Coelophysis have their fans, there are surely many more people who would name the likes of Tyrannosaurus, Brontosaurus, or Triceratops as their "favourite dinosaur". When it comes to fossil mammals, it's the mammoths and the sabretooth cats that hold the public imagination, and we also like to hear of giant bears or mighty prehistoric rhinos.
There's nothing wrong with that, and I'd have to confess to that bias myself. (I mean, come on, who doesn't love sabretooths?) Indeed, it seems that for much of Earth's history, the largest land animals were indeed larger than their counterparts today. But, at any given point in time, the small animals always outnumber the large ones. Today, over two thirds of known mammal species are either rodents, bats, moles, or shrews, and, while the details might have changed, it seems plausible that the majority of mammals in the past were at least roughly in that size range.
Smaller animals, especially those delicate skeletons, such as bats, don't always survive so well in the fossil record as their larger kin, despite there having been so many more of them. But even so, they are numerous enough that a lot of the less glamorous end of mammalian palaeontology consists of digging them up and examining them, and they can be useful for things like determining the exact age of a particular rock formation.
For example, the Denizli Basin is a rock formation in western Turkey with deposits ranging in age back to about 15 million years ago. Fossils uncovered there include those of horses, deer, rhinos, and short-necked giraffes. But a recently published paper looks at the numerous fossils found there of dormice, gerbils, and jumping mice, and we also know of chipmunks and animals related to hamsters from the same formation. The new paper also describes a new species of "true" mouse (as in, a member of the same genus as the common house mouse), Mus denizliensis, which, at 2 million years old, just before the Ice Ages, is older than anything of its kind known from Europe or Turkey. Apparently related to the fossil mice of North Africa, and with its oldest European relatives being known from the Greek islands, this gives us some idea as to how and from where such animals first entered the continent.
But one question that we often with these animals (and, to be fair, with larger ones, too) is just what size they actually were. A great many species, after all, are described from fragmentary remains. Often it's just the teeth, and, if you're lucky, whatever bits of skull the teeth happen to be attached to. It's not that you don't find, say, bits of arm bone, but the exact shape of teeth tend to be more distinctive, so an arm on its own is a lot less use that half a skull on its own.
So let's say you have a skull. You can tell it belonged to a rodent, because of the shape of the teeth, and you even know what general sort of rodent it was, because the cheek teeth, in particular, are helpful like that. But how big was the rest of that rodent?
Well, as a general rule, and all other things being equal, the larger the teeth the larger the animal they were originally attached to. We do at least know it is a rodent, so if we compare the size of these teeth to those of rodents that are still alive today, we ought to be able to make some sort of guess. The problem, however, comes from the "all other things being equal" part... because they often aren't.
There have been a number of attempts over the years to create formulas for estimating the size of a rodent from its teeth. Perhaps the most widely used was devised by Serge Legendre in 1986, and is based on the area of the first lower molar. However, there are reasons to suppose that it might not always work, especially for the oldest rodent fossils. For example, an animal that evolves to eat hard food, such as nuts or tough grains, will tend to have larger, crushing or grinding teeth than something that eats softer food, regardless of the size of the rest of the animal. Early rodents, in particular, tend to have less specialised teeth than modern, living rodents, so it could be that trying to compare the two as if they're the same won't always work. Indeed, it's been argued that it doesn't always work terribly well even for living species that weren't included in the original analysis.
There is, to be honest, no perfect solution to this. Apart from anything else, unless you happen to find a complete skeleton later on (and you usually won't) you'll never know whether or not you were right. But we can at least try to refine our methods, taking more features into account, and using a larger array of living species as our comparison point.
There have, of course, been many attempts to do this, and it's hard, if not impossible, to say definitively which one is "the best". One recent example, however, involved looking at over two hundred skulls belonging to 92 different living rodent species, comparing a range of different measurements with the animal's known body weights. Some, as one might expect, turned out to be better than others.
One of the least useful predictors, for instance, was the length of the palate. This is because different animals often have snouts of varying length, often depending on what it is that they eat. Voles and mice are similar in size, yet the former tend to have shorter, more rounded heads. Another measurement that didn't work terribly well was the width of the cranium, or braincase. If anything, this is likely to be even less useful when examining particularly old fossils, since it's entirely plausible that very early rodents had (on average) smaller brains than the modern sort.
The conclusion of this particular analysis, however, was that the best predictor of body weight was the total length of the skull. This matches with earlier studies on other groups of animals, such as monkeys, seals, and land-based carnivores, although it does rely on you having a reasonably complete skull to examine - which isn't always the case. Even then, of course, it will only give you a range of possibilities, rather than one specific answer, but it does seem to be the best you're going to get.
So, how can we apply to some real fossil rodents? The study's authors examined a number of skulls belonging to a group of rodents known as the ischyromyids. These lived during the early Eocene, around 50 million years ago. In fact, they are amongst the earliest rodents that we know of, with a number of primitive features, despite already having the clearly identifiable rodent incisors. They were hardly mouse-sized, however, and, in this particular case, their estimated body weights turned out to vary between 856 g (1 lb. 14 oz.) for Reithroparamys to 2914 g (6 lbs 6 oz.) for Paramys delicatus. Interestingly, the latter does have some fossils with a reasonably complete skeleton, and it's actually about the size of a marmot; since 3 kg is only just above average for marmots, this gives some confidence that we're in the right ballpark.
But, in fact, there are other things we can tell from the size and shape of the skull beyond how big the animal as a whole was. Applying statistical analysis to a range of figures based on different measurements and proportions, we can see whether or not the skulls of animals (or whatever else we happen to be measuring) fall into distinct groups, and whether or not those groups match anything else that we might know about.
Ischyromyids, while primitive, are generally thought to be related to squirrels, so this more detailed part of the study looked only at the skulls of living squirrel species. It showed that the skull shapes tended to fall into three groups, depending on whether the living animal was a tree squirrel, ground squirrel, or flying squirrel. Each group was quite distinct making it possible to predict, from the shape of the skull alone, which the animal happened to be.
Applying the same calculations to the fossil skulls showed that all but one of them had the same general shape as ground-dwelling squirrels, such as prairie dogs. This may be a little controversial, since ischyromyids have generally thought to live up trees, at least some of the time, and we are, after all, looking at the skulls, not the limbs. Still, it's not the first time this has been proposed, and it is plausible that the shape of the skull could correlate with burrowing habits.
Reithroparamys, the smallest of the fossils examined, also had the strangest shape to its skull. It wasn't outlandish or anything, but the statistical analysis, while still placing it, broadly speaking, as being like a ground squirrel, came up with numbers that didn't match those of any living species, as if it had some other adaptation that nothing has today. There's honestly no way of knowing what that might be, and it's as likely to be due to its diet as anything else... but it might be significant that it was once suggested, based on the shape of what we have of its legs, that it moved by hopping.
A hopping squirrel. There's something you don't see every day.
[Photo by Ryan Somma, from Wikimedia Commons.]