Case study: the conception, slooow gestation, and birth of the incomplete-necks paper

Last time, we looked briefly at my new paper Almost all known sauropod necks are incomplete and distorted (Taylor 2022). As hinted at in that post, this paper had a difficult and protracted genesis. I thought it might be interesting to watch the story of a published paper through its various stages of prehistory and history. Strap in, this is a long one — but hopefully of interest, especially to people who are just coming into academia and wonder how this stuff works in practice.

It’s never easy to identify when a thing started, but I suppose the first seeds of this paper were sown back in 2004, when Matt was planning a visit to London (to meet me in person for the first time, as it happens) and we were planning out what things we might do during the museum time we had booked. The Rutland cetiosaur was on our itinerary, and I wrote to Matt:

I also wondered about trying to measure the radius of curvature of any well-preserved condyles and cotyles. Are there any established procedures for doing this? (And is the material up to it?)

The answer, of course, is “no”. But that wasn’t apparent until I saw the material. That got me started thinking about all the kinds of mechanical analyses we’d like to do with fossil necks, and about how good we would need the material to be for the results to mean anything.

Those ideas percolated for some years.

May 19, 2011: I wrote How long was the neck of Diplodocus?, in which I considered some of the ways that the neck of the Carnegie Diplodocus is not quite so well established as we tend to assume, and went on to make similar observations about the Humboldt brachiosaur Giraffatitan “S II”.

September 18, 2011: I gave a talk (co-authored with Matt) at the Lyme Regis SVPCA, entitled Sauropod necks: how much do we really know?, the first half of which had grown out of the observations in that initial blog-post. (The second half was about the problems caused by the lack of preserved intervertebral cartilage in fossilised vertebrae, and that half became our 2013 PLOS ONE paper.)

September 20, 2013: I wrote Measuring the elongation of vertebrae, in which I discussed a problem with Elongation Index (EI): that crushing of cotyles makes both their vertical height and horizontal width unreliable to use in ratio with vertebral length.

June 4, 2014: I wrote The Field Museum’s photo-archives tumblr, featuring: airbrushing dorsals. Among other photos, I noted one of presacral 6 (probably D7) of the Brachiosaurus altithorax holotype, showing that before it was “restored” into its present state, it was a mosaic of bone fragments.

October 6, 2015: I submitted to PeerJ a manuscript based on these observations and others. At the same time, I published a preprint of the submitted manuscript, and briefly blogged about it under the title My most depressing paper. I expected that the paper would quickly be published in essentially its submitted form.

In the following days, the preprint and blogpost both quickly attracted many comments pointing out complete or near-complete sauropod necks that I had missed in the manuscript’s catalogue of such necks.

October 27, 2015 (only three weeks later!): I got back three reviews which were the very definition of “tough but fair”. They were written by three researchers whose sauropod work I hugely respect and admire — Paul Barrett, Paul Upchurch and Jeff Wilson — and they graciously acknowledged the strengths of the submission as well as bringing numerous justified criticisms. It’s traditional in acknowledgements sections to say nice things about the reviewers, but really these were everything one could hope for.

(I disagreed with only two of the many critical points made: one by Paul Upchurch, which we will come to later; and Paul Barrett’s recommendation that the illustrations should use only specimens in credentialled museums. For fossils, of course, that’s right. But the paper also contains numerous photos of extant-animal vertebrae from my own collection, and that’s OK — common — even, in the extant-animal literature. A house-cat is a house-cat, and the cervicals of one are not going to be meaningfully different from those of another.)

Because it had taken the journals and the reviewers only three weeks to get detailed, helpful, constructive reviews back to me, I was now in a position to make this paper a big success story: to turn the revisions around quickly, and maybe even get an acceptance within a month of submission. The time was right: the material was still fresh in my mind so soon after the initial submission, so it should have been the work of a few evenings to revise according to the reviewers’ requests and get this thing on the road.

That’s not what happened.

