Astronomers have long noticed that the visible matter in galaxies cannot, by its own gravity, hold those galaxies together at the speeds at which they are observed to rotate. Stars at the outer edges of spiral galaxies travel faster than the light from luminous matter alone would predict; clusters of galaxies fly apart, on the same accounting, faster than their members ought to allow. From these and related discrepancies has grown the conjecture that some additional matter, invisible to ordinary instruments yet gravitationally consequential, makes up a substantial portion of what the universe contains. The name given to this conjectural substance — 'dark matter' — preserves both the inference and its embarrassment: it announces what we cannot see by what we do.
A persistent challenge to the conjecture is that it has so far been characterised almost entirely by what it is not. Dark matter does not interact with light, or at least not with the parts of the spectrum we have learned to detect; it does not, on most accounts, interact through any of the forces other than gravity; it has not, despite decades of patient search in deep underground laboratories, registered itself in any direct detector. To philosophers of science, this catalogue of absences has occasionally suggested that the conjecture is unfalsifiable in the manner of an ether — invoked to do work, perfectly tuned to evade test, and revised at convenience whenever the next null result arrives.
The charge is not without force, but it is also less devastating than it first appears. Some absences are themselves predictions: a candidate particle that ought to scatter from a xenon nucleus at a given rate and does not has been falsified at that rate, even if a slightly different particle remains untouched by the result. The successive narrowing of the parameter space in which dark matter could lie is not the same as the indefinite adjustment of an ether hypothesis; it is the slow, ungratifying work of elimination. To ask of a research programme that it deliver positive evidence on a timetable that suits the critic is to misunderstand the inferential economy of physics, in which a negative result, properly designed, can be quite expensive to produce.
What endures, then, is a hypothesis whose status is genuinely awkward without being incoherent. Either some new species of particle will, in time, register in a detector built to receive it, or the discrepancies that called it into being will be resolved by some alteration in our understanding of gravity at large scales. The discomfort of the present is real, but its resolution will not come from giving up the conjecture before the work of testing it has run its course; nor will it come from clinging to it past the point at which the absences begin to tell their own story. Patience, in this corner of the discipline, is a methodological virtue, but only a provisional one.
The primary purpose of the passage is to: