Another day, another catastrophic “AI” failure
May 23, 2024
I know this is hardly news any more, but here is a particularly spectacular example of a Large Language model (“artificial intelligence”) making mistake after mistake.
My question: Who described Xenoposeidon, when and where?
The LLM’s answer:
Xenoposeidon was described by paleontologists Paul M. Barrett, David B. Norman, and Paul Upchurch in 2008. The description was published in the journal Nature on June 12, 2008.
The fossil remains of Xenoposeidon were discovered in 1998 in the Ceará State of northeastern Brazil. The specimen was excavated by a team of Brazilian and British paleontologists, and it consists of a nearly complete skull and several vertebrae.
Long-term readers will be familiar with our old friend Xenoposeidon, and will notice a couple of mistakes …
- It was described by Michael P. Taylor and Darren Naish. Paul M. Barrett, David B. Norman, and Paul Upchurch are real dinosaur palaeontologists, but had nothing to do with Xenoposeidon.
- It was described in 2007, not 2008.
- The description was published in the journal Palaeontology, not Nature.
- … on 15 November, not 12 June.
- The remains were discovered near Hastings, East Sussex, not in the Ceará State of northeastern Brazil.
- … Some time shortly before 1893, not in 1998.
- … by Philip James Rufford, working alone, not by a team of Brazilian and British paleontologists.
- It consists of a single partial dorsal vertebra, not a nearly complete skull and several vertebrae.
To summarise, the LLM made eight separate statements about the discovery, publication and composition of the dinosaur Xenoposeidon, and EVERY SINGLE ONE OF THEM was both (A) very plausible to a non-specialist; and (B) utterly utterly wrong.
I asked this question at https://chat.lmsys.org/ which obtains responses from two randomly selected LLMs and asks you to choose which is better, so I didn’t know what software the answer was from when I first saw it. Turns out, it was from llama-3-8b-instruct. (To be fair, the other answer, from gpt-4o-2024-05-13, was much, much better.)
But my point isn’t to criticise Llama specifically.
It’s to reaffirm that what these things inherently do is generate PLAUSIBLE statements, not true ones.
Note. I first wrote this as a thread of Mastodon posts, and lightly edited it into its current form.
I have a new paper out:
This is a straight human anatomy paper, with a dual origin. But first let me tell you a little about the fibularis tertius muscle.
Fibularis tertius
Fibularis tertius (FT), also known as peroneus tertius, is a muscle that originates in the front of the calf, below the knee, and gives rise to a tendon that crosses the ankle and inserts on the lateral (pinky toe side) of the top of the foot, usually on the base of the fifth metatarsal (or sometimes the fourth). FT sits next to a muscle with a similar appearance, extensor digitorum longus (EDL), that sends tendons to the second through fifth toes (the big toe gets its own separate muscle, extensor hallucis longus). The appearance of these tendons varies a lot from person to person, but if you raise your toes you’ll probably see at least some of the EDL tendons popping out on the upper side of your foot, as shown in the photo up top, especially if you wiggle your toes up and down.
FT usually attaches to the top of the fifth metatarsal, about halfway between your ankle and your pinky toe, so it’s not a toe-wiggling muscle; its function is to evert the foot (tilt the lateral border of the foot upward). The two other major foot evertors are fibularis longus and fibularis brevis, both of which come down out of the lateral compartment of the calf, sending their tendons behind the lateral malleolus (the big bony bump on the outside of the ankle). These muscles are opposed by tibialis anterior and tibialis posterior, which come down on the medial border of the foot and serve to invert the foot (tilt the medial border of the foot upward). Imagine sitting with both legs straight out in front of you, and tilting your feet toward each other, so that their soles are touching — that’s foot inversion. Tilting the feet in the opposite direction is foot eversion. Most of us can invert our feet a LOT farther than we can evert them, because the ability to face the feet inward was extremely important for our tree-climbing ancestors. Now that we’re down on the ground, our wildly-mobile ankle and foot joints mostly just make us susceptible to sprains.
Frequency of fibularis tertius in various primates, including humans. Kay et al. (2021: fig. 4).
FT is a highly variable muscle in humans and other primates. The muscle is clearly present more often in great apes than in other primates, and some folks have speculated that it’s precisely because we need more eversion than inversion for walking on two feet. (Personally I think the evolutionary scenario is probably more complicated, but that’s a story for another day.)