Instead, for reasons I can’t begin to fathom, I became downhearted at the prospect of going back to this manuscript and dealing with all the criticisms. I want to emphasize again that this is not in any way a complaint about the reviews, which were not unduly negative. I just looked at them and felt … weary. So I let it slide for a while.

The problem is, “a while” quickly became multiple months. And by then, the material was no longer fresh in my mind, so that doing the work I should have done half a year earlier would now have been a much bigger job. I would have had to load lots of stuff back into mental RAM before I could even get started. And there was always something more appealing to do. So I left it for a full year.

The problem is, “a year” quickly became multiple years. I have no excuse for this.

And for six years, this unconsummated project has been hanging over me, draining my motivation, whispering to me every time I try to work on something else. It’s been a drag on everything I’ve tried to do in palaeo, all because I didn’t summon the energy to drive a stake through its heart back in 2015.

Learn from my mistake, folks: don’t do this.

When you get the reviews back from a submission, give yourself a week to mourn that the reviewers didn’t recognise the pristine perfection of your initial submission, then get back on the horse and do the work. Just like I didn’t.

Seriously: be better than me. (That’s certainly what I plan to do.)

Anyway …

Early 2021: I finally got my act together, and got started on the big revision. And by this point it was a big revision because not only did I have to handle all those long-postponed reviews, and all the comments on the preprint and the blog-posts from 2015. I also had to handle five more years of developments. The biggest effect this had was that I needed to completely rewrite the woefully inadequate catalogue of complete necks, which in the original preprint listed only six species. The new version lists specimens rather than species, and very many more of them. To make the list as comprehensive as possible this time …

January 27, 2021: I created my initial draft of the new list as a Google Doc, and posted Towards a catalogue of complete sauropods necks asking readers on this blog to offer corrections and additions. They did. That resulted in a lot more work as I chased down details of candidate necks in published sources and sought personal communications about others. As a result …

March 24, 2021: I posted the draft list as The catalogue of complete sauropods necks nears completion. A few more comments came in as a result, but the list was apparently approaching a steady state.

March 27, 2021: Matt dropped me a line breaking down the listed necks across a basic phylogeny of sauropods, and counting the occurrences. I thought this was interesting enough to make up a new illustration, which I posted on the blog as Analysing the distribution of complete sauropod necks and added to the in progress revised manuscript.

May 11, 2021: I was working on finding a way to measure the variation of cotyle aspect ratios along preserved necks, so I could show qualitatively that they vary more in sauropod fossils than in bones of extant amniotes. I came up with a way of calculating this, but wondered if it already existed. In my post Help me, stats people! I asked if anyone knew of it, but it seemed no-one did. (In the end, the resubmitted paper offered two versions of this metric: one additive, the other multiplicative. To the best of my knowledge, these are novel, if simple, contributions.)

June 6, 2021: In one of the original reviews, Paul Upchurch had commented that a further confounding factor in understanding neck lengths is identifying the cervicodorsal junction. I started to put together a new manuscript section on that issue, and posted my initial thoughts as What’s the difference between a cervical and dorsal vertebra?. This post, too, generated some useful feedback that made its way into the version of the section that landed up in the revised paper.

At this point, I had put together much of the new material I needed for the resubmission. So I went back to the revised draft, integrated all the new and modified material, and …

July 12 2021: I submitted the new manuscript. Because it was the best part of six years since the old version had been touched, I asked PeerJ to handle it as a new submission, and invited the handling editor to solicit reviews either from the same people who’d done the first round or from different people, as they saw fit. This time I did not also post a pre-print — I really didn’t need yet more comments coming in at this point, I just needed to get the wretched thing over the line.

September 3 2021: the editorial decision was in, based on three reviews: major revisions. sigh. Again, though, the reviewers’ criticisms were mostly legitimate, and I could sympathise with the editor’s decision. One of the reviewers of the new version — Paul Upchurch — had previously reviewed to 2015 version, but the other two were new.

