A footnote to the 1967 Economica paper, addressing the econometric evidence then read as showing a stable Phillips Curve.
On my assumptions, the only steady-state Phillips Curve is a vertical line intersecting the horizontal axis at u*. Now some econometric work over the past ten years might suggest that, especially on a fairly aggregative level, the Phillips Curve is a tolerably stable empirical relationship. But these studies probably estimate some average of different Phillips Curves, corresponding to different expected rates of inflation and of wage change which have varied only over a small range.
The vertical long-run Phillips Curve — the geometric signature of the natural-rate hypothesis — appears here in a footnote, a year before Friedman’s address made the claim famous from the AEA podium. The footnote also anticipates the empirical rejoinder and answers it in advance: the econometric studies that seemed to establish a stable trade-off (Lipsey on British data, Perry on American) were, on Phelps’s reading, averaging across a family of short-run curves indexed by expected inflation, which had simply not yet varied enough to reveal itself. That is exactly the diagnosis the 1970s would vindicate — when expected inflation finally moved a long way, the estimated curves came apart. The placement is telling of the two routes to the same result: what Friedman staged as the centrepiece of a presidential address, Phelps derived as a corollary and filed under the line.