140 TRANSACTIONS LIVERPOOL BIOLOGICAL SOCIETY. 



The fit will not be exact, but the difference will not usually 

 exceed one or two days. The following precautions are 

 necessary in regard to the temperatures before the composite 

 high-water curve can be used. By comparing the daily 

 readings taken from the ten-daily curves the average daily 

 rate of change of temperature can be found. Thus, corrections 

 can be made. 



Suppose that it is required to apply the graph of high 

 waters for 1908 to that of 1907. It was found that, in order 

 to get the best fit, the whole graph for 1908 had to be moved 

 forward by four days : thus, January 6th of 1908 will comcide 

 with January 2nd of 1907. Then we see, from the ten-daily 

 average charts that the average temperature for seven years, 

 for January 2nd, is higher than that for January 6th by 

 0*33° C, and so this amount has to be added to the value taken 

 for January 6th, 1908. Each daily temperature must, of 

 course, be altered likewise before 1908 can be compared with 

 1907 and its tidal graph. The same method was applied to 

 each of the years, and the daily corrections were applied to the 

 particular readings. The corrected values for each day were 

 then added up and daily average fluctuations were thus obtained 

 for the years 1907-12 and 1914. These were smoothed by 

 taking three-daily averages, and the results were applied to 

 the high- water graph just as hi the case of the individual years. 

 The result seems to be satisfactory, though the effect of the 

 resultant fluctuations is toned down, so to speak, and the greater 

 part of the range is no more than - 28° C. above or below the 

 mean line. But in outline the graph of composite fluctuations 

 follows that of 1907 so closely that one must look on it with 

 some suspicion. If the year 1907 is omitted and the com- 

 bination of all the others is tried the result is very different. 

 This curve for 1908-12 and 1914 shows little variation from 

 January lst-5th. In February and March there are signs of 

 the same regular succession of maxima and minima that are 



