MEMOIRS OF THE

QUEENSLAND MUSEUM BRISBANE

© Queensland Museum PO Box 3300, South Brisbane 4101, Australia

Phone 06 7 3840 7555 Fax 06 7 3846 1226

Email qmlib@qm.qld.gov.au Website www.qm.qld.gov.au

National Library of Australia card number

ISSN 0079-8835

NOTE Papers published in this volume and in all previous volumes of the Memoirs of the

Queensland Museum maybe reproduced for scientific research, individual study or other educational purposes. Properly acknowledged quotations may be made but queries regarding the republication of any papers should be addressed to the Editor in Chief. Copies of the journal can be purchased from the Queensland Museum Shop.

A Guide to Authors is displayed at the Queensland Museum web site

A Queensland Government Project Typeset at the Queensland Museum

Mem. Qd Mus. 21(2): 413-517. [1984]

DINOSAUR TRACKWAYS IN THE WINTON FORMATION (MID-CRETACEOUS)OF QUEENSLAND

RICHARD A. THULBORN,Department of Zoology, University of Queensland

andMARY WADE

Queensland Museum

ABSTRACT

Dinosaur trackways have been discovered at several closely-grouped sites in mid-Cretaceoussediments of the Winton Formation, central W. Queensland. At one of these sites about 209 rri 2

of bedding plane was exposed to reveal trackways of more than 150 bipedal dinosaurs. One ofthese trackways is very much larger than any of the others; it is attributed to a large theropoddinosaur (carnosaur) and is identified as cf. Tyrannosauropus. The remaining trackways arereferred to two new ichnotaxa — Wintonopus latomorum and Skartopus australis — which areattributed to ornithopods and coelurosaurs (respectively). The sizes of the track-makers areestimated by means of allometric equations derived from osteological data; the speeds of thetrack-makers are estimated by using the mathematical relationships of size, speed and gait thathave been determined for living tetrapods. The carnosaur is estimated to have been about 2.6m high at the hip, and to have been walking at a speed of about 7 km/h. The ornithopod track-makers ranged from 14 to 158 cm in height at the hip; these animals were using a fast runninggait equivalent to cantering or galloping in mammals, and their mean speed is estimated at 16km/h. The coelurosaur track-makers ranged from 13 to 22 cm in height at the hip; these toowere using a fast running gait, and their mean speed is estimated to have been 12 km/h. Thetrackways of the ornithopods and coelurosaurs are interpreted as those of animals caught up ina stampede — which was presumably triggered by the approach of the carnosaur. It issuggested that relative stride length (i.e. stride length relative to height at the hip) is the bestavailable criterion for appraising the locomotor performances of dinosaurian track-makers. Bythis criterion the performances of the Winton ornithopods and coelurosaurs are outstandinglygood. There is an indication that these animals were running at or near their maximum speeds— with relative stride lengths in the range 3.9 to 5.0. If the most highly adapted of cursorialdinosaurs (the ornithomimids or 'ostrich dinosaurs') attained such figures for relative stridelength their speeds would have been up to about 60 km/h.

INTRODUCTIONIn June 1976 Mr Ron McKenzie showed us

some well-preserved dinosaur footprints that hehad collected from a site about 120 km SW ofWinton, central west Queensland. The footprintsite, which was later named Seymour Quarry, wasin sediments of the Winton Formation (mid-Cretaceous) and its existence was known to manyresidents in the Winton area. In 1971 a small fieldparty including Dr R.H. Tedford (AmericanMuseum of Natural History) and Dr A.Bartholomai (Queensland Museum) paid a briefvisit to the site; this party established that thefootprint horizon extended to a second site some100 metres away (Knowles 1980). This second

site, which was later named Lark Quarry,subsequently proved to be of very great interest.

The footprints at these localities attracted ourattention because of their abundance, theirexcellent preservation and their remarkably smallsize (by dinosaurian standards). In 1976 wecarried out preliminary excavations at both sites,and in the following year a large labour-force ofvolunteers co-operated in a major excavation atLark Quarry. This excavation revealed severalthousands of footprints representing thetrackways of well over 100 bipedal dinosaurs - -many of them apparently no bigger th inchickens. In preliminary accounts the LarkQuarry trackways have been interpreted as

414^ MEMOIRS OF THE QUEENSLAND MUSEUM

evidence of a dinosaur stampede (Thulborn andWade 1979, Wade 1979). If this interpretation iscorrect it may carry important implications forcurrent understanding of dinosaur biology. Thepresent work has three aims: 1) to provide asystematic account of the trackways at LarkQuarry and its surroundings; 2) to offer someinterpretations regarding the sizes, speeds andbehaviour of the track-makers; 3) to justify thoseinterpretations and to consider their implicationsfor the understanding of dinosaur biology.

LOCALITIESThe footprints described in this paper were

found at three localities on Mt Cameronproperty, SW of the town of Winton in centralwest Queensland. About 95 km SW of Winton theroad to Jundah and Stonehenge runs along thecrest of a west-facing scarp (the Tully Range)which is formed by sediments of the WintonFormation capped by duricrust. The localities areclose to the foot of the scarp, alongside a trackleading NW to Cork Station. The maximumdistance between any two of the localities is about200 m (see map, Fig. 1).

Seymour Quarry. A deep hillside cuttingalongside the track to Cork Station. The site isidentified as an opal mine on the Brighton Downssheet of the BMR 1:250,000 series of geologicalmaps (sheet SF/54-15; map reference 23°01'S,142°24'W). Footprints occur as natural castsbelow a thin bed of red arkosic sandstone thatoutcrops at the foot of the hill. This friablesandstone overlies a weathered mudstone, and itslower surface is infiltrated by dark brownironstone which prevents the footprints fromcrumbling on exposure. Traces of plant rootletsare preserved along with the footprints, whilethese themselves are very well preserved and mayeven show indications of skin texture (see p.422, Pl. 1). The footprints are attributed tosmall bipedal dinosaurs of two types(coelurosaurs and ornithopods), and they appearto represent continuations of trackways atanother site to the SW (Lark Quarry). This firstsite is named for Mr Glen Seymour, its discovererand former manager of Cork Station.

Lark Quarry (P1. 3). A large excavationrevealing more than 200 m of a single beddingplane. This site is located to the SW of SeymourQuarry, and has been the subject of preliminarydescriptions (Thulborn and Wade 1979, Wade1979). The Lark Quarry bedding plane carrieswell over 3000 footprints, representing thetrackways of at least 150 bipedal dinosaurs. The

trackways are almost entirely unidirectional: onetrack-maker was headed to the SW, whereas allthe others were headed to the NE (in the directionof the present Seymour Quarry). The footprintsoccur as natural moulds in a bed of laminatedclaystone which varies between 6 and 12 cm inthickness. The footprint itself will be referred toas a mould, and the filling of the footprint as acast, in conformity with standard ichnologicalusage. The claystone is generally bright pink incolour (though individual laminae range frompink through red to purple), and its upper surfaceappears to be stained dark red-brown byironstone infiltration. This 'surface stain' is infact an extremely thin adhesion from the base ofthe overlying sandstone. Below the claystone is athick bed of arkosic sandstone; this is buff incolour and finely cross-bedded. Similarsandstone/claystone couplets occur above andbelow the trackway horizon. The next claystonebed below the trackway horizon also bearsfootprints in the form of natural moulds, thoughthese are uncommon and seem to have nopreferred orientation. The footprint horizon atSeymour Quarry seems to be an extension of themain trackway surface at Lark Quarry; it ispossible to trace the footprint horizon throughintermediate outcrops, though there is a completebreak of about 30 metres caused by a creek bed.Moreover there is a uniform dip to the NW ofabout 4 0 , and by taking direct line of sight alongthe Lark Quarry bedding plane this will be foundto coincide with the footprint horizon at SeymourQuarry. It may be mentioned that the reverseprocedure (extrapolating the dip of the beds atSeymour Quarry) was used to locate the LarkQuarry trackways in the first instance (Knowles1980). At both sites the footprints are similar indiversity, in abundance, in morphology and intheir singular orientation. Lark Quarry has beendesignated an Environmental Park by theNational Parks and Wildlife Service, Queensland,and is now roofed for its protection. The site isnamed for Mr Malcolm Lark, of Miles, whoplayed a leading role in its excavation.

New Quarry. One of a series of small hillsideexposures scattered from 100 to about 120 m dueS of Lark Quarry. At the New Quarry site thetrackway of a single bipedal dinosaur wasmeasured in situ. In its preservation this trackwayis identical to those at Lark Quarry. There is amajor erosional gap between New Quarry andLark Quarry, but at both sites the footprintsoccur at equivalent levels in similar sequences ofsandstone/claystone couplets. Moreover at both

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS 415

LARK QUARRY

FIGURE 1. Map showing location of footprint sites. Contours are at 5 m intervals, and dashed lines indicate drycreek-beds. Inset map shows location of quarry in Queensland. B, Brisbane; C, Cairns; M, Mt. Isa; R,Rockhampton; T, Townsville.

416^ MEMOIRS OF THE QUEENSLAND MUSEUM

sites there is a marked change in sediment typeabout 5 or 6 m above the footprints — theappearance of a yellowish arkose containingscattered plant fragments. The evidence is notconclusive, but it suggests that the New Quarrytrackway is at about the same stratigraphic levelas the Lark Quarry trackways. Footprints alsooccur in the next claystone layer below the NewQuarry trackway; at this lower level the claystoneis thoroughly trampled and churned up by deepfootprints without preferred orientation.

METHODSEXCAVATION AND COLLECTION OF FOOTPRINTS

At Seymour Quarry the footprint horizon wasreached by digging through the overburden of soiland weathered rock. The thin sandstone layerbearing the footprints (natural casts) provedrather fragile; most footprints collected from thissite are on small slabs or are in the form ofdetached casts (see Pl. 1). One large slab (QMF12266) is approximately 95 by 40 cm and carriescasts of at least 28 footprints — representingabout 19 trackways.

At Lark Quarry the footprints were exposed bybreaking up and removing a thick overburden ofsandstone. Fortunately the sandstone was welljointed (as is the footprint surface — see Pl. 4),and it could be removed in blocks once these hadbeen levered out with crowbars and jack-hammer.About 60 tons of overburden was removed,exposing an area of more than 209 m. It was thennecessary to clean the footprints (natural moulds)be removing the sandstone that filled them. Thissandstone filling was soft enough to be broken upwith an awl. At the bottom of each footprintmould the colour of the sandstone filling changedfrom orange-red to bright yellow-green — auseful guide to ensuring that the footprints werefully excavated. More than 3300 footprints wereexposed and cleaned in this way. A portion of thefootprint bed was removed from the eroded NEmargin of the site and was transferred to theQueensland Museum (QM F10321). In additionseveral individual footprints were collected —including holotypes of the new ichnotaxadescribed below.

FIBREGLASS REPLICASAfter the Lark Quarry bedding plane had been

exposed, and its footprints were thoroughlyexcavated, it was swept free of dust and rockdebris; large parts of the bedding plane were thencoated with liquid latex, which was reinforcedwith a cloth backing. Once it had set, the latexwas stripped off in the form of large 'peels'

(Wade 1979). These latex 'peels' were later usedas a basis for moulding a fibreglass replica of thebedding plane and its footprints. The entire areashown in Fig. 3 was included in this replica.Individual footprints were also replicated (see, forexample, Pl. 5, Fig. A). These high-fidelityreplicas are much lighter and more durable thanplaster casts; they enabled us to undertake a long-term study of the Lark Quarry footprints, eventhough our total expenditure of time at the sitewas no more than a few weeks.

ILLUSTRATIONSThe Lark Quarry bedding plane is almost

horizontal, and its footprints are under naturallow-angle lighting for only a few minutes afterdawn and before dusk. Even at these times of dayit may be difficult to obtain worthwhilephotographs because the direction and intensityof lighting cannot be adjusted. We obtained fewgood quality photographs of individual footprintsin situ. Most of our illustrations (Pl. 5 to 16) showfibreglass replicas of the footprints — thoughsome do show original material (including typespecimens; see plate captions).

Most of the area exposed at Lark Quarry wasmarked out with a grid of chalk lines, and eachquadrat was photographed from a height of 1metre (with the camera mounted on a rigid ironframe). The resulting photographs were thenassembled into an accurate and detailedphotomosaic — a representative portion of whichis reproduced as Pl. 4. The same photographswere later used in drawing up a chart to show thedistribution of footprints at Lark Quarry (Pl. 17).

DESCRIPTIONSThere are no universally accepted methods for

describing footprints and trackways (Sarjeant1975, Leonardi 1979a), and it is necessary todefine the measurements and statistics weemploy. All linear measurements are expressed incentimetres.

Footprint length (abbreviation FL) — themaximum footprint dimension measured along,or parallel to, the axis of the longest digit (seeFigs. 2A, B).

Footprint width (FW) — the maximumfootprint dimension measured at a right angle tofootprint length (Figs. 2A, B).

The ratio footprint width / footprint length(ratio FW/FL) is used to express footprintproportions.

We discovered that FL and FW were quitevariable within each trackway, so that neither ofthese measurements could be regarded as acompletely reliable indicator of the track-maker's

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^417

size. Consequently we calculated an index offootprint size (SI) for each footprint:

SI = (FL x FW)"This index was found to be remarkably consistentwithin each trackway, and it seems to be a usefulguide to the relative sizes of two or more track-makers. The index is expressed in centimetres.

Interdigital angles (expressing divarication ofdigits) are commonly cited in descriptions ofvertebrate footprints, but they are difficult tomeasure consistently (see Sarjeant 1975) and areoften so variable that they are of questionablevalue (see comments of Welles 1971). We havenot attempted to compile detailed measurementsof interdigital angles, and they will be mentionedonly as approximate averages.

Pace length (PL) — the distance betweencorresponding points in two successive footprints(left and right, or right and left; see Fig. 2C).

Stride length (SL) — the distance betweencorresponding points in two successive prints of asingle foot (see Fig. 2C).

With measurements of two successive paces(PL and PL b), and of the stride (SL) that theyencompass, it is possible to calculate paceangulation (ANG) as follows:

(PLb) 2 — (SL)2cos ANG = (P 1-92+ 2 x (PLO) x (PO

The more nearly pace angulation approaches to180 0 the narrower is the trackway and the lessobvious is the zig-zag arrangement of itsfootprints (see Fig. 2C).

The ratios pace length / footprint length(PL/FL) and stride length / footprint length(SL/FL) are often used in definitions ofichnotaxa (see, for example, Lull 1953, Haubold1971) and are also provided here. These ratiostend to increase as a track-maker accelerates, andthey can therefore give a useful indication of atrack-maker's gait. In calculating these ratios FLwas taken to be the mean for the two footprintsdefining each pace or stride.

To calculate means and other statistics it wasusually necessary to reduce sample sizes (N) byexcluding data from damaged or badly distortedfootprints.

DESCRIPTIONSAt first glance the dinosaur trackways at Lark

Quarry present a rather confusing picture (Pl. 4).However, it soon becomes apparent that thetrackways can be sorted into several naturalgroups on the basis of size, orientation,preservation and footprint shape. Five suchgroupings may be recognized.

A. Remnants of a few trackways made by fairlylarge bipedal dinosaurs. These remnants comprisescattered footprints which are very poorlypreserved and have no preferred orientation. Thefootprints appear to have been tridactyl, withrather short, thick and bluntly-rounded toes, andthey are tentatively attributed to ornithopoddinosaurs. They seem to have been formed, theneroded and filled with water-laid sediment, wellbefore the substrate was exposed to the air andthe other footprints were formed at Lark Quarry.It was not possible to obtain any accuratemeasurements, and these remnants of oldtrackways will not be considered further.

B. A single trackway of a medium-size bipedaldinosaur (B in Fig. 3). This trackway extendsacross the southern part of Lark Quarry fromWSW to ENE, and is attributed to an ornithopoddinosaur. The tridactyl footprints have relativelyshort, broad and well-rounded digits (see examplea in Fig. 4) and are referred to the sameichnotaxon as the many small footprints in groupD (below). However, this trackway is much largerthan any of those in group D, and it was certainlyformed at an earlier date: some of its footprintswere deeply impressed in soft waterlogged mudand others (in lower-lying areas) were partlydestroyed by scouring. This trackway seems tohave been formed at about the time the substratewas draining free of surface water and wasbecoming exposed to the air.

C. A single trackway of an exceptionally largebipedal dinosaur (C in Fig. 3). This trackwayextends across the northern part of Lark Quarryfrom NE to SW, and is attributed to a carnosaur— a large representative of the Theropoda. Thefootprints are very obvious basin-like structures(Pl. 5, Fig. B), and some of them show cleartraces of three tapering or V-shaped digits (Fig.4).

D. Numerous trackways of small to medium-sized bipedal dinosaurs; extending from SW toNE (Fig. 3, Pl. 4). The footprints are wellpreserved and each of them comprises three fairlyshort, thick and bluntly rounded digits. Thesetrackways are attributed to ornithopod dinosaurs,and their footprints may be found superimposedupon those of the carnosaur (C, above) and uponthose of coelurosaurs (E, below).

E. Numerous trackways of small (andsometimes very small) bipedal dinosaurs;extending from SW to NE (Fig. 3, Pl. 4). Each ofthese trackways comprises footprints with threefairly long, narrow and sharply-pointed digits.The trackways are attributed to coelurosaurs(small dinosaurs of the suborder Theropoda), and

418^ MEMOIRS OF THE QUEENSLAND MUSEUM

their footprints may be found superimposed uponthose of the carnosaur (C ) and upon those of theornithopods (group D).

Footprints in the latter three groups are equallywell preserved, and all of them seem to have beenformed at about the same time. These footprintswere formed after the muddy substrate had beenexposed long enough to have attained a firmplastic consistency. From the evidence ofsuperimposed footprints it is clear that thecarnosaur traversed the Lark Quarry area beforesome, at least, of the ornithopods andcoelurosaurs did so.

The trackways attributed to ornithopods(group D) and coelurosaurs (group E) extend in asingle direction, and among them it is common tofind trackways coinciding, intersecting at lowangles, or weaving together inextricably (Pl. 4 and14). Moreover in some places the trackways ofsmall, and even medium-size, individuals quitesimply disappear: apparently these smallerdinosaurs were so light that their feet failed tobreak through the firmer patches of surfacesediment. These discontinuities are especiallynoticeable among the coelurosaur trackways(group F ); the coelurosaur track-makers seem tohave had relatively large feet (by comparison withthe ornithopod track-makers), and their widely-spread and probably rather springy toes seem tohave functioned as analogues of snow-shoes.Many of the track-makers, both ornithopods andcoelurosaurs, seem to have been roughly similarin size (the majority having hip height estimatedat less than 50 cm), and the footprints in any onetrackway are not always consistent in their shapeor spacing. This combination of factors makes itdifficult to trace any single trackway, withconfidence, for more than a few strides.Consequently our descriptions and analyses arebased, in the main, on relatively short sections oftrackways. For the ornithopod dinosaurs (groupD ) we examined 57 sections of trackways; onaverage each of these comprises 3 strides (asequence of 5 footprints). The longest section ofornithopod trackway studied here comprises 17strides (a sequence of 19 footprints). For thecoelurosaurs (group E) we examined 34 sectionsof trackways; here the average number of stridesper section of trackway is between 3 and 4(between 5 and 6 footprints). The longest sectionof coelurosaur trackway studied here comprises22 strides (a sequence of 24 footprints). Thedifference in sample size (57 ornithopodtrackways as opposed to 34 coelurosaurtrackways) does not indicate that ornithopods

were more abundant than coelurosaurs. On thecontrary, the ornithopod track-makers wereprobably outnumbered by the coelurosaur track-makers (see p. 443). The sample sizes differ fortwo reasons. First, the coelurosaur trackways areaffected by so many discontinuities that it isdifficult to find sequences of more than a fewpaces. Second, the coelurosaur footprints showmuch less variation in size and shape than do theornithopod footprints: coincident or intersectingtrackways of ornithopods could usually beseparated through differences in footprint size orfootprint shape, but coincident or intersectingtrackways of coelurosaurs were usuallyinextricable.

