PART I. : TEMPERATURE OBSERVATIONS IN THE MADUSEE, POMERANIA. 623 



§ 8. The method of discussing the observations was as follows : — For each series of 

 temperatures a temperature-depth curve was drawn, and from these curves could be 

 read off the depths at which particular temperatures occurred. When all these curves 

 had been drawn for each station, the depths at which temperatures of 16°, 14°, 12°, 10°, 

 and 8° C. occurred were plotted against the time of occurrence, a separate curve being 

 drawn for each temperature and each station. This final curve is reproduced in Plate I. 

 The continuous curve represents Moritzfelde observations, the dotted line Seelow, and 

 the dashed line Werben. Of course the interpolations in drawing these curves are con- 

 siderable, for, except from 30th to 31st July, and from 9th to 14th August, no observa- 

 tions were made during the night. On the Moritzfelde curve for 16° C. the hours of 

 actual observations are marked by a dot, but these dots have been omitted from the 

 other curves to avoid complication. Usually the observations at the various stations 

 were simultaneous. 



§ 9. That there are periodic temperature oscillations appears clearly from these 

 curves, the period being between 24 and 25 hours. E.g. take the period from 8 a.m. 

 on 7th to 11 a.m. on 14th August — 7 days 3 hours, with seven oscillations — period 

 24"4 hours. Or take the period at the beginning of the observations, from 8 p.m. on 

 25th July to midnight on the 28th — 3 days 4 hours, with three oscillations — period 25*3 

 hours. Thus the period seems to have been shorter in August than in July. An 

 examination of the observations shows the reason of this, for in August the temperature 

 discontinuity was much more marked than in July, and the difference of density 

 between the upper and lower layers being sharper, a shorter period is to be expected. 

 The opposition in the phase of the oscillations at the two ends of the lake appears very 

 clearly. The oscillations at Seelow near the centre of the lake are very much smaller 

 than at the ends. This sufficiently proves that the oscillations are due to a standing 

 wave and are not due to travelling waves reflected backwards and forwards from one 

 end to the other, for then the amplitude at Seelow would have been nearly the 

 same as at Moritzfelde and Werben. On the whole, the oscillations at Seelow are 

 of the same phase as those observed at Moritzfelde, indicating that the node is towards 

 Werben. 



§ 10. At first the fact that the period was so nearly 24 hours raised suspicion that 

 the oscillations might be caused by some diurnal influence. This suspicion was all the 

 stronger because oscillations observed by Dr Exner * had also approximately a 24-hour 

 period. It is possible that the diurnal variation in the direction of the wind above 

 referred to may have caused a forced seiche, or at least have helped to maintain the 

 amplitude of an existing seiche, but that each oscillation was caused by some change 

 in meteorological conditions seemed impossible even to observers who came prepared 

 to doubt. 



§ 11. The starting of a seiche by the strong winds on 5th, 6th, and 7th August is 

 well marked, as is also the dying away of an existing seiche on 31st July and 1st 



* Op. cit. 



