How Should Museums Organize Their Fossils?


Below ground on 77th Street and Central Park West in New York there is a door. It is a door that remains locked more often than not. Once unlocked, the door opens to paleontological wonders literally untold. All one needs is to find a way to unlock the door.

Sending an email and asking for permission is a really good start.

While visiting family in New York last week, I made time to visit the American Museum of Natural History. This trip was a fortuitous one for one of my labmates at the University of Oregon: he needed photographs of a particular specimen and that specimen just happened to be stored behind that door on 77th and Central Park West, in the AMNH. He and I therefore made arrangements to get me behind the scenes at the museum, where I learned about collections management systems—and how different museums reflect the theoretical virtues of the fields they represent.

A collections management system (abbreviated CMS) is a method for arranging the items that come into a collection. We all use these systems. Here’s the one I use for books in my office: science and philosophy of science books are alphabetized by author in the leftmost bookcase; general philosophy books are alphabetized by author in the center bookcase; all other books are alphabetized by author in the rightmost bookcase. You can tell from this arrangement how I think of academia: there’s science, there’s philosophy, and then there’s the stuff I toss into a corner for whenever I have time to get around to it. Similarly, the AMNH uses a CMS that reveals a particular way of thinking about paleontology and other sciences. As we’ll see, their way of thinking is perhaps a bit too idealized.

False cats and false names

The collections manager led me through the door and into an elevator. She explained that the AMNH divided its fossil mammals between seven floors and pushed the button for the fifth floor. When we got off the elevator she led me to a cabinet labeled “Nimravidae—Eusmilus,” unlocked the cabinet door, and pulled out a drawer. The specimen that my labmate wanted photographed laid there—funnily enough, next to an old loan card with the labmate’s name on it.

Nimravids are an extinct sister taxon to modern cats. The two are similar enough that nimravids sometimes take the colloquial name “false cats.” In fact, a good number of the animals popularly called “sabre-toothed cats” are nimravids. One such sabre-tooth was the domestic cat-sized Hoplophoneus cerebralis. My labmate specializes in the evolutionary history of nimravids; it was Hoplophoneus’ jaws that I had been dispatched to photograph.

Problem: the AMNH has no Hoplophoneus specimens catalogued in its collection.

The CMS in the AMNH is phylogeny-based. When searching through the museum’s extensive collections, locating a particular specimen—both in the catalogue and in the collections’ physical space—depends on your knowledge of where that specimen fits into the tree of life. Looking for a T. rex tooth will take you to a particular floor in one building; looking for Mongolian mammals unearthed by Roy Chapman Andrews will take you to a particular floor in another building; looking for nimravids will take you either a few floors up or a few floors down from Andrews’ mammals, depending on how those mammals are related to nimravids.

You’d have trouble with this CMS if you weren’t well versed in evolutionary history. That isn’t normally a problem: the AMNH probably won’t let you into its collections if you don’t already spend your time researching such things. The more significant problem—and the relevant one here—is what happens when the managers of the museum’s collections don’t know where a specimen fits on the tree of life.

The reason that the AMNH doesn’t have any catalogued Hoplophoneus specimens is that Hoplophoneus is a relatively new name. My labmate—the nimravid specialist—recently found that the nimravid taxon Eusmilus included a variety of specimens that should be split off into different genera and species, one of which became H. cerebralis. He therefore published a paper that rendered Eusmilus a false name for false cats. Nimravids aren’t an especially popular subject of paleontological study—that old loan card showed that this labmate had been the last visitor to the AMNH’s nimravid collection, back when he was researching the phylogenetic revision—and so no one at the AMNH had much reason to rearrange their collection. Finding the right specimen would have been really difficult if this labmate hadn’t already found it a few years ago.

This highlights a problem with a phylogeny-based CMS, which is a relatively common system in paleontological collections. Our understanding of evolutionary history changes in light of ongoing paleontological study, and so the specimens in a phylogeny-based would have to be rearranged with some frequency. Think about how often the bestseller shelf at your local bookstore changes; bestseller lists change on a weekly basis, and so a CMS based on those lists would require items to be rearranged with the same frequency. Now think about my labmate’s research, or—in a more extreme case—the recent phylogenetic revision of the Dinosauria. If a CMS is determined by phylogeny, as the AMNH’s collections are, then thousands of specimens might have to be rearranged with the publication of any given journal volume.

