Spinosaurus aegyptiacus is one of many people’s favorite theropods, but it is one that few people know a whole lot about. New discoveries over the last ten or so years have hinted at the potential that we may know more about this peculiar theropod dinosaur soon, but as it stands, what you see above is what we have at current. This is especially important because all that we have left is von Stromer’s original description  and the plates from which I cropped and modified the above image.
Astute observers will note that I’ve arranged the material differently than von Stromer did (you can see two sets of labels below the illustration, A referring to von Stromer’s original designations, intended to correpsond to his perceived series, and B referring to my personal argument.
A quick note: Andrea Cau of Theropoda has not only predated me on the placement of element “i” as a caudal, but in the separation of element “i” centrum as an anterior (or mid) dorsal. Now I feel a little flabbergasted. Bravo, Andrea!
At the time I created this, I used a variety of criteria to imply identity to the spines, which are listed below, after noting that von Stromer did not always associate vertebral centra (some unfused to their respective neural arches) and also that some were associated in part based on similar size, closeness of fit, or closeness as found in the ground. After splitting all material that wasn’t fused, I applied the following formulae:
1) Central anterior convexity. Centra would be arranged from front (first cervical, not preserved) to back (last dorsal, probably not preserved) based on the convexity of the anterior ball. The last preserved dorsal centrum is relatively flat, but has a high, shallow convexity, while the anteriormost preserved centrum (the fourth or fifth cervical) has an extremely semi-spherical “ball.”
2) The position of the parapophysis on the centrum or the neural arch (only in dorsals). In the theropod cervical-dorsal transition, the parapophysis migrates from a position at the ventral end of the centrum up to the dorsal end of the centrum, crosses the neurocentral suture, and makes its way up the neural arch. In posterior dorsals, the parapophysis has joined the diapophysis on the transverse process. Thus, the relative position of the parapophysis should indicate the relative position of a centrum (if its present) or a neural arch (same). If either lack such, and it’s relatively clear the element is not eorded (which is likely according to Margraff and von Stromer [1,2] to have occured to several elements as the skeleton was partially exposed prior to recovery).
3) The relative position of the transverse process. In cervicals, the tramsverse process is low and dependant (“hangs”) such that the distal end points ventrally; the orientation of the this process becomes horizontal or nearly so in the cervico-dorsal transition, and elevated or above the horizontal in the posterior dorsals and anterior caudals. Thus, this orientation, when relatively secure, should indicate the position of the neural arch in the series. However, much of the original material was distorted , rendering the skeleton asymmetrical, and often warped (several neural spines had a wavy or sinusoidal margin, and the transverse processes were variably defelcted dorsally).
4) Shape of the centrum in anterior view. When viewed from the front, the centra of most theropod vertebrae are nariably ovate: They transition from horizontal ovals (broader than deep) in the cervicals, to nearly circular in the cervico-dorsal transition, to vertical ovals (deeper than broad) in the middle to posterior dorsals. As for 3, this conditio0n prone to distortion, and the condition is given limited weight, but it seems that what is there suggests that the fossils agree with this argument.
5) Extent of the neurocentral laminae. While sauropods are infinitely more famous for the complexity of their laminae , the vertebral laminae in theropods is no less distinctive, if less complex generally. For theropod dorsals, the diapophyseal and parapophyseal laminae connecting to the zygapophyses and centra are the most distinctive, and they are no less so in Spinosaurus aegyptiacus. For the most part, this system is graded relative to the extent to which the laminae extend outward from the transeverse processes, which are regarded as the laminar “hubs” for the purpose of simplicity. The relative extent should decrease as the vertebrae progress posteriorly, as the parapophyses progress above the centra and onto the transverse processes; absence of the parapophysis, by merging with the diapophysis as expected in the posterior dorsals, sacrals, and in the caudals, should reflect loss of laminae.
