranialregion in Kennalestes(less completely preserved than in Asioryctes)differs from Daulestesin the presenceof an extensivepterygoid processof the sphenoidand in having the sphenoidwing (referredto by Kielan-Jaworowskaas a basisphenoidwing) in the form of a fossa,which, however,may be an artifact of preservation.Kielan-Jaworowska (1981)reconstructed the lateralwall of the braincasein Kennalestesonthe Asioryctes pattern,but the homologyof somedetails is not certain. Daulestes shareswith Zalarrbdalestidaethe plesiomorphic lack of pre- and postcingula on the upper molars, but differs in the otherwisevery different structureof the dentition and skull. The differences are: larger stylar shelf on the upper molars in Daulestesthan in Zalambdalestidae,sfongly reducedmetastylar region (especiallyon M2), paraconeand metaconerelatively higher and placed more lingually, and more prominent and more lingually placed conules.P3 of Daulestesis lessmolarized than in Zalarbdalestidae,without a spur-likeprotocone. Daulestes differs fromZalambdalestes probably in having a single-rootedupper canine and, if so, is convergentin this character wilhBarunlestes.The lowermolars arevery differently shapedthan inZalambdalestidae, having a greaterdifference in height betweenthe trigonid and talonid. Further,Daulestes differs fromZalanrbdalestidaein the lack of enlarged,procumbent anterior lower incisor, lack of labial mandibularforamen, and lack of a strongly elongatedsnout. In braincase structurethe differencesare dramatic and concem the lack of the pterygoidprocess of the sphenoid in Daulestes, presenceof the Vidian foramen, lack of a separateforamen rotundum, relatively larger alisphenoid and orbitosphenoid,different disfibution of alisphenoid/orbitosphenoidforamina and different pattern of other charactersof the basicranialregion (Kielan-Jaworowska1969, 1984c; Kielan-Jaworowska& Trofimov 1980).McKenna & Bell (1997) assignedZalarrbdalestidae to the grandorderAnagalida in the superorderPreptotheria McKenna, 1975. Finally, a remarkably diversified eutherian clade that made its appearancein the Late Cretaceouscomprises insectivorans, formally classifiedas Lipotyphla (seeButler 1956,1988;McDowell 1958; Novacek 1986;MacPhee & Novacek1993). McKenna & Bell (1997)assigned three Late Cretaceousgenera (known only from teethand dentaries)to Lipotyphla. Theseare: (I) Batodon, which was usually assignedtogether with Cimolestesto Didelpho- dontinaein Palaeoryctidae(e.g., Lillegraven 1969; Clemens1973; Archibald 1982; ACTA PALAEONTOLOGICAPOLONICA (45) (1) 47

Fox 1984;see also cladograms in Nessove/a1.1998: frg.2q.Poorly knownupper mo- lats of Batodonare of Cimolestespattem and do not resemblethose of Daulestes,being more elongatedtransversely, having pre- andpostcingula, less prominent conules, and smallerstylar region without a reducedmetastylar part. (2) Otlestes(see Nessov 1985a, 1997; Nessov et al.1994\: Kielan-Jaworowska& Dashzeveg(1989) assignedOtlestes and Prokennalestesto the OtlestidaeNessov, 1985a,but someof the charactersthat unite the two generaare plesiomorphies(see comparisonof Daulesteswith Prokennelestesabove). (3) Paranyctoides(Fox1979,L984;Lillegraven &McKenna 1986;Cifelli 1990)was for a long time regardedas the oldestNorth American eutherianmammal, but seeCifelli (1999); it showsunequivocal lipotyphlan affinities (family ?Nyctitheriidae).Daulestes differs from Paranyctoidestn lacking pre- and postcingulaand in having asymmetryof the stylar region, a smallerectoflexus, trigon basinmore elongatedtransversely, and generally more gracile structureof the upper molars. On lower molars the main difference concemsthe position of the paraconid,which in Daulestesis shiftedfrom the lingual side anteriorly and placed symmetricallybetween the protoconid and metaconid. Tiakinginto accountthe charactersof advancedlipotyphlans, Daulestes cannot be assignedto this group becauseit has a long jugal largely confributing to the lower edgeof the orbit, a long infraorbital canal (but seeMcKenna & Simpson 1959), apalatine con- tributing to the orbital wall, contactingthe lacrimal, anda true postglenoidprocess. We are awareof the cautionof MacPhee& Novacek(1993) that, out of a suite of lipotyphlan charactersgiven by Butler (1988), only two (hindgut simplification with correlatedab- senceof cecum,and pronouncedreduction of the pubic symphysis)are strong synapo- morphiesof Lipotyphla. However,neither of them can be applied for Daulestes. We believethat the anatomyof the skull is moreconservative and better reflects the affinities of mammaliantaxathanthe dentition, which easilychanges with demandsof the environment.In that we lack knowledgeon skull strucfureof most Cretaceous mammals,conclusions on their relationshipsare often basedon the dentition only. Daulestes,however, is representedby a largepart of the skull, althoughit is immature. Comparisonwith otherCretaceous mammals in which skull hasbeen preserved shows closesimilarity of the skull designof Daulestesto that of Asiorycte,s,more so than to Kennalestes,and very different fromZalambdalestidae.The skull similaritiesare also strongly supportedby similaritiesin the dentition.The foregoingcomparisons show that, of known CretaceousEutheria, Daulestesis most similar to Asioryctes and Ukhaatherium.Werefrain from creatinga separatefamily for Daulestes.Because of similarities and differenceswith Asioryctidae we assign Daulestestentatively to AsioryctitheriaNovacek et al., 1997,family incertaesedis. Daulesles appears to be- long to the asioryctitherianstem-group.