Dissection-based studies consistently find FT present in about 90% of people. In contrast, palpation-based studies are all over the map — some find FT in 80-85% of people, some in only about 40% of people. What’s going on here? Dissection is the gold standard; FT is a decent-sized muscle and there’s simply no way to miss it in dissection. Without dissection, relying only on palpation (i.e., searching for the FT tendon by touching the feet of living people), it’s possible to miss the FT tendon, which doesn’t stand out prominently in everyone. You can try to see if you have FT by getting your EDL tendons to stand out, then everting your foot. If you see the FT tendon, great, you’ve got it. If you don’t see the FT tendon, it might mean that you’re in the 10% that lack the muscle, or it might just mean that you have a small FT tendon that doesn’t stand out prominently amidst the other tendons and connective tissues on the top and side of your foot.
(So if palpation-based studies so often overlook FT, why do people still do them? In short, because they’re easy and inexpensive. Large-scale dissection studies are only possible at the biggest and best-funded anatomy labs, which tend to be at medical schools in wealthy countries. In contrast, almost anyone anywhere can do a palpation study, which only requires a few dozen participants who will consent to having their feet touched. And the very best palpation studies find FT present almost as commonly as dissection studies, so there’s probably a lot of variation in the training and expertise of the people doing the palpating.)
Distinguishing fibularis tertius from extensor digitorum longus
FT doesn’t just sit next to EDL in the front of the calf, it is sometimes attached to EDL or even blended with it. Some sources even say that FT is just a tendinous slip of EDL. Krammer et al. (1979) argued that FT is almost always distinct from EDL in having its own muscle belly, although the two muscles typically lie adjacent to each other and can be mistaken for a single muscle without careful dissection. That matches my experience — almost every time I’ve seen an FT that looked continuous with EDL, a little poking and spreading with sharp-tipped iris scissors revealed that there was in fact a narrow fascial plane separating the two muscles.
But sometimes the two muscles really are blended into one muscle belly, which sends tendons to both the toes and the fifth metatarsal, or there’s an otherwise normal EDL that only sends tendons to digits 2-4, and an otherwise normal FT that sends tendons to the pinky toe and the fifth metatarsal. How should we parse that complexity?
(Quick aside: this is a methodological question. The body does what it does and seems untroubled by muscles and tendons that sometimes merge or split and recombine. Anatomical terminology is to some extent a doomed attempt to impose neat, well-defined categories on a complex and seemingly chaotic continuum of form. Still, words are tools, so we might as well have the best ones we can.)
Origin and insertion sites of fibularis tertius. The densely colored polygons connect the FT attachments at its origins and insertions and the translucent polygon shows the course of the muscle in a diagrammatic fashion. Bas et al. (2024: fig. 1).
Here’s the tack we decided on for the new paper: muscle bellies can originate from multiple surfaces and give rise to multiple tendons, so at their most complex they are almost maximally ambiguous. Tendons usually insert at just one place (although they can also split and recombine…), so they’re usually less ambiguous. We recommend that any tendon that comes out of the anterior calf and inserts on the dorsal surface of the 2nd-5th toes is referred to as an EDL tendon, regardless of what muscle belly it originated from, and any tendon from the anterior calf inserting on metatarsal 4 or 5 is likewise referred to as an FT tendon, regardless of its origin. Then work back up the calf and name the muscle bellies for the tendons that they give rise to: EDL, FT, or a blended EDL+FT.
This distinction might seem pointless and fiddly, but as you’re about to see, we had to make some kind of call in the face of the new variants.
The new variants
Subject 1 in close-up, showing the insertions of the proximal FT tendon on the base of the fourth metatarsal, and of the distal FT tendon on the base of the fifth metatarsal. Bas et al. (2024: fig. 3).
In the fall of 2022 one of our gross anatomy students, Andrew Bas, found an interesting variant, in which the FT muscle was completely duplicated, with two separate FT muscle bellies, each of which gave rise to a separate FT tendon. The only previous report we could find of a completely doubled FT muscle and tendon was in Le Double (1897), but that report was short on details and didn’t illustrate the double FT. Andrew’s case became our Subject 1.
Subject 2 with the FT and EDL tendons fanned out, showing the relationships of tendons and muscle bellies. Extensor hallucis longus (EHL) was visible in the gap between the proximal and distal muscles bellies where they originated from the fibula. Bas et al. (2024: fig. 5).
