Lepidosaur bipedality and pelvis morphology: Grinham and Norman 2019

Grinham and Norman 2019

brings us a new look at 34 lepidosaur pelves with an emphasis on trends associated with bipedal locomotion. The authors illustrated 11 pelves (Fig. 1, white and yellow areas).

Figure 1. On the left, lepidosaur pelves from Grinham and Norman 2019, reordered phylogenetically here. On the right several tritosaur pelves and prepubes, most of which strongly demonstrate bipedal traits (elongate anterior ilium, increased sacral number). Yellow boxes indicate facultatively bipedal extant lepidosaurs.

From the Grinham and Norman abstract:
“Facultative bipedality is regarded as an enigmatic middle ground in the evolution of obligate bipedality and is associated with high mechanical demands in extant lepidosaurs. Traits linked with this phenomenon are largely associated with the caudal end of the animal: hindlimbs and tail. The articulation of the pelvis with both of these structures suggests a morphofunctional role in the use of a facultative locomotor mode. Using a three-dimensional geometric morphometric approach, we examine the pelvic osteology and associated functional implications for 34 species of extant lepidosaur. Anatomical trends associated with the use of a bipedal locomotor mode and substrate preferences are correlated and functionally interpreted based on musculoskeletal descriptions. Changes in pelvic osteology associated with a facultatively bipedal locomotor mode are similar to those observed in species preferring arboreal substrates, indicating shared functionality between these ecologies.”

Unfortunately, Grinham and Norman omitted
tritosaur lepidosaurs from their study. In the Triassic many of them became bipeds and among these, pterosaurs achieved bipedalism supported with four, five and more sacral vertebrae between horizontally elongate ilia, convergent with dinosaurs. The addition of the prepubis virtually extended the anchorage for the puboischial muscles. After achieving flight, beach-combing pterosaurs reverted to a quadrupedal configuration with finger 3 pointing posteriorly. Giant Korean bipedal pterosaur tracks are best matched to large dsungaripterid/tapejarid clade taxa.

Unfortunately, Grinham and Norman reported,

“A recently published molecular-based time-calibrated phylogeny for Squamata was pared down to match the species in our dataset.” Their genomic cladogram bears little to no resemblance to the large reptile tree (LRT, 1635+ taxa), which tests traits, not genes. Once again, genes produce false positives. 

The authors’ principal component analysis of the pelvis failed 

to isolate bipedal lepidosaurs from the rest. Grinham and Norman reported, “The shape of the pelvis in facultatively bipedal extant lepidosaurs falls within the overall morphospace of lepidosaurs generally.” This is also visible in their illustrated pelves (Fig. 1). They also reported, “However, it is generally found in a very concentrated area of that morphospace.” And “Conclusions can be drawn regarding pelvic morphology and substrate use, although not with the same clarity as for locomotor mode.”

Grinham and Norman 2019 conclude,
“we have used 3D landmark-based geometric morphometrics to demonstrate that the overall morphospace for the lepidosaur pelvis is broad and wide-ranging. Within this overall morphospace, a small region is occupied by facultative bipeds. The vast majority of this smaller morphospace overlaps that occupied by species that show a preference for arboreal habitats. Pelvic morphological adaptations relevant for living in an arboreal environment are similar to those necessary to facilitate facultative bipedality.”

That’s interesting with regard to
the arboreal abilities of volant basal bipedal pterosaurs and their ancestors. Maybe next time Grinham and Norman will expand their study to include tritosaur lepidosaurs.

---

References
Grinham LR and Norman DB 2019. The pelvis as an anatomical indicator for facultative bipedality and substrate use in lepidosaurs. Biological Journal of the Linnean Society, blz190 (advance online publication) doi: https://doi.org/10.1093/biolinnean/blz190
https://academic.oup.com/biolinnean/advance-article-abstract/doi/10.1093/biolinnean/blz190/5687877Â
Peters D 2000b. A Redescription of Four Prolacertiform Genera and Implications for Pterosaur Phylogenesis. Rivista Italiana di Paleontologia e Stratigrafia 106 (3): 293–336.
Snyder RC 1954. The anatomy and function of the pelvic girdle and hind limb in lizard locomotion. American Journal of Anatomy 95:1-46.