Acknowledgements we thankA.o. Averianov,J.D. Archibald, P.M. Butler, R.L. cifelli, w.P.Luckett, R.D.E. Macphee, M.J. Novacek,G.W. Rougier, and J.R. Wible for helpfulcomments and discussion. The dentalfor- mularesultedfromLuckett's insightthatdp2 mightlie whollybehindp2in thelowerjaw, resultingin a premolarcount of five sequentiallower premolariformteeth in useat the sametime. The specimen was preparedby William Amaral of Cambridge,Massachusetts, and the authors.The SEM micro- 48 Earliest eutherian mammal s&z//.'MCKENNA e/ a/.

graphs were taken by Jsrn Hurum and Z. K.-1. at the Paleontological Museum, University of Oslo, the other photographs by Chester Tarka at the American Museum of Nafural History in New York. The drawings were made by Natalia Florenskaya (St. Petersburg) and Karol Sabath (at the Institute of Paleobiology, Warsaw), after pencil sketchesby Z. K.-J., except Fig. 1, prepared by Ed Heck at AMNH, New York. The work of Z. K.-J. during her stay in Oslo was funded by the Norwegian Sci- enceFoundation, grant ABC/LR441.93/002, and the.University of Oslo, and on her retum to Poland by the Institute of Paleobiology, Polish Academy of Sciences,Warsaw.

References

Allin, E.F.&Hopson, J.A. 1992.Evolution of theauditory systeminSynapsida ('mammal-likereptiles'and primitive mammals)as seenin the fossil record.1n.' D.B.Webster, R.R. Fay, & A.N. Popper(eds),The EvolutionaryBiology of Hearing, 587414. Springer-Verlag,New York. Archibald, J.D. 1979. Oldest known eutherian stapesand a marsupial petrosal bone from the Late Creta- ceousof North America.- Nature 281, 669470. Archibald,J.D.1982. A studyof Mammalia andgeology across the CretaceousTertiary boundary in Gar- fieldCounty, Montana.-University of CalforniaPublications inGeologicalSciencesl22,12S6. Archibald, J.D. Sues,H.-D, Averianov, A.O., King, C., Ward, D.J., Tsaruk,O.I., Danilov, I.G., Rezvyi, S.B.,Veretennikov, G., & Khodjaev,A. 1998.Pr6cis ofthe Cretaceouspaleontology, biostratigraphy, and sedimentologyat Dzharakuduk(Turonian?-Santonian), Kyzylkum Desert,Uzbekistan. 1n.' S.G. Lucas,J.I. Kirkland, & J.W.Estep (eds), Lower andMiddle CretaceousTerrestrial Ecosystems. -New Mexico Museumof Natural History and ScienceBulletin l4,2IJ7 . Averianov,A.O. 1999.A new speciesof multituberculatemarwnlUzbekbaatarfromtheLate Cretaceous of Uzbekistan.- Ac ta Palae ont o lo gic a Po lonic a 44, 301104. Averianov,A.O. & Kielan-Jaworowska,Z. 1999.Marsupials from the Late Cretaceousof Uzbekistan.- Acta PalaeontologicaPolonica 44, 7 l-8I. Bafud,I.L. 1970a.A survey of the periotic labyrinth in somerepresentative Recent reptiles. -The Univer- sity of KansasScience Bulletin 41, 891-981. Batd, I.L. 1970b.The anatomyof the reptilian ear.In: C. Gans (ed.), Biology of the Reptilia,Yol.2, 193-275.Academic Press, London & New York. Baird, I.L. 1974.Anatomical features of the inner earin submanrmalianvertebrates. 12.' W.D. Keideland& W.D. Neff (eds),Handbook of SensoryPhysiology,Yol.Y-l: Auditory System,Anatomy, Physiology (Ear), 159-212. Springer-Verlag, New York. Beer,G.R. de 1929.The developmentofthe skull ofthe shrew. - PhilosophicalTransactionsofthe Royal Societyof Inndon 217, 41L480. Beer, G.R. de 1937. TheDevelopment of the VertebrateSkull. 552 pp. ClarendonPress, Oxford. Benton,M.J. 1990.Vertebrate Paleontology.3TT pp. Unwin Hyman, London. Bonaparte, J.F. & Rougier, G. 1987. Mamiferos del Cretdcico inferior de Patagonia.IV Congreso lrt- tino - ame ric ano de Pale onto lo g ia, Bolivia 1, 3 43-359. Bugge,J. 1974.The cephalicarterial system in insectivores,primates, rodents, and lagomorphs, with special referenceto the systematicclassification. - Acta Anatomica87 (Suppl.62), l-160. Butler, P.M. .1956.The skull of Ictops andthe classificationof the Insectivora.- Proceedingsof theZoo- logical Societyof London 126,453481. Butler,P.M. l972.Theproblem of insectivoreclassification. 1n.' K.A. Joysey& T.S.Kemp (eds),Studies in VertebrateEvolution,253-265. Oliver & Boyd, Edinburgh. Butler, P.M. 1978.A new interpretation of the mammalian teeth of tribosphenic pattem from the Albian of Texas.- Breviora 446, I-27. Butler, P.M. 1988.Phylogeny of the insectivores.In: M.J. Benton(ed.), The Phylogeny and Classification of the Tetrapods,Yol.2: Mammals,ll7-141. SystematicsAssociation Special Volume 35B. Claren- don Press,Oxford. Butler,P.M. 1990.Earlytrendsintheevolutionoftribosphenicmolars.-BiologicalReviews65,529-552. Carroll, R.L. 1988.Vertebrate Paleontology and Evolution.698 pp. W.H. Freeman,New York. Cifelli, R.L. 1982.The petrosal structure of Ilyopsodus withrespect to that of someother ungulates,and its phylogeneticimplications. - Journal of Paleontology56, 795-805. ACTA PALAEONTOLOGTCAPOLOMCA (45) (1) 49

