Kenneth Wilson and Michael Fisher, both at Cornell University, publish in January 1972 (paper received in October 1971) "Critical Exponents in 3.99 Dimensions" in Physical Review Letters. The central idea, today known as the epsilon expansion, consists of treating the system's spatial dimension d as a continuous, variable parameter, instead of fixing it at an integer such as 2, 3, or 4. Wilson and Fisher observe that the problem of critical phenomena becomes mathematically simple exactly at d = 4 dimensions, where mean-field theory works almost perfectly; by moving slightly away from that special dimension — writing d = 4 − ε, with epsilon (ε) a small number — they can calculate the critical exponents as a power series in ε using the renormalization-group equations Wilson himself had formalized the previous year. At the end of the calculation, they set ε = 1 to obtain a numerical estimate of critical behavior in the three real physical dimensions. The 1972 result obtains the critical exponent γ as 1 + ε/6 for an Ising-type model and 1 + ε/5 for an XY model — the first concrete, verifiable numerical predictions derived directly from the renormalization-group formalism. The non-trivial fixed point appearing in this calculation has since been known as the "Wilson-Fisher fixed point", and the technique would become a standard tool both in statistical physics and, later, in quantum field theory. The idea of treating dimension as a continuous parameter was developed in parallel and independently, almost simultaneously (1971-1972), by Gerardus 't Hooft and Martinus Veltman in their proof of the renormalizability of electroweak theory (see separate entry thooft-veltman-electroweak-renormalization-1971): that work gave rise to the sister technique known as dimensional regularization, which solves a different problem — the mathematical infinities of Feynman diagrams in quantum field theory — via a formally analogous mechanism but with its own genealogy and purpose, not derived from Wilson and Fisher's work nor vice versa.