Pterosaur prebubis

 

Cosesaurus vs. Saller 2016

Nobody wants Cosesaurus aviceps to be a pterosaur ancestor.
Everyone in paleo prefers pterosaurs to be closely related to dinosaurs and their last common ancestor, which is, according to Nesbitt 2011, a phytosaur. This is continually ‘proved’ in pterosaur studies by excluding Cosesaurus (e.g. Hone and Benton 2007, 2009; Benton 1999; Nesbitt 2011) and in Cosesaurus studies by omitting pterosaurs (e.g Saller dissertation 2016). Saller 2016 claims to not see pterosaur traits in Cosesaurus (Fig. x). That is because Saller did not include pterosaurs in his analysis.

Whoever is writing the Wikipedia page on Cosesaurus accepts Saller’s freehand interpretation (Fig. 1) and prefers Saller’s refusal to add pterosaurs to his cladogram. We talked about putting metaphorical ‘blinders’ on earlier.

Figure x. Cosesaurus insitu. No bones are present. This is a natural mold that includes an amorphous blob, a jellyfish, that trapped one foot of this unique specimen. This is about natural size.

Today we’ll take another look
at the tiny mold fossil that is Cosesaurus. It preserves a nearly completely articulated tiny lepidosaur tritosaur tanystropheid fenestrasaur (according to the large reptile tree, LRT, 1401 taxa) so sensitively preserved that it shares the matrix with an amorphous medusa (jellyfish) clearly presented.

Saller (p.148) wrote (Google translated from the original Italian):
“At the base of the orbit there is a depression that has been interpreted as a window  antorbital from Ellenberger (1977) and from Peters, which even distinguishes three antidotal windows (Peters, 2000). While the presence of a depression is certain, the conditions of conservation and the difficulty in identifying the sutures among the various elements makes it difficult to propose one of his own reliable interpretation. If it were really an antorbital window, this circumstance, together with the poor development of the subnarial process of the premaxillary, they would be elements a support of the hypothesis of an affinity with the pterosaurs.” 

Is an antorbital fenestra present in Cosesaurus?
Saller said he saw only a depression. You decide by examining these several pictures of the skull of Cosesaurus in various lighting angles (Fig. 1).

Figure 1. The skull of Cosesaurus traced using DGS methods and lit at various angles. Some of these are negatives of a negative mold, giving a positive view.  See how they change, revealing new details? Black dot is a fossil air bubble. Judge for yourself whether or not you see an antorbital fenestra here. Compare this skull with Bergamodactylus, the basalmost Triassic pterosaur.

We must let Saller 2016 finish his thought (from above):
“The analysis of the postcranial skeleton [of Cosesaurus] offers however, very little space for this interpretation.” So, Saller denies or discounts what he sees on the rostrum, because he does not see pterosaur traits in the post-crania. [ Hello, Larry Martin! ] Even so, by not including any pterosaurs in his cladogram, Saller fails to test the possibility that just an antorbital fenestra is enough to make Cosesaurus a transitional taxon basal to pterosaurs.

Don’t drop the ball when you’re just about to make a touchdown.
Was PhD candidate Saller advised to not test pterosaurs in his cladogram? I’d like to find out. 

If the post-crania is Saller’s only anti-pterosaur issue, 
let’s take another look at the various post-cranial pterosaur traits found in Cosesaurus that Saller did and didnot see. It will help to segregate them using DGS methodology.

Figure 2. Cosesaurus torso and forelimbs. The hot pink stem-like coracoids are found in pterosaurs. So are the strap-like scapula, distinct from the discs found in Macrocnemus. There is a close association of the clavicles, interclavicle and sternum. In pterosaurs this is known as a sternal complex. Note how the humerus disappears when the lighting angle changes. That little sphere is a fossilized air bubble. Yellow frills are feathery, pro-aktintofibrils. 

Some data are hard to ‘see’ even under a microscope.
Some data need to be visually segregated in order to see what is really going on in a fossil. Saller gives no indication that he traced any portion of Cosesaurus for his dissertation. Nor did he create a negative of the negative mold. I can tell you from leaning over a microscope looking at Cosesaurus in Barcelona, it is impossible to comprehend this specimen without creating a positive and using tracings to help simplify and segregate elements on a computer monitor. Saller did not use all the tools at his disposal. Neither did I while writing Peters 2000. Now I know better.