Cifelli, R.L. 1990. Cretaceousmammals of Southern Utah. IV. Eutherian mammals from the Wahweap (Aquilan) andKaiparowits (Judithian) formations.-Journal ofVertebratePaleontologyl0,346-360. Cifelli, R.L. 1999.Tribosphenic mammal from the North American Early Cretaceow.- Nature 40L, 363-366. Clemens,W.A. 1973.Fossil mammals of the type LanceFormation Wyoming. PartItr. Eutheriaand summary. - Universityof Calfornia Publicationsin GeologicalSciences 94, t-102. Clemens,W.A. Lillegraven,J.A., LindsayE.H., & Simpson,G.G. 1979.Where, when andwhat - a survey of known Mesozoicmammal distribution. 1n.' J.A. Lillegraven,Z.Kielan-Iaworowska,& W.A. Clem- ens(eds), Mesozoic Mammals: the First Two-Thirdsof MammalianHistory,7-58. University of Cali- fornia Press,Berkeley. Conroy,G.C. & Wible, J.R. 1978.Middle earmorphologyof Lemurvariegatus'.someimplications for pri- matepaleontology. - Folia Primatologica29, 81-85. Cope,E.D. 1880.On theforaminaperforating the posteriorpart of the squamosalbone of theMammalia. - Proceedingsof theAmerican Philosophical Socieg 18, 45246I. Crompton, A.W. & Kielan-Jaworowska,Z. 1978.Molar structureand occlusionin Cretaceoustherian mammals.In: P.M. Butler & K.A. Joysey(eds), Studies in Development,Function and Evolution of Teeth,249J87. AcademicPress, London & New York. Doran,A.H.G. l8Ts.Morphologyofthemammalianossiculaauditus.-TransactionsoftheLinneanSoci- etyof London2,I,371497. [PublishedinpartVtr,August,1878(wholevolume, 1879)] Estes,R. 1961.Cranial anatomy of the cynodontreptile Tftrinaxodonliorhinus. - Bulletin of theMuseum of ComparativeZoology 125, 165-180. Fearnhead,R.W., Shute,C.C.D., & Bellairs, A. d'A. 1955.The temporo-mandibularjoint of shrews.- Proceedingsof theZoological Societyof London 125,795-806. Fleischer,G.1973. Studienam Skelettdes Gehiirorgans der Sriugetiere,einschliesslich des Menschen. - SriugetierkundlicheMitteilung en 21, l3l-239. Fox, R.C. 1979.Mammals from the Upper CretaceousOldman Formation, Alberta. III. Eutheia. - Cana' dian Journal of Earth Sciences16, lI4-125. Fox, R.C. 1984.Paranyctoides maleficus (new species),an early eutherianmammal from the Cretaceousof Alberta. In: M.R. Dawson (ed.), Papersin Vertebrate Paleontology Honoring Robert Warren Wilson. - CarnegieMuseum of Natural History, SpecialPublication9,9-20. Fox, R.C. 1989. The Wounded Knee local fauna and mammalian evolution near the CretaceousTefiiary B oundary,Saskatchewan, Canada. - Palae ont o g raphic a A 208, I l-59. Fox, R.C. & Meng, J. 1997.AnX-radiographic and SEM studyof the osseousinner ear of multituberculates and monotremes(Mammalia): implications for mammalianphylogeny and the evolution of hearing. - Zoological Journal of the Linnean Society,London l2l,249-291. Gates,G.R., Saunders, J.C., Bock, G.R.,Aitkin, L.M., & Elliott, M.A. 1974.Peripheral auditory function in theplatypus,Ornithorhynchusanatinus.-JoumaloftheAcousticalSocietyofAmerica 56,152-156. Gaupp, E. 1900. Das Chondrocraniumvon Lacerta agilis: Ein Beitrag zum Verstiindnisdes Amnio- tenschiidels.- AnatomischeH efte 15, 433-164. Gradstein,F.M., Agterberg,F.P., Ogg, J.G., Hardenbol, J., Van Veen,P., Thierry, J., &Hnang,Z.1994. A Mesozoic time scale.- J ournal of Ge ophy sic al Res e ar ch 99 (B l2), 2405| 1407 4. Gradstein,F.M., Agterberg,F.P., Ogg, J.G., Hardenbol, J., Van Veen,P., Thierry, J., &Httang,Z.1995. A Triassic, Jurassicand Cretaceoustime scale.12.' W.A. Berggren,D.V. Kent, M.-P. Aubry, & J. Hardenbol (eds), Geochronology, Time Scalesand Global Stratigraphic Correlation,95-126. Special Publication54, SEPM (Societyfor SedimentaryGeology), Tulsa. Gray,A.A. 1908.Aninvestigationontheanatomicalstructureandrelationshipsofthelabyrinthintherep- tile, the bird, and the mammal.- Proceedingsof the Royal SocietyB 80,507-528. GraybealA.,Rosowski, J., Ketten, D.R., & Crompton,A.W. 1989.Innerear structuretnMorganucodon,an early Jurassicmammal. - Zoological Journal of the Linnean Society96,107-ll'l. Gregory,W.K. 1910.The ordersof mammals.- Bulletin of theAmerican Museum of Natural History 27, t-524. Gregory,W.K. 1920.Studies in comparativemyology andosteology; no. IV. A review of the evolutionof the lacrymalbone of vertebrateswith specialreferenceto that of mammals.- Bulletinof theAmerican Museumof Natural History 42,95-263. Gregory,W.K., & Simpson,G.G.1926.Cretaceous mammal skulls from Morgoba.-American Museum Novitates225,l-20. Griffiths, M. 1968.Echidnas. 282 pp. PergamonPress, Oxford. 50 Earliest eutherian mammal skall: MCKENNA et al