Speaking as someone with an ever-growing collection of natural history toys, I can say with some authority that having to rearrange items in a large collection is a pain. Why, then, would a museum use a CMS that requires frequent rearrangement of specimens?

I can’t speak for the AMNH or its collections managers, but I can speculate as to the utility of a phylogeny-based CMS. Our understanding of phylogeny changes because it’s the product of scientific research; science is, after all, self-correcting. By maintaining a phylogeny-based CMS, the AMNH communicates a particular set of values: it is a scientific institution that frames the world in terms of scientific research outcomes. Whether or not those outcomes are convenient for researchers themselves is irrelevant; the process itself is what matters.


This is all well and good in the abstract. But let’s not forget that the practice of science is not abstract: somebody’s got to move those damn specimens around when new research demands it. Some institutions mitigate the kinds of difficulties facing the AMNH collections by adding stipulations to their CMS. The Smithsonian and the University of California Museum of Paleontology at Berkeley, for example, arranges its collections primarily by geological period, and then phylogenetically within those divisions. Collections managed in such a way will still have to be rearranged regularly, but those rearrangements are less extensive; it’s incredibly rare that new research will change the period in which a fossil was found. By analogy: if you have to rearrange your library, it’s much easier to rearrange a single shelf than it is to rearrange an entire bookcase.

The University of Oregon’s Museum of Cultural and Natural History (MCNH) uses a similar CMS for its paleontological specimens. On the “cultural history” side of the museum, however, a different CMS governs specimen organization. Archaeological and anthropological finds are arranged not by information intrinsic to the specimens themselves, but rather by the times when and places where those specimens were collected. There’s one significant advantage to this CMS: presuming that there aren’t any mistakes along the way, such a collection will never have to be rearranged.

Learning about this methodological difference in museum curation reminded me of Thomas Kuhn’s work. This is unfortunate: I’m a contrarian at heart, and so I chafe against Kuhn’s omnipresence in philosophy of science. Still, I appreciate (what I take to be) Kuhn’s fundamental insight that science is an activity practiced by scientists. It seems to me that a CMS like the one used by anthropologists at MCNH communicates a valuation of that fact: there are actual people doing this work. Kuhn was an anthropologist, after all.

Make no mistake: I love the AMNH and the Smithsonian and the paleontological side of the MCNH. But this is one case in which I think that the archaeologists and anthropologists have things figured out better than then paleontologists do. Collection time and place is information extrinsic to a specimen—you’d never know who collected a fossil, or when, just by looking at the fossil itself—but that information is nevertheless invaluable. This is particularly true in paleontology, wherein the prevailing theories of the day influence the preparation of a fossil. Knowing who collected a fossil from which place at what time therefore provides important context for understanding the fossil. That information can of course be recovered by other means, but encoding the information in the CMS communicates that it’s information that researchers should keep in mind.

I did note one caveat: a practice-based CMS like the one used by anthropologists is only constant if no one makes a mistake. People make mistakes, of course, so shouldn’t those collections also be subject to reorganization?

Yes, a collection organized by a history-based CMS will have to be rearranged every now again, but that rearrangement is accidental. It’s a bug that only comes up when someone does something she’s not supposed to do. With a theory-based CMS, rearrangement is a feature, not a bug: scientific theories are made to be revised, after all. There’s consequently no way to maintain a theory-based CMS without having to rearrange periodically, even if no one makes any mistakes.


A fair (if reductive) summary of my argument above is that paleontologists ought to arrange their collections in a way that’s easier for collections managers to maintain because collections managers are people, too. To idealize science as an impersonal process that somehow simply happens leaves out a great deal about what’s interesting and informative about scientific research. A phylogeny-based CMS idealizes science in just such a way. By contrast, a history-based CMS recognizes the role of scientists in science. I’d think that the scientists would be grateful for the recognition.