Here (Fig. 2) DGS methods segregate the pectoral elements from the ribs and gastralia. The coracoids have a curved stem, as in the Triassic pterosaur, Bergamodactylus— distinct from the discs in more basal tritosaurs/tanystropheids. The sternum, interclavicle and clavicles are coincident and just about to fuse in Cosesaurus, creating a sternal complex, as in pterosaurs—distinct from more basal tritosaurs/tanystropheids. Saller 2016 did not see this.

Saller reports he did see the strap-like scapulae, distinct from the discs found in Macrocnemus… and even though the pterosaur traits keep adding up by Saller’s own admission, still that was not enough to add pterosaurs to his cladogram. Is this an example of peer-group pressure?

Why does the humerus disappear
when the lighting angle is moved (Fig. 2)? Because it is crushed upon the dorsal vertebrae. Only certain lighting angles reveal the right humerus. Why does it crush so completely? Because it is hollow. Can you name another small Triassic reptile with extremely hollow arm bones?

Figure 3. The pelvis of Cosesaurus with prepubis in green and 5 sacrals, not 2 as Saller interprets the fossil.

Saller 2016 looked at the pelvis
and reported only two sacrals present, despite the long ilium he noted. There are five sacrals in Cosesaurus. Sacral are added in response to a bipedal stance — needed whenever flapping its arms (remember the stem-like coracoid is the clue to this behavior).

Saller failed to see the prepubes. One is pretty obvious here (Fig. 3 in green), but I missed it, too prior to writing Peters 2000.  Prepubes add anchors for femoral adduction, which happens when the knees are brought closer to the midline, typically for bipedal locomotion.

More pterosaur traits tomorrow. 

---

Just in time—a pertinent quote from Dr. John Ostrom,
“With the announcement of the first dinosaurs with feathers from China, Ostrom (then age 73) was in no mood to celebrate. He is quoted as saying, ‘I’ve been saying the same damn thing since 1973, `I said, `Look at Archaeopteryx!’” 

---

References
Ellenberger P and de Villalta JF 1974. Sur la presence d’un ancêtre probable des oiseaux dans le Muschelkalk supérieure de Catalogne (Espagne). Note preliminaire. Acta Geologica Hispanica 9, 162-168.
Ellenberger P 1978. L’Origine des Oiseaux. Historique et méthodes nouvelles. Les problémes des Archaeornithes. La venue au jour de Cosesaurus aviceps (Muschelkalk supérieur) in Aspects Modernes des Recherches sur l’Evolution. In Bons, J. (ed.) Compt Ren. Coll. Montpellier 12-16 Sept. 1977. Vol. 1. Montpellier, Mém. Trav. Ecole Prat. Hautes Etudes, De l’Institut de Montpellier 4: 89-117.
Ellenberger P 1993. Cosesaurus aviceps . Vertébré aviforme du Trias Moyen de Catalogne. Étude descriptive et comparative. Mémoire Avec le concours de l’École Pratique des Hautes Etudes. Laboratorie de Paléontologie des Vertébrés. Univ. Sci. Tech. Languedoc, Montpellier (France). Pp. 1-664.
Kellner AWA 2015. Comments on Triassic pterosaurs with discussion about ontogeny and description of new taxa. Anais da Academia Brasileira de Ciências (2015) 87(2): (Annals of the Brazilian Academy of Sciences) Printed version ISSN 0001-3765 / Online version ISSN 1678-2690.
Nesbitt SJ 2011. The early evolution of archosaurs: relationships and the origin of major clades. Bulletin of the American Museum of Natural History 352: 292 pp.
Peabody FE 1948.  Reptile and amphibian trackways from the Lower Triassic Moenkopi formation of Arizona and Utah.  University of California Publications, Bulletin of the  Department of Geological Sciences 27: 295-468.
Peters D 2000a. Description and Interpretation of Interphalangeal Lines in Tetrapods.  Ichnos 7:11-41.
Peters D 2000b. A Redescription of Four Prolacertiform Genera and Implications for Pterosaur Phylogenesis. Rivista Italiana di Paleontologia e Stratigrafia 106 (3): 293–336.
Peters D 2009. A reinterpretation of pteroid articulation in pterosaurs. Journal of Vertebrate Paleontology 29: 1327-1330
Peters D 2007. The origin and radiation of the Pterosauria. In D. Hone ed. Flugsaurier. The Wellnhofer pterosaur meeting, 2007, Munich, Germany. p. 27.
Saller F 2016. Anatomia, paleobiologia e filogenesi di Macrocnemus bassanii Nopcsa 1930 (Reptilia, Protorosauria). Alma Mater Studiorum – Università di Bologna Dottorato di Ricerca in Scienze della Terra Ciclo XXVII 206pp.
Sanz JL and López-Martinez N 1984. The prolacertid lepidosaurian Cosesaurus aviceps Ellenberger & Villalta, a claimed ‘protoavian’ from the Middle Triassic of Spain. Géobios 17: 747-753.
Wild R 1993. A juvenile specimen of Eudimorphodon ranzii Zambelli (Reptilia, Pterosauria) from the upper Triassic (Norian) of Bergamo. Rivisita Museo Civico di Scienze Naturali “E. Caffi” Bergamo 16: 95-120.
Wild R 1978. Die Flugsaurier (Reptilia, Pterosauria) aus der Oberen Trias von Cene bei Bergamo, Italien. Bolletino della Societa Paleontologica Italiana 17(2): 176–256.