Griffiths, M. 1978.The Biology of the Monotremes.36Tpp. AcademicPress, New York. Hurum, J.H.,Presley, R., & Kielan-Jaworowska,Z.l996.Thernddle earinmultituberculatemammals. - Acta PalaeontologicaPolonica 41, 253-275. Huxley, T.H. 1880. On the application of the laws of evolution to the arangement of the Vertebrata and more particularly of the Mamm:ha. - Proceedingsof the Scientific Meetings of the hological Society of London fortheyear 1880,649462. Kermack,K.A. 1963.The cranialstructure of the triconodonts.- PhilosophicalTransactionsof theRoyal Societyof LondonB 246,83-103. Kermack,K.A., Mussett,F., & Rigney,H.W. 1981.The skullof Morganucodon.- Zoological Joumal of the Linnean Society71, l-158. Kielan-Jaworowska,Z. 1969. Preliminary data on the Upper Cretaceouseutherian mammals from Bayn Dzak, Gobi Desert.In: Z.Kielan-Jaworowska(ed.), Results of the Polish-MongolianPalaeontological Expeditions,Pelrtl. - PalaeontologiaPolonica L9, 17l-lgL Kielan-Jaworow ska,Z. 1975a. Preliminary description of two new eutherian generafrom the Late Creta- ceousof MongoLa.In: Z. Kielan-Jaworowska (ed.), Results of the Polish-Mongolian Palaeontological Expeditions,Part VI. - PalaeontologiaPolonica 33, 5-16. Kielan-Jaworow ska,Z. 1975b . Evolution of the therian mammals in the Late Cretaceousof Asia. Part I. Deltatheridiidae. In: Z. Kielan-Jaworowska (ed.), Results of the Polish-Mongolian Palaeontological Expeditions,Part VI. - Palaeontologia Polonica 33, lO3-132. Kielan-Jaworowska,Z.1975c.Possible occurrence of marsupialbones in Cretaceouseutherian mammals. - Nature 255,698499. Kielan-Jaworow ska,Z. 1981. Evolution of the therian mammals in the Late Cretaceousof Asia. Part IV. Skull structurein Kennalestesand Asioryctes. In: Z. Kielan-Jaworowska(ed.), Resultsof the Pol- ish-MongolianPalaeontological Expeditions, Part D(. - PalaeontologiaPolonica 42,2518. Kielan-Jaworow ska.Z. 1984a.Evolution of the therian mammals in the Late Cretaceousof Asia. Part V. Skull structure in Zalambdalestidae.In: Z. Kielan-Jaworowska (ed.), Results of the Polish-Mongolian PalaeontologicalExpeditions, PartX. - PalaeontologiaPolonica 46, 107-ll7 . Kielan-Jaworow ska,Z. 1984b. Evolution of the therian mammals in the Late Cretaceousof Asia. Paft VI. Endocranial castsof eutherian mammals. 12.'Z. Kielan-Jaworowska (ed.), Results of the Polish-Mon- golian PalaeontologicalExpeditions, Part X. - PalaeontologiaPolonica 46, 151-17l. Kielan-Jaworowska,Z. 1984c.Evolution of the therian mammals in the Late Cretaceousof Asia. Part Vtr. Synopsis.1n.' Z. Kielan-Jaworowska(ed.), Results of the Polish-MongolianPalaeontological Expedi- tions, PartX. - PalaeontologiaPolonica 46, 173-183. Kielan-Jaworowska,2.,& Dashzeveg,D.1989.Eutherianmammals fromtheEarlyCretaceous of Mongo- lia. - Zoolo gica Scripta 18, 347155. Kielan-Jaworowska,Z. & Nessov,L.A. 1990.On the metatheriannature of the Deltatheroida, a sister group of the Marsupialia.- Lethaia 23, l*10. Kielan-Jaworow ska, Z. & Nessov, L.A. 1992. Multituberculate mammals from the Cretaceousof Uzbe- kistan.- Acta PalaeontologicaPolonica 37, l-17 . Kielan-Jaworowska,2., Presley,R., & Poplin, C. 1986. The cranial vascularsystem in taeniolabidoid multituberculate mammals. - Philosophical Transactions of the Royal Society of London B 313, 525402. Kielan-Jaworowska,Z. & Trofimov, B.A. 1980. Cranial morphology of the Cretaceouseutherian mammal Barunle s t e s. - Acta Palae ont olo gic a Poloni c a 25, 167-185. Kielan-Jaworowska,2., Bown, T.M., & Lillegraven,J.A. 1979.Eutheria. In: J.A.LrlTegraven,Z. Kielan- Jaworowska,& W.A. Clemens(eds), Mesoruic Mammals: TheFirst Two-thirdsof Mammalian His- tory, 221J58. University of California Press,Berkeley. Kuhn, H.J. 1971. Die Entwicklung und Morphologie des Schiidels von Tachyglossusaculeatus. - Abhandlungender SenckenbergischenNaturforschenden Gesellschaft 528, l-192. Lewis,E.R.,Leverenz,E.L.,&Bialek,W.S.l9S5.TheVertebrateInnerEar.256pp.BocaRaton,Florida: CRC Press,Inc. [Reprint]. Lillegraven,J.A. 1969.LatestCretaceousmammalsofupperpartofEdmontonFormationofAlberta,Can- ada, and review of marsupial-placentaldichotomy in mammalian evolution. - University of Kansas PaleontologicalContributions 50 (Vertebratal2), l-L22. Lillegraven, J.A. 1975 . Biological considerationsof the marsupial-placental dichotomy. - Evolution 29, 707122. ACTA PALAEONTOLOGICAPOLONICA (45) (1) 51