The trackways of the carnosaur, theornithopods and the coelurosaurs are described inturn. But before proceeding to these descriptionsit will be useful to consider the circumstancesunder which the trackways were formed. It isimportant to determine these circumstancesbecause some of them (e.g. the consistency of thesubstrate) have a direct bearing on footprintmorphology, while others (e.g. the physicalgeography) are pertinent to the behaviour of thetrack-makers.

The Winton Formation is a series ofcontinental sediments that reaches a thickness ofmore than 1000 feet in the area around Winton(Casey 1966). The sediments are mainlysandstones, siltstones and mudstones, thoughthere are some intraformational conglomeratesand coal seams. Fragments of fossil wood arecommon, but other well preserved fossils are rare;these include angiosperm leaves, conifers,freshwater bivalves, lungfish toothplates andfragmentary remains of sauropod dinosaurs(Senior, Mond and Harrison 1978, Coombs andMolnar 1981). The Winton Formation is mid-Cretaceous in age (probably uppermost Albian toCenomanian), and the conditions under which itssediments accumulated were well summarized bySenior et al. (1978, p. 15): 'Terrestrial-fluviatile,paludal, lacustrine. Low relief, wide river flats,local development of short lived lakes andswamps.'

The sediments at and around Lark Quarry areof lacustrine and fluviatile origin. At the time thedinosaur trackways were formed the Lark Quarrysite seems to have represented part of a majordrainage channel; it was most probably part of aplatform lying on point bar deposits, a sand-spit,that had built out into a lake which deepened tothe SW (see Fig. 25A). The Lark Quarry beddingplane now dips NW at 4 0 , but it originally had a

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^419

run-off to the SW — as is indicated by theorientation of drag marks and prod-marksproduced by floating vegetation (Pl. 4). At timesof flood the lake would have spread over a widearea (including all three footprint localities) todeposit sand followed by muddy sediments. In theintervening periods the lake would have recededto become little more than a remnant water-holesurrounded by newly-exposed mud. It was duringsuch a period that the trackways seem to havebeen formed.

The mud began to compact under water, andbefore it was fully exposed a single dinosaurtraversed the southern part of the future LarkQuarry site from WSW to ENE. In this trackway(B ) some footprints were formed as the animalcrossed slightly elevated and newly-exposedpatches of very soft sediment; the other footprintswere formed in lower-lying areas of mud whichwere still covered by water. These lower-lyingfootprints were subsequently scoured and erodedas the remaining surface water drained off to theSW. At the future site of Lark Quarry therecently-laid sand and overlying mud werepenetrated by narrow, vertical and unbranchedtubes that probably mark the escape of buriedarthropods (Pl. 2, Fig. B). Traces of similarescape burrows may be found at the SeymourQuarry site, along with horizontal and obliquetubular structures that seem to represent plantrootlets of various sizes (Pl. 1, Figs. A, B). Thepresence of plant rootlets might indicate that themud was exposed sufficiently long for terrestrialvegetation to take hold. After the mud had beenexposed for some time it was traversed by a singlecarnosaur and by numerous ornithopods andcoelurosaurs (trackway groups C, D and Eabove). The mud was exposed long enough toachieve a firm plastic consistency, but not longenough for desiccation cracks to appear. Theperiod of exposure would certainly have been amatter of hours, if not of days or weeks.Evidently the mud was not waterlogged at thetime it was traversed by the dinosaurs: none of thethousands of footprints collapsed or slumpedafter withdrawal of the track-maker's foot. Nordoes the mud seem to have been very tenacious,for there are very few instances in which itadhered to a track-maker's foot. In one of these(footprins, No. 8 in Fig. 4) the mud adhering to theunderside of a single toe was drawn up into alongitudinal crest; in another (Pl. 8, Fig. C) mudadhered to one toe in the form of a 'cusp' or'bubble-like' structure. However, we suspect thatin many cases the imprints of one or more digitshave been narrowed by suction created during

withdrawal of the track-maker's foot. Overall itseems that the mud may have had the consistencyof potter's clay at the time it was traversed by thedinosaurs.

CARNOSAUR TRACKWAYIchnogenus cf. Tyrannosauropus Haubold 1971

Eleven footprints at Lark Quarry are far biggerthan any others, and form a single trackwayextending from NE to SW (Figs 3 and 4, Pls 4 to6). This trackway is attributed to a carnosaur — alarge bipedal predator of the dinosaur suborderTheropoda (order Saurischia).

It was not feasible to collect any of thecarnosaur footprints, for to do so it would havebeen necessary to destroy many other trackways.In any case, the footprints do not show sufficientdetail to warrant their assignment to any new orexisting ichnospecies. Measurements of thecarnosaur trackway were taken directly from thebedding plane at Lark Quarry and were checkedon fibreglass replicas (QM F10322) at theQueensland Museum.

DESCRIPTION.The trackway comprises deep, basin-like and

rather 'messy' footprints, often with poorlydefined margins. Evidently the track-maker's feetplunged right through the muddy surface layerand churned up the underlying sandy sediment. Inmost cases the impact of the foot caused sedimentto bulge up between the toes and behind the footto leave a prominent raised rim to the footprint(Pl. 5, Fig. B). The sandy sediment in the floor ofthe footprint is usually raised into a series ofirregular ripples. Two of the carnosaur footprintshave been illustrated elsewhere (Thulborn andWade 1979, fig. 2), and two more examples areshown here (Pls 5 and 6). The followingdescription is a generalized one, based oninformation from all better-preserved footprintsin the trackway.

Each footprint is tridactyl, with clear imprintsof digits 2, 3 and 4, but with no trace of the hallux(digit 1). The three digits are relatively short,emerging from a large basin-like depressionrepresenting a 'sole' or 'pad' to the foot; infootprint No. 3, for example, the length of digit3, as a free entity, represents only about 41% oftotal footprint length (Pl. 5, Fig. A). The digitsare usually quite sharply defined, but often thereis no clear outline to the back of the foot, makingit difficult to obtain a measurement of totalfootprint length (see, for example, Pl. 5). All thebetter-preserved footprints are slightly narrower

420^ MEMOIRS OF THE QUEENSLAND MUSEUM

than long, with footprint width equivalent tosome 85-90% of footprint length. The digits arebroad and straight, and taper sharply to V-shapedtips. In none of the footprints are there traces ofdigital swellings or nodes that might indicate thephalangeal formula. Digits 2 and 4 are distinctlyshorter than digit 3, and are roughly mirror-images — being almost complementary in shape,size and angle of divergence from digit 3. In oneof the best-preserved footprints (No. 3; Pl. 5, Fig.A) the interdigital angles are about 33° (2-3) and30° (3-4).

Mean measurements of footprint size, pacelength and stride length are as follows (each withstandard deviation and coefficient of variation):

Mean FL: 51.4 ± 6.5 cm (CV 13 07o; N 7)Mean FW: 46.1 ± 4.0 cm (CV 9 07o; N 10)Mean PL: 166.6 ± 26.5 cm (CV 16%; N 10)Mean SL: 330.6 ± 37.4 cm (CV 11%; N 9)

Mean ratio FW/FL: 0.88 ± 0.05 (CV 6%; N 7).

The near-symmetrical footprints are arrangedwith slight positive rotation (i.e. they point notonly forwards, but slightly inwards). They form anarrow trackway, with mean pace angulationcalculated at 170°47' (SD 9°26'; CV 5.5% N 9).The trackway has a slightly sinuous course fromNE to SW. From the spacing and orientation ofthe first few footprints we suspect that the animalactually approached the present Lark Quarry sitefrom the NNE; the orientation of the last (11th)footprint indicates that the animal made anabrupt right-hand turn and moved off to the NW(see Fig. 3). There is no trace of a tail drag.

Some of the footprints show additional detailsof interest. Footprint No. 7, for example, consistsof little more than shallow imprints of the threedigits (PI. 6), and seems to have been formed on arelatively resistant patch of sediment. In thisfootprint there is evidence that the tips of thedigits extend forwards, beneath the surface of thesediment, as conical tunnels about 4 cm in length.These tunnels appear to be marks of long, robustand sharply-pointed claws. Traces of similarclaws occur in several other footprints of thecarnosaur. In footprint No. 8 the central digit (3)is unusually broad and contains a longitudinalcrest of mudstone in the midline (Fig. 4; see alsoThulborn and Wade 1979, fig. 1B). This crest waspresumably formed by mud adhering to theunderside of the middle toe as the animal's footwas lifted from the substrate; other prints fromthe same foot do not show this feature. Fig. 4illustrates variation in shape of the carnosaur'sfootprints.

STATUS AND AFFINITIES.The occurrence of a carnosaur trackway at

Lark Quarry is not unduly surprising. Carnosaursseem to have had an almost world-widedistribution during the Cretaceous period, withtheir footprints having been reported as far afieldas Spitzbergen (Edwards et al. 1978) and WesternAustralia (Colbert and Merrilees 1968). Noskeletal remains of carnosaurs are recorded fromQueensland, though footprints of large theropoddinosaurs are well known in the Jurassic rocks ofthe state (Ball 1933, 1934a, 1934b, 1946;Anonymous 1951, 1952a, 1952b; Staines 1954;Bartholomai 1966). For the sake of conveniencewe may distinguish two major groupings ofcarnosaur footprints in general: those withrelatively long and slender toes, and those withcomparatively short and thick toes. Examples ofthese two groupings are, respectively,Megalosauropus and Tyrannosauropus (seeHaubold 1971 and references cited therein). Theformer are probably footprints of smaller andmore gracile carnosaurs, such as Allosaurus andMegalosaurus, while the latter probably representbigger and more robust forms likeTyrannosaurus. The footprints of the LarkQuarry animal have rather short thick toes, andthey appear to be closer in appearance toTyrannosauropus than to any other form ofcarnosaur footprint so far described. The LarkQuarry footprints resemble Tyrannosauropus ingeneral shape and proportions (mean ratioFW/FL of 0.88 as opposed to approximately 0.86in Tyrannosauropus ), but they differ in thefollowing respects: in size (FL up to 80 cm inTyrannosauropus ), in pace angulation (170° asopposed to approximately 150°), and in the ratioSL/FL (6.4 as opposed to 5.0). On the basis ofthese similarities and differences we recommendthat the Lark Quarry footprints should bereferred to as cf Tyrannosauropus. Thisidentification does not imply that the theropoddinosaur Tyrannosaurus was responsible for theLark Quarry trackway; the track-maker can beidentified no more precisely than `carnosaue.

It must be mentioned here that footprints ofcarnosaurs have often been confused with thoseof ornithopod dinosaurs (bipedal herbivores ofthe suborder Ornithopoda, order Ornithischia).The source of this confusion is partly historical:the first footprints to be attributed to a particulargenus of dinosaur — the ornithopod Iguanodon— happened to be large tridactyl examples fromthe Lower Cretaceous of Europe. Subsequently,there grew a common tendency for any largetridactyl footprints to be ascribed to Iguanodon

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^421

or to some similar ornithopod dinosaur (seecomments by Beckles 1862, Charig and Newman1962, Sarjeant 1974). Such confusion has, in fact,occurred over footprints from the AustralianCretaceous: large tridactyl prints from theBroome Sandstone (Lower Cretaceous) ofWestern Australia were attributed to iguanodontsby McWhae et al. (1958), but were later identifiedas those of theropod dinosaurs (Colbert andMerrilees 1968). Iguanodontids and sometheropods both produced large tridactylfootprints which may, in some circumstances, bedifficult to distinguish — particularly ifpreservation is poor. It has been suggested to us(by Dr D.B. Norman, pers. comm.) that somedoubt may attach to the footprints of the LarkQuarry animal, and that these might actually befootprints of an iguanodontid ornithopod.However, several distinctive features of thefootprints lead us to conclude that they are almostcertainly those of a carnosaur. First, thefootprints are slightly longer than broad, whereasthose of ornithopods are commonly broader thanlong (see, for example, Langston 1960, Currieand Sarjeant 1979, and the many ornithopodfootprints described below). Next, the three digitshave an almost symmetrical arrangement; inmany ornithopod footprints digit 2 is more widelyspaced from digit 3 than is digit 4 (see sameexamples). Third, the Lark Quarry footprintshave traces of a large pointed claw on each toe;the ungual phalanges of the larger ornithopodswere blunter, and sometimes rather hoof-like,structures. Finally, the central digit (3) is V-shaped in outline; in ornithopod footprints digit 3tends to have roughly parallel margins that curveround to form a U-shaped extremity. These basicdistinctions seem to confirm that the largetrackway at Lark Quarry is that of a carnosaur(see Fig. 4).

ORNITHOPOD TRACKWAYSIchnogenus Wintonopus ichnogen. nov.

Type and only ichnospecies W. latomorumichnosp. nov.

HOLOTYPE: single right footprint, preserved asnatural mould; QM F10319 (Pl. 7, Fig. A).REFERRED MATERIAL: QM F10320 (singleleft footprint, as natural mould; Pl. 11, Fig. A);QM F10321 (rock slab with footprints andtrackways as natural moulds); QM F12264 (singleright footprint, as natural cast; Pl. 1, Figs. A, B);QM F10322 (fibreglass replicas of footprints andtrackways preserved as natural moulds; Pl. 8 to

10; Pl. 11, Figs. B, C, D; Pl. 13, Fig. C; Pl. 14,Fig. A; Pl. 16, Figs. B, C).LOCALITIES: Lark Quarry (QM F10319, QMF10320, QM F10321, QM F10322); SeymourQuarry (QM F12264). See Fig. 1 for location ofquarries.HORIZON: interbedded sandstones andmudstones about the middle of the WintonFormation; early Upper Cretaceous(Cenomanian).ETYMOLOGY: Ichnogenus name derived fromthe name Winton and the Greek pous (foot);ichnospecies name (from Latin latomus, stone-mason) as tribute to the many volunteers whoworked at the Lark Quarry excavation.DIAGNOSIS (ichnogenus and ichnospecies):narrow trackway of small to medium-sizedigitigrade biped, with pace angulation about160°. Footprint size index (SI) usually between3.2 and 11.1 cm, but occasionally as high as 26.6cm. No imprints of hand or tail. Footprintstridactyl (digits 2, 3 and 4), slightly broader thanlong (ratio FW/FL about 1.15), showing distinctpositive rotation. Digits broad, with rounded orbluntly angular tips, without indications ofphalangeal pads. Digit 3 longest, with sub-parallelsides. Digit 4 shorter and slightly narrower thandigit 3, extended as blunt posterior salient. Digits3 and 4 close together, parallel or only slightlydivergent. Digit 2 shortest, and widely separatedfrom digit 3 (with interdigital angle often about60°). Imprints of digits 2 and 3 sometimescompletely separated. Posterior margin of footconvex forwards. Ratio PL/FL usually between8.0 and 13.5, rarely as low as 4.0 or as high as15.0; ratio SL/FL usually between 16.0 and 24.0,rarely as low as 8.0 or as high as 27.0.DESCRIPTION. The holotype is a sharplydefined footprint, probably formed by the footpenetrating and leaving the substrate withminimal disturbance (Pl. 7, Fig. A). Few of thefootprints referred to Wintonopus latomorum areidentical to the holotype: the majority have lesscomplete imprints of the digits, and many appearto have been disfigured by withdrawal of thetrack-maker's^foot^from^the sediment.Nevertheless all these disfigured and less completeexamples can be interpreted as variants of thefootprint pattern exemplified by the holotype (seeFig. 5). Variation in shape of the footprints isdescribed first; thereafter we describe variation insize, proportions and spacing of the footprints.

All the footprints are tridactyl. They are oftenseveral centimetres deep, yet none of them showsany trace of digit 1. If this digit was present in the

422^ MEMOIRS OF THE QUEENSLAND MUSEUM

track-maker's foot it must have been quite shortand non-supportive, so that it failed to touchdown even when the three weight-bearing digitssank quite deeply into the mud. Digit 1 wasprobably no longer than it is in ornithopods suchas the Lower Cretaceous Hypsilophodon (Fig.6B). The imprint of digit 4 extends farther backthan the imprints of digits 2 and 3, giving theimpression of a distinct salient or 'spur' at theposterolateral corner of the footprint. This 'spur'is unlikely to indicate the presence of a functionaldigit 5 (which is vestigial in even the earliestornithopod dinosaurs) and probably reflects thatthe distal end of metatarsal 4 was located well tothe rear of the foot, in standard ornithopodfashion (see, for example, the foot ofFabrosaurus — Thulborn 1972, fig. 12B).

Nearly all the footprints are asymmetrical, witha strongly divergent digit 2, so that isolatedexamples are readily identified as left or right.This method of identification has been verified inat least 60 Wintonopus trackways. Only a fewfootprints (in otherwise typical trackways) showany close approach to a symmetrical arrangementof the three digits. These near-symmetrical printscould have been produced in any of several ways:by some degree of spreading and/or closure ofdigits in the foot; by partial flexion of thedivergent digit 2; by rotation of the foot aroundthe long axis of digit 3 (so that digit 2 was carriedlaterally and slightly underneath digit 3); by thefoot meeting the substrate at an unusual obliqueangle (with the long axis of digit 3 directedforwards, downwards and slightly outwards).

