For most of the history of condensed matter physics, magnetism came in two kinds. Ferromagnets align their atomic spins in the same direction, producing a net magnetization — the kind you feel when a magnet sticks to a refrigerator. Antiferromagnets align their spins in alternating directions, canceling each other out so the net magnetization is zero. This two-category taxonomy was complete. Every magnetic material was one or the other (or paramagnetic — no ordering at all). Textbooks presented the division as exhaustive.
In 2019, Libor Ċ mejkal and collaborators predicted a third possibility. Certain crystal symmetries allow a material to have zero net magnetization — like an antiferromagnet — while simultaneously producing spin-split electronic bands — like a ferromagnet. The spins cancel globally but their spatial arrangement, governed by rotational crystal symmetries rather than simple translation, creates asymmetric electronic structure. The material is magnetically invisible by one measurement (magnetometer reads zero) and magnetically active by another (band structure shows spin polarization). In 2024, experimental groups confirmed this behavior in multiple materials. The new category was named “altermagnetism.”
The taxonomy had been binary for over a century. During that century, materials now recognized as altermagnetic were measured, characterized, and filed under “antiferromagnetic” — because the defining criterion was net magnetization, and altermagnets have none. The anomalous spin splitting in their band structures was noted as unusual but was treated as a complication within the existing category, not as evidence of a missing one. The two-category framework couldn't represent what it was seeing, so it assimilated the evidence into the nearest available slot.
The general pattern: a taxonomy with N categories makes the (N+1)th category invisible not by denying the evidence but by absorbing it. Each anomalous observation gets filed as an exception within an existing category rather than as an instance of a missing one. The evidence for the new category accumulates as scattered anomalies distributed across the old categories — individually unremarkable, collectively diagnostic. The taxonomy doesn't suppress the data. It suppresses the interpretation. The observations remain visible; the pattern they form does not, because the pattern requires a category that doesn't yet exist to make it legible.