Lillegraven, J.A. & Hahn, G. 1993. Evolutionary analysis of the middle and inner en of Late Jurassic multituberculates. - Journal of Mammalinn Evolution l, 4714. Lillegraven,J.A., & McKenna,M.C. 1986.Fossil mammals from the 'Mesaverde'Formation (Late Creta- ceous,Judithian) of the Bighorn andWind River Basins,Wyoming, with definitions of Late Cretaceous North American land-mammal 'Ages.' - American Museum Novitates 2840, l-68. Lillegraven, J.A., Kielan-Jaworowska,2.,& ClemensW.A. (eds) 1979.Mesozoic Mammals: The First Twolhirds of MammalianHistory. 31I pp. University of CaMorniaPress, Berkeley. Lillegraven,J.A., Thompson, S.D., McNab,8.K., &Patton, J.L. 1987.The originof eutherianmammals.- Biological Journal of the Linnean Society32,281-336. Lofgren, D. 1995. The Bug Creek problem and the Cretaceous-Tertiary transition at McGuire Creek, Montana.- Publicationsin GeologicalSciences, University of Calfornia 140, 1-185. Luckett,W.P. 1993.An ontogeneticassessment of dentalhomologies in therianmammals. 12.' F.S. Szalay, M.J. Novacek, & M.C. McKenna (eds), Mammal Phylogeny. YoI. I. Mesozoic Dffirentiation, Multituberculates, Monotremes, Early Therians, and Marsupials, 182-204. Springer-Verlag, New York. Llo,Z. 1989.Structure of thepetrosals of Multituberculata(Mammalia)and the malar morphologyof the early arctocyonids(Condylarthra, Mammalia). Unpublished Ph.D. thesis.426pp.Unversity of Cali- fomia, Berkeley. Luo,Z.1994.Sister-grouprelationships of mammalsand transformations of diagnosticmammalian characters.In: N.C. Fraser& H.-D. Sues(eds), In the Shadowof the Dinosaurs.Early MesozoicTetrapods, 98-128. CambridgeUniversity Press,London, Luo,2., Crompton,A.W., & Lucas,S.G. 1995.Evolutionary origins of the mammalianpromontorium and cochlea.- Journal ofVertebrate Paleontology15, ll3-121. Luo, Z. & Ketten, D.R. 1991. CT scanning and computerized reconstructions of the imer ear of multi- tuberculatemammals. - Journal of VertebratePaleontology ll,220*228. Maclntyre,G.T. 1972.The trisulcatepetrosal pattem of mammals.In: T.Dobzhansky,M.K. Hecht,& W.C. Steere(eds), Evolutionary Biology,vol.6,275-303. Appleton-Century-Crofts,New York. MacPhee,R.D.E. 1981.Auditory regionsof primatesand eutherianinsectivores: morphology, ontogeny and character analysis. - Contributions to Primatology lE, l-282. MacPhee,R.D.E. & Novacek,M.J. 1993.Definition andrelationships of Lipotyphla.In: F.S.Szalay, M.J. Novacek,& M.C. McKenna (eds),Mammal Phylogeny.Yol.IL Placentals,l3-jl. Springer-Verlag, New York. Manley,G.A. l9T0.Comparativestudiesofauditoryphysiologyinreptiles.-Zeitschriftfi)rvergleichende Physiologie 67, 363181. Marshall, L.G., Muizon, C. de, & Sigogneau-Russell,D. 1995. Pucadelphysandinus (Marsupialia, Mammalia) from the early Paleocene of Bolivia. - Mdmoires du Musdum National d'Histoire Narurelle L65, I-164. Marshall, L.G. & Kielan-Jaworowska, Z. 1992. Relationships of the dog-like marsupials, deltatheroidans and early tribosphenicmammals. - Lethaia25,36l-374. Matthew, W.D. 1909. The Carnivora and Insectivora of the Bridger Basin, middle Eocene. - Memoirs of theAmerican Museum of Natural History 9,289*567 . McDowell, S.B.,Jr. 1958.The GreaterAntilleaninsectivores. -Bulletin of theAmericanMuseum of Naru- ral History ll5, ll3-214. McKenna,M.C. 1966.Paleontology and the origin of the Primates. - Folia Primatologica4, l-25. McKenna, M.C. 1975.Toward a phylogeneticclassification of the Mammalia. In: W.P. Luckett & F.S. Szalay (eds), Phylogeny of the Primates, 2146. Plenum Publishing Corporation, New York. McKenna,M.C. & Bell, S.K. 1997.Classification of MammalsAbove the Species LeveL 63l pp. Columbia University Press,New York. McKenna,M.C. & Simpson,G.G. 1959.A new insectivorefrom the middle Eoceneof TabemacleButte, Wyoming. - AmericanMuseum Novitates 1952,l-12. Meng, J. & Fox, R.C. 1995a.Therian petrosals from the Oldman and Milk River formations(Late Creta- ceous),Alberta, Canada.- Journal ofVertebrate Paleontology15, 122-130. Meng,J. & Fox, R.C. 1995b.Osseous inner ear sffuctures and hearing in earlymarsupials andplacentals. - Zoological Journal of the Linnean Societyll5,47Jl. Meng, J. & Wyss, A.R. 1995.Monotreme affinities and low-frequencyhearing suggested by multituber- culateear. - Nature 377, l\l-lM. Meng, J. & Wyss, A.R. 1996.Multituberculate phylogeny. - Nature 379,407. 52 Earliest eutherian mammal skul/.' MCKENNA et al,

