Argentinosaurus may actually still be the biggest.

I'd been meaning to get around to this for a while...

We all know Argentinosaurus is woefully incomplete. But for the first time we can get a mostly solid ides of what it looked like. For a long time, most Argentinosaurus reconstructions had been either purely speculative (i.e. Greg Paul - though he wasn't too far off the mark given the data available in the 1990s) or based on "cloning" the body of a distant relative (Ken Carpenter most notoriously used Saltasaurus, which as a low-grazing dwarf species, is among the worst models for restoring any fragmentary titanosaur over 20m).

More recently some speculative skeletals have cloned Malawisaurus and scaled it up. While a better animal to use for Argentinosaurus gap-filler than Saltasaurus ever was, seeing as it's so much closer to the basal position of Argentinosaurus, Malawisaurus still has the problem of being a relatively small titanosaur with only a moderately long neck by sauropod standards - basically titanosauria's answer to Camarasaurus. Additionally, the extremely short-faced skull of Malawisaurus - whose cheekbones and surangulars appear to make up nearly half its length - appears to be a unique feature of this species. The oddly serpentine look of its rear jaws has no parallel in other titanosaurs, most of which have the jaw joint far forward of the braincase, giving the impression of a Habsburg jaw.

Fortunately, the recent discovery of Patagotitan, and the less recent one of Rukwatitan (which few have taken notice of, despite it probably being Argentinosaurus' closest known relative) have cleared up a lot of confusion. But until now nobody had actually used them to reconstruct the missing parts of Argentinosaurus... yes, until now.

While doing just that with the Mk-II skeletal, I realized one shocking thing. When people like Mazzetta and Carpenter had reduced the estimates for Argentinosaurus size below 30m, they had gotten it dead wrong. Totally off the mark, in fact. It turned out even my own estimate of 33m was low. If Argentinosaurus had neck proportions anything like Rukwatitan (which contains both lower cervicals and a partial pelvis, which can be cross-scaled with that of Argentinosaurus and is nearly identical in shape), this means Argentinosaurus had a bigger neck than any previous silhouette had ever shown - a neck approaching euhelopodid proportions. If you scale the Patagotitan cervical to fit into the series, and scale up Malawisaurus only for the gaps that are left, you end up with one of the longest necks for any sauropod - and that's assuming it had only 14 cervicals, which is only average for a titanosaur. Some derived titanosaurs like Rapetosaurus had as many as 17 cervicals, while some basal ones (which are not known from complete necks) may have maintained the Euhelopodid tradition of having 17 or even 18 of them. Even if it had just 14 of them, Argentinosaurus would have had a roughly 16m (53 ft.) long neck! Proportionally the euhelopodids and mamenchisaurs still have more extreme necks, but in terms of raw length, Argentinosaurus is right up there. Not only that, but if its tail proportions were anything like Rukwatitan (again, relative to the hips), it would have exceeded 40m long. As it stands, because of the small centra in the rear sacrum of Argentinosaurus, I scaled down the caudals to about 85% of what they would have been with exact Rukwatitan proportions (using Rukwatitan caudals as the basis for the tail and Malawisaurus caudals to fill whatever gaps were left). The skeletal still comes out at 37m.

This tells us a few things.
 
First, Argentinosaurus turns out to be considerably bigger than I previously restored it. 4m longer and at least 10 tons heavier.
Second, with the largest of the Alamosaurus specimens being around 30-33m long based on Scott Hartman's restoration (which I don't disagree with too much), Argentinosaurus is considerably longer. Though it probably was not heavier, due to the likely shape of its rib cage based on the likely narrower dorsals, the difference is still well within the margin of error with such incomplete giants.
Third, Puertasaurus may not outclass it either, as the only Puertasaurus cervical found is around 1.2m long, whereas the longest cervicals of Argentinosaurus would be around 1.8m long with the new Rukwatitan-like neck proportions.
Fourth, even Patagotitan probably was not as big. Its cervicals indicate similar neck proportions, but overall even the largest Patagotitan specimens don't outclass Argentinosaurus in femur length, dorsal vertebra dimensions, or any other measurement I've been able to find. Thanks for getting our hopes up with that forklift pallet photo, Pablo Puerta.
Fifth, Ruyangosaurus may have rivaled it for neck length (the longest Ruyangosaurus cervicals approach 1.5m) but definitely not for body mass (Lu et al. 2014, which I did not have access to when working on our paper, shows that the dorsal column of Ruyangosaurus is considerably shorter, and even with a wider rib cage, it's hard to imagine this animal topping 60 tons - contrary to my own reconstruction, which will need heavy revision).
Sixth... it appears that Paralititan, Dreadnoughtus, the FMNH argyrosaurs, "Antarctosaurus" giganteus and even Notocolossus don't come close.
Seventh, Rukwatitan indicates that "argentinosaurids" already had crazy-long necks even at smaller sizes, and so likely filled in a different niche than "malawisaurids". While the two groups have roughly similar shapes to their cervical vertebrae, the cervical series in argentinosaurids is considerably larger relative to the hips (and thus the body overall) than in malawisaurids.
Eighth, Rukwatitan also shows that artists should NEVER overlook a related species when restoring an incomplete giant - even if, like Rukwatitan, the relative doesn't get a whole lot of press coverage. This one animal completely changes our perspective on Argentinosaurus, even though it's far from complete itself. Every additional bone is valuable data, and far more useful for a reconstruction than simply cloning and scaling up a more complete and well-known but more distant relative (i.e. Malawisaurus).
Ninth... disturbingly, this means every previous restoration of Argentinosaurus (including my old one) is wrong. My Ruyangosaurus is also wrong. And in fact, just about every giant titanosaur mount (other than the Perot Museum's Alamosaurus) is wrong too.
Tenth, if you're in hardcore TL;DR mode, all of this means that Argentinosaurus is still the king.