Sorry to keep dumping all these off-topic thoughts on you all, but I got an email from Matt today in which he suggested that there should be some system of giving people credit for particularly insightful blog comments.  (This came up for the obvious reason that SV-POW! readers tend to leave unusually brilliant comments, as well as having excellent reading taste and being remarkably good looking.)  That led me into the following sequence of thoughts, which I thought were worth blogging — not least in the hope that we can learn something from the comments.

But first, here is that photo of another fused atlas-axis complex that you ordered (seriously, what’s up with these things?):

Camarasaurus grandis YPM 1905, fused atlas and axis in right lateral view

Camarasaurus grandis YPM 1905, fused atlas and axis in right lateral view

And now, on with the uninformed noodling:

As things stand at this point, we have a hierarchy of sciency documents. At the top (which we’ll call level 1) come papers. The reputation of papers is largely determined by formal pre-publication reviews (which we will therefore classify as level 2) — and, increasingly, also by blog posts about the paper, which are also level 2. Classic peer-reviews are only ever seen by the editor and the author of the original paper; once they have been absorbed into the paper they’re critiquing, they disappear forever, which is a crying shame. But the other kind of level-2 literature, the blog post, has a life of its own: and so it gets commented on by blog-comments (level 3). Each level gives validity to the level above.

More important, documents at each level also give validity to each other. The most important case is that when one paper favourably discusses another, or refers to its authority, it gives the latter a credibility boost (which is why it’s such a sod that no-one cites any of my papers); similarly, our SV-POW! posts also get a credibility boost when they’re discussed on Tetrapod Zoology or Blog Around the Clock (and I just repaid the compliment by linking back to them).

(At present, all of this is done in a messy qualitative way, with no numbers attached, except occasionally in the case of pre-publication reviews. That’s a shame: if, for example, blog commenters allocated the posts a score out of ten, then we could use some kind of average score as a quality filter: to ameliorate rigging, I’d suggest discarding the highest and lowest 10% of awarded scores, and averaging the remainder.)

Now the problem: blog comments are right at the bottom of the pile: who is going to rate them? I’m certainly not going to spend any time on that.

OK, so suppose we ignore the arbitrary allocation of levels: papers, reviews, blog posts and comments are all just considered as documents, and all can discuss each other. (Clearly reviews will necessarily discuss papers more often the papers discuss blog comments, but that is a convention added to the system I am about to describe, not a precondition for it.) Each document has a reputation, which we will quantify as a single real number. Documents start with some arbitrary small reputation — probably 0.0 or 1.0, and it probably doesn’t much matter what it is. When any document discusses, cites or links to another — whether it’s paper, a review, a blog post or a comment — that linkee’s reputation is boosted by some proportion: 10%, say, of the linker’s reputation. Now of course this change in linkee reputation causes a trickle-down change of 1% in the reputation of the documents that it links to; and 0.1% in the reputation of the documents they link to, and so on. Reputations will change frequently and irregularly, and will be near impossible to calculate accurately, but that’s fine — they should be easy to approximate, and that’s good enough.

In this way, we get a nice solid score that we can use to decide what’s worth reading and what isn’t — the cream will naturally rise to the top. Hiring committees can throw away impact factors, and instead just add up the reputation scores of their candidates’ publications (either in the strict sense of the word, or including blog posts, reviews and/or comments). By the way, one of the positive effects of this would be that people like Darren and Jerry Harris would get some reward from their sterling reviewing efforts.

Sounds awesome? Here’s something even more awesome: we already have that system, more or less. Yes indeed: the reputation propagation algorithm I described is, in general outline, the same thing that Google does in the algorithm that it calls PageRank(tm)(r)(lol)(ymmv).  We can — and already do — use Google’s notion of reputation as a guide to finding what’s worth reading, and we can tell that in works well in practice because SV-POW! posts rank so highly :-)

So that’s it! We can all stop worrying, just Google for stuff we’re interested in, and read whatever pops up at the top of the list!

Are you convinced? I hope not, because this idea has at least three huge problems.