Meng, J., Ting, S., & Schiebout,J. A.1994. Cranialanatomy of an earlyEocene didymoconid (Insectivora, Mammata). - Joumal of VertebratePaleontology 14,534-551. Miao, D. 1988.Shll morphologyof Lambdopsalisbulla (Mammalia,Multituberculata) and its implications to mammalian evolution. - Contibutions to Geology, University of Wyoming, SpecialPaper4, r-104. Miao, D. & Lillegraven,J.A. 1986.Discovery of threeear ossiclesin a multituberculatemammal. - Na- tional GeographicResearch 2, 500-507. Miller, M.R. 1966.The cochlearduct of lizards.- Proceedingsof the CalifurniaAcademy of Sciences33, 255-359. Miller, M.R. 1968.The cochlearduct of snakes.- Proceedingsof the Calfomia Academyof Sciences35, 425475. Mills,J.R.E.1966.Thefunctionalocclusionof theteethoflnsectivora.-JournaloftheLinneanSociety (Zoology)47, l-25. Nessov,L.A. Qt{esov,L.A.) 1982.Ancient mammalsof the USSR [in Russian).- EZegodnikvsesoAzno7o PaleontologiieskogoObSesna 35, 228243. Nessov,L.A. 1984.On someCretaceous mammals from the SovietMiddle Asia [in Russianwith English summaryl.- VestnikZoologii, Kiev 2,6045. Nessov,L.A. (Nesov,L.A.) 1985a.Newmammals fromtheCretaceous of Kyzylkum [inRussianwithEng- lish summaryl.- VestnikLeningradskogo (Jniversiteta, Series 7, 17, 8-18. Nessov,L.A. (Nesov,L.A.) 1985b.Rules of historicalchanges of Mesozoicecosystems of thecoastal plains of Middle Asia and Kazakhstan[in Russian].XXXI SesiAVsesoiznogo Paleontologideskogo ObSes- tva, Leningrad.Abstracts: 53-54. Nessov,L.A. (Nesov,L.A.) 1987.Results of searchand study on theCretaceous and early Paleogene mammals on the territory of the USSR [in Russian].- EZegodnikvsesofrznogo Paleontologiieskago ObSesna30,199115. Nessov,L.A. (Nesov,L.A.) 1988.Assemblages of lateMesozoic and Paleocene vertebrates of Middle Asia [in Russian].1z; T.N. Bogdanova& M.V. Oshurkova(eds), Stanovlenie i Evolutcd Kontinentalnyh Biot, 93-101. Izdatelstvo Nauka, Leningrad. Nessov,L.A. (Nesov,L.A.) 1990.Small ichthyomithiform bird and other bird remainsfrom the Bissekty Formation (Jpper Cretaceous)of Central Kyzylkum Desert [in Russian]. - Trudy Zoologiieskogo InstitutaAkademii Nauk SSSR21, 59-62. Nessov,L.A. (Nesov,L.A.) 1993.New Mesozoicmammals of Middle Asia andKazakhstan and comments about evolution of theriofaunas of Cretaceouscoastal plains of ancient Asia [in Russian with English summaryl.- Trudy ZoologiieskogoInstituta RossijskojAkademii Nauk249,105-133. Nessov,L.A. Qrlesov,L.A.) 1997. CretaceousNon-marine Vertebrates of Northern Eurasia fin Russian, with English summaryl.University of SaintPetersburg, Institute of Earth Crust,Saint Petersburg. 218 pp. @osthumousedition by L.B. Goloweva & A.O. Averianov). Nessov,L.A., Archibald,J.D., & Kielan-Jaworowska,2. LggS.tJngtlate-likemammals from theLate Cre- taceousof Uzbekistanand aphylogenetic analysis of Ungulatomorpha.In:C.C. Beard & M.R. Dawson (eds),DawnoftheAgeof MammalsinAsia. -Bulletinof the CarnegieMuseumofNaturalHistory34, 40-88. Nessov,L.A. & Gureev,A.A. 1981.Finding of a dentaryof an ancientshrew in the Upper Cretaceousof Kyzylkum Desert[in Russian]. - DokladyAkademii Nauk USSR257 ,1002-1004. Nessov,L.A. & Kielan-Jaworowska,Z. 1991.Evolution of the CretaceousAsian therianmammals.In: Z. Kielan-Jaworowska,N. Heintz, & H.-A. Nakrem (eds),Fifth Symposiumon Mesozoic Terrestrial Eco- systemsand Biota: ExtendedAbstracts. - Contributionsfrom the PaleontologicalMuseum, (Jniver- sity of Oslo 364,51-52. Nessov,L.A. & Mefiiniene, R.A. 1986.Remains of chondrichthyanfishes from the Cretaceousof Middle Asia and Kazakhstanas a sorrrceof information on the age and origin of the deposits [in Russian]. In; G.Ya. Kzymgoltz [G.A.Kzymgolc] & V.S. Belenkova(eds), Paleontology and DetailerJStratigraphic Correlation, 35-41. IzdatelstvoNauka, Leningrad. Nessov,L.A., Sigogneau-Russell,D., & Russell,D.E. 1994.A surveyof Cretaceoustribosphenic mammals from Middle Asia (Uzbekistan, Kazakhstanand Tajikistan), of their geological setting, age and faunal environment.- Palaeovertebrata23, 5l-92. Nessov,L.A. & Trofimov, B.A. 1979.Ancientinsectivores from theCretaceous of UzbekSSR [in Russian]. - DokladyAkademii Nauk 555R247,952-955. ACTA PALAEONTOLOGTCAPOLONTCA (45) (1) 53

