128 TRANSACTIONS LIVERPOOL BIOLOGICAL SOCIETY. 



(equinoctial tides). An annual sea-temperature oscillation 

 ought, then, to decompose into three component oscillations : — 

 (1) Annual ; (2) Six-monthly ; (3) Fortnightly. On adding, 

 algebraically, these three oscillations there should result one 

 that is very similar to the observed oscillation. There ought, 

 of course, to be a " residue " unaccounted for by the analysis. 



The explanation, it has been said, assumes that heat is 

 transferred between the sea-position investigated and the 

 adjacent land by means of tidal oscillatory streams, and it is 

 easy to see that this must be the case. In winter the extensive 

 sand-banks, uncovered by the ebb-tide, lose heat by radiation 

 and evaporation, so that they become colder than the sea- water 

 that flows over them at next flood-tide : therefore, that water 

 is cooled, and when it ebbs out seawards to the sea-position 

 under investigation it dilutes the warmer water there by colder 

 water. A reversal of these conditions occurs in the summer 

 months. 



The question presented itself : could the annual tempera- 

 ture oscillation, at a point in mid-channel in the Irish Sea, be 

 subjected to Fourier analysis so as to bring out the three 

 periodic oscillations mentioned above and then leave a residue 

 which might be traced to other periodic or unperiodic causes. 

 It is probable that Fourier analysis is not theoretically justi- 

 fiable, for the method assumes that the function (the annual 

 sea-temperature oscillation at a fixed station) is periodic — that 

 is, repeats itself exactly from year to year. But it does not ; 

 and there are quite well-marked variations in (1) the dates of 

 maxima, minima and means, and (2) amplitude from year to 

 year. These may, of course, be due to an oscillation, the period 

 of which is a number of years, but about that we do not know 

 yet. 



Anyhow, it appears to be worth while to look for the 

 existence of a fortnightly component of sea-temperature 

 oscillation before plunging into Fourier analysis. This is the 