The imprints of the digits are usually quiteshort and relatively broad. In most footprints thethree digits are about equally broad (as, forexample, in the holotype), but in a few cases digit3 is much broader than digits 2 and 4 (e.g. Pl. 9,Fig. D; Pl. 10, Fig. D). In these examples digit 3seems to have borne most of the track-maker'sweight, while the flanking digits splayed out toform smaller and shallower imprints. A verysimilar effect was described by Sternberg (1932) inan ornithopod footprint (Gypsichnites pacensis )from the Lower Cretaceous of British Columbia.Digit 3 is longest, and in the least-disfiguredfootprints digits 2 and 4 extend about equally farforwards. The imprint of digit 3 is often straight,but sometimes shows very slight curvature(convex laterally; e.g. Pl. 11, Fig. A; Pl. 15, Fig.B). The hindmost margins of digits 2 and 3 lie ona line approximately normal to the long axis ofdigit 3, whereas digit 4 extends well behind thisline to form the posterior salient or 'spur'. In

consequence the posterior margin of the footprint(or the line connecting the hindmost points of thethree digital imprints) is arched forwards. In somefootprints, such as the holotype, the three digitalimprints are joined together posteriorly, and thearched rear margin is continuous. Evidently thesefootprints were formed by the foot sinking intothe mud up to, or beyond, the distal end of themetatarsus. In many other examples the foot didnot sink so deeply, so that the three digitalimprints are partly or completely separated andthere is no continuous rear margin to thefootprint (e.g. Figs 5D, E, F; Pl. 8, fig. B). Theimprint of digit 2 is often completely separatedfrom that of digit 3, whereas the imprints of digits3 and 4 are usually joined together (e.g. Pl. 8, FigsA, B, D). This difference probably indicates thatdigit 2 diverged from digit 3 higher up themetatarsus than did digit 4 (see foot skeletons ofornithopods, Fig. 6). There are no certainindications of phalangeal pads in any of thefootprints. The tips of the digital imprints aregenerally well-rounded in outline — except wherethey have been extended forwards as scrape-marks (see below) — but are sometimes a littlesharper in the smallest footprints (Pl. 11).Interdigital webbing is limited in extent; theholotype shows traces of a small web betweendigits 3 and 4, and less certain traces of anotherbetween digits 2 and 3.

Win tonopus material from Seymour Quarrycomprises natural casts reinforced by superficialinfiltration of ironstone. The surfaces of the castsare wrinkled and finely tuberculate, andsomewhat reminiscent of reptilian skin texture(Pl. 1). However, it is not certain that thesespecimens do show preservation of skin texture:an apparently identical texture is found onfootprint casts attributed to the coelurosaurs andon some areas of seemingly undisturbedsediment, and it may be no more than a by-product of ironstone formation.

Practically every Wintonopus footprint at LarkQuarry seems to have been disfigured to someextent as the track-maker's foot was withdrawnfrom the mud. Basically, each foot sank quitedeeply into the mud, and as it was lifted clear atthe end of the stride the tips of one or more digitstended to drag and scrape through the rim of thenewly-formed footprint. So, in many cases, thereare scrape-marks extending forwards from one ormore of the digital imprints. Digit 3 was longest inthe foot and, for that reason, tended to produce ascrape-mark most frequently (e.g. Pl. 8, Fig. A).Digit 4 was intermediate in length between digits 2

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^423

and 3, and was next most likely to produce ascrape-mark, whereas digit 2 was shortest(parallel to the long axis of digit 3) and rarely didso (e.g. Pl. 10, Fig. D). In most cases the scrape-marks are fairly short, but in a few examples theyare longer than the digital imprints (Pl. 8, Fig. A,and Pl. 10, Fig. D). The scrape-marks are notstraightforward extensions of the digital imprints,but veer away from them at a distinct angle.Sections of trackways show that each foot wasplanted into the mud with slight positive rotation(i.e. with the toes pointing forwards andinwards). But as the foot was lifted from the mudit evidently turned to face directly ahead, so thatthe tips of the toes swept forwards and slightlyoutwards. In other words the foot was placed inthe mud at one angle and was withdrawn at adifferent angle, and it is for this reason that thedigital imprints have a different orientation fromtheir scrape-marks. In some footprints this effectis so marked that the tip of digit 3 appears to beforked or Y-shaped (e.g. Pl. 8, Fig. D, and, to alesser extent, in the holotype). In such examplesthe medial branch of the fork was formed whenthe digit was placed into the mud, and the lateralbranch is a scrape-mark produced when the digitwas withdrawn in a different direction. Anexactly comparable type of scrape-mark wasillustrated by Sarjeant (1970, Fig. 5e) in anornithopod footprint (? Satapliasaurus cf. S.dsocenidzei ) from the Middle Jurassic ofYorkshire, England.

Clear examples of backwardly-directed scrape-marks are less common. Again, the longest digit(3) seems to have produced a scrape-mark mostfrequently whereas the shortest digit (2) rarelyproduced one. These scrape-marks are alsoaligned at a slight angle to the long axes of thedigital imprints (e.g. Pl. 8, Fig. B), confirmingthat the tips of the digits swung laterally as thefoot was lifted from the substrate.

The development of these scrape-marks(whether forwards or backwards) is bestunderstood in relation to the sequence of eventsduring the track-maker's stride cycle (Fig. 7). Atthe start of this cycle the forwardly-extended footwould have been planted into the sediment withslight positive rotation (Stage 1 in Fig. 7). Theinitial footprint would have been quite shallow.At mid-stride the track-maker's centre of gravitywould have passed forwards above the foot,which would then have sunk deeper into thesubstrate (Stage 2 in Fig. 7). In many instances thefoot also slipped backwards a little, so that thefront margins of the footprint are distinctly

'stepped' or 'terraced' (see, for example, Pl. 8,Figs. A, B; PI. 9, Fig. D). Shortly thereafter thefoot began to rotate (so that the long axis of digit3 was directed straight ahead), and the rear partof the foot started to lift clear of the substrate.Sometimes the toes continued to slip backwardsas they were lifted from the footprint: in thesecases the toe-tips incised deep slots in the floor ofthe footprint (Stage 3A in Fig. 7) or evenbreached the rear wall of the footprint to leavebackwardly-directed scrape-marks (Stage 3B).More commonly there was limited back-slip ofthe toes (Stages 1 to 2) and the toes-tips draggedthrough the front wall of the footprint to produceforwardly-directed scrape marks (Stage 3C).

Wintonopus footprints are typically broaderthan long, even though many examples have theirtotal length exaggerated by scrape-marks. Insome cases the track-maker's foot was plantedinto the mud at a steep angle, to leave relativelyshort and stubby imprints of the toes (e.g. Pl. 8,Fig. B). In other cases the foot seems to have lostits purchase in the muddy substrate, and the toesslithered back to form deep scratches thatexaggerate the total length of the footprint (e.g.Pl. 9, Fig. B). In still other cases only the distalparts of the toes entered the mud, and thenskidded backwards to produce a footprintconsisting of little more than three long scratch-marks (e.g. Pl. 10, Fig. A). A few footprintsconsist of three puncture-marks apparentlyformed by the toes entering and leaving the mudalmost vertically (e.g. Pl. 15, Fig. C).

Measurements of footprints, paces and strideswere taken from parts of 57 different trackwaysof Wintonopus (see 'Methods' for descriptions ofmeasurements). Fifty-six of these trackways areon the Lark Quarry bedding plane; the othersection of trackway is at a different site — NewQuarry. The 56 trackway sections at Lark Quarrycomprise 284 footprints, representing 228 pacesand 172 strides. This sample provides thefollowing mean figures for dimensions offootprints, paces and strides:

Overall meansMean FL: 6.71 ± 3.39 cm (CV 51Wo; N 200)Mean FW: 7.58 ± 4.51 cm (CV 60 07o; N 214)Mean PL: 68.3 ± 32.2 cm (CV 47 07o; N 215)Mean SL: 131.7 ± 63.4 cm (CV 48 ; N 162)

The high coefficients of variation reflectconsiderable ranges in size. Nearly all thetrackways are those of small animals, withfootprint lengths between 2 cm and 16 cm, andstride lengths in the range 49-271 cm; but the size

424^ MEMOIRS OF THE QUEENSLAND MUSEUM

distribution, as a whole, is attenuated by thepresence of a single very large trackway withfootprints up to 33 cm long and stride lengthsreaching 345 cm (trackway '13' in Fig. 3; see Figs 8and 10).

The index of footprint size, the pace angulationand various ratios were calculated from the basicmeasurements listed above (see `Methods'); theyhave the following means:

Overall meansMean SI: 7.20 ± 3.93 (CV 55%; N 173)Mean ANG: 161°24' ± 11°20' (CV 7% N145)

Mean ratio FW/FL: 1.15 ± 0.25 (CV 22%; N173)

*Mean ratio PL/FL: 10.18 ± 2.00 (CV 20%; N120)*Mean ratio SL/FL: 19.84 ± 3.69 (CV 19%; N88)(* FL is mean for two footprints defining eachpace or stride)

Footprint proportions (FW/FL) and paceangulation appear to show relatively littlevariation, and are probably good diagnosticcharacters; that is, Wintonopus trackways arecharacterized by being narrow, and fairlystraight, and by having footprints that are usuallybroader than long.

There is a strong positive correlation betweenany two measurements of size in the Wintonopustrackways and footprints; for example:

product momentcorrelation coefficients

untransformed log transformedvariables^data^dataFL : FW (N 174)

^0.87^

0.89*FL : PL (N 112)

^0.86^

0.87*FL : SL (N 79)

^0.87^

0.89*FW : SL (N 89)

^0.89^

0.93*SI : SL (N 59)

^0.90^

0.92(*mean for two footprints defining each pace or

stride)

The correlations between stride length andfootprint dimensions (FL, FW or SI) are worthyof note. It is only to be expected that biggeranimals would take bigger strides, but stridelength varies according to the gait and speed of ananimal - and not simply to its size alone. Theimpressive correlations between foot size andstride length imply that the Wintonopus track-makers at Lark Quarry were all using a similar

gait; in a random sample of dinosaur trackwaysone might expect to find a somewhat loosercorrelation between stride length and footprintdimensions.

The correlation between footprint size (SI) andfootprint proportions (ratio FW/FL) is muchpoorer: 0.22, with untransformed data, N 172.So, too, is that between footprint length and paceangulation (0.23, with untransformed data, N73). These poor correlations seem to confirm ourobservation that footprint proportions and paceangulation tend to remain fairly constantthroughout the entire size range of Wintonopustrackways (see further discussion below).

All the preceding estimates and statistics arebased on pooled data from the Wintonopustrackways (i.e. on every available example of the284 footprints and their paces and strides). If thedata are grouped, and mean figures are taken foreach of the 56 trackways studied, there emerges asomewhat similar pattern of size distribution andcorrelations (see Figs 9 and 10). Means based onthe grouped data may be summarized as follows:

Means per trackwayMean FL: 6.64 ± 3.07 cm (CV 46%; N 56)Mean FW: 7.55 ± 4.36 cm (CV 58%; N 56)Mean PL: 67.1 ± 30.6 cm (CV 46%; N 55)Mean SL: 128.0 ± 59.6 cm (CV 47%; N 52)Mean SI: 7.05 ± 3.58 cm (CV 51%; N 56)Mean ANG: 162°47' ± 8°46' (CV 5%; N 51)

Mean ratio FW/FL: 1.14 ± 0.19 (CV 18%; N 56)Mean ratio PL/FL: 10.18 ± 1.97 (CV 19%; N49)Mean ratio SL/FL: 19.75 ± 3.50 (CV 18%; N44)

Most coefficients of variation remain very high.An analysis of variance (see Sokal and Rohlf1969, p. 204 et seq. ) will reveal how much of thisvariation lies within the Wintonopus trackways:

^variation^variation

^

within^among

^

trackways^trackways

^

variable^(07o)^(M)^FL ^15.1^84.9

^

FW^6.0^94.0

^

PL^7.5^92.5

^

SL^2.4^97.6

^

SI^0.8^99.2

^

ratio FW/FL^52.1^47.9

^

ANG^77.9^22.1

Evidently footprint dimensions, pace length andstride length remain fairly constant within theWintonopus trackways. Footprint length is

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^425

somewhat more variable than footprint widthwithin the trackways, but the index of footprintsize is virtually constant. (Footprint length ismore variable than footprint width because it ismore strongly affected by the angle at which thefoot enters and leaves the substrate, and by thedevelopment of scrape-marks; see Fig. 5). Stridelength appears to be remarkably consistent withintrackways, though pace length is more variable.Footprint proportions (ratio FW/FL) and paceangulation appear to vary more within trackwaysthan they do between trackways; but these twofeatures show little overall variation in the firstplace — so they may still be regarded as gooddiagnostic characters. In general, most of thevariation in footprint dimensions, paces andstrides can be attributed to the difference in sizebetween one trackway and another.STATUS AND AFFINITIES. The footprintsreferred to Wintonopus latomorum vary a gooddeal in size and in appearance, yet they all showsome at least of the following diagnosticcharacters: a widely spaced or divergent digit 2, abackwardly-projecting 'spur' behind digit 4, aforwardly-arched rear margin, and a width thatequals or exceeds footprint length. Where thefootprints can be connected into trackways theyare arranged with distinct positive rotation, andthe trackways are always narrow and ratherstraight (with pace angulation about 160 0 ).Moreover, footprint dimensions are stronglycorrelated one with another, and with stridelength. In short, all the footprints may beregarded as those of animals sharing onedistinctive pattern of foot structure (see Fig. 6C)and using the same gait. For these reasons itseems justifiable to assemble all these footprintsin a single ichnospecies. Differences in shapebetween one footprint and another appear to beno more than circumstantial (see precedingdescriptions and Fig. 5). According to recentrecommendations for the nomenclature of tracefossils (see Article 40 in Basan 1979) it might belegitimate to define several ichnospecies ofWintonopus on the basis of footprint shape alone— e.g. a 'scratchy-toed' species, a 'stubby-toed'species, and so on. In the present circumstances,where footprint shape varies within a singletrackway, such a measure would be ratherconfusing. Moreover there is no clear evidencethat Wintonopus footprints of differentmorphology were 'produced in different phasesof behavior' on the part of the track-maker (asBasan's Article 40 seems to require). In addition itmay be noted that the range of variation in

Wintonopus is no greater than that in someexisting ichnotaxa — e.g. the ichnogenusAnomoepus (as defined by Lull 1953), and theichnospecies Grallator variabilis and G. olonensis(as defined by de Lapparent and Montenat 1967).

The makers of the Win tonopus trackways werealmost certainly dinosaurs of the suborderOrnithopoda (bipedal herbivores of the orderOrnithischia). Ornithopods had a world-widedistribution during the Mesozoic era: theirskeletal remains and their footprints have beenreported from every continent except Antarctica.The following features of Wintonopuslatomorum seem to be characteristic of very manyornithopod footprints: the footprint is tridactyl,and its width rivals or exceeds its length; thedigital imprints are relatively short, thick andblunt (indicating the presence of 'hoof-like'unguals rather than sharp claws); the spacebetween digits 2 and 3 is distinctly greater thanthat between digits 3 and 4; the outer margins ofdigits 2 and 4 diverge only slightly from thelongitudinal axis of digit 3, so that the footprinthas sub-parallel sides; there are sometimes tracesof small interdigital webs. In all these features W.latomorum resembles other footprints attributedto ornithopod dinosaurs — e.g. Amblydactylusichnospp. from the Lower Cretaceous of Canada(Sternberg 1932, Currie and Sarjeant 1979);unnamed types from the Late Jurassic/EarlyCretaceous of Mexico (Ferrusquia-Villafranca etal. 1979); footprints of Iguanodon, from theLower Cretaceous of Europe (Beckles 1862, Dollo1906). However, the footprints described here aredistinctly smaller than many others attributed toornithopod dinosaurs (see Table 1). Of the 57Wintonopus trackways examined in this studyonly two have mean SI greater than 12 cm(actually 12.7 cm and 26.6 cm); among otherfootprints attributed to ornithopods only those ofAnomoepus ichnospp. are commonly found to beso small. Aside from this Wintonopus differsfrom most other ornithopod footprints in oneother respect — in the absence (or weakdevelopment) of an imprint representing a 'sole'or 'heel' to the foot. The rear margin of thefootprint is concave (arched forwards) ratherthan convex (arched backwards) and presumablycorresponds to the natural arch formed by thedistal ends of the metatarsals. This distinctivefootprint shape seems to indicate that theWintonopus track-makers were thoroughlydigitigrade, whether they were walking or running(see later discussion of speeds and gaits). Thesedifferences in size and shape are sufficient to

426^ MEMOIRS OF THE QUEENSLAND MUSEUM

distinguish Wintonopus from most other tracksattributed to ornithopod dinosaurs. Anomoepusichnospp. are comparable in size to Wintonopus,but are distinguished by narrower and moreacutely pointed digits with obvious phalangealnodes (see Lull 1953). In addition most examplesof Anomoepus have the ratio PL/FL much lowerthan it is in Wintonopus (Fig. 11).

However, two types of footprint described byde Lapparent and Montenat (1967) from theRhaeto-Liassic of Vendee (W France) bear somedefinite resemblances to .Wintonopus. One ofthese types, Anat opus palmatus, was alsoattributed to an ornithopod dinosaur but is,unfortunately, represented by only three isolatedfootprints. In all three cases footprint size index isabout 9.0 cm — well within the range describedfor Wintonopus; the ratio FW/FL is about 1.14

— practically identical to the mean forWintonopus (1.15). Anatopus appears toresemble Wintonopus not only in size, shape andarrangement of the three digits, but also inpossessing what seem to be anterolaterallydirected scrape-marks at the tips of digits 3 and 4(see de Lapparent and Montenat 1967, fig. 16, butnote the different identification of digits).Nevertheless Anatopus certainly differs fromWintonopus in that the digits are relativelynarrow and show distinct outlines of phalangealpads. Moreover, de Lapparent and Montenatidentified traces of very extensive interdigitalwebbing in the type specimen of Anatopus. Asecond type of footprint, Saltopoides igalensis,was attributed to a theropod dinosaur but is, onceagain, rather similar to Wintonopus (see deLapparent and Montenat 1967, Fig. 15).

TABLE 1: A COMPARISON OF SIZE AMONGFOOTPRINTS ATTRIBUTED TO ORNITHOPODDINOSAURS.

mean index offootprint size

(cm)68.5 `Ornithopoda', Jurassic of Brazil

(Leonardi 1980)61.4 Amblydactylus gethingi, L Cretaceous of

Canada (Sternberg (1932)51.3 Iguanodon, L Cretaceous of Portugal

(Antunes 1976)46.3 Irenesauripus acutus, L Cretaceous of

Canada (Sternberg 1932)28.6 Gypsichnites pacensis, L Cretaceous of

Canada (Sternberg 1932)24.7 `Ornithopod morphotypes',

Jurassic/Cretaceous of Mexico(Ferrusquia-Villafranca et al. 1978)

23.5 Satapliasaurus cf S. dsocenidzei, MJurassic of England (Sarjeant 1970)

18.5 Satapliasaurus dsocenidzei, L Cretaceousof Georgia, USSR (Gabouniya 1951)

18.3 Amblydactylus kortmeyeri, L Cretaceousof Canada (Currie and Sarjeant 1979)

17.6 Iguanodon, U Jurassic of England (Delairand Brown 1974)

17.5 ?cf Satapliasaurus, M Jurassic of England(Sarjeant 1970)

15.7 Irenichnites gracilis, L Cretaceous ofCanada (Sternberg 1932)

15.6 Sauropus barrattii15.0 Anomoepus crassus^Triassic of10.7 Anomoepus isodactylus^Connecticut

(Lull 1953)9.2 Anomoepus intermedius9.0 Anatopus palmatus, Rhaeto-Liassic of

France (de Lapparent and Montenat 1967)8.7 Anomoepus curvatus, Triassic of

Connecticut (Lull 1953)7.7 Anomoepus scambus, Triassic of

Connecticut (Lull 1953)7.2 Wintonopus latomorum6.2 Anomoepus gracillimus, Triassic of

Connecticut (Lull 1953)5.4 Anomoepus minimus, Triassic of

Connecticut (Lull 1953)

Saltopoides has a footprint size index about 13.4cm, but the footprints differ from those ofWintonopus in being distinctly longer than wide(FW/FL ratio about 0.75). In addition the lateraland medial margins of the footprints aredivergent (rather than parallel as in Wintonopus),and there is no very marked positive rotation ofthe footprints. Saltopoides also differs in showing

faint indications of phalangeal pads, but in twoother respects it is very like Win tonopus — inhaving high values for pace angulation (almost180°) and for the ratio PL/FL (11.1). Insummary, Wintonopus is similar to bothAnatopus and Saltopoides in some features, butin neither case is there an exact correspondence infootprint morphology.