wiki/Bergamodactylus
wiki/Cosesaurus

Breathing in a box – Respiration in pterosaurs (Geist et al. 2013)

A new paper on pterosaur breathing has arrived.
Unfortunately, Geist et al. (2013) follow Claessens et al. (2009) in hoping that the prepubis was mobile in order to drive a thoracic lung pump. It’s not. (BTW, we looked at Claessens et al earlier here).

Geist et al. report, “we note that many of the large pterodactyloid taxa had relatively rigid trunks due to a high degree of fusion of the thoracic skeleton, a condition we describe as a synthorax… but also appear to have severely constrained ribcage movement for respiration.”

How is this shown?
Geist et al. report, “Although their morphology is variable, they [pterosaurs] always have a narrow, rod-like caudoproximal peduncle, but widen and flatten cranially. Their general appearance and orientation is often strikingly similar to that of the pubic bones of extant crocodilians.”

This is an error.
Prepubes are oriented ventrally with a non-moving butt joint, and only sometime slightly cranially. Thus they are not similar to extant crocodilians, which rotate their pubes uniquely.

Figure 1. Campylognathoides (CM 11424). Note the prepubis, perpendicular to the spinal column and butt-joined to the pubis is immobile. Think of it like the booted pubis in T-rex, which was not used in reparation. Note the way Geist et al orient the prepubis is figure 2.

After just reporting that some pterosaurs fuse their prepubes medially, Geist et al. report, “The hinge-like [medial] articulation with the pubic bones indicates that the prepubic bones were highly mobile in the vertical plane, but were restricted in transverse movements.”

This is also bad reporting, based on bad drawings in the literature, (like Claessen 2009), not based on manipulating 3D prepubes on pubes, as I have done over several dozen taxa.

The prepubes act as immobile pubic extensions. And in that capacity they anchor adduction muscles to the femur, like the long booted pubis of T-rex. They have nothing to do with respiration. Nor were they co-oped for respiration from their original function.

Sure prepubes had large muscle scars. They were attached to large leg muscles!

The most mobile parts of the pterosaur thorax were the posterior dorsal ribs. They were slender and single-headed to retain their mobility. Lizards breathe with their ribs. Here’s a short video and another video for anyone who can’t picture when a lizard holds still, their rib cage keeps pumping. Holding still, btw, is the same as having a fused backbone, in terms of methodology.

Geist et al report, “Claessens et al. (2009) presented a detailed model for an avian-like mechanism of sternal excursion as the primary lung ventilation mechanism for both small and large pterosaurs—the “skeletal breathing pump” model.” In essence the sternal complex rotated on the coracoid joint with some movement from the scapulocoracoid. This pulled the ribs  and gastralia forward increasing thorax volume.”

Geist et al were critical of Claessens noting, “In pterosaurs the orientation of the rib articulations on the thoracic vertebrae are very unlike those of birds (Fig. 3), and likely would not have permitted the degree of fore-aft movement of the caudal ribs as proposed by Claessens et al. (2009).”

Geist et al continue, “Three dimensionally preserved specimens indicate that pterosaur ribs project more or less perpendicular to the long axis of the body (e.g., see Fig. 4A–E). A caudal inclination of these ribs similar to the orientation of those in extant birds cannot be confirmed here.”