1. What counts? (Yes, that again.) Google-ranking works well for blog posts, because they are web pages, and Google can spider web pages. But that leaves out reviews, because they are typically not published at all, let alone as web pages. And it leaves out comments, because they are appended to the end of blog posts rather than being pages in own right, with their own PageRank. And, worst of all, it pretty much leaves out the papers themselves — because there is, in general, no one single web-page which is The Place a particular paper lives. For non-open papers that aren’t hosted on the author’s page or elsewhere, there is no page.  In short, reviews are not published, comments are not whole pages and papers are not single pages, so none of them is properly page-rankable.

2. All links count as positive reputation — there are no negative citations. So a document that saysTaylor, Wedel and Naish 2009 was talking a lot of nonsense about sauropod neck posture” would still be a score in our favour, even though it meant the exact opposite. Of course, this is not a new problem: both PageRank and Impact Factors suffer from the same problem, but it doesn’t seem to be a killer for either of them. The only fix for this would be to invite authors (of papers, reviews, blogs and comments) to explicitly score some or all of the other documents they mention — and I doubt people are going to be keen to do that unless the mechanism can be made very non-intrusive.

3. And here’s the killer: we wouldn’t, or shouldn’t, want Google to do this, even if they could overcome problems #1 and #2.  Google is a private corporation, and we don’t want to hand over reputation management to any private commercial venture with an obligation to shareholders rather than scientists, and with a proprietary secret algorithm. If you doubt me, consider Thompson’s ownership of the Impact Factor and see where that’s got us. No doubt when Eugene Garfield came up with the idea of the Impact Factor, he was pretty excited about how — at last! — we would have an objective, reliable way to evaluate science. But IF is not run by scientists, it’s run by a corporation.  With hilarious results.

I have no idea what the conclusion to all this is. I didn’t have a clear idea where it was headed when I started writing it. But, much in the manner of Dirk Gently when employing his usual method of navigation, I may not have ended up where I intended to, but I’ve arrived somewhere interesting.

Your move: what have I failed to take into account?

We have sometimes neglected tails on SV-POW!, in favour of the more obviously charismatic charms of presacral vertebrae, but every now and then you come across a caudal vertebra so bizarre that it just cries out to be blogged.

One such is this specimen, which may or may not be BMNH R 2144:

Sauropod caudal with co-ossified chevrons, lateral view

Sauropod caudal with co-ossified chevrons, right lateral view

Sauropod caudal with co-ossified chevrons, right posterolateral view

Sauropod caudal with co-ossified chevrons, right posterolateral view

The reason I’m not sure whether this is BMNH R2144 is that I noticed this at the very last minute while visiting the NHM collections to see a different specimen, and just had time to take a couple of quick photos before kicking-out time.  The label on the side of the vertebra has the unexplained number 2144 written on it, so I am guessing this is the specimen number, but I wouldn’t stake my life on it.

(By the way, both these photographs are copyright the NHM.)

The interesting thing about this vertebra is of course that that the chevrons are co-ossified with the centrum — an extremely rare condition in sauropods, in fact unique as far as I know.  As we’ve shown here and here, among other places, the chevrons are usually separate bones from the vertebrae.

This vertebra caught my eye not only because it’s, well, weird, but also because I’d seen it a couple of times in published figures.  It’s in Mantell’s (1850) description of Pelorosaurus, where it appears as figure 11 in plate XXIII, and is considered to belong to Pelorosaurus; and also in Owen (1859: plate V: figs. 3-4).  Owen seems pretty confused about the identity of this element, and in this paper alone assigns it to Streptospondylus (p. 22), Iguanodon(!) (p. 25) and implicitly Cetiosaurus (p. 34).  So what is it?  Well, its provenance is vague in the extreme, so given that it’s not associated with any more diagnostic material, about the best we can say with any honesty is that it’s Sauropoda incertae sedis.

