Water becomes ice and ice becomes steam, but at no point does the H₂O molecule change. What changes is the relationship between the molecules — how tightly they bind, how freely they move, how much energy is in the system. Long love is a phase-change phenomenon in this exact sense. The two people are mostly the same people; what changes, several times over a long relationship, is the structure of how they are bound. Each phase has its own physics. The mistake most couples make is treating the physics of one phase as a permanent description of love, then declaring the relationship broken when the phase changes.
The classical model — Sternberg's triangular theory of love, with its three components of intimacy, passion, and commitment — gives us a useful first cut. Different combinations produce different kinds of love: infatuation (passion alone), empty love (commitment alone), companionate love (intimacy plus commitment), consummate love (all three). What Sternberg makes explicit is that the components are independent variables that fluctuate over time, and that the felt quality of love is the joint vector. When passion drops, you don't fall out of love — you move to a different point in the triangle. Most couples experience this drop as failure because they were never told the geometry.
Helen Fisher overlays a three-system biological model: lust (testosterone-driven), attraction (dopamine-driven, the limerent phase), and attachment (oxytocin- and vasopressin-driven). The three systems can run in parallel or in sequence, can fire for the same person or different people, and obey different time courses. Attraction has a half-life of two to three years; attachment can run for decades. A long relationship is, biologically, a sequence in which attraction dims and attachment compounds. The felt experience of this is "we don't have the spark anymore," which is true and also incomplete — what you have instead is the bond that the spark was building toward.
The phases are not a linear staircase. Daniel Stern's work on developmental stages, transposed to couples, suggests that each phase is stable for a period, destabilizes around a transition point, and reorganizes into the next phase. The destabilization always feels like crisis. Couples who haven't been told about phase changes interpret the destabilization as evidence that they chose wrong; couples who have been told about it can use it as scaffolding. The transitions tend to cluster around specific life events: cohabitation, marriage, first child, midlife, empty nest, retirement, illness. Each is a phase change. Each can be survived, and each requires a slightly different relationship to emerge on the other side.
A useful list, drawn from the convergent observations of Hatfield, Fisher, Sternberg, Aron, and Gottman: passionate beginning (months 0–24), the first reckoning (years 2–5, when limerence ends), companionate consolidation (years 5–12), the midlife reorganization (years 12–25, often triggered by children's launch or career inflection), late attachment (years 25+, characterized by what the longitudinal studies call "mutual interdependence"), and end-of-life love (when one partner is dying or both are facing it). Each has its own emotional signature. Each requires a different set of behaviors. None is the "real" love that the others are deviations from.
The Law 5 implication is concrete. Long love is not a thing you preserve; it is a sequence you revise through. The couples that last are not the ones with the strongest initial bond — Eli Finkel's review of the marriage research finds essentially no correlation between courtship intensity and long-term satisfaction. The couples that last are the ones who handle phase transitions well. The skill is meta: noticing when the physics has changed, naming it, and updating the operating assumptions. Couples who never update keep trying to operate the current relationship using the manual from three phases ago, which fails progressively and then catastrophically.
There is one further insight worth holding. The phase changes are not symmetric — they don't hit both partners at the same time. One partner enters the midlife reorganization at thirty-eight while the other is still in companionate consolidation; one wants to revise the relationship while the other still wants to maintain it. This asynchrony is one of the central sources of divorce in long marriages. It is not pathology; it is two clocks. The couples that survive it are the ones who can hold the asynchrony as a phenomenon to navigate rather than as evidence of betrayal.
To love someone for decades is to love several different people in sequence — including several different versions of yourself loving them. The relationship is the continuity that makes the sequence narratable. The phases are the actual content.