6) Relative height of the neural spines. Based on comparison to most other theropods, dorsal neural spines generally increase posteriorly in the dorsal series and are highest above the hips. Thus the talled spines should be the sacrals (unless otherwise indicated). Not all of the spines are complete; in fact, only the tallest and the first few “dorsal” vetrebral spines seem to be complete, while all other exhibit missing material, even when it seems (as in element “f”) the distal end is preserved. Here, a proxy for length is the anteroposterior (craniocaudal) length of the blade at its greatest, normally very distal; the further distal this greater diameter, while being grossly larger than other spines, the further posteriorly in the series the spine may be located.
7) Relative size and extend of the basal neural spine “expansions.” Several neural spines have a protrusion on the anterior, posterior, or both sides of the base of the neural spine above the zygapophyses. This feature is unique to Spinosaurus aegyptiacus, so comparisons to other taxa fail; however, it is argued here that extent of the protrusion should increase posteriorly. There is more of an anterior protrusion than a posterior protrusion in most vertebrae, and this is considered supplemental to the premise.
8) Relative orientation of the neural spines on the neural arches. Some of the smallest neural spines are inclined cranially, as in Acrocanthosaurus atokensis , and these vertebrae are also the most anterior of the dorsal series; the tallest neural spine is also the only caudally inclined spine. It is thus assumed that the spines should form a radial pattern, with anteriorly-inclined elements in the anterior vertebra and posteriorly-inclined elements in the posterior vertebrae.
9) Orientation of the prezygapophyses. Although it is subjected to the same level of distortion as in 3 and 4, the orientation of the prezygapophyses are generally inclined dorsally in the cervicals and anterior dorsals, and become essentially horizontal in middle and posterior dorsals.
Based on the above, the series shown in the figure is presented. The identification of von Stromer’s “element i” as a caudal is novel; no other researchers has published this argument, and this is based on the apparent lack of parapophyses, the extent of the laminae, and the extreme caudal inclination of the spine itself. It can certainly be a posterior dorsal, although it is unlikely to be an anterior sacral based on the preserved sacral series, and the condition of the neural arch indicates it is not a posterior sacral.
Several of the centra can certainly be attached to adjacent dorsals, although as shown some of them are unlikely to go to von Stromer’s original identifications (“element i” centrum is extremely cervical-like, while the neural arch and spine are not), although the element is also lacking in features indicating parapophyseal development, laminae, etc, and could very well be a caudal centrum with an anterior convexity.
Some of the variables (3, 4, 8 & 9) used to grade the vertebrae are distortion-dependant, making them less useful, but this leaves five of the nine criteria with higher degrees of confidence left. A caudal vertebra (von Stromer’s “element k”) does not bear a large resemblance to typical theropod vertebrae, as it is highly anteroposteriorly compressed, being broader and higher than it is craniocaudally long; it may not be theropodan. However, it is also hollow , indicating the presence of pneumatic chambers within the centrum, a feature absent in non-saurischian dinosaurs, but present only in derived theropods and sauropods; it differs from sauropods in having transverse processes dorsoventrally flattened, elongated, and blade-like. It is probably a middle caudal (around the 20th to the 25th).
 von Stromer, E. 1915. Wirbeltier-Reste der baharije-Stufe (unterstes Cenoman).3. Das Original des Theropoden Spinosaurus aegyptiacus nov. gen. et nov. spec. [Vertebrate remains of the Baharije Formation (lowest Cenomanian). 3. The original of the theropod Spinosaurus aegyptiacus nov. gen. et nov. spec.] Abhandlungen der Königlichen Bayerischen Akademie der Wissenschaften Mathematisch-physikalische Klasse Abhandlung 28:1–32.
 Nothdruft, W. & Smith, J. 2002. The Lost Dinosaurs of Egypt. (Random House, New York).
 Wilson, J. A. 1999. A nomenclature for vertebral laminae in sauropods and other saurischian dinosaurs. Journal of Vertebrate Paleontology 19(4):639-653.
 Stovall, J. W. & Langston, W. 1950. Acrocanthosaurus atokensis, a new genus and species of Lower Cretaceous Theropoda from Oklahoma. American Midland Naturalist 43:696–728.