In mammals — certainly the most-studied vertebrates — regional differentiation of the vertebral column is distinct and easy to spot. But things aren’t so simple with sauropods. We all know that the neck of any tetrapod is made up of cervical vertebrae, and that the trunk is made up of dorsal vertebrae (subdivided into thoracic and lumbar vertebrae in the case of mammals). But how do we tell whether a given verebra is a posterior cervical or an anterior dorsal?

Here two vertabrae: a dorsal vertebra (D3) and a cervical vertebra (C13) from CM 84, the holotype of Diplodocus carnegii, modified from Hatcher (1901: plates III and VII):

It’s easy to tell these apart, even when as here we have only lateral-view images: the dorsal vertebra is tall, its centrum is short, its neural spine is anteroposteriorly compressed and its parapophysis is up on the dorsal half of the centrum; but the cervical vertebra is relatively low, its centrum is elongated, its neural spine is roughly triangular and its parapophysis hangs down well below the centrum (and has a cervical rib fused to it and the diapophysis).

But things get trickier in the shoulder region because, in sauropods at least, the transition through the last few cervicals to the first few dorsals is gradual — the vertebrae become shorter, taller and broader — and tends to have no very obvious break point. In this respect, they differ from mammals, in which the regional differentiation of the spinal column is more abrupt. (Although even here, things may not be as simple as generally assumed: for example, Gunji and Endo (2016) argued that the 1st thoracic vertebra of the giraffe behaves functionally like an 8th cervical.)

So here are those two vertebrae in context: the sequence D3 D2 D1 C15 C14 C13 in CM 84, the holotype of Diplodocus carnegii, modified from Hatcher (1901: plates III and VII):

Given that the leftmost is obviously a dorsal and the rightmost obviously a cervical, where would you place the break-point?

The most usual definition seems to be that the first dorsal vertebra is the first one that has a free rib, i.e. one not fused to the vertebra: in the illustration above, you can see that the three cervicals on the right all have their cervical ribs fused to their diapophyses and parapophyses, and the three dorsals on the left do not. This definition of the cervical/dorsal distinction seems to be widely assumed, but it is rarely explicitly asserted. (Does anyone know of a paper that lays it out for sauropods, or for dinosaurs more generally?)

But wait!

Hatcher (1903:8) — the same dude — in his Haplocanthosaurus monograph, writes:

The First Dorsal (Plate I., Fig. 1). […] That the vertebra now under consideration was a dorsal is conclusively shown not by the presence of tubercular and capitular rib facets showing that it supported on either side a free rib, for there are in our collections of sauropods, skeletons of other dinosaurs fully adult but, with the posterior cervical, bearing free cervical ribs articulating by both tubercular and capitular facets as do the ribs of the dorsal region. The character in this vertebra distinguishing it as a dorsal is the broadly expanded external border of the anterior branch of the horizontal lamina [i.e. what we would now call the centroprezygapophyseal lamina]. This element has been this modified in this and the succeeding dorsal, no doubt, as is known to be the case in Diplodocus to give greater surface for the attachment of the powerful muscles necessary for the support of the scapula.

Hatcher’s illustrations show this feature, though they don’t make it particularly obvious: here are the last two cervicals and the first dorsal, modified from Hatcher (1903:plate I), with the facet in question highlighted in pink: right lateral view at the top, then anterior, and finally posterior view at the bottom. (The facet is only visible in lateral and anterior views):

Taken at face value, Hatcher’s words here seem to imply that he considers the torso to begin where the scapula first lies alongside the vertebral column. Yet if you go back to the Diplodocus transition earlier in this post, a similar scapular facet is not apparent in the vertebra that he designated D1, and seems to be present only in D2.

Is this scapular-orientation based definition a widespread usage? Can anyone point me to other papers that use it?

Wilson (2002:226) mentions a genetic definition of the cervical/dorsal distinction

Vertebral segment identity may be controlled by a single Hox gene. The cervicodorsal transition in many tetrapods, for instance, appears to be defined by the expression boundary of the Hoxc-6 gene.

But this of course is no use in the case of extinct animals such as sauropods.

So what’s going on here? In 1964, United States Supreme Court Justice Potter Stewart, in describing his threshold test for obscenity, famously said “I shall not today attempt further to define the kinds of material I understand to be embraced within that shorthand description, and perhaps I could never succeed in intelligibly doing so. But I know it when I see it.” Is that all we have for the definition of what makes a vertebra cervicals as opposed to dorsal? We know it when we see it?

Help me out, folks! What should the test for cervical-vs-dorsal be?


  • Gunji, Mego, and Hideki Endo. 2016. Functional cervicothoracic boundary modified by anatomical shifts in the neck of giraffes. Royal Society Open Science 3:150604. doi:10.1098/rsos.150604
  • Hatcher, Jonathan B. 1901. Diplodocus (Marsh): its osteology, taxonomy and probable habits, with a restoration of the skeleton. Memoirs of the Carnegie Museum 1:1-63 and plates I-XIII.
  • Hatcher, J. B. 1903b. Osteology of Haplocanthosaurus with description of a new species, and remarks on the probable habits of the Sauropoda and the age and origin of the Atlantosaurus beds; additional remarks on Diplodocus. Memoirs of the Carnegie Museum 2:1-75 and plates I-VI.
  • Wilson, Jeffrey A. 2002. Sauropod dinosaur phylogeny: critique and cladistic analysis. Zoological Journal of the Linnean Society 136:217-276.