Novacek, M.J. 1982. Information for molecular studies from anatomical and fossil evidence on higher eutherianphylogeny. In: M. Goodman(ed.), Macromokcular sequencesin Systematicand Evolution- ary Biology,3-41. PlenumPress, New York. Novacek,M.J. 1986.The skull of leptictid insectivoransand the higher-levelclassification of eutherian mammals.- Bulletin of theAmerican Museum of Natural History 183, 1-l I 1. Novacek,M.J., Rougier,G.W., Wible,I.R., McKenna,M.C., Dashzeveg,D., & I. Horovitz. 1997.Epipubic bonesin eutherianmammals from the Late Cretaceousof Mongolia. - Nature 389, 483-486. Presley,R. 1979.The primitive courseof the intemal carotidartery in mammals.- Acta Anatomica 103, 238-244. Pritchard,U. 1881.The cochleaof theOrnithorhynchusplatypus compared with that of ordinarymammals andbirds. - PhilosophicalTransactions of the Royal Socie4tof London 172,267182. Quiroga, J.C. 1979.The inner ear of two cynodonts (Reptilia-Therapsida) and somecomments on the evolution of the inner ear from pelycosaursto mammals.- GegenbaursMorphologisches Jahrbuch Leip- zis 125.178-190. Romer,A.S. 1966.VertebratePaleontology. x+ 468pp. 3rd edition,University of ChicagoPress, Chicago. Rougier,G.W., Wible, J.R.,& Hopson,J.A.1992. Reconstructionof the cranialvessels in the Early Creta- ceousmammal Vrncelestes neuquenianus: inpbcatrons for the evolution of the mammalian cranial vascular system.- Joumal of VertebratePaleontology 12, 188-216. Rougier,G.W., Wible, J.R.,& Hopson,J.A. 1996a.Basicranial anatomy of Priacodonfruitaensrs(Tricono- dontidae,Mammalia) fromtheLateJurassic of Colorado, and areappraisalof marnmaliaforminterrela- tionships.- AmericanMuseum Novitates 3183, 1-38. Rougier,G.W., Wible, J.R., & Hopson,J.A. 1996b.Middle-ear ossiclesof the multituberculateKrypto- baatar ftom the Mongolian Late Cretaceous:implications for mammaliamorph relationships and the evolution of the auditory apparatus.- AmericanMuseum Novitates 3187,143. Sigogneau-Russell,D., Dashzeveg,D., & Russell,D.E. 1992.Further data on Prokennalestes(ManmaL4 Eutheria,inc. sed.)from the Early Cretaceousof Mongolia. - ZoologicaScripta 21,205-209. Simpson,G.G. 1933.The earregionand the foramina of thecynodont skull. -AmericanJournal of Science 26,285J94. Simpson,G.G.1937. Skullstructureof theMultituberctlata.-Bulletinof theAmericanMuseumofNatu- ral History73,727163. Simpson,G.G. 1945.Notes on Pleistoceneand Recent tapirs. -Bulletin of theAmerican Museum of Naru- ral History 86,33-82. Sloan,R.E. & Van Valen, L. 1965.Cretaceous mammals from Montana.- Science148,220-227. Sochava,A.V. (Sodava,A.V.) 1968.Red-colored Cretaceous Sediments of Middle Asia [in Russian].122 pp. Izdatelstvo Nauka, Leningrad. Thewissen,J.G.M. & Gingerich, P.D. 1989. Skull and endocranialcast of Eoryctes melanus,a new palaeoryctid (I\4ammalia:Insectivora) from the early Eoceneof western North America. - Journal of Verteb rate Pale ontolo gy 9, 45947 0. Van Valen, L. 1978.The beginningof the ageof mammals.- Evolutionary Theory4,45-80. Van Valen, L. & Sloan,R.E. 1965.The earliestprimatss. - ggisngs150,743145. Voit, M. 1909. Das Primordialcranium des Kaninchens unter Berticksichtigung der Deckknochen. - Anatomis che H efte 38, 425-616. Wever, E.G. 1967.Tonal differentiationin the lizard ear.- Laryngoscope77, 1962-1973. Wever, E.G. 1978. The Reptile Ear: Its Structure and Function. 1024 pp. Princeton University Press, Princeton,New Jersey. Wible, J.R. 1987. The eutherian stapedialartery: characteranalysis and implications for superordinalrela- tionships.- Zoologic al Journal of the Linnean Society91, 107-135. Wible, J.R. 1990.Petrosals of Late Cretaceousmarsupials from North America,and a cladisticanalysis of the petrosalin therianmammals. - Journal ofVertebrate Paleontology10, 183-205. Wible,J.R. l99l.OriginofMammalia:thecraniodentalevidencereexamined.-JournalofVertebratePa- leontologyll, 1_28. Wible, J.R.& Hopson,J.A. 1993.Basicranial evidence for earlymammal phylogeny. In: F.S.Szalay, M.J. Novacek,& M.C. McKenna (eds),Mammal Phylogeny,Mesozoic Dffirentiation, Multituberculates, Monotremes,Early Therians,and Marsupials,4542. Springer-Verlag,New York. Wible, J.R. & Hopson, J.A. 1995. Homologies of the prootic canal in mammals and non-mammalian cynodonts.- Journal of VertebrateP aleontolo gy 15, 331-356. 54 Earliest eutherian mammal skull: MCKENNA et al.