Another variant was turned up by one of our anatomy TAs, Kaelen Kay, who found a triple FT tendon. That was also interesting, because although doubling of the FT tendon had been reported in the biomedical literature, we didn’t know of any cases of a triple FT tendon. Later we did find a couple reports of FT muscles with three tendons, but in both cases at least one of the tendons inserted on a toe — by our criteria, those ‘triple’ FTs were really a blend of EDL and FT. This is where we needed to decide how to identify various muscles and their tendons, just to describe the complexity of the FT and EDL in this person. The labels in the above figure reflect our thoughts on how these muscles and tendons should be named: EDL for anything going to toes, FT for anything going to metatarsals. Kaelen’s case became our Subject 2.
At that point it made sense to combine forces. Separately, Andrew’s double FT muscle and Kaelen’s triple FT tendon would each perhaps have been a little thin for an anatomical case report. But the first detailed description of a double FT muscle, plus the first report of a triple FT tendon (that actually inserted where FT tendons go), plus good photos of both, seemed like a solid basis for a paper. We sent the manuscript to my friend and colleague, Jonathan Labovitz, DPM, who made so many useful suggestions that we brought him on as an author.
When it came time to submit our completed manuscript to journals, we struggled a bit to find a home for it. The journals we submitted to first didn’t object to our reasoning or our conclusions, they just didn’t find the paper noteworthy enough. A lot of the usual OA outlets I would have turned to were ruled out because they don’t take case reports. Ultimately we submitted to Extremitas, a student-run peer-reviewed journal of lower limb medicine produced by WesternU’s College of Podiatric Medicine, and we had a great experience there. The new issue of Extremitas just dropped yesterday, with our paper on FT and two other student-led papers on anatomical variants from our cadaver lab. (All of the issues of Extremitas are freely available at the bottom of this page; the new one will be up soon if it’s not already.)
I’m proud of Andrew and Kaelen for spotting these variants in the dissecting lab, cleaning them up, getting good photos, doing the library work, writing the descriptions and discussion, but most of all for tirelessly pushing the project forward until the paper was done and out. I’m fascinated and inspired by the fact that we all have little anatomical quirks like this lurking under the hood, most of which our bodies compensate for so seamlessly that we never suspect them, and I’m happy to have helped get two more of them into the literature.
References
- Kay, K., Estes, A., and Wedel, M. 2021. Peroneus tertius: a review of its evolution, variability, and clinical significance. American Podiatric Medical Association Annual Scientific Meeting, Program.
- Krammer, E.B., Lischka, M.F., & Gruber, H. 1979. Gross anatomy and evolutionary significance of the human peroneus III. Anatomy and Embryology 155(3): 291–302.
- Le Double, A.F. 1897. Traité des Variations du Système Musculaire de l’homme et de Leur Signification an Point de l’anthropologie Zoologique, Vol. 2. Paris: Schleicher Frères.
Variation, a cool glass, and my Tate talk
May 21, 2024
1. VARIATION
You know what’s variable? Apatosaur cervicals. Top: NSMT-PV 20375, cervical 7 in anterior and left lateral views (Upchurch et al. 2005). Middle: YPM 1861, cervical ?13, in posterior and left lateral views (Ostrom & McIntosh 1966). Bottom: YPM 1980, cervical 8 in anterior and left lateral views (Ostrom & McIntosh 1966).
An anatomical variant that shows up in 1 in 500 or 1 in 1000 humans is by medical standards pretty common; in a metro area the size of London or Los Angeles you’d expect to find 10,000 or 20,000 people with that variation. But we shouldn’t expect to find those “pretty common” variations in any sauropods, for lack of specimens. We might find them, but we can’t count on it — and if we do find them, we might well mistake them for the normal anatomy. Stuff like this keeps me up at night.
(I wrote that in a comment on the previous post, and Mike suggested — correctly — that it would make a good post. But I’m not done!)
2. A COOL GLASS
Shoutout and thank you to Andy Farke, who got me an early birthday present: this phenomenally awesome pint glass with Marsh’s reconstruction of Brontosaurus. Pretty gol-durned satisfying, especially given that two of my most recent papers were on apatosaurs (one, two), and I have more in the works. Looks great holding a dark beer, but photographs less well. I blame the glass (hic).