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^427

COELUROSAUR TRACKWAYSIchnogenus Skartopus ichnogen. nov.

Type and only ichnospecies S. australis ichnosp.nov.

HOLOTYPE: single left footprint, preserved asnatural mould; QM F10330 (Pl. 7, Figs. B, C).REFERRED MATERIAL: QM F10321 (rockslab with footprints and trackways as naturalmoulds); QM F12265 (single right footprint, asnatural cast; Pl. 1, Figs. C, D); QM F10322(fibreglass replicas of footprints and trackwayspreserved as natural moulds; PI. 10, Figs B, D;Pl. 12; Pl. 13, Figs A, B; Pl. 14; Pl. 15, Fig. A; Pl.16).LOCALITIES: Lark Quarry (QM F10330, QMF10321, QM F10322); Seymour Quarry (QMF12265). See Fig. 1 for location of quarries.HORIZON: interbedded sandstones andmudstones about the middle of the WintonFormation;^early^Upper^Cretaceous(Cenomanian).ETYMOLOGY: Ichnogenus name derived fromGreek skartes (nimble) and pous (foot);ichnospecies name refers to southern (Australian)provenance.DIAGNOSIS (ichnogenus and ichnospecies):trackway of small digitigrade biped, with paceangulation about 150°. Footprint size index (SI)between 2.9 and 5.7 cm. No imprints of hand ortail. Footprints tridactyl (digits 2, 3 and 4),slightly longer than broad (ratio FW/FL about0.95) showing distinct positive rotation. Digitimprints narrow, straight and sharply pointed,without indications of phalangeal pads. Digit 3longest; digits 2 and 4 about equal in length, andalmost equally divergent from digit 3 (bothinterdigital angles between 25° and 30°). Imprintof digit 4 extends slightly farther back thanimprint of digit 2, but does not form a posteriorsalient or 'spur'. Traces of small interdigital webssometimes present. Posterior margin of footprintis an oblique line (posterolateral to anteromedial),either straight or arched forwards. In someexamples there is an imprint of the metapodium:this is sub-rectangular in outline and roughlyequivalent in length to digit 3. Ratio PL/FLusually between 5.5 and 8.5, rarely as low as 5.2or as high as 9.1; ratio SL/FL usually between11.0 and 16.0, rarely as low as 10.6 or as high as17.3.

DESCRIPTION: The holotype is a well-definedfootprint, impressed in the substrate withminimal disturbance. In general the footprintsidentified as Skartopus show much less variation

in shape than do those referred to Wintonopus.Once again, however, all the variations that doexist may be interpreted as circumstantialmodifications of the footprint pattern shown bythe holotype. Variation in footprint shape isdescribed first; thereafter we describe variation insize, proportions and spacing of the footprints.

All footprints referred to Skart opus aretridactyl, with clear traces of digits 2, 3 and 4. Thefootprints sometimes reach a depth of 2 cm, ormore, but none of them shows any certain traceof the hallux (digit 1). If the hallux was present inthe track-maker's foot it must have beenrelatively short and without a major supportiverole; presumably it extended no farther than theline of the metatarso-phalangeal contacts in digits2, 3 and 4 (see Figs 6D-F). Where the three digitsare deeply impressed, as in the holotype, it may beseen that their rear ends do not fall on a straightline. Digits 2 and 4 extend slightly farther backthan digit 3, to form a curve (convex forwards)that presumably reflects the natural arch formedby the distal ends of the metatarsals.

Skartopus footprints are almost bilaterallysymmetrical. It is sometimes difficult to identifyisolated examples as left or right, but thefollowing features are often useful guides: thespace between digits 2 and 3 is slightly greaterthan that between 3 and 4; digit 2 is slightly moredivergent than digit 4; scratch-marks, oftenpresent at the tips of the digit imprints, extendsforwards and laterally. All these features are wellshown in the holotype (Pl. 7, Figs B, C). Wherethe footprints can be connected into sections oftrackway they are easily identified as left andright on account of their positive rotation (see Pl.14, Fig. B).

The imprints of the digits are straight, andrelatively long and narrow (by comparison withthose of Wintonopus). In all cases the three digitimprints are about equally broad. Digits 2 and 4are roughly equal in length, and digit 3 is longer.In nearly all examples the tips of the digit imprintsare quite sharply pointed. The interdigital anglesare small, and in some footprints digits 3 and 4are sub-parallel (e.g. the holotype). There are nodefinite indications of phalangeal pads in any ofthe footprints. The holotype shows traces of smallinterdigital webs, as do several other footprints inthe referred material (e.g. Pl. 12, Fig. D).Skartopus material from Seymour Quarrycomprises natural casts with a finely wrinkledironstone surface; it is not certain if this wrinklingis a representation of original skin texture (see Pl.1 and p. 422).

428^ MEMOIRS OF THE QUEENSLAND MUSEUM

Variation in footprint shape is less marked inSkartopus than in Wintonopus. Divarication ofthe digits is slightly more pronounced in somefootprints than in others, but in all cases the digitsform a near-symmetrical pattern. In mostSkartopus footprints the digits terminate in sharpscratch-marks; apparently similar marks havebeen illustrated (though not described as such) incoelurosaur footprints from the mid-Cretaceousof Israel (Avnimelech 1966, pl. 7, fig. 2).Variation in the shape of Skartopus footprints ismost easily explained by reference to eventsduring the track-maker's stride cycle (Fig. 12). Atthe start of this cycle the forwardly-extended footwould have been planted on the sediment, andthere would have been a very shallow initialfootprint, or no footprint at all (Stage 1 in Fig.12). At mid-stride the track-maker's centre ofgravity passed forwards above the foot, which insome cases sank into the substrate (Stage 2A inFig. 12). Later, as the rear part of the foot startedto lift clear of the sediment, the claws presseddown and slightly backwards to produce sharpoutlines to the tips of the digital imprints (Stage3A in Fig. 12). At this point the toes sometimesstarted to slip backwards, their claws incisinggrooves in the floor of the footprint (Stage 4A inFig. 12). This sequence of events produced sharp-toed tridactyl footprints such as the holotype (Pl.7, Figs. B, C), some of which are secondarilydeepened by backwardly-directed scratch-marks(e.g. Pl. 12, Figs. C, D; Pl. 13, Fig. A). However,in many other cases the foot did not sink into thesubstrate at mid-stride (Stage 2B in Fig. 12). Tojudge from the number of discontinuities (or'missing' footprints) in Skartopus trackways thisseems to have been a very frequent occurrence.There are at least two obvious reasons why thefeet of Skartopus track-makers did not alwaysleave recognizable footprints. First, the track-makers seem to have been remarkably small, andpresumably light, dinosaurs (see Fig. 15). Second,it appears that coelurosaurs have bigger feet(relative to hip height) than many other bipedaldinosaurs (see later discussion concerning sizes oftrack-makers). It seems, then, as if the Skartopustrack-makers may have been lightweightdinosaurs with large spreading feet that acted asanalogues of snow-shoes. Even if the entire footdid not sink into the substrate the tips of the toessometimes left imprints as the track-maker'kicked off' at the end of its stride (Stage 3B inFig. 12). The toes then slithered back through themud to leave a series of curved parallel scratches(Stages 4B and 5B in Fig. 12). In some cases only

one or two of the toes left such traces (e.g. Pl. 16,Fig. A).

A few Skartopus footprints are noteworthy inthat they appear to include an imprint of themetapodium (e.g. Pl. 12, Figs A, B; Pl. 13, Fig.B). In these examples the imprint of themetapodium is a large sub-rectangular depressionbehind the three digit imprints. The metapodiumimprint is no wider than the maximum spread ofthe digits, and it is roughly as long as the imprintof digit 3; it is widest at the rear, where it isbroadly rounded in outline (convex backwards).Footprints with such traces of the metapodiumare uncommon, and most of them occur singlyand at random in the Skartopus trackways.However, one short section of trackway (asequence of 3 paces) is composed entirely of suchfootprints (PI. 14, Fig. B).

Measurements of footprints, paces and strideswere taken from parts of 34 trackways ofSkart opus australis (see 'Methods' for descriptionof the measurements). All 34 trackways are on theLark Quarry bedding plane, and they comprise atotal of 191 footprints (representing 157 pacesand 123 strides). This sample provides thefollowing mean figures for measurements offootprints, paces and strides:

Overall meansMean FL: 4.46 ± 0.70 cm (CV 16%; N 131)Mean FW: 4.10 ± 0.56 cm (CV 14%; N 158)Mean PL: 32.1 ± 4.0 cm (CV 12%; N 151)Mean SL: 61.7 ± 7.8 cm (CV 13%; N 122)

The coefficients of variation are considerablylower than those for equivalent measurements inWintonopus — in consequence of the muchsmaller size range in Skartopus. The index offootprint size, the pace angulation, and standardratios were calculated from the basicmeasurements listed above (see `Methods'); theyhave the following means:

Overall meansMean SI: 4.29 ± 0.52 cm (CV 12%; N 126)Mean ANG: 152°38' ± 11 044 (CV 8%; N112)

Mean ratio FW/FL: 0.94 ± 0.16 (CV 17%; N126)*Mean ratio PL/FL: 7.27 ± 1.33 (CV 18%; N97)*Mean ratio SL/FL: 14.03 ± 2.26 (CV 16%; N85)(* FL is mean for two footprints defining eachpace or stride)

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^429

Pace angulation appears to show relatively littlevariation, and is probably a good diagnosticcharacter. In general terms pace angulation isabout 10 0 greater in Wintonopus than inSkartopus - so that trackways of the latter tendto be slightly broader and to have a more obviouszig-zag arrangement of the footprints. In additionthe footprints of Skartopus are commonly longerthan broad whereas the reverse is true inWin tonopus.

There is generally a poor correlation betweenany two measurements of size in the Skartopusmaterial; for example:

product momentcorrelation coefficients

untransformed log transformedvariables^data^dataFL : FW (N 131)

^0.36^

0.34*FL : PL (N 70)

^0.04^

0.16*FL : SL (N 64)

^0.07^

0.07*FW : SL (N 88)

^0.23^

0.20*SI : SL (N 57)

^0.15^

0.10(*mean for two footprints defining each pace or

stride)

The correlations are not improved bytransformation of the data. These poorcorrelations may be attributed, once again, to thelimited size range of the footprints and trackways(see further discussion below). The correlationbetween footprint size index (SI) and footprintproportions (ratio FW/FL) is also poor (-0.19,with untransformed data; N 126), as is thatbetween footprint length and pace angulation(-0.18, with untransformed data; N 61). Thesignificance of these poor correlations will beexamined later (p. 430).

A somewhat similar pattern of size distributionand correlations emerges if the data are groupedand mean figures are taken for each of the 34Skartopus trackways (see Figs 14 and 15). Meansderived from the grouped data may besummarized as follows:

Means per track wayMean FL: 4.46 ± 0.50 cm (CV 11 07o; N 34)Mean FW: 4.14 ± 0.49 cm (CV 12 07o; N 34)Mean PL: 32.0 ± 2.6 cm (CV 8 07o; N 34)Mean SL: 61.8 ± 4.8 cm (CV 8 07o; N 34)Mean SI: 4.28 ± 0.43 cm (CV 10 07o;, N 34).Mean ANG: 153°37' ± 9°35' (CV 6 07o; N 34)

Mean ratio FW/FL: 0.94 ± 0.12 (CV 13 07o; N34)Mean ratio PL/FL: 7.19 ± 1.08 (CV 15 07o; N29)Mean ratio SL/FL: 13.73 ± 1.73 (CV 13 07o; N32)

Analysis of variance (below) reveals that there isas much, or more, variation within trackways asthere is among trackways:

variation^variationwithin^among

track ways^track ways

^

variable^(To)^(07o)^FL ^57.9^42.1

^

FW^46.2^53.8

^

PL^79.9^20.1

^

SL^50.6^49.4

^

SI^52.1^47.9

^

ratio FW/FL^74.5^25.5^ANG^49.9^50.1

However, all these characters show little overallvariation in the first place (see coefficients ofvariation) - so that most, if not all, of them maystill be regarded as of diagnostic value.STATUS AND AFFINITIES. The footprintsdesignated Skartopus australis do not vary a greatdeal in size or in shape. They consistently showthe following distinctive features: a near-symmetrical arrangement of three long, relativelynarrow and sharply pointed digits, a forwardlyarched rear margin (except where there is animprint of the metapodium), and a length thatequals or exceeds footprint width. Where thefootprints can be connected into trackways theyare found to be disposed with distinct positiverotation; the trackways are moderately broad,with pace angulation about 150°. By comparisonthe trackways of Wintonopus appear to benarrower, with pace angulation about 160°.Variation in shape of the Skartopus footprintsappears to be circumstantial - the occasionalappearance, in otherwise normal trackways, offootprints represented only by scratches or offootprints including a trace of the metapodium.The scratch-like footprints were probably formedwhen the track-maker's foot slipped backwardsacross the surface of the muddy substrate (Stages2B to 5B in Fig. 12); footprints with a trace of themetapodium were presumably formed when thetrack-maker inadvertently came down 'flat-footed', or perhaps when the foot sank deeply inthe mud. (Note, however, that one short sectionof trackway consists entirely of footprints withtraces of the metapodium (Pl. 14, Fig. B). Thissequence of footprints could be fortuitous, or itcould derive from any of several factors - e.g. apathological condition of the track-maker, or anaccumulation of mud on the animal's feet.)

There is limited variation in size of theSkartopus footprints, paces and strides. Thebiggest footprint is less than twice the size of the

430^ MEMOIRS OF THE QUEENSLAND MUSEUM

smallest (in terms of SI); by contrast the biggestexample of Wintonopus is nearly 12 times the sizeof the smallest. The coefficients of variationindicate that dimensions of footprints, paces andstrides vary much less in Skartopus than they doin Win tonopus — yet the correlation between anytwo of these dimensions is in most cases verymuch poorer in Skartopus (compare Figs 10 and15). These poor correlations do not necessarilyindicate that the Skartopus material is aheterogeneous assortment of footprints andtrackways: rather, they reflect the very limitedrange in size. For comparative purposes the entiresample of Skartopus tracks might be regarded asequivalent to a small size class selected from theWintonopus sample. A very similar relationshipbetween two ichnotaxa has been well illustratedby de Lapparent and Montenat (1967, fig. 8). Inother words the Skartopus footprints andtrackways are all roughly similar in theirdimensions, so that there is as much (or more)variation within a trackway as there is betweenone trackway and the others (see analysis ofvariance). In addition it must be noted that theSkartopus footprints and trackways are, on thewhole, much smaller than the Wintonopusfootprints and trackways — yet both have beenmeasured within the same limits of error. Smallmeasurement errors would be of negligibleimportance in the large Wintonopus tracks, butthey would certainly tend to blur correlations inthe absolutely smaller Skartopus tracks. Overallthe Skartopus footprints are quite consistent in

size, shape, and their spacing within trackways;for these reasons it seems justifiable to assemblethem in a single ichnospecies. The range ofvariation seen in this assemblage is no greaterthan that in several other ichnospecies attributedto coelurosaurs (e.g. Grallator olonensis and G.variabilis, de Lapparent and Montenat 1967).

The Skartopus footprints were very probablymade by coelurosaurs — small representatives ofthe dinosaur suborder Theropoda. Skeletalremains and footprints of theropods have beenrecorded from every continent except Antarctica;theropod body fossils have not yet been reportedfrom Queensland, though their footprints are wellknown in the state (for references see p. 420). Intheir size and general appearance the Skartopusfootprints are comparable with those attributedelsewhere to coelurosaurs (see review byHaubold, 1971). The examples listed in Table 2will illustrate the basic agreement in size. All theseexamples (including Skartopus) share thefollowing similarities: the imprints of digits 2, 3and 4 are rather narrow and quite sharplypointed; the digits diverge (usually) at low angles,and often have a near-symmetrical arrangement.However, Skartopus differs from nearly all othertrackways attributed to coelurosaurs in havingexceptionally high values for the ratios PL/FLand SL/FL (Fig. 16). Skartopus is also distinctivein its footprint morphology. It differs fromColumbosauripus ichnospp. in the lesserdivarication of the digits; in addition the digital

TABLE 2: A COMPARISON OF SIZE AMONG FOOTPRINTSATTRIBUTED TO SMALL THEROPODDINOSAURS.

mean index offootprint size

(cm)

24.3 Anchisauripus minusculus, Triassic ofConnecticut (Lull 1953)

14.1 Grallator formosus, Triassic ofConnecticut (Lull 1953)

12.2 Columbosauripus (2 ichnospp.),Cretaceous of Canada and Algeria (bothichnospp. illustrated by Haubold 1971,q.v.)

10.2 Coelurosaurichnus (5 ichnospp.), Triassicof Europe (all ichnospp. illustrated byHaubold 1971, q.v.)

8.1 Grallator cf G. variabilis, Triassic ofAlgeria (Bassoullet 1971)

7.5 Anchisauripus hitchcocki, Triassic ofConnecticut (Lull 1953)

5.8 Otouphepus magnificus, Triassic ofConnecticut (Lull 1953)

4.3 Skartopus australis

3.8 Plesiornis pitulatus, Triassic of Connecticut(Lull 1953)

3.7 Wildeichnus navesi, Jurassic of Argentina(Casamiquela 1964)

3.4 Grallator gracilis, Triassic of Connecticut(Lull 1953)

2.5 Stenonyx lateralis, Triassic of Connecticut(Lull 1953)

0.4 Coelurosaurichnus ichnosp., Triassic ofEngland (Wills and Sarjeant 1970)

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^431

imprints of Columbosauripus tend to be broaderthan those of Skartopus, and digit 4 is oftenconsiderably longer than digit 2 (see Haubold1971, fig. 47). Coelurosaurichnus ichnospp. aredistinguished from Skartopus by very distinctimprints of claws and phalangeal pads, by thecommon preponderance of digit 4 over digit 2,and (in many cases) by having digit 3 noticeablybroader than the flanking digits. In a fewinstances, however, Coelurosaurichnus doesresemble Skartopus in having the rear margin ofthe footprint arched to the front. SimilarlyGrallator ichnospp. differ from Skartopus inpossessing well developed phalangeal pads, tracesof acuminate claws and, very often, a pronounced'spur' formed by the backwards extension of digit4 (see Lull 1953). Stenonyx may be distinguishedfrom Skartopus by the same features and, in someinstances, by the presence of a hallux imprint.Wildeichnus appears to differ in possessing aprominent 'spur' behind digit 3 and in showing aclear imprint of the hallux. To summarize,Skartopus is similar in basic morphology to othercoelurosaur footprints, but it may bedistinguished from these through differences indivarication and relative lengths of the digits,through lacking imprints of phalangeal pads or ofacuminate claws, through the absence of aposterior 'spur', through the absence of a halluximprint, through the relative straightness of thedigits, and through its high values for the ratiosPL/FL and SL/FL.