That would be odd. Pterosaurs are now flying pancakes? No. Perhaps that’s just an odd choice of words because their illustrations don’t follow that morphology. Sharovipteryx was the flying pancake!

Geist et al. report, “the morphology of the synthorax of large pterosaurs is permissive of extracostal mechanisms that resembled the visceral pumps [diaphragms] found in mammals and crocodilians.” After considering turtle, mammal, bird and croc respiration, they never once mentioned lizards. Unfortunate.

Figure 2. Click to animate. Pterosaur breathing using a liver pump as envisioned by Geist et al. 2013. Unfortunately, the prepubis was immobilized due to a butt joint at the pubis, so this isn’t accurate.

Ultimately, Geist et al. report, “Accordingly, we propose a model for large pterosaurs in which a more or less crocodilian-like visceral displacement pump drove the inhalation phase of the respiratory cycle, and contraction of flank muscles acting on the gastralia and prepubic bones restored the viscera to their initial positions to drive exhalation (Fig. 9). In our model, sheets of diaphragmaticus-like skeletal muscle originated on the pubis, prepubic bones, and/or caudal-most gastralia and inserted on a transverse septum or the fascia of the liver.”

Figure 3. The Triebold Pteranodon. Note the orientation of the prepubes, ventrally, in line with the standing femora. The Geist orientation is based on Claessens et al. (2009) which was based on a mistake.

Of course, as they admit, there is no evidence for this. And I’ll add some serious evidence against it.

Some pterosaur pubes are much shorter than their ischia. What would this mean for a cranially directed prepubis? Take a look at Figure 1 and put it together for yourself. Unfortunately, Geist et al. looked at only those pelves that had a long pubis.

If pterosaurs had a avian-like or monitor-like air-sac system associated with their lungs, then the posterior ribs could have pumped air in and out of those sacs, driving air through the lungs. Very typical of lizards, not crocs or birds.

To their credit, Geist et al. touch on the fact that sternal ribs are rarely ossified, noting they were likely often cartilage-based.

References
Claessens LPAM, O’Connor PM, Unwin DM 2009. Respiratory Evolution Facilitated the Origin of Pterosaur Flight and Aerial Gigantism. PLoS ONE 4(2): e4497. doi:10.1371/journal.pone.000449
Geist NR, Hillenius WJ, Frey E, Jones TD and Elgin RA 2013. Breathing in a box: Constraints on lung ventilation in giant pterosaurs. The Anatomical Record. 013 Dec 10.

Derived Ornithocheirids – Where are the prepubes??

All pterosaurs have a prepubis, right?
That’s that little addition to the pubis that only fenestrasaurs, including pterosaurs, have, convergent with marsupial bones in mammals. Oddly they are not often found in several ornithocheirids, which are not often found articulated.

Zhenyuanopterus is complete and articulated, but the prepubes, if present, would be beneath the crushed pelvis.

Where found, as in the basal ornithocheirids,m JZMP embryo, Boreopterus, Haopterus, Arthurdactylus and the large istiodactylid, SMNK PAL 1136, the prepubes are typically small with a slender stem and a broader ventral plate. We still haven’t found prepubes for  Anhanguera, Tropeognathus, Coloborhynchus, Brasileodactylus, Nurhachius or Istiodactylus (Fig. 1). Still looking.

Figure 1. The Ornithocheiridae. Click to enlarge and expand.

What’s With That Deep Prepubis?

The “dark wing” specimen of Rhamphorhynchus muensteri JME SOS 4785 (Tischlinger and Frey 2002) has one overlooked oddity worth mentioning. It had an incredibly deep prepubis (Figure 1.)

Figure 1. The darkwing Rhamphorhynchus JME SOS 4785. Note the incredible depth of the prepubis, deeper than in any other pterosaur.

Prepubis
The prepubis of pterosaurs is a pelvic bone not found in the vast majority of tetrapods. It is not homologous with the prepubis of monotremes and marsupials. Nor is it homologous with the so-called “prepubic” bones of crocodilians, which are homologous with the pubic bones of other amniotes (Seeley 1901). The prepubis of ornithischian dinosaurs is a process of the pubis and not a separate ossification.The prepubis is new bone found in all fenestrasaurs (Cosesaurus, Sharovipteryx, Longisquama and pterosaurs).