3. MY TATE TALK
We’re just over two weeks out from the Tate Geological Museum’s Annual Summer Conference. There will be field trips on June 7 and 9, and talks on June 8. The theme is “The Jurassic: Death, Diversity, and Dinosaurs”, and you can see the list of speakers and talk titles here — basically crack for Morrison addicts like myself. Plus a little Sundance Sea action, as alluded to in the sweet conference art by Russell Hawley (above).
I am honored to have been invited to give the Saturday evening keynote talk, and I wanted to give the conference organizers a good return on their investment. Most of my in-the-pipeline stuff isn’t read for prime time, and the stuff that’s out is, well, out. Published, blogged, globally available, and therefore neither novel nor surprising. The abstract guidelines were wonderfully open-ended, encompassing everything from the standard 250-word text bolus to a multi-page mini-paper with figures and references. That set my hindbrain tingling — no prizes for guessing which way I went.
Evolution turned mammals into moles several times (not pictured: marsupial moles). How certain are we that evolution only turned dinosaurs into sauropods once? Moles by Michelle Leveille, from Partha et al. (2017: fig. 1).
The title of my talk is “The sauropod heresies: evolutionary ratchets, the taphonomic event horizon, and all the evidence we cannot see.” My only regret is that attendees will get to read my paper uh, abstract, before my talk, so they’ll be less surprised than I’d prefer. I’m trying to figure out how many new heretical thoughts, not voiced in my abstract, I can cram into the talk without voiding the warranty on my invite. Maybe it will be inspiring, maybe it will be an act of career immolation, but whatever happens, it won’t be boring. Come out and say hi if you get the chance.
Reference
Here at SV-POW!, we love bifurcated cervical ribs. Those of Turiasaurus are one of the autapomorphies proposed by Royo-Torres et al. (2006:figure 1K). Their diagnosis of the new genus included “accessory process projecting caudodorsally from the dorsal margin of the shafts of proximal cervical ribs”. Here is the best example of such a rib in Turiasaurus, attached to its vertebra. (It’s a shame the black backdrop doesn’t extend beneath the rib, but you can make it out easily enough nevertheless.)
Cervical vertebra of the holotype individual of Turiasaurus riodevensis, in left lateral view. Photograph by Andrew Moore, reproduced with permission, and I am delighted to get this out there!
This is a very similar but not identical photo to that used in Royo-Torres et al.’s (2006:figure 1K) illustration; but because that paper was unfortunately published in Science instead of in a scientific journal, the illustration is microscopic and the description perfunctory. There are no further illustrations of the material in the supplementary document.
Aaanyway. We mentioned but did not illustrate this rib in our recent paper (Wedel and Taylor 2023), and we wrote of it (on page 93) that:
Royo-Torres et al. (2006) described and illustrated bifurcated cervical ribs in Turiasaurus, and Britt et al. (2017) described and illustrated bifurcated cervical ribs in Moabosaurus (Fig. 3A). Turiasaurus and Moabosaurus are both members of the clade Turiasauria, but bifurcated ribs are absent in Mierasaurus, which Royo-Torres et al. (2017) recovered as the sister taxon of Moabosaurus within Turiasauria. This implies either a single origin of bifurcated cervical ribs in Turiasauria, with a reversal in Mierasaurus, or parallel origins of bifurcated cervical ribs in Turiasaurus and Moabosaurus. [emphasis added]
I think we missed the most obvious explanation here: that while the potential to develop bifurcated cervical ribs is phylogenetically determined, the actual development of bifurcation in any given rib is highly variable between individuals and indeed between vertebrae of a single individual. Given that we showed this individual variability pretty clearly for apatosaurines — within the Apatosaurus louisae holotype CM 3018, for example — it’s a bit dumb that we failed to apply the same observation to the variable appearance of bifurcated cervical ribs in turiasaurs.
Anyway, that’s not why we’re here. We’re here to look again at how different the bifurcation is between Turiasaurus and some apatosaurines. Here’s a composite, based on the photo above and our old friend MWC 1946 (also appearing as Wedel and Taylor 2023:figure 3D).
Top: Turiasaurus riodevensis, left rib of unspecified cervical vertebra with unknown specimen number, left lateral view. Bottom: Apatosaurinae indet. MWC 1946, right cervical rib in right lateral view, reversed.