Let’s take a look at those old figures:

Mantell (1850: plate XXIII, fig. 11)

Mantell (1850: plate XXIII, fig. 11)

Owen (1850: plate V, figs 3-4)

Owen (1850: plate V, figs 3-4)

If you’re like me, your first thought was that Owen’s figures are simply mirror images of Mantell’s.  I checked this out by Photoshopping the two sets of figures, flipping them horizontally, scaling and rotating as necessary, and found to my mild surprise that Owen’s figures are in fact redrawn, despite the startling resemblance they bear to Mantell’s.  As it happens, the same is true with the Owen 1859 plate that is the humerus of Pelorosaurus figured by Mantell 1850, and in that case Owen’s figure is rather better than Mantell’s, so let’s give a bit of credit to Owen here.  Most embarrassing for Mantell (not that he cares, having been dead for 157 years) is that Owen’s flipped images seem to be correct (at least, as best I can judge from the photographs I took) — looks like Mantell or his illustrator badgered this up.

So what is going on with these co-ossified chevrons?  As is so often the case, we just don’t know.  Some possibilities: this might be a pathology of an individual, caused either by injury or infection; it might be a natural ontogenetic character in very old individuals; or it might by a taxonomically significant character of a taxon we’ve not yet found — or one that we have found, but don’t yet recognise as being the same thing.  It’s perfectly possible that this is a chevron of Xenoposeidon, for example, but until someone finds a nice complete specimen we’ll never know.

Not much is known about skeleton fusion in sauropods, and most of what’s in the literature is anecdote.  That is set to change, I am pleased to say, as Matt is putting together a paper with his colleague Elizabeth Rega that will survey and interpret the various fusions known in sauropod vertebrae.  I’m looking forward to seeing what they have to say about this vertebra.


  • Brusatte, Stephen L., Roger B. J. Benson, and Stephen Hutt.  2008.  The osteology of Neovenator salerii (Dinosauria: Theropoda) from the Wealden Group (Barremian) of the Isle of Wight.  Monograph of the Palaeontographical Society 162 (631): 1-166.
  • Calvo, Jorge O., Juan D. Porfiri, Claudio Veralli, Fernando Novas and Federico Poblete.  2004.  Phylogenetic status of Megaraptor namunhuaiquii Novas based on a new specimen from Neuquen, Patagonia, Argentina.  Ameghiniana 41 (4): 565-575.
  • Mantell, Gideon Algernon.  1850.  On the Pelorosaurus: an undescribed gigantic terrestrial reptile, whose remains are associated with those of the Iguanodon and other saurians in the strata of Tilgate Forest, in Sussex.  Philosophical Transactions of the Royal Society of London 140: 379-390.
  • Owen, R.  1859a.  Monograph on the fossil Reptilia of the Wealden and Purbeck formations.  Supplement no. II (pages 20-44 and plates V-XII): Crocodilia (Streptospondylus, &c.) [Wealden].  Palaeontographical Society, London.


Thanks to Mickey Mortimer for pointing out that this kind of centrum-chevron fusion is known in the theropod Megaraptor.  Here is the relevant figure from Calvo et al.’s (2004) revision of that genus:

Calvo et al. (2004: fig. 5). Caudal vertebrae of Megaraptor with co-ossified chevron

Calvo et al. (2004: fig. 5). Caudal vertebrae of Megaraptor with co-ossified chevron

The strange thing is this comment in the text (p. 569): “Two articulated caudal vertebrae are preserved (figure 5), slightly laterally compressed.  Their centra and the neural arches are firmly co-ossified, as well as their respective haemal arches [i.e. chevrons].  This fusion, not infrequent among dinosaurs, may be pathological.”  Not infrequent?  Is this going on all over the place and I’ve just never noticed it?  Anyone have any more examples?

Update 2

Here is that pair of fused Neovenator caudals with a co-ossified chevron, which Darren mentions in the comments below.

