The only other skeleton was bombed into dust in 1944
Spinosaurus was named from bones that no longer exist. Richard Markgraf, a fossil collector working for Ernst Stromer von Reichenbach, dug the original material out of Cenomanian-age rock at the Bahariya Oasis in Egypt in 1912: part of a lower jaw with a squared-off tip and unusually round, widely spaced, unserrated teeth, along with neural spines from the back, hip and tail region, some of them close to 2 metres long. Stromer described it in 1915 as Spinosaurus aegyptiacus, "Egyptian spine lizard," and the sail became one of the most recognisable silhouettes in palaeontology despite the fact that almost nothing else about the animal was known.
The specimen was catalogued as BSP 1912 VIII 19 and kept in Munich's Bavarian State Collection of Palaeontology. Stromer, an outspoken opponent of the Nazi Party, repeatedly asked the collection's director to move the fossils out of the city as Allied bombing intensified. According to the Linda Hall Library's account of his life, the director, "a fervent Nazi," adhered to the party line that the Luftwaffe would protect German cities from attack, and refused to relocate the collection. On the night of 24–25 April 1944, a Royal Air Force raid hit the museum directly. The Spinosaurus holotype was destroyed along with Stromer's other Bahariya specimens, including the only known material of Aegyptosaurus, Bahariasaurus and much of his Carcharodontosaurus.
The war cost Stromer more than his fossils. All three of his sons were conscripted to the Eastern Front; two were killed, in 1941 and 1944, and the third did not return from Soviet captivity until 1951. Stromer died the following year, having outlived both the specimens and most of the family that might have carried on his work. What survived were his own notes, measurements, drawings and photographs, which is why every reconstruction of the original holotype since 1944, including the plate reproduced below, is a copy of a copy: nobody alive has seen the actual bones.
A new skeleton from Morocco, and a decade of doubt about whether it was one animal
For nearly a century after Stromer's holotype was lost, Spinosaurus was reconstructed mostly by borrowing proportions from its relatives, particularly the African spinosaurid Suchomimus. That changed in 2014, when Nizar Ibrahim and colleagues published a new partial skeleton in Science, designated FSAC-KK 11888 and recovered from the Kem Kem beds of Morocco. The paper argued for a genuinely strange animal: nostrils retracted toward the middle of the skull, a trunk and neck long enough to shift the center of mass forward, a reduced pelvis, unusually short hindlimbs, and solid limb bones lacking the open medullary cavity most land dinosaurs have. Their conclusion was that Spinosaurus was not just semiaquatic but possibly a facultative quadruped, an interpretation that broke sharply with a century of bipedal restorations.
The skeleton immediately drew a specific kind of skepticism. The Kem Kem beds are worked by local commercial collectors who sell loose bones into the fossil trade rather than excavating articulated skeletons, and multiple spinosaurid species are known from the same formation. Critics questioned whether FSAC-KK 11888 was really one individual rather than a chimera of bones from different animals, possibly different species, assembled after the fact. The doubt was reasonable: essentially the entire basis for a radically new body plan rested on a skeleton bought in pieces.
The case for a single individual came from bone histology rather than field notes. Papers examining the specimen's stratigraphy and internal bone structure in 2020 found that the vertebrae, ribs, gastralia and the diminutive pelvic elements all independently pointed to the same conclusion: a subadult animal still growing, with immature bone texture running consistently across skeletal regions that would be hard to fake by combining unrelated individuals. That did not settle every question about the reconstruction's proportions, but it took the chimera objection off the table as the leading explanation for why the skeleton looked so unusual.
A tail tested in a flume like a machine part
The most direct evidence for Spinosaurus in water came six years later, from a part of the skeleton nobody had. Ibrahim and colleagues reported a nearly complete tail in Nature in 2020, roughly 80 percent of the caudal series, and it turned out to carry tall neural spines and long chevrons that together built a deep, flexible, fin-like structure capable of a wide side-to-side sweep, unlike the stiff, narrow tail of a typical theropod.
Rather than argue from shape alone, the team, working with Harvard biomechanists Stephanie Pierce and George Lauder, built physical models and tested them. Scaled cutouts of the Spinosaurus tail, plus comparison tails from the land-dwelling theropods Coelophysis and Allosaurus and the semiaquatic Nile crocodile and crested newt, were mounted on a robotic flapping rig and driven through a water flume at speeds meant to simulate slow, steady swimming. The Spinosaurus shape generated over eight times the thrust of the other theropod tails and more than 2.5 times their propulsive efficiency, performing close to the crocodile and newt rather than the land-dwelling dinosaurs. It was the first physical, testable evidence that a non-avian dinosaur's tail was actually built for pushing a body through water rather than just balancing one on land.