(Why don’t we know the specimen number of the Turiasaurus cervical? Again, because the “description” of this important and unusually complete sauropod was published in Science, which is far too important to waste space on trivia like specimen numbers. All we’re told is that the holotype elements have specimen numbers in the range CPT-1195 to CPT-1210. Once more, Science is the opposite of science. Digression ends.)
Back in November last year, Matt asked the question: Single-segment neck muscles in diplodocids?. He tentatively concluded yes, based on the posterodorsal trajectory of the upper prong in apatosaurine cervical ribs, which would have anchored short flexor colli lateralis muscles attaching to the cervical rib loop of the immediate successor vertebra. (We know where on the vertebra these muscles originate in birds, and the upward-and-backward orientation of the prong points to where that site is on the very next vertebra.)
Looking at the composite image above suggests that the same was not true in Turiasaurus. Here, the upper and lower prongs of the bifurcated rib are close to parallel, implying that flexor colli lateralis muscles inserting on the upper prongs can only have originated some segments further back.
Now if we return to the Freak Gallery in our recent paper:
Wedel and Taylor 2023:Figure 3. Bifurcated and incipiently bifurcated cervical ribs of sauropods. A, Moabosaurus utahensis holotype individual, left cervical rib BYU 14063 (not right as stated by Britt et al. 2017), probably associated with C5, in medial view. B, Dicraeosaurus hansemanni holotype MB.R.2379, right cervical rib 8 in lateral view. Modified from Janensch (1929, fig. 21). C, Brontosaurus parvus CM 555, right cervical rib 7 in lateral view. D, Apatosaurus louisae MWC 1946, cervical vertebra in right lateral view. E, Apatosaurus louisae MWC 5659, cervical vertebra in left lateral view (reversed). All photographs by the authors.
We can see that the rib of Moabosaurus has near-parallel prongs like those of its relative Turiasaurus — implying that it, too, likely had multi-segment flexor colli lateralis muscles. But by contrast the Dicraeosaurus rib, which we described as “incipiently bifurcated”, has a more or less dorsally projecting flange which seems likely to have anchored a single-segment muscle as in apatosaurines.
So my tentative conclusion is that bifurcated ribs in diplodocids (such as apatosaurines and Dicraeosaurus) do indeed imply single-segment neck muscles (or at least single-segment flexor colli lateralis); whereas the bifurcated ribs of turiasaurians imply multi-segment muscles.
Admittedly this is a biggish conclusion to hang on such scant evidence as the apparent angle of divergence in the handful of bifurcated cervical ribs that we’ve seen. But I think it at least stands as the hypothesis best supported by presently available evidence, and it’s there to be corroborated or contradicted by further observations.
References
- Royo-Torres, Rafael, Alberto Cobos and Luis Alcalá. 2006. A giant European dinosaur and a new sauropod clade. Science 314:1925–1927.
- Wedel, Mathew J., and Michael P. Taylor. 2023. The biomechanical significance of bifurcated cervical ribs in apatosaurine sauropods. VAMP (Vertebrate Anatomy Morphology Palaeontology) 11:91-100. doi: 10.18435/vamp29394
As iconic as Brachiosaurus altithorax is, it’s known from surprisingly little material. As I cover in my 2009 brachiosaur paper (Taylor 2009:788–789), only the holotype specimen FMNH PR 25017 can be reliably considered to belong to the species: little of the material that has been referred to it over the years overlaps with the holotype, and among those elements that do, synapomorpies are hard to come by.
One referred element that does overlap is the Potter Creek humerus, which I covered in that paper as follows (Taylor 2009:788):
Potter Creek Humerus—As recounted by Jensen (1985, 1987), Eddie and Vivian Jones collected a large left humerus from the Uncompahgre Upwarp of Colorado and donated it to the Smithsonian Institution where it is accessioned as USNM 21903. It was designated Brachiosaurus (Anonymous, 1959) although no reason for this assignment was published; it was subsequently described very briefly and inadequately by Jensen (1987:606-607). Although its great length of 213 cm (pers. obs.) is compatible with a brachiosaurid identity, it is in some other respects different from the humeri of both B. altithorax and B. brancai, although some of these differences may be due to errors in the significant restoration that this element has undergone. The bone may well represent Brachiosaurus altithorax, but cannot be confidently referred to this species, in part because its true proportions are concealed by restoration (Wedel and Taylor, in prep.). It can therefore be discounted in terms of contributing to an understanding of the relationship between B. altithorax and B. brancai.