DISCUSSIONSIZES AND SPEEDS OF TRACK-MAKERS

From studies of locomotion in living terrestrialvertebrates Alexander (1976) determined thefollowing relationship between stride length (A, inmetres), height at the hip (h, also in metres) andspeed (u, in metres per second):(1) A/h L:a 2.3 (u'/gh )"In this equation A represents our measurement forstride length (SL), and g is a constant — theacceleration of free fall; the ratio A/ h is termed'relative stride length' (Alexander 1976).Alexander indicated that this relationship seemsto hold true, at least in general terms, for largeand small animals, both bipeds and quadrupeds,at gaits from slow walk to fast run. In addition heobserved that the relationship does not seem to beseriously affected by variation in the consistencyof the substrate. With additional data from fast-moving African ungulates Alexander, Langman

and Jayes (1977) refined expression (1) to give:(2) A/h 1.8(u2/gh )0 "These authors concluded that expression (2 ) isbest applied to animals that are cantering orgalloping, whereas expression (I ) is appropriatefor animals using slower gaits. In mammals thechange from a walking gait to a trotting gaitoccurs when A/h is approximately 2.0 (Alexander1976); the change from trotting to gallopingfollows when A/ h has increased to about 2.9. Thislatter figure is derived from two generalizationspresented by Alexander (1977). The first of theseis that mammals tend to shift from a trotting orracking gait to a galloping or cantering gait whenthe quantity i4 reaches a value of about 1.5. Thequantity i, or 'dimensionless speed', was definedby Alexander as follows:(3) = u(ghrwhere h is expressed in metres and u is in m/s.The second generalization, which seems to applyto a wide variety of animals through a wide rangeof speeds, is that(4) A/h 2.3 06

where A/h represents mean relative stride length.From these generalizations it may be assumedthat the shift from trotting to galloping occurswhen

(5) A/h a_ 2.3 (1.5)"X/ h -a 2.9

Consequently it is possible to identify the gaits ofdinosaurian track-makers on the basis of relativestride length, as follows:walk: A/h < 2.0; locomotor performanceequivalent to walking in mammals.trot: A/ h = 2.0 to 2.9; locomotor performanceequivalent to trotting or racking in mammals.run: X/ h > 2.9; locomotor performanceequivalent to cantering, galloping or sprinting inmammals.

To estimate the speeds of certain dinosaursAlexander (1976) transformed expression (1) togive:(6) u 0.25g "A'"h -'''This equation was then applied to data fromdinosaur trackways, where A could be measureddirectly (Alexander's A = SL of our descriptions)and where h might be estimated from the size ofthe footprints. This method has since beenapplied to more than 50 dinosaur trackways,including a sample of those at Lark Quarry(Russell and Beland 1976; Tucker and Burchette1977; Coombs 1978; Thulborn and Wade 1979;

432^ MEMOIRS OF THE QUEENSLAND MUSEUM

Farlow 1981; Kool 1981; Thulborn 1981, 1982).Table 3 presents a summary of the speeds so farestimated in this way.

Figures listed in Table 3 for the Lark Quarrydinosaurs are preliminary estimates, and they willbe revised in the present work. Our discussion willfocus on the problem of estimating h on the basisof footprint dimensions. It is desirable that hshould be estimated with reasonable care, becausean underestimate will generate an overestimate ofthe track-maker's speed; conversely anoverestimate of h will generate an underestimateof speed. In some instances the hindlimb length ofa dinosaurian track-maker has been estimatedfrom the evidence of pace length or stride length(e.g. Avnimelech 1966); an estimate of this type isof questionable value, simply because pace lengthand stride length vary according to the gait andspeed of the track-maker (Lull 1953, p. 146). Inother cases hindlimb length has been estimated on

the basis of footprint dimensions; for example,Avnimelech (1966, p. 5) suggested that in thefootprints of bipedal dinosaurs the length of digit3 represented about 18% of hindlimb length.Elsewhere Alexander suggested (1976) that hcould be calculated as approximately four timesfootprint length for a variety of dinosauriantrack-makers, both bipeds and quadrupeds, andthis suggestion has been rather widely accepted(see all sources cited in Table 3). However,Coombs (1978) expressed some reservationsabout this generalization, and Alexander (1976)did mention that footprint length could representanything between 0.23h and 0.28h in the bipedaldinosaurs that he examined. In discussing thesizes and speeds of the Lark Quarry track-makerswe will investigate some other methods forestimating h.

In comparing the sizes, weights and speeds ofvarious dinosaurs Coombs (1978, Table 2) drew a

TABLE 3: SUMMARY OF SIZES AND SPEEDS PREVIOUSLY ESTIMATED FOR DINOSAURIAN TRACK-MAKERS.

ichnotaxa or track-makers

Bipedal dinosaurs

(N)

(6)(m)

0.6-2.1

(m/s)

1.2-3.6

(km/h)

4.3-13.0

X/h

1.2-2.5

gaitwalk (4)trot (2)

source

Alexander 1976

Sauropods (2) 1.5-3.0 1.0-1.1 3.6-4.0 0.8-1.1 walk

Ornithomimid (1) 1.2 1.8 6.4 1.5e walk Russell and Wand 1976aGiant ornithopod (1) 3.4 7.5 27.1 2.7e trotaGiant ornithopod (1) 3.4 2.4 8.5 1.3 walk Thulborn 1981

?Anchisauripus (2) 0.4-0.6e 4.5-7.9 1.3-1.4e walk Tucker and Burchette1977

Carnosaur (1) 2.6 2.3 8.2 1.4 walk Thulborn and Wade 1979Ornithopods (10) <•1.0 4.3m 15.5m >2.0 trot/run (Lark Quarry)Coelurosaurs (10) <1.0 3.6m 13.0m >2.0

Gypsichnites pacensis (1) 1.2 2.0 7.2x 1.8 walkIrenesauripus spp. (3 ) 1.5-2.1 1.4-2.7 5.0-9.7x 1.2-1.6 walkIrenichnites gracilis ( 1 ) 0.6 2.8 10.1 x 2.3 trot Kool 1981Amblydactylus kortmeyeri (1) 0.5 1.1 4.0 1.5 walkTetrapodosaurus borealis (1) 1.4 0.9 3.2x 0.9 walkfTheropods (3) 1.2-1.5 e 8.3-11.9 29.9-42.8 3.7-4.9 run

sTheropods (3) 1.8-2.5 6.4-8.9 1.5-1.8 walk Farlow 1981Theropods (15) 1.5 me 4.2m 15.2m 2.3m trot

a: two interpretations of single trackway.f: three fastest of Farlow's 15 track-makers.s: three slowest of Farlow's 15 track-makers.e: figures estimated from published data.m: mean.x: Coombs (1978) provides different speed estimates,

apparently through computational error (Farlow 1981).

(12.6)10.6-11.810.5-11.9

(11.6)

(11.3)(11.2)

(4.0)3.5-3.83.3-3.6

(3.5)

(3.6)(3.4)

12.1-14.0 (13.5) 3.7-4.1 (4.0)10.6-11.3 (11.0) 3.3-3.4 (3.3)13.7-15.8 (14.7) 3.9-4.4 (4.2)11.2-13.8 (12.4) 3.5-4.1 (3.8)13.6-15.2 (14.4) 3.9-4.2 (4.1)9.4-11.5 (10.5) 2.7-3.2 (3.0)

14.5-16.1 (15.5) 3.7-4.0 (3.8)15.8-17.6 (16.7) 4.0-4.4 (4.2)16.1-18.0 (17.0) 3.9-4.2 (4.0)13.2-15.3 (14.0) 3.7-4.1 (3.9)

^(14.6)^ (3.5)

^

17.3-18.5 (17.9)^3.9-4.1 (4.0)

^

15.3-17.8 (16.8)^3.8-4.3 (4.1)

^

14.7-19.3 (16.5)^3.6-4.4 (3.9)

^

(22.7)^ (5.0)

^

15.6-15.8 (15.6)^(3.9)

^

13.7-15.6 (14.6)^3.3-3.7 (3.5)

^

12.5-12.7 (12.6)^(3.0)

^

14.6-17.8 (16.4)^3.3-3.9 (3.7)

^

12.7-16.9 (15.2)^3.2-4.0 (3.7)

^

(14.8)^ (3.4)

^

13.3-14.1 (13.7)^3.3-3.4 (3.3)

^

13.7-18.0 (15.5)^3.2-4.0 (3.5)(15.5) (3.7)

15.6-16.2 (15.8) 3.6-3.7 (3.6)14.0-18.0 (16.2) 3.1-3.8 (3.5)11.6-12.3 (11.9) 3.0-3.1 (3.1)13.3-16.5 (14.9) 3.2-3.7 (3.4)

(14.0)^ (3.3)13.8-19.8 (17.7)^3.5-4.6 (4.2)

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^433

distinction between 'height at hips' (as had beenestimated by Colbert 1962) and 'standard [orskeletal] hindlimb length' (the sum of the lengthsof femur, tibia and metatarsal 3). He indicatedthat there was sometimes a considerabledifference between these two dimensions -particularly among large dinosaurs. In addition arather similar distinction has been made betweenskeletal hindlimb length (h ) and 'height of thehindlimb' (H) - the latter being defined as 'the

combined lengths of femur, tibia and longestmetatarsal, plus an increment of 9 07o to accountfor ankle bones and for soft tissues at knee, ankleand sole' (Thulborn, 1982, p. 228). For presentpurposes these various dimensions are assumed tobe roughly equivalent, on the grounds that theyare likely to be of great significance only in largedinosaurs. Our estimates of h for the Lark Quarrytrack-makers (Tables 4 and 5) are based onosteometric data and may be regarded as

TABLE 4: ESTIMATES OF SIZE, SPEED AND RELATIVE STRIDE LENGTH FOR 57 W1NTONOPUS TRACK-MAKERS.

estimated h^ estimated speed(cm)^(na/s)^(km/h)^estimated X/h

^

10.0-12.5 (11.2)^3.6-4.2 (3.9)

^

11.9-14.3 (12.9)^3.8-4.3 (4.0)1. 13.7 2.8-3.5 (3.1)2. 16.8 3.3-4.0 (3.6)3. 16.3 (3.5)4. 16.8 3.0-3.3 (3.1)5. 18.9 2.9-3.3 (3.1)6. 18.6 (3.2)7. 18.4 3.4-3.9 (3.7)8. 19.7 3.0-3.1 (3.0)9. 20.1 3.8-4.4 (4.1)

10. 18.4 3.1-3.8 (3.4)11. 20.4 3.8-4.2 (4.0)12. 23.7 2.6-3.2 (2.9)13. 27.1 4.0-4.5 (4.3)14. 25.5 4.4-4.9 (4.6)15. 29.1 4.5-5.0 (4.7)16. 22.0 3.7-4.3 (3.9)17. 29.6 (4.1)18. 32.6 4.8-5.2 (5.0)19. 27.4 4.3-5.0 (4.7)20. 29.7 4.1-5.4 (4.6)21. 29.2 (6.3)22. 26.4 4.3-4.4 (4.3)23. 30.7 3.8-4.3 (4.1)24. 33.2 (3.5)25. 34.7 4.1-4.9 (4.6)26. 30.1 3.5-4.7 (4.2)27. 33.6 (4.1)28. 31.0 3.7-3.9 (3.8)29. 34.2 3.8-5.0 (4.3)30. 29.5 (4.3)31. 32.4 4.3-4.5 (4.4)32. 38.8 3.9-5.0 (4.5)33. 29.2 3.2-3.4 (3.3)34. 33.4 3.7-4.6 (4.1)35. 33.7 (3.9)36. 28.1 3.8-5.5 (4.9)

434

37.38.39.

45.644.637.1

MEMOIRS OF THE QUEENSLAND MUSEUM

5.8-6.3 (6.1)^20.9-22.7 (21.9)(5.7)^ (20.4)(4.2)^ (15.2)

4.0-4.2 (4.1)(3.9)(3.4)

40. 38.1 4.2-4.8 (4.5) 15.0-17.3 (16.2) 3.3-3.7 (3.5)41. 35.8 3.2-4.9 (4.0) 11.4-17.7 (14.4) 2.7-3.8 (3.3)42. 39.6 3.4-4.5 (3.9) 12.3-16.0 (14.0) 2.8-3.4 (3.1)43. 49.3 5.6-6.5 (6.0) 20.0-23.2 (21.6) 3.7-4.2 (3.9)44. 41.8 2.5-3.3 (3.1) 9.2-12.0 (11.0) 2.2-2.7 (2.5)

*45. 47.4 (1.1) (4.0) (1.5)46. 49.4 (7.8) (28.2) (4.9)47. 58.7 (4.0) (14.4) (2.7)48. 61.5 6.0-7.6 (6.8) 21.8-27.3 (24.4) 3.6-4.3 (4.0)49. 53.6 6.1-6.6 (6.4) 22.1-23.7 (22.9) 3.9-4.1 (4.0)50. 54.9 5.0-7.0 (5.7) 18.0-25.0 (20.4) 3.3-4.2 (3.6)51. 52.0 5.4-5.7 (5.5) 19.4-20.5 (19.9) 3.6-3.7 (3.6)52. 60.3 6.1-6.5 (6.3) 22.1-23.4 (22.8) 3.7-3.9 (3.8)53. 53.9 8.2-8.3 (8.2) 29.4-29.9 (29.7) (4.9)54. 61.9 (5.1) (18.5) (3.2)55. 54.7 5.0-6.3 (5.7) 18.2-22.6 (20.7) 3.3-3.9 (3.7)56. 70.0 5.4-7.0 (6.3) 19.6-25.1 (22.6) 3.2-3.9 (3.6)57. 158.4 4.6-5.0 (4.8) 16.7-18.1 (17.2) 2.1-2.2 (2.1)

Means 4.3-4.8 (4.6) 15.4-17.4 (16.4) 3.5-3.9 (3.7)

*New Quarry trackway.For each track-maker we show the range and the mean (in parentheses) of speed and relative stride length. A singlefigure (in parentheses) indicates that only one stride could be measured, or that there was little or no variation instride length. Figures for the New Quarry track-maker (No. 45) were excluded from calculations of overall means.

equivalent to skeletal hindlimb length. If theseestimates were increased by 9 07o (to provideestimates of H, or 'height of the hindlimb') themean increment for the ornithopod track-makerswould be 3.1 cm; for the coelurosaur track-makers the mean increment would be 1.5 cm.These small increases in estimated body sizewould not affect the general conclusions that wedraw regarding the speeds and gaits of the track-makers.

CARNOSAUR TRACKWAYIn the trackway of the Lark Quarry carnosaur

footprint length ranges from 41 cm (estimated) to64 cm; mean footprint length is 51.4 cm. With theassumptions used by Alexander (1976) h could beestimated to lie in the range 1.64 to 2.56 cm -with the mean estimate at 2.06 m. However, itseems legitimate to base our estimate of h on thebest-preserved and most complete footprint; thisparticular footprint (number 3 in the trackway) is64 cm long, providing estimated h of 2.56 m. Thefootprint has a well-defined rear margin, and itslength is not exaggerated by scrape-marks; otherfootprints in the carnosaur trackway appear tohave less complete impressions of the rear part of

the foot, and they would probably generateunderestimates of h (and, in consequence,overestimates of the carnosaur's speed).

There are at least two other ways to estimate hfor the Lark Quarry carnosaur. First it is possibleto compare the sizes of carnosaur footprints tothe sizes of carnosaur skeletons. Footprintsattributed to tyrannosaurs are reported to reach amaximum length of about 80 cm (Haubold 1971),while the largest well-known tyrannosaur,Tyrannosaurus rex, has a skeletal hip heightabout 3.17 m (representing the sum of the lengthsof femur, tibia and longest metatarsal). With theadmittedly untestable assumption that the largestknown footprints were made by animals aboutthe size of Tyrannosaurus, skeletal hip heightcould be predicted as 3.96 times footprint length.In the case of the Lark Quarry carnosaur thismethod would indicate a skeletal hip height ofabout 2.54 m.

