The observation of a triangular-shaped galactic cosmic ray (GCR) intensity maximum during the mid-1980s solar minimum was successfully predicted by drift models of cosmic ray modulation. The assumption underlying this prediction was that the evolution of solar activity, represented in drift models by the time development of the ''tilt'' angle of the heliospheric current sheet (HCS), does not vary greatly from cycle to cycle. Tilt angles derived from coronal brightness distributions in the 1970s show that this assumption, seemingly supported by the successful prediction, may not be valid. The evolution of the HCS during the 1970s, when the 11-year GCR maximum exhibited a broad peak, may have been significantly different (with a broader and less regular period of low tilt angles) from that inferred for the preceding mid-1960s solar minimum and from that determined by the Wilcox Solar Observatory during the following mid-1980s minimum. Had the 1970s corresponded to an A<0 epoch (in the drift formulation), it appears that the resultant GCR intensity maximum would have been double peaked, with a deep rift corresponding to the 1974 minicycle. The different sensitivities of GCR intensity to tilt angle changes in A positive and A negative solar cycles, as demonstrated by other authors, provide support for drift models of modulation. The point we make here is that nonsystematic evolution of the HCS from one cycle to the next can be an additional source of variation in the shapes of consecutive GCR intensity maxima. ¿ American Geophysical Union 1993