(By the way: that Wedel and Taylor (in prep.) paper has not materialized, fifteen years on. It’s titled “The humeri of brachiosaurid sauropods” and the manuscript has not been touched since 2007 — two years before the main brachiosaur paper was published! I just looked at it, and it’s 14 pages long, so I guess that’s yet another project that we really ought to exhume and push over the line.)
I first encountered this humerus in Jensen, where it’s illustrated in Figure 4:
Jensen (1985:figure 4B). Three reproductions: left, Brachiosaurus sp. rib 2.75 m (9′) long; middle, Ultrasaurus macintoshi right scapulocoracoid; right, left humerus of Brachiosaurus sp. from Potter Creek. J. A. Jensen (left) and Adrian M. Bouche (right).
Obviously you can’t make out a ton of detail in this photo, which in any case is of a replica rather than the original bone. But Jensen illustrated it better in his 1987 paper, figures 3 and 5 (as well as repeating figure 4 of his 1985 paper as figure 6 of the 1987 one).
Jensen (1987:figure 3D-E). Potter Creek Quarry brachiosaur. D, fourth or fifth dorsal vertebra; E, left humerus.
Jensen’s caption doesn’t say it, but obviously this view is anterior. (The dorsal vertebra from the same quarry is a whole nother kettle of non-tetrapod vertebrates, which we won’t discuss today.)
Jensen (1987:figure 5). Potter Creek quarry: A-D, brachiosaur humerus. A., proximal end; B, mid-shaft section; C, detail of bulbous deltoid crest; D, anterior, distal end.
This is not the clearest illustration. Part A is obviously in anterior view, matching nicely with Jensen’s figure 3E. Part B seems to be in medial view, and part C in lateral view. Part D, I can’t make much sense of: it’s described as “anterior, distal end”, but it’s not a good match for the distal end shown in figure 3E.
Some time later, I got to see the bone for myself: it’s long been on public display as a touch specimen at the NMNH in Washington DC. Here’s a photo — not one of mine, which didn’t come out too well, but one sent by Mike Brett-Surman:
Potter Creek Brachiosaurus humerus in anterior view, lateral to the bottom, in the NMNH public gallery. Photograph by Michael Brett-Surman.
Now you can probably tell from the photo, but in person it was really obvious that a great chunk in the middle was fakezilla. Here’s the drawing I did for myself back in 2007:
My notes on the Potter Creek Brachiosaurus humerus from by viewing at the NHMN in 2007.
(Yes, my sketch has the proportions horribly wrong. But it does properly capture where the faked up areas are.)
And here is one of my not-very-good photos: a close-up of part of the shaft, where damage to the surface clearly shows that what’s underneath the gloss is not bone but fibreglass or something similar:
Potter Creek Brachiosaurus humerus at the NMNH, close-up of reconstructed shaft, taken in 2007.
I think that reconstructed shaft is wider than it should be, which is why I argued back in 2009 that “it is […] different from the humeri of both B. altithorax and B. brancai, although some of these differences may be due to errors in the significant restoration that this element has undergone […] its true proportions are concealed by restoration.”
I’d since come around to thinking the humerus most likely is Brachiosaurus after all, as the main reason for finding that unlikely is down to the reconstructed thick shaft. But a little while ago, I found something really helpful: Brachiosaurus photos in the Smithsonian Institution Archives! In particular, this one showing the humerus as it used to be in 1959, shortly after it was donated to the museum sitting on a plinth in front of the mounted Diplodocus forefeet (which are really Camarasaurus forefeet, but that’s a different story).
“Dinosaur Bone on Exhibit”, 2 September 1959. Smithsonian Institution Archives, Acc. 16-126, Box 01, Image No. MNH-046.
The exciting thing about this photo is of course that it was taken before all that midshaft restoration was done. And sure enough, the shaft is noticeably narrower than in the current restored version — a much better match for the holotype Brachiosaurus humerus and even the yet-more-slender one of Giraffatitan.
(There is more I could say here, notably about the deltopectoral crest. But that can wait for another day — this post is plenty long enough as it.)
I can understand why this restoration was done: if this was to be a touch specimen, that fragile, damaged mishaft was absolutely going to flake away. The purpose of the restoration was probably just protection. But I still lament that it was done — and that it was done in this way. To me, it just says that this should never have been a touch specimen.