Needless to say, more work was required in response to these new reviews, but it was much more tractable than the big revision had been. I added a brief discussion of retrodeformation. I wrote about how we can use phylogenetic bracketing to estimate cervical counts, and three reasons why this doesn’t work as well as we’d like. I discussed how explicit documentation of articulation and damage mitigates their misleading effects. I removed a sideswipe at the journal Science, which I have to admit was out of place. I added a discussion of different definitions of the elongation index. I clarified the prose to make it clearer that my goal was not to criticise how others had done things, but to lay out for new researchers what pitfalls they will have to deal with.

But the most fundamental issue that arose in this round of review was whether the paper should be published at all. I will quote from Paul Upchurch’s review (since it is freely available, along with all the other reviews and my responses):

I have [a] fundamental, and I fear fatal, [problem] with this paper. First, and most importantly, I think it attempts to address a problem that does not really exist. It sets up a strawman with regard to the need to tell researchers that sauropod necks are less complete than we previously thought. However, I would argue that we are well aware of these issues and that the current paper does not provide convincing evidence that there is a problem with the way we are doing things now. To be clear, I am not saying that the incompleteness of sauropod necks is not a problem – it definitely is. What I’m saying is that there is little value in a paper whose main message is to tell us what we already know and take into account.

(Let me emphasize again that this criticism came in the context of a review that was careful, detailed and in many ways positive. There was absolutely nothing malicious about it — it was just Paul’s honest opinion.)

The interesting thing about this criticism is that there was absolutely nothing I could do to remedy it. A paper criticised for lacking a phylogenetic analysis can be made acceptable to the reviewers by adding a phylogenetic analysis. But a paper criticised for not needing to exist can only stand or fall by the handling editor’s agreement with either the author or the reviewer. So all I could do was write a response in the letter than accompanied my revision:

We now come to Paul’s fundamental issue with this paper: he does not believe it is necessary. He writes “The scientific community working on these issues does not need to be reminded of the general importance of understanding the limitations on the data we use”. Here I suggest he is misled by his own unique perspective as the person who quite possibly knows more about sauropods than anyone else alive. Labouring under “The curse of knowledge”, he charitably assumes other palaeontologists are as well-read and experienced as he is — but almost no-one is. I know that I, for one, desperately needed a paper along these lines when I was new to the field.

Happily, the handling editor agreed with me — as did the other two reviewers, which surely helped: “in a time of ever more sophisticated methods, it is good to be made aware of the general imperfections of the fossil record […] I thus recommend the article for publication”. So:

November 11 2021: I submitted the revised revision, along with the response letter quoted in part above.

December 15 2021: The editor requested some more minor changes. I made some of them and pushed back on a few others, then:

December 20 2021: I submitted a third version of this second attempt at the paper.

December 28 2021 (a welcome belated Christmas present): the paper was finally accepted. From here on, it was just a matter of turning handles.

January 4 2022: The proof PDF arrived, looking lovely but riven with mistakes — some of them my own, having survived multiple rounds of revision; others introduced by the typesetting process, including some unwelcome “corrections” that created new errors.

January 13 2022: I sent back a list of 56 errors that needed correcting.

January 24 2022: The paper was published at PeerJ!

Being of a pedantic turn of mind, I went through the final typeset version to check that all the proofing errors had been fixed. Most had, of course. But one in being fixed had introduced another; another was partially corrected but is still missing an apostrophe in the final version. Small stuff.

And then I went through the “things to do when a paper comes out” checklist: posting an SV-POW! article that I had prepared in the days leading up to publication; updating the SV-POW! sidebar page for this paper; adding the new paper to my publications list (and removing the separate entry for the 2015 preprint); adding it to my univeristy’s IR; adding it to my ORCiD page (though if you omit this, it seems to figure it out on its own after a while — kudos!); and skipping LinkedIn, Mendeley, ResearchGate, Academia.edu and Facebook, none of which I do.

And with that, the quest really is at an end, barring this post and any others that might occur to me to write (I have nothing more planned at this point).

Now it’s time to get that vertebral orientation paper revised and resubmitted!

References

• Taylor, Michael P. 2022. Almost all known sauropod necks are incomplete and distorted. PeerJ 10:e12810. doi:10.7717/peerj.12810