It is shown analytically that if f(0) represents the frequency of the 22-year sunspot cycle or that of the 22-year sunspot cycle or that of the 35- to 40-year sunspot cycle or that of the seasonal cycle, then large heat/temperature oscillations should exist in the surface-atmosphere system (SAS) at frequencies 1/3f(0), 2/3f(0), 1 1/3f(0) and 2f(0). Also if f(0) represents the frequency of the seasonal cycle alone, then large heat/temperature oscillations should exist in the SAS not only at the frequencies listed above but also at the additional frequencies 1 2/3f(0) and 1/2f(0) Actual existence of heat/temperature oscillations in the SAS at the frequencies given above has been amply verified by means of past temperature records. Furthermore, we illustratively show that both the Ii-year sunspot cycle and corresponding heat/ temperature oscillations in the SAS at a period of ii years maintain approximately stable phase relationships with each other as long as each of the latter oscillations steadily keeps to one amplitude-modulation state. The phase relationship may change if the heat/ temperature oscillation involved switches into a different amplitude-modulation state. One of the spin-offs of the analysis reported herein is a realisation that the quasi-biennial oscillation is apparently composed of individual components at periods 1 1/2, 2 and 3 years.