Bone density says diver, a flesh model says it would have drowned trying
The tail paper made the case for propulsion; it did not settle how the whole animal behaved in water, and the next two years produced directly opposed answers from two different lines of evidence. In March 2022, Matteo Fabbri and colleagues published a survey in Nature of femur and rib bone compactness across 291 living and extinct species, land-dwellers and water-dwellers alike, establishing that animals which regularly submerge to forage tend to have nearly solid bone through the cross-section, while land animals show the hollow, doughnut-like profile typical of terrestrial locomotion. Spinosaurus and its relative Baryonyx both showed the dense, near-solid pattern associated with diving; Suchomimus, despite eating fish and belonging to the same family, showed the hollower, more terrestrial pattern. Ibrahim, a co-author on the study, called the Suchomimus result "a bit of a surprise, because Baryonyx and Suchomimus look rather similar." The result implied real ecological differences within a single dinosaur family, with Spinosaurus and Baryonyx as genuine subaqueous foragers.
Eight months later, Paul Sereno, Nathan Myhrvold and seven co-authors published the opposite conclusion in eLife, working from an entirely different method. Instead of bone density, they built a CT-based digital skeletal reconstruction of both Spinosaurus and Suchomimus and added a posable flesh model with internal air spaces and muscle, then tested its physical behavior in water. Their center-of-mass calculation placed it over the hind feet in a bipedal stance, not shifted forward into the trunk as the 2014 paper had proposed. In deep water the model was unstable, tending toward a stable equilibrium floating on its side, and righting itself would require roughly 5,000 newton-metres of torque, more than the animal could plausibly generate. Its estimated top swimming speed came out to somewhere between 0.8 and 1.4 metres per second, well under a tenth of the 10 to 33 metres per second real pursuit predators like dolphins reach. Diving, they calculated, would take about 17,000 newtons of propulsive force, roughly 25 times the maximum force the 2020 flume tests had measured from the tail. Their alternative was a bipedal, wading ambush predator that hunted large fish from the margins of shallow coastal and inland waterways rather than pursuing them underwater.
A 2024 rebuttal: is bone density even the right test
The disagreement did not end with two competing papers reaching opposite conclusions from different data; it extended into a direct challenge to the 2022 bone-density method itself. In 2024, Myhrvold and colleagues published a methodological critique in PLOS ONE, "Diving dinosaurs? Caveats on the use of bone compactness and pFDA for inferring lifestyle," arguing that the statistical technique behind the Fabbri result, a phylogenetic flexible discriminant analysis trained on bone compactness, does not reliably separate divers from non-divers in the first place. Their central example: elephants, animals with no meaningful aquatic lifestyle, show bone density overlapping with confirmed divers in the same dataset, and the paper reports broader overlap between the compactness of known diving and non-diving groups generally, plus small sample sizes within some of the comparison categories that widen the uncertainty further. If the underlying statistical method cannot cleanly separate a known non-diver like an elephant from a known diver, the argument goes, it cannot be trusted to classify an extinct animal that nobody can watch enter the water.
Fabbri's original paper is not undefended; the bone-density link between compactness and diving behavior draws on an established pattern in living birds, and the 2022 result was peer-reviewed in Nature rather than a preprint. What the 2024 paper establishes is narrower and more procedural: that the statistical confidence the original study placed in its classification may be overstated, which is a different claim from showing Spinosaurus definitely did not dive. As of this writing, no single study has reconciled the tail-propulsion evidence, the bone-density evidence and the buoyancy-and-swim-speed evidence into one settled picture, and the dinosaur most people can picture instantly, sail and all, remains one of the least settled questions in vertebrate palaeontology.
How big was it? Depends which decade you ask
Spinosaurus is routinely described as the longest known carnivorous dinosaur, longer even than Tyrannosaurus rex, but the number behind that claim has moved substantially over the past two decades. Cristiano Dal Sasso and colleagues, working in 2005 from an isolated skull and assuming Spinosaurus shared body proportions with the better-known Suchomimus, estimated a length of 16 to 18 metres and a mass of 7 to 9 tonnes. Two years later, François Therrien and Donald Henderson applied square-cube scaling to similar comparisons and reached figures running up to 20 tonnes, a number that circulated widely in documentaries and popular accounts for years afterward.
The 2014 and 2020 papers, built on an actual (if partial) skeleton rather than proportional guesswork, pushed the consensus mass back down considerably, with most recent estimates clustering around 6.4 to 7.5 tonnes and a length nearer 13 to 15 metres. Nathan Myhrvold, presenting preliminary results at PALEODAYS in Pisa in 2024 using 3D-sculpted flesh models over CT-scanned bone, arrived independently at a similar figure of roughly 7 tonnes. The pattern, an early estimate inflated by proportional guesswork and later walked back once real material and volumetric modelling arrived, is close to what happened with the long-inflated size record of Mosasaurus, and it means Spinosaurus's claim to the length record is on firmer ground today than its claim to the weight record: even at 7 tonnes it is lighter than the largest known T. rex individuals, and length, not mass, is what modern estimates actually support.
That the length record survives revision better than the mass record is itself a comment on how thin the primary evidence still is. As with the Ankylosaurus skeletons largely reconstructed from relatives rather than their own type material, most of what the public knows as "Spinosaurus" is inference built outward from a single subadult skeleton, a destroyed holotype, and a shrinking handful of isolated bones from a fossil trade that rarely preserves context.