Various Internet rumours have suggested that the Archbishop is a super-giant sauropod one third larger than the mounted Giraffatitan specimen MB.R.2181 (formerly HMN SII). This is incorrect.

Figure E. Skeletal inventory of NHMUK PV R5937, “The Archbishop”, showing which bones were excavated by Migeod’ expedition. Based on a skeletal reconstruction of Giraffatitan brancai kindly provided by Scott Hartman: note that this image does not illustrate the shapes or proportions of the Archbishop material. Bones prepared and available for study are shown in white; those still in jackets awaiting preparation in light grey; those excavated by Migeod but apparently lost or destroyed in dark grey.

Migeod’s assessment of the size of the animal was based on the vertebrae: “The [neck] vertebrae found give a 20-foot [6.10 m] length […] The length of the back including the sacral region was about 15 feet [4.57 m]. The eight or nine caudal vertebrae cover about 6 feet [1.83 m]” (Migeod 1931a:90). This gives the total preserved length of the skeleton as 41 feet (12.50 m). By comparison, Janensch (1950b:102) gives lengths of portions of the mounted skeleton of MB.R.2181 as 8.78m (neck), 3.92m (torso) and 1.07m (sacrum) for a torso-plus-sacrum length of 4.99m. On this basis, the preserved neck of NHMUK PV R5937 is only 69% as long as that of MB.R.2181, but since the first four vertebrae were missing and omitted from Migeod’s measurement, this factor cannot be taken at face value. More informative is the torso-plus-sacrum length, which in NHMUK PV R5937 is 92% the length of MB.R.2181.

This is consonant with measurements of individual elements, which compare as follows:

Table 4. Comparative measurements of Archbishop and Giraffatitan elements

ElementMeasurement (cm)ArchbishopGiraffatitanRatio
Torso plus sacrumtotal length4574990.916
C10 (mC4)centrum length991000.990
C11 (mC3)centrum length104100[1]1.040
D4 (mD3)centrum length27360.750
Longest riblength over curve2352630.894
Left scapulocoracoidlength over curve221238[2]0.929
Right humeruslength1462130.685
Right humeruswidth51590.864
Right iliumlength98123[3]0.797
Right iliumheight7996[4]0.823

Archbishop measurements taken from Migeod (1931a) and converted from imperial; Giraffatitan measurements are for MB.R.2181 except where noted, and are taken from Janensch (1950a:44) and Janensch (1961).
[1] Janensch (1950a) did not report a total centrum length for C11, as its condyle had not been removed from the cotyle of C10; but since the length of its centrum omitting the condyle was, at 87 cm, identical to that of C10, it is reasonable to estimate its total length as also equal to that of C10.
[2] Janensch (1961:181) did not include measurements for the right scapula of MB.R.2181, which is incorporated into the mounted skeleton, but does give the proximodistal length of its right coracoid as 45 cm. Using the 193 cm length given for the similarly sized scapula Sa 9, we can deduce a reasonable total estimate of 238 cm for the scapulocoracoid.
[3] Estimated by Janensch (1950b:99) based on cross-scaling from the fibula and ilium of Find J from the Upper Saurian Marl.
[4] This is the measurement provided by Janensch (1961:199) for the ilium Ma 2, which is incorporated into the mounted skeleton, and which Janensch (1950b:99) considered to match MB.R.2181 very precisely.
[5] Based on a restoration of the midshaft which Janench (1950b:99) calcuated based on other finds.

Individual lines of this table should each be treated with caution: Migeod’s measurements may have been unreliable, and in any case are underspecified: for example, we do not know whether, when he gave a vertebra’s length, he included overhanging prezygapophyses or the condyle. Similarly, we know that Migeod (1931:96) wrote that a rib “was as much as 92.5 inches long”, but we do not know for certain that, like Janensch, he measured the length over the curve rather than the straight-line distance between the ends. And when Migeod says that the ilium “measured 38.5 by 31 inches” we do not know that the height was measured “at the public process”, as Janensch (1961:199) specified.

With those caveats in place, nevertheless, a picture emerges of a sauropod somewhat smaller than MB.R.2181, though by no means negligible. On average, the measurements come out about 15% smaller than those of Giraffatitan.

But this average conceals a great deal of variation. The cervical vertebrae are comparable in length to those of MB.R.2181 (The total of 203 cm for C10 and C11 in the Archbishop, only 1.5% longer than 200 cm for MB.R.2181, is a difference well within the margin of measurement error). The Archbishop’s scapulocoracoid may have been 93% as long as in MB.R.2181. But the limb bones are signficantly shorter (87% for the humerus and a scarcely credible 62% for the femur), and the humeri at least bseem to be have been proportionally more robust in the Archbishop: only 2.86 times as long as wide, whereas the ratio is 3.61 in MB.R.2181. If Migeod’s measurements can be trusted, we have here an animal whose neck is as long as that of Giraffatitan, but whose limbs are noticably shorter. These proportions corroborate the hypothesis that the Archbishop is not a specimen of Giraffatitan.