See that skeleton above? It’s a display at the Natural History Museum in Oulu, Finland, and it shows a domestic dog in mid-stride. The only problem?

That’s not how dogs walk. If you actually closely observe a dog — or any other quadruped — while it’s walking slowly, you see that they step with their left hind leg, followed by their left foreleg, then the right hind leg and the right foreleg. This gives them maximum “static stability”: They always have three paws on the ground, so they won’t trip or get easily knocked over. But that dog in the Finland museum is shoving its right front paw forward, followed by its back left paw. What gives?

It used to be that people argued to the point of fistfights over how quadruped legs moved. Then in the 1870s, the photographer Edward Muybridge began settling these fights by pioneering serial freeze-frame photography that revealed how horses and the like walked and galloped, and pretty soon naturalists had this stuff all figured out. But recently a team of biological physicists noticed that they were seeing errors in museum exhibits and taxidermy models, so they decided to see how commonplace these goofs were. They randomly gathered a representative sampling of 307 depictions of quadrupeds walking in museum exhibits, taxidermy catalogues, animal-anatomy books and toys. The result?

Museums screwed things up a stunning 41% of the time. Taxidermy catalogues got it wrong 43% of the time, toys 50% of the time, and animal-anatomy catalogues were the worst, with 63.6% errors. As the scientists dryly pointed out in their paper — which you can download free here …

This high error rate in walking illustrations in natural history museums and veterinary anatomy books is particularly unexpected in a time where high-speed cameras and the internet offer ideal possibilities to obtain reliable quantitative information about tetrapod walking.

So why exactly do we get dog-walking wrong so often? Dogs, after all, are kind of all over the place, and thus pretty easy to observe. But the fact is quadruped leg-motion isn’t intuitive: When you close your eyes and visualize it, it makes more sense for the legs to alternate steps left and right, much like the screwed-up skeleton above. What we see in our mind’s eye doesn’t match what we actually see in the world around us — so we ignore the evidence in front of our eyes. It’s kind of like how Aristotle maintained that men had more teeth than women because it made more sense to him, and never bothered to actually check inside an actual woman’s mouth.

Of course, given that we now actually do empirical science and stuff, it’s still pretty alarming that museums mess up quadrupedal gait almost as often as toy companies do. It’s possible, the scientists suggest, that modern media is partly to blame. When museum researchers and taxidermists do a quadruped exhibit, they probably just refer to existing illustrations and models, even when those are wrong — so the errors simply compound themselves. They don’t think to re-investigate the correct gait of dogs, because they presume this was settled a hundred years ago. It was! But even in science, errors of human psychology can creep back in.

Interesting coda: The scientists found that CGI movies like Jurassic Park buck this trend — they tend to correctly represent quadruped gaits.

Interesting coda, 2: I think I’ve broken my world record for “time between blogging posts.”