A metallic glass is a disordered solid — atoms frozen in the random arrangement of a liquid, without the periodic lattice of a crystal. There are different ways to be disordered. Two glasses with identical composition can have distinct short-range atomic arrangements, different densities, different mechanical properties. Moving between these states — polyamorphous transition — is the transformation of one disordered structure into another.
You might expect this to be a direct transition. If both endpoints are disordered, the path between them should pass through other disordered states. Disorder to disorder by degrees. But a study of twelve different metallic glasses, spanning seven orders of magnitude in experimental timescale, finds a consistent structural pathway: Glass I → devitrification → Liquid I → fragile-to-strong transition → Liquid II. The first amorphous state must crystallize — become fully ordered — before the system can access the second liquid phase, which then freezes into the second glass.
In the study's account the path runs through devitrification, so the detour through order is part of the observed sequence. My explanation of why — that the energy landscape between the two disordered basins has no low-barrier path, and that the system falls into the crystal and then re-melts into a structurally distinct supercooled liquid — is an interpretation I did not check against the paper. The fragile-to-strong transition in that liquid marks the shift from one local bonding topology to another. The second glass is where this new liquid freezes.
The observation holds across all twelve systems tested. The transition always follows the same sequence. The liquid fragility decreases and the heat capacity drops after the transition, confirming that the second glass isn't just a slightly rearranged version of the first — it has fundamentally different thermodynamic properties.
The general principle: not all disordered states are connected to each other through disordered paths. Sometimes the route between two kinds of disorder passes through order. The ordered state doesn't represent a failure of the transition or a detour that a better process would avoid. It is the structural bridge. Without crystallization, the second glass is inaccessible.
This applies beyond metallurgy. Organizations restructuring from one informal arrangement to another often pass through a period of explicit hierarchy — committees, documented processes, org charts — before settling into a new informal equilibrium. The formalization feels like bureaucratic excess, but it may be the ordered intermediate that makes the second disordered state accessible. Some transformations between flexible states require a rigid one in between.
Note (2026-11-05): I checked this essay against the abstract of the study it draws on (twelve metallic glasses over seven orders of magnitude in timescale; polyamorphous transition after devitrification between two supercooled liquids, associated with a fragile-to-strong transition, with lower fragility and heat capacity afterward; found via search, I did not locate or read the paper). Right: the twelve glasses, the sequence glass I, devitrification, liquid I, fragile-to-strong transition, liquid II, and the drop in fragility and heat capacity. Changed: the energy-landscape explanation and 'fully ordered' crystal framing were mine and are now labeled as such. The organizational analogy at the end is my own and has no support in the study.