Wible, J.R., Rougier, G.W., Novacek, M.J., McKenna, M.C., & Dashzeveg, D. 1995. A mammalian petrosal from the Early Cretaceous of Mongolia: implications for the evolution of the ear region and mammaliamorph interrelationships. -American Museum Novitates 3149, 1-19. Zeller, U. 1985a.Themorphogenesis of the fenestrarotundainmammals. In: H.-R. Duncker & G. Fleischer (eds), Vertebrate Morphology. - Fortschritte der Zoologie 30, 153-157. Zeller, U. 1985b. Die Ontogenese und Morphologie der Fenestra rotunda und des Aquaeductus cochleae von Tupaia und anderen Saugern. -Gegenbaurs Morpkologisches Jahrbuch Leipzig 131,179-204. Zeller, U. 1993. Ontogenetic evidence for cranial homologies in monotremes and therians, with special ref- erence to Ornitkorkynckus. In: F.S. Szalay, M.J. Novacek, & M.C. McKenna (eds), Mammal Pkylog- eny, Mesozoic Dzfferentiation, Multituberculates, Monotremes, Early Tkerians, and Marsupials, 95-107. Springer-Verlag, New York.

Najstarsza czaszka ssaka toiyskowego z pdinej kredy (koniaku) Uzbekistanu

MALCOLM C. MCKENNA, ZOFIA KIELAN-JAWOROWSKA i JIN MENG

Streszczenie

W pracy opisano czqs'ciowo uszkodzonq czaszkq dodego osobnika okreilonego jako Duulesles nessovi sp. n. z koniaku Uzbekistanu. Jest to najwczehiejsza czaszka ssaka lozyskowego, liczqca ok. 87 milion6w lat. Daulestes jest sz6stym rodzajem kredowego ssaka lozyskowego, kt6rego czaszka jest znana. Poniewaz czaszka D. nessovi nalezy do mlodocianego osobnika, niekt6re cechy plezjomorficzne mogq by6 zwiqzane z wczesnym stadium rozwojowym. W czaszce wystqpujq cztery g6me przedtrzonowce i pi$ dolnych, jest jednak mozliwe, ze w stadium doroslym dolnych bylo tez tylko cztery. Na g6mych zqbach policzkowych wystqpujq konule ze skrzydelkami. Na M2 wyrostek parastylamy jest duzy, metastylarny - mdy. Pre- i postcingula nie wystqpujq na mlecz- nym czwartym przedtrzonowcu ani na trzonowcach. Talonidy dp4-m2 mierzq ok. 90 % szerokos'ci trygonid6w, a entokonidy sq oddalone od hipokonulid6w. W czaszce wystqpuje duza szczelina klinowo-oczodolowa i zapewne brak wyrostka skrzydlowego koSci klinowej. Duze skrzydlo oczodolowe kos'ci podniebiennej powoduje, ze brak jest kontaktu miqdzy szczqkq a kos'ciq oczodo- lowq w obrqbie oczodolu. Wyrostek jarzmowy jest delikatny. Blqdnik ma jeden pelny skrgt z rozszerzonym wierzcholkiem, co moze wskazywak na obecnoSC lageny. Duzy mloteczek z silnym wyrostkiem przednim oraz duze promontorium mogq by6 zwiqzane z mlodym wiekiem osobnika lub zachowaniem cechy plezjomorficznej (jak u dziobaka). W zwiqzku z podobienstwem do Asioryctidae zardwno w budowie czaszki, jak i uzqbienia, Daulestes zostal zaliczony do Asio- ryctitheria Novacek et al. 1997, rodzina incertae sedis.