A spectral reset law explains and predicts periodic recurrence in the elements

<jats:p> The recurrence of chemical properties is the defining feature of the periodic system, yet it is conventionally explained only indirectly, through inferred electron configurations and heuristic filling rules rather than being directly derived from atomic observables. Building on an empirical spectral framework in which elements occupy a bounded spectral space defined by two measured frequencies, the K-edge frequency (𝜈 <jats:sub>max</jats:sub> ) and the first ionization frequency (𝜈 <jats:sub>ion</jats:sub> ), we show that recurrence arises from simple quantitative constraints operating directly on these observables. As 𝜈 <jats:sub>max</jats:sub> increases across the elemental sequence, 𝜈 <jats:sub>ion</jats:sub> rises toward the upper boundary of the spectral space. At this limiting condition, a spectral mode bifurcates and 𝜈 <jats:sub>ion</jats:sub> undergoes a sharp reset toward a nearly invariant baseline, cancelling the accumulated rise to within 1%. Repetition of this rise–split–reset structure partitions spectral space into successive bounded intervals that emerge as periods, with lower and upper edges corresponding to elements conventionally classified as Groups 1 and 18, respectively. The rise–split–reset structure yields the recurrence 𝜈 <jats:sub>ion</jats:sub> (𝑛 + 1) = 𝜈 <jats:sub>ion</jats:sub> (𝑛) + Δ𝜈 <jats:sub>ion</jats:sub> (rise) − Δ𝜈 <jats:sub>ion</jats:sub> (reset) , with |Δ𝜈 <jats:sub>ion</jats:sub> | = 𝐴𝜈 <jats:sub>𝑚𝑎𝑥</jats:sub> <jats:sup>𝑏</jats:sup> , so that successive period anchors recur at the same relative position in spectral space. Across the four experimental period boundaries, reset magnitude scales linearly with the spectral splitting amplitude (𝑅 <jats:sup>2</jats:sup> = 0.962), empirically linking the recurrence to the mode differentiation that establishes the periods. Seeded with a single measured value (lithium), the recursion regenerates the experimental G1 and G18 period anchors with a median error of 1.4%, and extrapolates beyond the known table to hypothetical periods 8 and 9, including element 119 at 𝜈 <jats:sub>ion</jats:sub> = (1.12 ± 0.33) × 10 <jats:sup>15</jats:sup> s <jats:sup>−1</jats:sup> (4.6 ± 1.4 eV). These results establish periodic recurrence as a quantitative, generative spectral structure in measurement space and show that ionization frequencies are not independent quantities, but are determined sequentially from a single seed and a pair of spectral scaling power laws with only two measured observables. </jats:p>