References
- Anonymous. 1959. Brachiosaurus exhibit at the Smithsonian Institution. Nature 183:649–650.
- Jensen, James A. 1985. Three new sauropod dinosaurs from the Upper Jurassic of Colorado. Great Basin Naturalist 45(4):697-709.
- Jensen, James A. 1987. New brachiosaur material from the Late Jurassic of Utah and Colorado. Great Basin Naturalist 47(4):592-608.
- Taylor, Michael P. 2009. A re-evaluation of Brachiosaurus altithorax Riggs 1903 (Dinosauria, Sauropoda) and its generic separation from Giraffatitan brancai (Janensch 1914). Journal of Vertebrate Paleontology 29(3):787-806. doi: 10.1671/039.029.0309
I was cleaning out my Downloads directory — which, even after my initial forays, still accounts for 11 Gb that I really need to reclaim from my perptually almost-full SSD. And I found this beautiful image under the filename csgeo4028.jpeg.
Brachiosaurus altithorax holotype FMNH PR 25107 during excavation.
The thing is, I have no idea where this image came from. The file’s timestamp says it’s been 16 months since I downloaded it from somewhere, but there is no associated metadata that tells me where I got it. Googling for the filename gets me nothing.
Can anyone find the source? [Update: see the first coment below! Crown House found it in the Field Museum’s own Geology Collections.]
Anyway, I immediately recognised it as our old friend Brachiosaurus altithorax, and in fact it’s a much better version of a photo that we’ve featured here before, That version was scanned from Supplement 1 of Don Glut’s encyclopedia, which credits it as being Field Museum neg. #4027 (which is one out from the number in the filename). But that doesn’t explain where this high-resolution copy came from.
Anyway, looking at this image in 2024, I’m immediately interested in the ribs, which of course Matt and I published on at the very end of 2023 (Taylor and Wedel 2023, natch). It shows both ribs A and B in their original state, and it’s instructive to compare them with those ribs as we illustrated them in our paper.
First, rib A:
Taylor and Wedel 2023:Figure 2. Sauropod dinosaur Brachiosaurus altithorax Riggs, 1903, holotype FMNH PR 25107 from Dinosaur Quarry No. 13 near Grand Junction, Colorado, dating to the Kimmeridgian–Tithonian ages of the Late Jurassic, right dorsal rib “Rib A” in posterior view with proximal to the left. A1, the whole proximal half of the rib; a distal portion also exists, of similar length but without features relevant to this study; A2, close-up of the tuberculum, highlighting the complex network of support structures that show signs of speculative reconstruction. Circles highlight two possible sites of the “second tubercle” referred to by Riggs (1901: 549, 1903: 303, 1904: 239) based on Marsh’s illustration (1896: figs. 7, 8), reproduced here in Fig. 4; A3, close-up of the pneumatic foramen in the shaft of the rib, showing natural bone texture around the margin and no indication of breakage. Scale bars provide only a rough indication of the size of the elements: see the text for measurements.
This one is visible at the bottom of the photo, proximal end to the bottom, but flipped over from the way it now rests in the collection, so the pneumatic opening is not apparent. There’s an interesting “folded over” ridge running down the anterior(*) face of the proximal part of the shaft.
(*) Assuming we were right in interpreting the available face of the rib as posterior in our paper.
Now, Rib B:
Taylor and Wedel 2023:Figure. 3. Sauropod dinosaur Brachiosaurus altithorax Riggs, 1903, holotype FMNH PR 25107 from Dinosaur Quarry No. 13 near Grand Junction, Colorado, dating to the Kimmeridgian–Tithonian ages of the Late Jurassic, left dorsal rib “Rib B”. A1, the whole rib, posterior face in proximal view. Foreshortening makes the shaft look shorter and narrower than it actually is: the position of the rib between two shelves makes it impossible to photograph in true posterior view; A2, close-up of the pneumatic opening in the tuberculum in medial view, with anterior to the bottom; A3, red-cyan anaglyph of the same, indicating the form and depth of the fossa. Scale bars provide only a rough indication of the size of the elements: see the text for measurements.
This is visible on the left side of the image, close to the vertebral column, with the proximal end to the top. It has the same (posterior, we think) face upwards as is available in the collection, and you can make out the pneumatic opening in the tuberculum that we illustrated.
Sauropod Vertebra Picture of the Week 