Next it is possible to make use of the fact thatmetatarsus length (MT) is strongly correlated withskeletal hip height in carnosaurs - see Fig. 17,where the least squares regression line representsthe following allometric equation:(7) h 4.15MT + 28.52 cm

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^435

(In this equation all measurements are expressedin cm; Bartlett's three-group method (see Sokaland Rohlf 1969, p. 483) yields a virtually identicalequation). To apply equation (7) in the case ofthe Lark Quarry carnosaur it is first necessary toestimate MT on the basis of footprint size. Itshould be possible to make such an estimate witha fair degree of accuracy because in many bipedaldinosaurs MT is roughly equivalent to thesummed lengths of phalanges in digit 3 (EP; seeFigs 6 and 18, and various illustrations given byCoombs 1978); moreover, in a digitigrade animal

such as a dinosaur total footprint length (FL)should be slightly greater than EP (Fig. 18). Incarnosaurs MT seems to be a little greater than EP(see Russell 1970, Table 1, for data on Canadiancarnosaurs), but would probably have been lessthan FL (which comprised EP, claw sheath, jointcapsules and, possibly, a 'heel' region supportedby the distal part of the metatarsus). EvidentlyMT would have been somewhat less than totalfootprint length. For practical purposes we willassume that MT is roughly equivalent to footprintsize index (SI) - which is also less than FL

TABLE 5: ESTIMATES OF SIZE, SPEED AND RELATIVE STRIDE LENGTH FOR 34 SKARTOPUS TRACK-MAKERS.

estimated h^ estimated speed(cm)^(m/s)^(km/h)^estimated X/h

1. 13.3 3.0-4.2 (3.7) 10.8-15.1 (13.3) 3.8-5.0 (4.5)2. 13.3 3.8-4.4 (4.2) 13.6-15.9 (15.0) 4.6-5.2 (4.9)3. 14.4 2.8-3.3 (3.1) 10.2-12.0 (11.2) 3.6-4.0 (3.8)4. 14.5 3.2-3.9 (3.5) 11.4-14.0 (12.6) 3.9-4.5 (4.2)5. 14.8 3.5-4.1 (3.8) 12.5-14.7 (13.6) 4.1-4.7 (4.4)6. 14.9 2.8-3.4 (3.0) 10.2-12.2 (10.8) 3.5-4.0 (3.7)7. 15.0 2.9-3.3 (3.1) 10.6-11.9 (11.2) 3.6-3.9 (3.8)8. 15.1 3.4-3.6 (3.5) 12.1-13.0 (12.5) 4.0-4.2 (4.1)9. 15.2 3.1-3.8 (3.4) 11.3-13.8 (12.3) 3.8-4.4 (4.0)

10. 15.3 2.8-3.5 (3.0) 10.1-12.5 (10.9) 3.4-4.1 (3.7)11. 15.5 3.3-3.4 (3.3) 11.8-12.3 (12.1) 3.9-4.0 (3.9)

*12. 15.5 3.4-3.6 (3.4) 12.1-12.8 (12.4) 3.9-4.1 (4.0)13. 15.5 3.6-4.1 (3.9) 12.8-14.9 (14.0) 4.1-4.6 (4.4)14. 15.6 3.4-3.8 (3.6) 12.2-13.5 (13.0) 4.0-4.3 (4.2)15. 16.2 3.6-3.9 (3.7) 12.9-13.9 (13.3) 4.1-4.3 (4.2)16. 16.3 2.8-3.2 (3.0) 10.2-11.4 (10.6) 3.4-3.7 (3.5)17. 16.4 3.1-3.3 (3.2) 11.3-11.8 (11.6) 3.7-3.8 (3.7)18. 16.6 3.3-3.5 (3.4) 11.9-12.4 (12.1) 3.8-3.9 (3.8)19. 16.6 (3.2) (11.4) (3.7)20. 17.2 2.5-3.7 (3.1) 8.8-13.3 (11.3) 3.0-4.1 (3.6)21. 17.7 2.8-3.9 (3.3) 9.9-13.9 (12.0) 3.2-3.8 (3.7)22. 17.9 3.1-3.9 (3.3) 11.0-14.1 (12.0) 3.5-4.2 (3.7)23. 17.9 2.5-3.1 (2.8) 9.0-11.3 (10.1) 3.0-3.5 (3.2)24. 18.2 2.3-3.2 (2.6) 8.3-11.6 (9.5) 2.8-3.6 (3.1)25. 18.2 2.2-2.8 (2.5) 8.0-10.2 (9.1) 2.7-3.2 (3.0)26. 18.3 (3.2) (11.7) (3.6)27. 18.5 2.6-2.8 (2.7) 9.4-10.1 (9.7) 3.0-3.2 (3.1)28. 19.3 2.9-3.5 (3.2) 10.4-12.6 (11.4) 3.2-3.7 (3.5)29. 19.3 2.5-2.6 (2.5) 8.9-9.3 (9.1) 2.9-3.0 (2.9)30. 19.6 2.7-2.9 (2.8) 9.8-10.5 (10.1) 3.1-3.2 (3.1)31. 19.9 3.2-3.7 (3.5) 11.7-13.2 (12.6) 3.5-3.8 (3.7)32. 20.1 2.6-3.0 (2.8) 9.2-10.7 (10.1) 2.9-3.2 (3.1)33. 21.6 2.5-3.1 (2.9) 9.1-11.3 (10.3) 2.8-3.3 (3.1)34. 21.9 2.7-3.4 (3.0) 9.6-12.2 (10.6) 2.9-3.5 (3.1)

Means 3.0-3.5 (3.2) 10.7-12.5 (11.6) 3.5-3.9 (3.7)

For each track-maker we show the range and the mean (in parentheses) of speed and relative stride length. A singlefigure (in parentheses) indicates that only one stride could be measured, or that there was little or no variation instride length.* Trackway No. 12 was made by an animal with consistent 'flat-footed' gait (see Plate 14, fig. B). The length of eachfootprint is exaggerated by an imprint of the metatarsus, and to estimate the animal's size and speed ourmeasurements of total footprint length were reduced by 5007o.

436^ MEMOIRS OF THE QUEENSLAND MUSEUM

because the footprint is longer than wide. Thebest-preserved carnosaur footprint at LarkQuarry has SI of 57.69 cm; by substituting thisfigure for MT in equation (7 ) we can estimateskeletal hip height to have been about 2.68 m.

These various estimates are in close agreement,and they indicate that the Lark Quarry carnosaurwas between 2.54 and 2.68 metres in height at thehip. The mean figure, which we will use forestimating the animal's speed, is 2.59 m.Apparently the animal was about the same size asone specimen of the Canadian carnosaurAlbert osaurus libratus (National Museum ofCanada, No. 2120, with h about 2.63 m; Lambe1917); it would have been intermediate in sizebetween specimens of Daspletosaurus torosus (h2.40 m, estimated from data of Russell 1970) andTyrannosaurus rex (h about 3.17 m; Osborn1917).

The strides of the Lark Quarry carnosaur rangein length from 2.82 m to 3.74 m (mean 3.31 m).Consequently relative stride length (A/h) isestimated to range from 1.09 to 1.44 (mean 1.28).In every stride A/h is well below 2.0, whichindicates that the animal was using a walking gait(Alexander 1976) and that its speed is mostappropriately estimated with equation (6 ). Thecarnosaur's progress may be plotted in somedetail, as follows:

stride speed(m/s) (km/h)

A/h

1 2.34 8.43 1.442 2.13 7.65 1.363 2.35 8.47 1.444 2.34 8.43 1.445 1.76 6.32 1.216 1.67 6.03 1.187 1.74 6.26 1.208 1.54 5.54 1.129 1.47 5.28 1.09

means 1.93 6.93 1.28

The animal took a slightly weaving course (Fig.3), during which it showed a definite tendency todecelerate (i.e. to shorten its strides). Its first fourstrides are relatively long and are defined by verydeep footprints; the animal then switched quiteabruptly to a series of shorter strides (Nos. 5-9,Fig. 22A) and its footprints became noticeablyshallower. The last two strides are the shortestand were taken as the animal made a sharp turn toits right.

ORNITHOPOD TRACKWAYSIn our preliminary account of Lark Quarry

(1979) the makers of the ornithopod trackways

(Wintonopus ) were estimated to range from thesize of bantams to the size of ostriches; their meanspeed was calculated to be about 4.31 m/s (15.52km/h). These preliminary estimates of size andspeed were derived from a small sample (parts of10 trackways) with the methods described byAlexander (1976).

In the 57 trackways studied here meanfootprint length ranges from 2.40 cm to 22.75 cm(mean 6.68 cm); by excluding the earlier-formedtrackway of the exceptionally large animal (No.57 in Table 4) the range of means for footprintlength is reduced to 2.40 - 10.86 cm (with overallmean 6.39 cm). Alexander assumed (1976) thatheight at the hip (h) could be estimated asapproximately four times footprint length for avariety of dinosaurian track-makers; if we applythis assumption to the Wintonopus data thetrack-makers are estimated to have had h rangingbetween 9.60 and 43.44 cm (mean 25.58 cm), withthe single large individual at 91.0 cm. However,Alexander indicated that footprint length (FL)could represent anything between 0.23h and 0.28hin the bipedal dinosaurs that he studied (1976),and it seems worthwhile to investigate analternative method for estimating h.

Analysis of variance (p. 424) revealedconsiderable variation in footprint length withinthe Wintonopus trackways. Within a singletrackway there may occur foreshortened ('stubby-toed') prints, normal prints and attenuated toescratches from a single foot (Figs 5L and 7, Pl.10D). Consequently mean footprint length for atrackway may not be a satisfactory indicator ofthe track-maker's size - because the mean will beaffected by the relative frequencies offoreshortened and attenuated footprints.Fortunately footprint size index (SI) seems to be areliable guide to the relative sizes of two or moretrack-makers: the index differs from onetrackway to the next, but is virtually constantwithin any one trackway. Footprint size index isusually a little greater than footprint length inWin tonopus (because the footprints are usuallybroader than long), and it has the followingrange: 2.89 to 12.69 cm (mean 6.74 cm), with thesingle large individual at 26.59 cm. Twoobservations make it possible to estimate h on thebasis of SI: first, in the foot skeletons of manyornithopods the summed lengths of phalanges indigit 3 (EP) is roughly equal to the length ofmetatarsal 3 (MT; see Figs 6A, C); and, second,there exists a strong correlation between MT andskeletal hip height in ornithopods (see Fig. 19).So, to calculate h for the Wintonopus track-makers there seems to be only one prerequisite -

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^437

an estimate of MT (or of EP) derived from SI. Bydinosaurian standards the Wintonopus track-makers seem to have been fairly small animals, sothat absolute differences between FL and EP (andhence MT) are likely to have been small. For thesake of convenience we will assume that SI isroughly equal to MT. It may be noted that SI isusually a little greater than FL, because theWintonopus footprints are normally a littlebroader than long. Consequently our estimates ofh (based on SI) will tend to be slightly greater thansimilar estimates based on FL. This differencewill, on the whole, have a conservative effect — inthe sense that the sizes of track-makers mayoverestimated, and their speeds may beunderestimated.

Figure 19A shows the relationship between MTand skeletal hip height in a sample of 32ornithopod skeletons (with MT ranging from 6.3to 38.1 cm); the least squares regression linerepresents the following allometric equation:(8) h 3.76MTH 6

where both h and MT are expressed incentimetres. By substituting SI for MT in thisequation we estimate that the Lark Quarryornithopods ranged in skeletal hip height from12.86 to 71.57 cm (mean 34.41 cm), with thesolitary large individual at 168.81 cm. However,the regression equation for cursorial ornithopods(with femur shorter than tibia) is slightly differentfrom that for graviportal ornithopods (withfemur longer than tibia); for cursorialornithopods the equation is:(9) h 3.97MT'while for graviportal ornithopods it is:(10)^h^5.06MT'The two regression lines, which are shown in Fig.19B, have similar slopes but different intercepts.In general terms equation (10) provides estimatesof h about 20 to 25 07o greater than those derivedwith equation (9). The Lark Quarry ornithopodsseem to have been small animals, judging fromthe size of their footprints,and this might indicatethat equation (9 ), based on data from cursorialornithopods, should be used to estimate h. Itmust be admitted, however, that the exact identityof the Wintonopus track-makers is unknown —beyond that fact that they seem to have beenornithopods of some sort. For this reason we haveobtained three estimates of h for each track-maker (by substituting SI for MT in each of thepreceding equations), and we will use the meanfigure in calculating the speed of each animal.These mean values for h range from 13.70 to

69.98 cm (overall mean 34.82 cm), excluding thesingle large animal with h estimated to be 158.59cm.

Of the 57 Wintonopus trackways examinedhere only one appears to have been made by ananimal more than 1 metre high at the hip. Forty-six of the track-makers (81 07o) are estimated tohave h less than 50 cm, and 24 of them (42%) hless than 30 cm. In all cases Alexander'sassumptions (1976) would provide smallerestimates of h.

In terms of general body size the Lark Quarryanimals may be compared with cursorialornithopods of the families Fabrosauridae,Hypsilophodontidae and Heterodontosauridae(see skeletal reconstructions or flesh restorationsgiven by Galton 1974, Thulborn 1972, Santa Lucaet al. 1974); they might also be compared withjuvenile specimens of some bigger graviportalornithopods (e.g. the juvenile hadrosaursdescribed by Horner and Makela 1979) and withthe juvenile psittacosaurs recently described byCoombs (1980a, 1982). From the evidence oftrackways it does not seem possible to decide withcertainty whether the Lark Quarry ornithopodswere small cursorial forms ranging up to adultstatus, or whether they were juveniles of somebigger graviportal ornithopod. Neither of thesepossibilities can be dismissed entirely: two typesof small ornithopod, resemblinghypsilophodontids, are reported from theCretaceous of Victoria (Flannery and Rich 1981),and a large ornithopod with h about 2.44 m wasrecently described from the Lower Cretaceous ofQueensland (Muttaburrasaurus, Bartholomai andMolnar 1981). In either case it would still seemreasonable to regard the smallest (at least) of theLark Quarry track-makers as juvenile animals. Ithas sometimes been supposed that juveniledinosaurs were rare (Richmond 1965, Leonardi1981), but Horner and Makela (1979) found thatmore than 80 07o of dinosaur specimens collectedfrom the Two Medicine Formation (UpperCretaceous) of Montana could be identified asjuveniles or subadults. It seems that juveniledinosaurs may have been quite common, at leastin some localities.

The curve illustrating size frequencydistribution for the Wintonopus track-makers isdistinctly skewed and is similar to the type ofcurve derived by Boucot (1953) from 'lifeassemblages' of fossils (see Fig. 21A). The curvemight be interpreted in any of several ways. Firstit could simply be regarded as a survivorshipcurve for a population of small cursorialornithopods. According to this interpretation the

438^ MEMOIRS OF THE QUEENSLAND MUSEUM

animals would normally have grown to achieve hof about 30 cm (peak of curve); thereafter themortality rate would have reached its maximum(steepest part of curve), with fewer and feweranimals surviving to reach greater and greater size(by virtue of the indeterminate growth prevailingamong reptiles). This interpretation assumes, ofcourse, that the sample of 57 track-makers is trulyrepresentative of the dinosaur population fromwhich it was drawn; it also assumes the existenceand constancy of some strong correlation betweensize and age. Secondly it is possible to interpretFig. 21A as a survivorship curve based on asample drawn from a population of biggraviportal ornithopods (with assumptions asabove). In this case it would appear that there wasa high rate of mortality among juveniles orsubadults once these had attained h of about 40cm. The adults, presumably with h of 1.5 metresor more, would then have been comparativelyrare and relatively long-lived. Next there is thepossibility that our sample of 57 track-makers isnot a representative one: it could compriseanimals from two or more species, or there couldbe serious under-representation of bigger animalsin a single species. (It seems scarcely possible thatsmaller animals could be under-represented).There is no obvious reason to assume that theWintonopus trackways were produced by two ormore different types of dinosaur: none of thefrequency distributions is strongly bimodal (Figs8 and 9), there are impressive correlationsbetween any two dimensions of the footprints andtrackways (Fig. 10), and all the footprints can beinterpreted as those of animals sharing onedistinctive pattern of foot structure (Fig. 6C) andusing the same gait (Fig. 11). We cannot discountentirely the possibility that our sample of 57Wintonopus trackways might be heterogeneous(in which case we could deduce nothing about thepopulation structure and possible affinities of thetrack-makers), but this possibility does seemrather unlikely: it supposes the co-existence oftwo or more dinosaurian species, each of which isrepresented by juveniles or is characteristically ofsmall size, and each of which producedWintonopus -like trackways. The final possibilityis that bigger animals did exist, but that they areunder-represented in our sample. Under-representation of bigger animals could not beregarded as an effect of sampling: trackways oflarge animals are not common at Lark Quarry,and we ensured that our sample contained datafrom the largest and second-largest examples ofWintonopus. Consequently this final possibilitymust imply that large and small animals were

segregated, either fortuitously or through somedeliberate strategy on the part of the potentialtrack-makers. Of all these possible interpretationsthe simplest would certainly seem to be the first —that the sample of 57 trackways is representativeof a dinosaur population in which animals rarelygrew to hip heights estimated at greater than 70cm. Finally there is some evidence of three sizeclasses in the Wintonopus sample (note the threeclumps of data points in Fig. 10). In terms ofestimated hip height these three size classes haveapproximate limits of 13 to 20 cm, 25 to 40 cm,and 50 to 60 cm. There is no way to investigate thepossibility that these size groupings might beequivalent to age classes.

From our estimates of h it is calculated that themean value of A/h per trackway ranges from 2.69to 5.03 (overall mean 3.69); these figures excludevalues for the solitary large animal (A/h 2.10),and for the New Quarry track-maker. It seemsthat all the Wintonopus track-makers at LarkQuarry were using a gait faster than a walk (A/hgreater than 2.0). Two individuals wereapparently moving at a trot or a slow run (withmean A/h at 2.1 and 2.7), whereas all the others(96%) were using a fast running gait equivalent toa mammalian gallop (with A/h at 3.0 or greater).

The 56 ornithopod track-makers at LarkQuarry all have A/h estimated to be greater than2.0, so that their speeds are most appropriatelyestimated by means of equation (2 ) re-written as:(11) u [gh(A/1.8h "] O'In this equation A and h are expressed in metresand u is in metres per second. Mean speed pertrackway is estimated to range from 2.92 m/s(10.52 km/h) to 8.24 m/s (29.66 km/h), with theoverall mean for 55 animals (excluding solitarylarge individual) at 4.48 m/s (16.12 km/h). Theone exceptionally large animal has an estimatedmean speed of 4.78 m/s (17.22 km/h). Theminimum speed estimated for any of these track-makers, on the basis of its single shortest stride, is2.22 m/s (8.00 km/h); the maximum speedestimated for any of the track-makers, on thebasis of its single longest stride, is 8.30 m/s (29.88km/h).

The single Wintonopus trackway at NewQuarry has mean SI of 8.91 cm; for this track-maker h is estimated to have been 47.4 cm,indicating A/ h of about 1.54. The New Quarryanimal was not associated with those at LarkQuarry, and it seems to have been using adifferent gait (walking rather than running). Itsspeed, which is best calculated with equation (6),is estimated to have been 1.11 m/s (3.98 km/h).

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^439

Even if its speed is estimated, inappropriately,with equation (11 ) the New Quarry animal wouldstill seem to have been moving more slowly thanany of the Lark Quarry track-makers (i.e. at 1.76m/s, or 6.34 km/h).

The estimated sizes, speeds and relative stridelengths of all 57 Wintonopus track-makers aresummarized in Table 4. The animals seem to havemaintained fairly constant speeds, and in mostcases mean stride length for the second half of thetrackway differs only slightly from mean stridelength for the first half (see Fig. 23A). On averagethe difference between these two means in anincrease of 1.3%. An analysis of stride lengths inthe longest section of ornithopod trackway (No.41 in Table 4, comprising 17 strides) reveals noconsistent trend towards acceleration(lengthening of strides) or deceleration(shortening of strides). Mean stride length for thesecond half of this trackway is 6.8 07o less thanmean stride length for the first half. Shortsections of this trackway (Fig. 22B) might give theimpression of a strong tendency to lengthen orshorten the strides, yet the track-maker seems,overall, to have maintained a reasonablyconsistent stride length (116.82 ± 11.13 cm; CV9.5 07o).

To summarize, the Wintonopus trackways atLark Quarry seem to have been made by smallornithopod dinosaurs, mostly less than 60 cm inheight at the hip. All these dinosaurs seem to havebeen using a fast running gait, with A/h in mostcases between 3.2 and 4.1. For the majorityspeeds are estimated between 3.4 m/s (12.2 km/h)and 5.6 m/s (20.2 km/h), and there is no clearevidence that the animals were accelerating ordecelerating.

COELUROSAUR TRACKWAYSIn our preliminary study of Lark Quarry (1979)

we estimated that the makers of the coelurosaurtrackways (Skartopus) had a size range equivalentto that between bantams and half-grown emus;their mean speed was estimated to have beenabout 3.62 m/s (13.04 km/h). These preliminaryestimates of size and speed were obtained byapplying Alexander's method (1976) to parts of10 trackways.

With Alexander's working assumption (1976)that footprint length (FL) represents about 0.25hit may be estimated that the 34 trackways studiedhere were made by animals ranging from 14.5 to22.5 cm in height at the hip (overall mean 17.8cm). But again, as with the ornithopod track-makers, it may be worthwhile to investigateanother method for estimating h.

