dark matter refers to a component of the universe that we can infer from its gravitational effects, although nobody has identified what it actually consists of. in the standard cosmological model, it behaves as matter that interacts gravitationally but does not appear to interact appreciably with electromagnetic radiation. this is a little different from saying that it is simply invisible, since an enormous collection of ordinary objects too faint to observe would not account for the measurements we have.

the problem becomes apparent when we examine the rotation of galaxies. if most of a galaxy's mass were concentrated in the stars and gas we can see, the orbital velocities at sufficiently large distances from its center would be expected to decrease as the enclosed mass stopped increasing substantially. observations of many spiral galaxies instead reveal approximately flat rotation curves, meaning that stars and gas far from the galactic center continue moving at speeds that require more gravitating mass than is visible. this is usually described by introducing a dark matter halo extending well beyond the luminous part of the galaxy, although the precise distribution has to be inferred from the dynamics.

there were indications of a similar discrepancy in galaxy clusters before galactic rotation curves became such a familiar part of the argument. the velocities of galaxies within clusters suggest substantially more mass than is present in their luminous components, and observations of hot intracluster gas provide additional constraints on the gravitational potential. there is also gravitational lensing, which makes it possible to map the distribution of mass by measuring how the images of more distant objects are distorted. particularly interesting are collisions between galaxy clusters, where the hot gas can be displaced from the regions containing most of the mass inferred through lensing. the Bullet Cluster is a well-known example. the gas represents much of the ordinary matter in these systems, so its separation from the principal lensing mass is difficult to explain by simply assuming that astronomers have underestimated the amount of visible material.

the cosmic microwave background provides evidence from a completely different period in the universe's history. small variations in its temperature preserve information about the conditions of the early universe, and their angular distribution depends on the relative amounts of ordinary matter and dark matter, among other cosmological parameters. measurements of these fluctuations, together with observations of how large-scale structure developed, are consistent with a substantial nonbaryonic matter component. roughly five-sixths of the matter in the standard model is dark matter. this should not be confused with dark energy, which enters cosmology for rather different reasons and accounts for a larger fraction of the universe's total energy density.

there are several proposed explanations for what dark matter might be. weakly interacting massive particles have been investigated extensively, as have axions and other hypothetical particles, but searches have not established the existence of any particular candidate. some researchers instead investigate whether modifications to gravitational dynamics could account for the observations without introducing an additional matter component. modified gravity models can reproduce certain galactic phenomena, although explaining the cluster measurements and the cosmic microwave background simultaneously presents further difficulties. it is possible to obtain very precise constraints on the properties of something without having succeeded in identifying it.

i first learned about dark matter at a planetarium when i was sixteen. i remember being confused by the distinction between observing something and observing evidence of its existence, because the presentation seemed to describe a form of matter that astronomers could measure while admitting that they did not know what it was. i spent some time afterward reading about rotation curves, and eventually learned that the inference involves an entire collection of independent observations rather than a single unexplained feature of galaxies. sean carroll has written about the relationship between physical theories and the entities we infer from them, which is relevant here. the question is not simply whether something can be photographed or isolated in a laboratory; it is also what kind of physical explanation best accounts for the measurements.

i was interested in another limitation of cosmology around the same time, concerning the difference between the observable universe and the universe as a whole. the distance light has been able to travel since the early universe, together with the expansion of space, determines which regions can be observed from our location. the observable boundary is not necessarily a physical boundary, and the universe may extend considerably farther than we can measure. we do not know whether space is infinite. even in principle, observations from one location cannot provide direct access to every possible region of spacetime.

dark matter remains unidentified despite decades of increasingly sensitive experiments. astronomical observations continue to constrain how it is distributed and how it could interact with ordinary matter, while direct-detection experiments progressively exclude portions of the parameter space associated with particular particle models. i have been especially interested in how much can be established gravitationally before the underlying microscopic description is available, and in the rather difficult question of what would constitute a conclusive identification.

last edited: 31 october 2026