SEA-FISHERIES LABORATORY. 225 



polating values for intermediate times, since, obviously, 

 many curves could be drawn through the four given 

 points. It has, indeed, been shown* that such interpolated 

 values would really be very accurate. They would not take 

 account of the " accidental " variations in sea temperature 

 resulting from unusual winds, or the operation of other 

 factors, but would express only the generalised annual 

 temperature wave, which is indeed all that one attempts 

 in most discussions of such variations in regard to the 

 sea fisheries. It is assumed that the August sea tempera- 

 ture is that of the maximum, the February one that of 

 the minimum, and the May and November ones those 

 representing the half-amplitudes of the annual wave. 

 But whenever we have more than four observations, or 

 when, as in the present investigation, the date of the 

 cruises do not coincide with those times when we have 

 reason to suppose that maxima and minima occurred, 

 then the sine-function must be used cautiously in inter- 

 polation. The first Lancashire cruise had to be made at 

 the end of January, and the third at the end of July, and 

 we know, from other observations, that the actual 

 minimum fell in February and March, while the 

 maximum was in August. If, then, the above formula 

 had been employed, inaccurate results would have been 

 obtained, since the constants would have been incorrectly 

 deduced. 



I have, therefore, resorted to graphic interpolation. 

 We have daily observations for the various light vessel 

 stations, so that a smooth curve drawn as closely as 

 possible to points representing the monthly mean tem- 

 peratures will give a very reliable picture of the annual 

 variations. The observations made at the Carnarvon Bay 

 light vessel are regarded as indicating the variation in 



* Ibid. p. 186-8 