Analysis of variance (p. 429) reveals that FLis highly variable within and among the Skartopustrackways; so, too, are footprint width (FW) andfootprint size index (SI). Consequently thereseems to be little advantage in selecting SI, ratherthan FL or FW, as an indicator to the relativesizes of the track-makers. None of these variablesappears to be a very reliable guide to the relativesizes of the track-makers, but this fact is notparticularly important because all these animalsseem to have been much the same size anyway.Our estimates of size and speed for the track-makers will be based on mean FL per trackway. Itmight not be appropriate to base these estimateson mean SI (as was done for the ornithopodtrackways) because this is usually less than actualfootprint length — on account of the footprintsusually being longer than broad. This preferencefor FL, rather than SI, will have a conservativeeffect — in that estimates of h will tend to beincreased and estimates of speed will tend to bedecreased. In the 34 trackways studied here meanFL ranges from 3.62 to 5.62 cm, with the overallmean at 4.46 ± 0.50 cm.

In a variety of coelurosaurs MT (the length ofmetatarsal 3) is strongly correlated with skeletalhip height; this relationship illustrated in Fig. 20,where the least squares regression line representsthe following allometric equation:(12) h 3.06MT'Both MT and h are expressed in centimetres. Toestimate h for the Lark Quarry coelurosaurs wewill substitute FL for MT in this equation, on theassumption that these two dimensions wereroughly equal. This assumption is reasonablebecause both dimensions were probably a littlegreater than EP (summed lengths of phalanges indigit 3). The slight preponderance of FL over EPis apparent from Fig. 18; from variousillustrations of coelurosaur foot skeletons itappears that MT is also a little greater than EP —in a ratio about 11:10 (see Figs 6D, E, andreferences given in caption to Fig. 20).

By substituting FL for MT in equation (12) weestimate that the Lark Quarry coelurosaursranged from 13.27 cm to 21.93 cm in height at thehip (with mean for the 34 animals at 16.92 ± 2.23cm). These estimates are slightly smaller thanthose obtained with Alexander's assumption thath is approximately four times FL; for the smallesttrack-maker the two estimates differ by 12 mm,and for the largest they differ by 6 mm. Thesedifferences are unlikely to be of greatsignificance, especially since Alexander's work(1976) indicates that h could represent anything

440^ MEMOIRS OF THE QUEENSLAND MUSEUM

from about 3.6FL to about 4.3FL in the variousbipedal dinosaurs that he studied. Moreover thesedifferences will not affect the general conclusionsthat we draw regarding the gaits and speeds of thetrack-makers (see Table 6, where our estimatesfor h are increased by about 40 to 50%).

From these estimates is seems that theSkartopus track-makers were somewhat smallerthan the familiar coelurosaurs Coelophysis (hfrom 33.7 to 55.9 cm in 13 specimens listed byColbert 1964), Ornitholestes (h about 48.3 cm,Osborn 1917) and Podokesaurus (h 25.5 cm,Talbot 1911). The only well known coelurosaur ofcomparable size would seem to beCompsognathus, with h as little as 21.1 cm (forholotype, estimated from measurements given byOstrom 1978). However, the Skartopus footprintsdo agree in size with many other footprintsattributed to coelurosaurs (see Table 2).

The size frequency distribution for theSkart opus track-makers is distinctly skewed (Fig.21B) and is open to the several interpretationsthat were considered earlier for the Wintonopustrack-makers. The possible interpretations maybe summarized as follows:

1. The 34 Skartopus track-makers represent acoelurosaur population in which individualanimals grew to a maximum hip height estimatedat 22 cm. This interpretation assumes that thesample of 34 track-makers is truly representativeof the population from which it was drawn (i.e.that it includes both juveniles and adults), andthat there existed a strong unvarying correlationbetween size and age.

2. The 34 track-makers are juveniles of sometype of theropod dinosaur that grew to greatersize — though the bigger individuals are notrepresented at Lark Quarry. This implies thatlarge and small animals were segregated, either bychance or through their behaviour.

3. The sample of 34 trackways isheterogeneous. This assumes that two or moretypes of small (or juvenile) dinosaurs madeidentical trackways at a single site, and that theydid so at approximately the same time.

All these possible interpretations involveuntestable assumptions, but the first of themwould seem to be the simplest. In any case it isnoteworthy that the Skartopus track-makers musthave been remarkably small animals bydinosaurian standards. Even if it is assumed thatthe track-makers were exceptionally long-leggedanimals resembling ornithomimids (or 'ostrichdinosaurs') it may be estimated that the largest of

them was less than 32 cm in height at the hip (seeTable 6 and further discussion below).

With our estimates of h the mean value of A/hper Skartopus trackway is found to range from2.90 to 4.94, with the overall mean at 3.71. Inonly three trackways is minimum A/h (based onthe shortest stride) estimated to be less than 2.9,and in no case does it fall below 2.7. Evidently all34 track-makers were using a fast running gait.These figures for A/h indicate that the speeds ofthe track-makers are most appropriatelyestimated by means of equation (11). Mean speedper trackway is estimated to range from 2.53 m/s(9.09 km/h) to 4.16 m/s (14.98 km/h), with theoverall mean at 3.22 m/s (11.58 km/h). Theminimum speed estimated for any of the track-makers, on the basis of its single shortest stride, is2.23 m/s (8.03 km/h); the maximum speedestimated for any of the track-makers, on thebasis of the single longest stride, is 4.42 m/s(15.91 km/h).

Table 5 presents a summary of estimated size,speed and relative stride length for each of the 34track-makers. There is no clear indication that theanimals were either accelerating or decelerating.In most cases mean stride length for the secondhalf of a trackway differs only slightly from meanstride length for the first half (Fig. 23B). Onaverage this difference between the two means is adecrease of about 2.1%. From Fig. 23B it appearsthat the majority of track-makers showed a slightreduction in stride length during their progress.However, this tendency to shorten the strides isneither consistent nor well-marked, and it isprobably of little significance. By way ofillustration Fig. 22C shows an analysis of thelongest section of coelurosaur trackway (No. 20in Table 5, comprising 22 strides): in the latterhalf of this trackway mean stride length is 2.5%less than mean stride length in the first half, butthere is no consistent trend towards progressiveshortening of the strides. Short sections of thetrackway could give the impression of a strongtrend to shortening or lengthening the strides, yetthe track-maker seems, overall, to havemaintained a fairly consistent stride length (61.77± 5.40 cm; CV 8.7%).

In summary, the Skartopus trackways seem tohave been produced by small coelurosaurs, all ofwhich are estimated to have been less than 22 cmhigh at the hip. All of these animals seem to havebeen using a fast running gait; estimates of A/hare in most instances greater than 2.9, though afew trackways include short strides indicatingoccasional lapses of A/h as low as 2.7. Mean

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^441

TABLE 6: ESTIMATES OF SIZE, SPEED AND RELATIVE STRIDE LENGTH FOR 34 SKARTOPUS TRACK-MAKERS - WITHTHE ASSUMPTION THAT THESE WERE ANIMALS RESEMBLING ORNITHOMIMIDS. RANGES AND MEANS OFESTIMATES SHOWN AS IN TABLE 5.

h (cm)^speed (m/s)^speed (km/h)^X/h1. 19.5 2.2-3.1 (2.8) 8.0-11.1 (9.9) 2.6-3.4 (3.1)2. 19.6 2.8-3.3 (3.1) 10.1-11.8 (11.1) 3.1-3.5 (3.4)3. 20.9 2.1-2.5 (2.3) 7.6-8.9 (8.4) 2.4-2.8 (2.6)4. 21.1 2.4-2.9 (2.6) 8.5-10.4 (9.4) 2.7-3.1 (2.9)5. 21.5 2.6-3.0 (2.8) 9.4-10.9 (10.2) 2.8-3.2 (3.0)6. 21.6 2.1-2.5 (2.3) 7.6-9.1 (8.1) 2.4-2.8 (2.5)7. 21.7 2.2-2.5 (2.3) 7.9-8.9 (8.3) 2.5-2.7 (2.6)8. 21.9 2.5-2.7 (2.6) 9.0-9.7 (9.4) 2.7-2.9 (2.8)9. 22.0 2.3-2.9 (2.6) 8.4-10.3 (9.2) 2.6-3.0 (2.8)

10. 22.1 2.1-2.6 (2.3) 7.5-9.3 (8.2) 2.4-2.8 (2.5)11. 22.4 2.5-2.6 (2.5) 8.9-9.2 (9.1) 2.7-2.8 (2.7)

*12. 22.4 2.5-2.8 (2.6) 9.0-9.6 (9.3) 2.7-2.9 (2.8)13. 22.4 2.7-3.1 (2.9) 9.6-11.2 (10.5) 2.9-3.2 (3.1)14. 22.6 2.6-2.8 (2.7) 9.2-10.2 (9.8) 2.8-3.0 (2.9)15. 23.2 2.7-2.9 (2.8) 9.7-10.5 (10.0) 2.8-3.0 (2.9)16. 23.4 2.1-2.4 (2.2) 7.7-8.5 (8.0) 2.4-2.6 (2.4)17. 23.5 2.4-2.5 (2.4) 8.5-8.9 (8.7) (2.6)18. 23.8 2.5-2.6 (2.5) 9.0-9.4 (9.1) (2.7)19. 23.8 (2.4) (8.6) (2.6)20. 24.6 1.9-2.8 (2.4) 6.7-10.0 (8.5) 2.1-2.8 (2.5)21. 25.3 2.1-2.9 (2.5) 7.5-10.6 (9.1) 2.3-2.9 (2.6)22. 25.4 2.3-3.0 (2.5) 8.4-10.7 (9.1) 2.4-2.9 (2.6)23. 25.4 1.9-2.4 (2.1) 6.8-8.5 (7.7) 2.1-2.5 (2.3)24. 25.8 1.7-2.4 (2.0) 6.3-8.8 (7.2) 1.9-2.5 (2.2)25. 25.9 1.7-2.1 (1.9) 6.1-7.7 (6.9) 1.9-2.3 (2.1)26. 26.0 (2.5) (8.9) (2.5)27. 26.2 2.0-2.1 (2.0) 7.2-7.7 (7.4) 2.1-2.3 (2.2)28. 27.2 2.2-2.7 (2.4) 7.9-9.6 (8.7) 2.3-2.6 (2.5)29. 27.3 1.9-2.0 (1.9) 6.8-7.1 (6.9) 2.0-2.1 (2.0)30. 27.6 2.1-2.2 (2.2) 7.5-8.0 (7.8) 2.2-2.3 (2.2)31. 27.9 2.5-2.8 (2.7) 8.9-10.1 (9.7) 2.5-2.7 (2.6)32. 28.2 2.0-2.3 (2.1) 7.1-8.2 (7.7) 2.1-2.3 (2.2)33. 30.1 1.9-2.4 (2.2) 7.0-8.7 (8.0) 2.0-2.4 (2.2)34. 30.5 2.1-2.6 (2.3) 7.4-9.4 (8.2) 2.1-2.5 (2.2)

Means 2.3-2.6 (2.4) 8.1-9.5 (8.7) 2.4-2.7 (2.5)

estimated speeds of the track-makers aregenerally in the range 2.82 m/s to 3.61 m/s (10.16to 13.00 km/h), and there is no clear indicationthat the animals were either accelerating ordecelerating.

The preceding estimates of size and speeddepend on the assumption that the Lark Quarrytrack-makers had hindlimb proportions similar tothose in 'typical' coelurosaurs such asCoelophysis, Compsognathus and Ornitholestes.However, one group of small to medium-sizedtheropod dinosaurs - the ornithomimids or'ostrich dinosaurs' - is characterized by unusualhindlimb proportions: these animals haveexceptionally long hindlimbs terminating inrelatively short toes. For the sake of completeness

we will examine the possibility that Skartopustrackways might have been made by animalsresembling ornithomimids.

From published measurements of orni-thomimid skeletons (Russell 1972 - Orni-thomimus, Struthiomimus, Dromiceiomimus;OsmOlska et al. 1972 - Gallimimus) it appearsthat MT represents something between 1.45 and1.72 times EP (the mean figure derived from 9specimens being 1.57). Since FL was probably alittle greater than EP (see Fig. 18) we may assume,for the sake of convenience, that MT wasequivalent to about 1.5FL. From the sameosteometric data it seems that there is a strongpositive correlation between MT and skeletal hipheight (r = 0.997, N = 9); the latter may be

442^ MEMOIRS OF THE QUEENSLAND MUSEUM

predicted by means of the following allometricequation:(13) h 3.49MT' 'where h and MT are expressed in centimetres. Ifthe Lark Quarry coelurosaurs resembledornithomimids in limb proportions their skeletalhip heights might, then, be estimated as follows:(14) h 3.49(1.5FL)'This method provides estimates of h ranging from20.44 to 31.96 cm (with overall mean of 25.27cm). Consequently estimates of mean A/h pertrackway would range from 1.96 to 2.88 (withoverall mean 2.49). In only two of the 34 caseswould the estimate for mean A/h be less than 2.0(actually 1.99 and 1.96). These figures wouldindicate that all the Skartopus track-makers weretrotting or running (or, in two cases, were at leaston the point of breaking into a trot). Estimates ofmean speed per trackway, obtained with equation(11 ), range from 1.85 m/s to 2.97 m/s (6.67km/h to 10.69 km/h), with the overall mean at2.33 m/s (8.40 km/h). These particular estimatesof size, speed and relative stride length aresummarized in Table 6.

It seems difficult to escape the conclusion thatSkart opus trackways are those of runningdinosaurs — even with the assumption that thesemight have been exceptionally long-leggeddinosaurs resembling ornithomimids. There are atleast two good reasons for believing that thetrack-makers were not ornithomimids. First thereis simply the matter of size: the Skartopusfootprints are much smaller than the footskeleton in any specimen of ornithomimiddinosaur so far described. In the smallest of thecomplete foot skeletons listed by OsmOlska et al.(1972, p. 131) EP is 12.8 cm. One smaller exampleis listed by these authors, but it lacks thepenultimate phalanx in digit 3; for this specimenwe estimate EP to have been about 9.0 cm. In theornithomimids described by Russell (1972) EPranges from 21.5 cm to more than 25.5 cm.Footprints attributed to ornithomimids, or tounknown but presumably similar dinosaurs, haveFL from about 10 cm (Hopiichnus shingi, Welles1971) to about 28 cm (Ornithomimipus angustus,Sternberg 1926). By comparison the largest figurefor mean FL in any of the Skart opus trackways is5.62 cm. In the second place there seem to be nocertainly-identified skeletal remains ofornithomimids from the Gondwana continents. Afew bones (dorsal vertebrae and a phalanx) fromthe Lameta Formation of India were described byvon Huene and Matley (1933) asOrnithomimoides mobilis and 0. barasimlensis,

but OsmOlska et al. regarded the material as'systematically insufficient' (1972, p. 104).Russell (1972) considered the two species to benomina vana. Molnar (1980), examining theseand other records, concluded that there was noevidence suggesting that ornithomimids existedon the southern continents. In summary, theSkartopus footprints seem rather too small to bethose of ornithomimids, and there is littleevidence to suggest that such dinosaurs existed inthe southern continents. Consequently we mayassume the Lark Quarry trackways to have beenmade by some other type of theropod dinosaur —which presumably had 'typical' limb proportions.

A DINOSAUR STAMPEDE?The most distinctive features of the Lark

Quarry trackway site are: (1) that the dinosauriantrack-makers were very numerous; (2) that nearlyall these track-makers seem to have been small bydinosaurian standards; (3) that trackways of thetwo main types ( Wintonopus and Skartopus ) arein some places coincident, superimposed orinterwoven; (4) that all the trackways (except thatof the carnosaur) head in a single direction; and(5) that all the track-makers (except thecarnosaur) seem to have been running. Thiscombination of features appears to be unique,and it previously led us to interpret the LarkQuarry trackways as the result of a dinosaurianstampede (Thulborn and Wade 1979). Theevidence underlying this interpretation may nowbe examined in more detail.

The Lark Quarry bedding plane carries one ofthe densest accumulations of dinosaur footprintsyet reported (see Table 7). In our preliminarydescription of the site we estimated that thesefootprints represented the trackways of at least130 dinosaurs (excluding the carnosaur), withcoelurosaurs ( = Skartopus trackways)outnumbering ornithopods ( = Wintonopustrackways) in a ratio about 55:45. It provedimpossible to assign every footprint at LarkQuarry to a particular trackway, and so ourestimate for the number of track-makers must bechecked in some other way. In a 1 metre widetransect at a right angle to the direction of thetrackways (between points X-X in Fig. 3) wecounted a total of 350 footprints. In practically allcases mean pace length for a track-maker(whether ornithopod or coelurosaur) is less than 1metre — which means that nearly every animalcrossing the line of the transect should have left atleast one footprint in the metre-wide strip.Consequently the maximum number of animalsthat crossed the transect could be estimated at

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^443

350. However, mean pace length for Wintonopusis 68 cm (i.e. about 11/2 footprints per metre), andfor Skartopus it is 32 cm (i.e. about 3 footprintsper metre). By assuming that ornithopods andcoelurosaurs were present in equal numbers wecan estimate the number of animals to havecrossed the transect as:

350= 156 animals

(3 x 0.5) + (1.5 x 0.5)

This very rough estimate must be regarded as anabsolute minimum. We suspect that coelurosaursoutnumbered ornithopods, but it is impossible tomake allowance for this without identifying everyfootprint and assigning to it a particulartrackway. This proved an impossible task (forreasons described earlier; see p. 418). Moreoverit is certain that many smaller animals (bothcoelurosaurs and ornithopods) crossed ourtransect without leaving recognizable footprints:these animals would have been so light that theirbroad-spreading and rather springy feet simplyfailed to break through the surface of thesediment. This seems to have happened verycommonly, to judge from the number ofdiscontinuities or 'gaps' in the trackways at LarkQuarry. The area of bedding plane exposed atLark Quarry is delimited partly by erosion andpartly by undisturbed overburden (mainly to Sand SW, see PI. 3). Within this area the footprintsare fairly evenly distributed (Pl. 4), and it seemsalmost certain that more of them could berevealed by extending the quarry to the SSW. Ourestimate of 156 animals may well represent afraction of the number of dinosaurs thattraversed Lark Quarry and its environs.

Trackways are certainly abundant, but this factalone does not support the hypothesis of adinosaur stampede. However, this fact doesassume some significance when one considers thatall the trackways may have been formedsimultaneously by animals running in a singledirection.