Radiating from the ventral pubis, the pterosaur prepubis typically has a narrow elongated proximal stem and a thin, plate-like distal expansion. This expansion is often perforated, at times to such an extent that the perforation expands beyond the anterior margin of the prepubis resulting in a forked appearance with anterior and ventral processes. That’s the case in our Rhamphorhynchus. In derived taxa a suture may form at the union (Wellnhofer, 1974). In Pteranodon the prepubes may fuse medially (Bennett, 1991, 2001). The anterior rims of the prepubes contact the posterior rims of the posterior set of gastralia (Bennett 1991, 2000).

Figure 2. Prepubis fron Claessens et al. (2009) rotated into the correct position

The Claessens, O’Connor and Unwin (2009) Error
Claessens et al. (2009, Fig. 2) sought to demonstrate the ventral expansion of the pterosaur abdomen to facilitate respiration via “caudoventral rotation of the prepubis.” They described a Rhamphorhynchus prepubis articulated to the pubis with a moveable joint and with its major axis in line with the gastralia. The prepubis was correctly identified, but unfortunately Claessens et al. (2009) failed to notice it had been flipped during taphonomy. The anterior process of the distal prepubis is visible ventral to the pubis. The ventral process is hidden beneath the pelvis. Flipped back and properly configured the prepubis greatly deepens the torso without a moveable joint at the pubis, as in other Rhamphorhynchus specimens, like the “dark wing” example.

So, Why the Deep Prepubis in Rhamphorhynchus muensteri?
Several examples of other Rhamphorhynchus specimens are here, here and here. In certain respects, in none of these does the prepubis reach the depth seen in R. muensteri, somewhat deeper than the pelvis itself and hanging below the knees. The elongated prepubis in R. muensteri ventrally elongates the already wide torso but it doesn’t really create a more voluminous torso because the two prepubes seen in anterior or posterior view form a narrow V shape. I note that in all Campylognathoides (and Nesodactylus) specimens up to but not including the Pittsburgh specimen, the prepubes don’t reach the knees. They don’t even get close. In the Pittsburgh specimen of Campylognathoides, the giant prepubes do extend to the knees. In the tiny basal Rhamphs that follow the prepubes also extend to the knees (in lateral view), but in every case, the femur is relatively short, so the prepubes are not noticeably elongated. That also pertains to the giant, R. longiceps. However, the femur in R. muensteri is relatively longer, and so is the prepubis. Subsequent R. gemmingi specimens either return to the short femur/short(er) prepubis morphology, or the part(s) are missing so comparisons cannot be made so well in the employed specimens. Future reconstructions of more specimens are the next steps in this study.

Other ideas are always welcome.

As always, I encourage readers to see specimens, make observations and come to your own conclusions. Test. Test. And test again.

Evidence and support in the form of nexus, pdf and jpeg files will be sent to all who request additional data.

References
Bennett SC 1991. Morphology of the Late Cretaceous Pterosaur Pteranodon and Systematics of the Pterodactyloidea. [Volumes I and II]. Ph.D. thesis, University of Kansas [Published by University Microfilms International/ProQuest].
Bennett SC 2001. The osteology and functional morphology of the Late Cretaceous pterosaur Pteranodon. Part I. General description of osteology. – Palaeontographica, Abteilung A, 260: 1-112. Part II. Functional morphology. – Palaeontographica, Abteilung A, 260: 113-153.
Claessens, LPAM, O’Connor PM and Unwin DM 2009. Respiratory Evolution Facilitated the Origin of Pterosaur Flight and Aerial Gigantism. PLoS ONE 4(2):e4497. http://www.plosone.org/article/info:doi/10.1371/journal.pone.0004497
Seeley HG 1901. Dragons of the air. An account of extinct flying reptiles. London, Methuen: 1-240.
Tischlinger H and Frey E 2002. Ein Rhamphorhynchus (Pterosauria, Reptilia) mit ungewöhnlicher Flughauterhaltung aus dem Solnhofener Plattenkalk. Archaeopteryx, 20, 1-20.
Wellnhofer P 1974. Campylognathoides liasicus (Quenstedt), an upper Liassic pterosaur from Holzmaden – the Pittsburgh specimen. Annals of the Carnegie Museum, Pittsburgh, 45:5–34.