SEA-FISHERIES LABORATORY. 135 



March has been taken, and, first of all, the data for separate 

 years have been considered : we begin with 1907. The heights 

 of the tide at high water at Liverpool — which is the nearest 

 standard port — are plotted for the p.m. tides of each day, 

 thus we obtain a curve, the maxima of which represent springs 

 and the minima neaps. This curve is drawn as the dotted 

 line in Fig. 3 and the straight line marked 0° C. may be 

 taken as representing mean sea-level at high water for the 

 period in question. To compare the deviations from the mean 

 level at high water with the temperature variations we have 

 next to find the smoothed ten-daily temperature, and it is 

 assumed that in taking the means of overlapping groups of 

 ten days the smaller deviations of temperature, due to tidal 

 causes, are smoothed out. The straight line, " 0° C." in Fig. 3, 

 then represents the temperature graph as deduced from the 

 ten- daily smoothed averages, only it is straightened out, 

 so to speak, because all that we want to find is the deviations 

 of temperature from this smoothed value. 



The smoothed temperature values are read off directly 

 from a graph, and then the differences between these and 

 the actually-observed daily temperatures are taken. These 

 actual daily temperatures run rather irregularly, and so I have 

 taken three daily averages : this gets rid of some of the more 

 violent irregularities, but ought not to hide the variations due 

 to the influence of the tides. Thus, the value for January 2nd, 

 for instance, is taken as the mean of three days January 1st 

 to 3rd, and it is plotted against the date January 2nd. 



The difference between the smoothed mean ten-daily 

 temperature and the mean three-daily temperature can now 

 be found by adding or subtracting, as the case may be. The 

 former series is represented by a straight line and the latter 

 is represented by a graph, which runs above and sinks below 

 this straight line. It will be seen that there is a rather striking 

 similarity between the fluctuations of temperature and the 