There are several reasons for believing that allthe Wintonopus and Skart opus trackways at LarkQuarry were formed at, or about, the same time(excepting the single unusually large example ofWintonopus ). First, all these trackways are verysimilar in preservation; there is no evidence (suchas scouring or erosion) to indicate that sometracks are much older than others. Second, all thetrackways are impressed to about the same depth— evidently in sediment of uniform consistency.If the trackways had accumulated over a lengthyperiod one might expect to find evidence of a

change in the consistency of the substrate (i.e.some tracks more deeply impressed than others).Finally there is the evidence of superimposedfootprints; these are quite common, and in allcases the later-formed print is similar in its depthand state of preservation to the earlier-formedone. These similarities are apparent even wherethree or more animals have trodden the same spot(see Pl. 13, Fig. C; Pl. 16, Figs B, C). Ornithopodfootprints ( Wintonopus) may be foundsuperimposed on footprints of other ornithopodsor of coelurosaurs (Skartopus ), and the same istrue for the coelurosaur footprints. Footprints ofboth types may be found superimposed uponthose of the carnosaur. From the evidence ofsuperimposed footprints it may be deduced (a)that the carnosaur traversed the area before some(at least) of the ornithopods and coelurosaurs didso; (b) that some ornithopods preceded somecoelurosaurs; and (c) that some coelurosaurspreceded some ornithopods. But from theevidence as a whole it is possible to reach a moregeneral conclusion: that both ornithopods andcoelurosaurs traversed the Lark Quarry site at (orabout) the same time, and that they did so afterthe passage of the solitary carnosaur. This isexactly the chronological sequence to be expectedif the approach of the carnosaur had triggered astampede of the ornithopods and coelurosaurs.Of course there is no absolute proof that the LarkQuarry trackways were formed in exactly thissequence, but it is difficult to imagine any otherwhen one considers that all the ornithopod andcoelurosaur track-makers were running in a singledirection.

Perhaps the most striking feature of the LarkQuarry site is that all the animals responsible forWintonopus and Skartopus trackways wereheaded in a single direction — about 55 0 E of trueN, and in almost direct opposition to the coursetaken by the solitary carnosaur (see Pl. 4). Noneof these abundant ornithopod and coelurosaurtrackways deviates more than a few degrees froma single compass bearing, and in this respect theLark Quarry site appears to be unique. Dinosaurtracks uncovered at other prolific localities arerandomly oriented (e.g. see de Lapparent andMontenat 1967, Tucker and Burchette 1977) orshow some less obvious tendency to sub-parallelalignment (e.g. see Avnimelech 1966, Ostrom1972). And at sites where sub-parallel trackwaysdo predominate these often represent a `two-waytraffic' — with some trackways diametricallyopposed to others (e.g. Avnimelech 1966, pl.VIII; Ostrom 1972, fig. 4). By contrast the Lark

MEMOIRS OF THE QUEENSLAND MUSEUM

TABLE 7: COMPARISON OF STATISTICS FOR VARIOUS TRACK WAY SITES.

number of number ofsite age area(m2) footprints trackways^source

Ain-Sefra, Algeria: Triassic 1 12 Bassoullet 1971St Laurent de Treves,

France: L Jurassic 25 24 Thaler 1962Bendrick Rock, S Wales: U Triassic 25 400 Tucker and Burchette 1977Rocky Hill, Connecticut: Rhaetic 930 1000+ Ostrom 1972

Swanage, S England: U Jurassic 47e 46 3 Charig and Newman 1962Kerman area, Iran: Jurassic 5 8 ?6e de Lapparent and

Davoudzadeh 1972Tocantins River, Brazil: Jurassic —

Cretaceous 6-8000 47+ 6 Leonardi 1980Beth Zayit, Israel: Cenomanian 400 200+ ?10 Avnimelech 1966F6 Ranch, Texas: Aptian —

Albian 76 15 Farlow 1981Serrote do Letreiro, Brazil: Triassic 'spacious' 17 Leonardi 1979

Mt Tom, Massachusetts: Rhaetic 800 137 28+ Ostrom 1972Lark Quarry, Queensland: Cenomanian 209 3300+ 130+ Thulborn and Wade 1979

( e — estimated figures.)

Quarry trackways are more nearly parallel and(excepting the carnosaur trackway) entirelyunidirectional. The coincidence of so manytrackways certainly implies that some externalfactor controlled the behaviour of the track-makers (cf. Ostrom 1972). At the time thetrackways were formed the Lark Quarry siteappears to have been part of a broad drainagechannel, and it is conceivable that the track-makers might have been funnelled along acommon route by physical barriers such as leveesor steep banks. However, there is no directevidence of such barriers, and it may be recalledthat remnants of a few randomly orientedtrackways do occur at the site. These scatteredand eroded remnants of trackways testify thatsome medium-sized bipedal dinosaurs (probablyornithopods) traversed the area before thecarnosaur made its appearance, and that they didso randomly — that is, seemingly without thecontrol of physical barriers. Apparently thebehaviour of the Wintonopus and Skartopustrack-makers was influenced by some factor thathad not previously affected the movements ofdinosaurs across the same area. Next, it is obviousthat the Lark Quarry site cannot have been partof some established route along which dinosaurswere accustomed to move in either direction. Noris it possible to believe that the dinosaurian track-makers could have adhered with absolute fidelityto a system of 'one-way' routes. One generalconclusion seems inescapable: that the singular

orientation of trackways at Lark Quarry mustreflect some unusual behaviour on the part of thetrack-makers.

There can be little doubt that all theWintonopus and Skartopus trackways at LarkQuarry were made by running animals. In mostcases relative stride length is estimated to havebeen well over 2.9 — apparently indicative of afast running gait equivalent to a mammaliangallop or sprint. In the few remaining casesrelative stride length is estimated to have beenbetween 2.0 and 2.9 — indicative of a gaitequivalent to mammalian trotting. Our estimatesof relative stride length (and hence of speed)depend in turn upon estimates of hindlimb height.The piling of estimate upon estimate may, indeed,have introduced and multiplied some errors, butthese are unlikely to be of very great significancefor our general conclusions. For example, in thecase of the Wintonopus track-makers ourestimates of hindlimb height are consistentlygreater than those that would be obtained bystraightforward application of Alexander'smethod (1976). Consequently relative stridelength (and speed) may, if anything, be under-estimated for these animals. Our estimates ofhindlimb height for the Skartopus track-makersare slightly smaller than those that would beobtained with Alexander's method; but even ifour estimates are increased by as much as 40 or50 07o it still appears that these dinosaurs wouldhave been running (compare Tables 5 and 6). In

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^445

any event, it is not necessary to estimate hindlimbheight in order to demonstrate that theWintonopus and Skartopus track-makers had anexceptionally long-striding gait: this is fullyapparent from simple ratios of SL/FL andPL/FL (see Figs 11 and 16). Once again it appearsthat the Lark Quarry dinosaurs were indulging inunusual behaviour, for reports of trackwaysattributed to running dinosaurs are otherwise veryfew (see Farlow 1981, Thulborn 1982). Here it isworth recalling that the single Wintonopustrackway at another site (New Quarry) is that of awalking animal.

In summary, the Lark Quarry site has revealeda most unusual assemblage of dinosaurtrackways, and to account for their origin it seemslegitimate to postulate some exceptional patternof behaviour on the part of the track-makers. Thefindings of this study seem to confirm, or evenstrengthen, our earlier interpretation of thetrackways as evidence of a dinosaur stampede.Indeed, it is not easy to propose (let alone justify)an alternative interpretation. This difficulty arisesfor two reasons: (1) because all the Wintonopusand Skartopus trackways are unidirectional, and(2) because all these trackways seem to have beenmade by running dinosaurs. It seems very unlikelythat this assemblage of trackways could representa series of events — i.e. the passage, at intervals,of individual animals or of small groups. Anysuch series of events would have involved themost remarkable coincidences: at different timesvarious dinosaurs would have traversed the samearea, but always in exactly the same direction,and always at a run. Consequently one is forcedto conclude that the Lark Quarry trackwaysrepresent a single event — a conclusion supportedby the uniform preservation of the trackways. Wesubmit that a group of more than 150 animalsrunning in one direction must constitute astampede or some similar event. Severalexperienced stockmen have examined the LarkQuarry trackways; all of them agreed that thetrackways could well have resulted from astampede, though on two occasions we wereoffered an alternative explanation (as a joke) —that the track-makers were being 'herded' or'driven'.

There arise some intriguing questions. First,what caused the stampede ? Only one piece offossil evidence seems to hint at a plausible answer— the trackway of the single carnosaur. It is quiteconceivable that a gathering of ornithopods andcoelurosaurs, drinking or foraging round a water-hole, might have been startled by the approach of

a large predatory dinosaur. We have reservationsabout reading too much significance into theevidence of a single dinosaur trackway, but theonly alternative is to admit that the unusualbehaviour of the Wintonopus and Skartopustrack-makers is inexplicable. A second questionarises: why did the ornithopods and coelurosaursrun to the NE if this was the very direction fromwhich the carnosaur had approached them? Herewe can only offer speculations. The ornithopodsand coelurosaurs had reached the water-hole, tothe SW of the present Lark Quarry site, by someunknown and presumably preferred or 'normal'route. One might have expected these animals tohave made their escape by such a route. The factthat some, at least, did not do so seems to implythat their preferred route had been blocked —perhaps by the manoeuvres of the carnosaur,which certainly made a sharp right turn. Inmaking this turn to its right the carnosaur wouldsimultaneously have opened up a new escaperoute — to the NE, and along the broad drainagechannel that extended over the present LarkQuarry site towards Seymour Quarry. Thisreconstruction of events (Fig. 25) accords with allavailable evidence and seems to be fairlyparsimonious. The motives of the carnosaurremain unknown; it may simply have beenapproaching the water-hole to drink, or it mayhave been hunting — perhaps attempting tocorral its prey on the point extending SW into thewater-hole. If the animal were hunting it ispossible to speculate a little further. First, it seemslikely that this large predator would have selectedits prey from among the ornithopods; thecoelurosaurs, insofar as they are known fromtheir trackways, would seem to have been rathersmall game. Next it is conceivable that thecarnosaur may not have been hunting alone: itmight possibly have been assisted by anothercarnosaur (or perhaps even more than one)strategically placed to forestall the escapingornithopods and coelurosaurs. This is no morethan a speculation, but it might help to accountfor the remarkably close grouping of theornithopods and coelurosaurs — especially sincetheir progress over the Lark Quarry site seems notto have been constrained by physical barriers.These suggestions are not inconsistent withprevious speculations on the hunting behaviour oflarge theropod dinosaurs (Farlow 1976).

Whatever the carnosaur's maneouvres orintentions may have been, a number ofWintonopus and Skartopus track-makers did runto the NE, across the present Lark Quarry and

4,46^ MEMOIRS OF THE QUEENSLAND MUSEUM

Seymour Quarry sites. In doing so these animalsseem to have traversed an area that they had nottrodden before (or, at least, not in the immediatepast). Their trackways at Lark Quarry are strictlyunidirectional, and none of them seems to havebeen made by a walking animal. Evidently theseornithopods and coelurosaurs were not followingsome well-trodden path, and they may in facthave been moving across an area that wasformerly unattractive to them. It is not difficult toimagine in what sense the Lark Quarry area mighthave been unattractive to small bipedal dinosaurs:it was covered by a layer of soft mud, into whichthe animals might (and certainly did) sink to adepth of several centimetres. This would havebeen of little consequence to a very large animalsuch as the carnosaur; as this animal traversed thearea its feet plunged right through the mud to reston the firmer sandy sediments below. But it isconceivable that the small coelurosaurs (mean hipheight about 17 cm) and ornithopods (mean hipheight about 35 cm) ran some risk of becomingbogged. If these small dinosaurs were crossing anunattractive or even dangerous area, it isreasonable to suppose that they were doing sounder some quite unusual and compellingcircumstances.

Our reconstruction of the events at and aroundLark Quarry is illustrated in Fig. 25. Thisreconstruction takes into account all thepeculiarities of the Lark Quarry trackways, 4ndwe have found no evidence that conflicts with it.A central assumption of this reconstruction is thatthe behaviour of the carnosaur was responsible,at least in some measure, for the unusualbehaviour of the ornithopods and coelurosaurs.This assumption naturally implies that very littletime would have elapsed between the formationof the carnosaur's trackway and the formation ofthe ornithopod and coelurosaur trackways. Theuniform preservation of all these trackways seemsto indicate that they were formed at about thesame time ... but there is no certain way todiscover if the carnosaur preceded theornithopods and coelurosaurs by a matter ofminutes or by a matter of hours. However there isone suggestive clue, provided by ornithopod andcoelurosaur footprints superimposed on those ofthe carnosaur. In their preservation thesesuperimposed prints are identical to thoseelsewhere, yet they were formed in thin streaksand pockets of mud remaining in the floor of thecarnosaur's prints. The fact that these remnantpatches of mud had not dried out, despite theirthinness, suggests that little time elapsed (or thatdrying was very slow). If the intervening period

were to be estimated in terms of hours, ratherthan minutes, we could no longer maintain oursuggestion that the carnosaur's behaviourprompted the stampede of ornithopods andcoelurosaurs. Even so, it would remain clear thata stampede (or some similar event) did occur —though its cause would be unknown.

The minimum distance travelled by thestampeding ornithopods and coelurosaurs is morethan 95 metres — from the SW end of LarkQuarry to the NE end of Seymour Quarry. Fromthe speed estimates presented earlier it may becalculated that most of the ornithopods andcoelurosaurs covered this distance in less than 30seconds; at their minimum speeds the slowestornithopod and the slowest coelurosaur wouldhave done so in 38 seconds and 45 secondsrespectively. There is no indication of either thestart or the end of the stampede.

Finally, it might be objected that our use of theterm 'stampede' is inappropriate because theanimals involved seem to have been moving atrather low speeds (mean 16 km/h forornithopods, and mean 12 km/h forcoelurosaurs). In a present-day stampede,comprising ungulate mammals, one might expectto find animals moving several times faster thanthe dinosaurs at Lark Quarry. However, suchcomparisons of absolute speeds are of limitedsignificance — simply because the dinosaurs thattraversed Lark Quarry were, on the whole, muchsmaller than living ungulates. To measure andcompare the locomotor performances ofdifferent-sized animals it is necessary to selectcriteria that will reduce or eliminate the effects ofsize-differences. Such criteria will be examinedbelow.

IMPLICATIONS FOR THEUNDERSTANDING OF DINOSAURBIOLOGY

It has sometimes been supposed that juveniledinosaurs were rare (Richmond 1965, Leonardi1981), and that small (but adult) dinosaurs wereequally rare or 'unknown' (Bakker 1972).However, there have recently been reports ofjuvenile dinosaurs, some no bigger than rats orpigeons, and even embryonic dinosaurs (seeBonaparte and Vince 1979, Kitching 1979,Coombs 1980a, 1982, Carpenter 1982, amongothers). The Lark Quarry trackways providestriking confirmation that small dinosaurs —whether adults or juveniles, or both — may havebeen abundant in some localities. This fact shouldcertainly have some bearing on the debate overthermoregulation in dinosaurs, though at this

THULBORN AND WADE: WINTON DINOSAUR TRACKWAYS^447

point we cannot pursue so large and complicateda subject (see Thomas and Olson 1981 for acomprehensive review).

It seems impossible to distinguish with certaintybetween trackways made by small (but adult)dinosaurs and those made by juveniles.Consequently it is fruitless to use ichnologicaldata in speculating about the reproductivestrategies and population dynamics of dinosaurs.These aspects of dinosaur biology should moreproperly be investigated on the basis of bodyfossils — where it might prove possible todetermine relationships between age, sexualmaturity and body size. Nevertheless we haveoutlined some possibilities regarding populationstructure of the Wintonopus and Skartopus track-makers (Fig. 21). It must be emphasized that theseare only possibilities, for we cannot agree entirelywith the assumption that footprints of differentsizes were made by animals of different ages(Leonardi 1981). Such an assumption is valid onlyif growth proceeded at a constant rate through thelives of the track-makers. Moreover theassumption would seem to be particularly suspectwhere there is a limited range in size of footprints(e.g. Skartopus, where the largest footprint is lessthan twice the size of the smallest). Even so, itmay be legitimate to identify very small footprintsas those of juveniles in cases where there is a verygreat range in footprint size (e.g. Wintonopus,where the largest footprint is nearly 12 times thesize of the smallest).

Ostrom (1972) compiled data on dinosaurtrackways in order to examine the possibility thatsome dinosaurs might have been gregarious. Atvarious sites he found trackways grouped in near-parallel arrangement, and he concluded thatseveral different kinds of dinosaurs may well havebeen gregarious in habit. Ostrom's conclusion issupported, incidentally, by the identification ofdinosaurian adaptations for intra-specificcombat, display and vocalization (Galton 1970,Hopson 1975, Molnar 1977, Weishampel 1981).The trackways at Lark Quarry provide strongevidence of gregarious behaviour, for they seemto have resulted (with the exception of thecarnosaur trackway) from the movement of asingle large group of dinosaurs. This groupseems, however, to have been heterogeneous and,perhaps, rather disorganized. The trackways ofcoelurosaurs (Skartopus) and ornithopods(Wintonopus ) are thoroughly intermingled, andthere is no clear indication that the two types ofdinosaur were segregated. This could indicate thatthese coelurosaurs and ornithopods were in thehabit of moving together as a 'mixed herd', as

was suggested by Krassilov (1980). In this caseone might evisage the coelurosaurs asopportunists — ready to seize insects and othersmall animals as they were flushed fromvegetation by an ornithopod herd movingthrough its feeding grounds. Alternatively thetrackways at Lark Quarry could have resultedfrom accidental mingling of an ornithopod herdwith one or more foraging parties ofcoelurosaurs. Even so it would be reasonable toregard both the ornithopods and the coelurosaursas gregarious, for it seems unlikely that so manyindividuals could have gathered independentlyand at random in the vicinity of the Lark Quarrysite. Nevertheless it is just conceivable that all theLark Quarry track-makers were normally solitaryanimals: they might have been attracted to theLark Quarry water-hole during a period ofdrought. However, we have found no desiccationcracks or other evidence of drought, and it maybe recalled that all the trackways seem to havebeen formed in moist sediment. Moreover, themuch-trampled claystone layer at New Quarryseems to confirm that the Lark Quarry area wasquite commonly frequented by large numbers ofdinosaurs.

Several studies of dinosaur trackways haveemployed Alexander's method (1976) to estimatethe speeds of the track-makers (Russell andBeland 1976, Tucker and Burchette 1977,Coombs 1978, Thulborn and Wade 1979, Kool1981, Farlow 1981, Thulborn 1981, 1982). Thesestudies have made it possible to compare thespeeds of various dinosaurs under variouscircumstances (e.g. see Russell and Beland 1976),or to compare the speeds of dinosaurs to thoserecorded for mammals and ground-dwelling birds(e.g. see Farlow 1981). Such comparisons ofabsolute speeds are certainly interesting, but theyoften give a poor indication of relative locomotorperformances. A mammalian analogy will makethis clear: if a horse, a dog and a mouse all moveat 6 km/h, the horse will be walking, the dog willbe trotting, and the mouse will literally begalloping (Heglund et al. 1974). Absolute speed isidentical for all three animals, but theirlocomotor performances are obviously verydifferent. Exactly similar relationships betweensize, speed and gait would have prevailed indinosaurs, despite Kool's generalization (1981)that different-sized animals, including dinosaurs,'all walk at roughly the same speed'

To evaluate and compare the locomotorperformances of dinosaurs it is desirable to adoptsome criterion that will reduce or eliminate theeffect of differences in body size. In comparing