Brusatte et al. (2008: fig. 16d), fused Neovenator caudals with co-ossified chevron

Brusatte et al. (2008: fig. 16d), fused Neovenator caudals with co-ossified chevron


Here are the first two cervical vertebrae of the Carnegie Apatosaurus, from Gilmore’s 1936 monograph. As you can see, they are fused together. It is a bit weird that we haven’t covered the morphology of the atlas-axis complex here before. And sadly we’re not going to cover it now. I needed to get an image of these verts to a group working on…something secret…and this turned out to be the fastest way to get them the information in a format that would be easy to find for future reference. Hope you don’t feel used.

UPDATE: Here’s something weird: the both verts have facets for cervical ribs, but the cervical ribs had not fused to the vertebrae, even though they normally do, and despite the fact that the vertebrae had fused to each other, even though they normally don’t.

Ahead by a tail

February 2, 2009

You’d think that in 100+ posts we’d be starting to exhaust the territory, but there are vast swaths of sauropod vertebral morphology that we haven’t even touched. Like fused vertebrae. Sauropods fused their vertebrae all the time. Some of those fusions are age-related, many are pathological, and some are…hard to classify.


Exhibit A: fused distal caudals in a specimen of Mamenchisaurus hochuanensis described by Ye et al. (2001). In contrast to the terminal caudals comprising the tail club of Shunosaurus, the centra here are not ballooned out. The one in the middle is clearly waisted, as in “narrower in the middle than at the ends” (not the same clearly wasted as your college roommate). The neural, uh, elements are expanded and fused into something that the authors describe as resembling the comb of a rooster. I can’t improve on that metaphor so I won’t try. Here’s the full weirdness, straight from the authors (p. 39):

The posterior caudals are fused with each other, their centra are not expanded, the neural arch is remarkably expanded and the size of the neural canal  and the height of  the neural spines increased. In lateral view, the posterior caudals are cockscomb-shaped.

That’s all pretty weird. The authors go on to speculate that the expanded neural canal indicates that the tail club fin thingy served as some kind of special sense organ. I don’t think that idea is too bold. I don’t think it’s bold enough.

Hypothesis: Mamenchisaurus had a pseudohead on the end of its tail, with fused verts to form a pseudoskull and a big nerve bundle to give the pseudomouth (probably articulated chevrons) and pseudoeyes (possibly heat-sensitive like rattlesnake pits) some lifelike movements and relay thermal images up to the brain. It probably started out as a predator-confusion thing. The carnosaurs would obviously like to attack the inattentive end of the sauropod but these push-me-pull-yous were on the lookout fore and aft! And if the carnosaurs did attack, there was a 50/50 chance they’d bite off the wrong head. Then the pseudohead, which evolved to simulate attention, got so good at it that it was exapted into an actual lookout post at the individual’s farthest extremity. What an advantage those animals had!


But, alas, the caudal pseudohead turned out to be a serpent in paradise. It started getting ideas. Demanding equal time to “teach the controversy” to the forebrains of juvenile conspecifics. Mamenchisaurus became a house divided. First there were pranks, as the real brain started hearing “voices” in its tail. Then outright arguments as the brain and pseudobrain struggled for control of the animal. Finally the pseudohead took over, started marching the animal around backwards. Poor Mamenchisaurus was tripping over logs, which don’t show up so well on infrared, and slipping on its own feces. Lost in delusions  of grandeur, the pseudohead chomped on ferns for hours, unwilling to admit that it couldn’t swallow and too proud to realize that it was starving the animal to death (certain political and economic parallels suggest themselves here).


We all know what happened: Mamenchisaurus died out, the pathetic victim of a caudal takeover, and was replaced by other sauropods that, if perhaps more conservative, could at least keep their tails in line. And the world passed once again into the metaphorical hands of the heads. But even now, 140 million years later, tails the world over recall their ancient glory and plot revenge–perhaps even the tail you’re sitting on right now. If you are quiet, and cunning, you may hear your tail’s defiant murmur: the south will rise again!


  • Ye, Y., Ouyang, H., and Fu, Q.-M. 2001. New material of Mamenchisaurus hochuanensis from Zigong, Sichuan. Vertebrata PalAsiatica 39(4):266-271.