It is shown that if data on the latitude variation of winter storm tracks in the North Atlantic are sorted according to the phase of the quasi-biennial oscillation (QBO) of stratospheric winds in the tropics, then a clear solar cycle variation emerges without any need to smooth the data. For the west phase of the QBO the peak to peak amplitude of the solar cycle variation is about 6° of latitude, with storm tracks occurring equatorward at solar maximum and poleward at solar minimum. This is consistent with the recent work of Labitzke and van Loon, and extends to dynamical effects at the surface the pattern of atmospheric responses that they found. Problems with the acceptance of the concept of solar variability effects on weather are discussed. To relate the solar variable energy inputs to the weather effects which are energetically several orders of magnitude larger, scenarios for a class of mechanisms with two or three stages of amplification are outlined, involving middle atmospheric energy inputs, in contrast to the tropospheric inputs of solar luminosity changes.